Advances in Psychology Volume 112 The Self in Infancy: Theory and Research

Z Z Z 9 ADVANCES IN PSYCHOLOGY 112 Editors: G. E. STELMACH R A. VROON ELSEVIER Amsterdam - Lausanne - New York -...

Author: P. Rochat

12 downloads 483 Views 56MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Z Z Z 9

ADVANCES IN PSYCHOLOGY 112 Editors:

G. E. STELMACH R A. VROON

ELSEVIER Amsterdam

- Lausanne

- New

York - Oxford

- Shannon

- Tokyo

THE SELF IN INFANCY Theory and Research

Edited by Philippe ROCHAT Department of Psychology Emory University Atlanta, GA, U.S.A.

1995 ELSEVIER Amsterdam

- Lausanne

- New York-

Oxford - Shannon

- Tokyo

NORTH - H O L L A N D ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 P.O. Box 211, 1000 AE Amsterdam, The Netherlands

Library

of Congress C a t a l o g i n g - I n - P u b l i c a t i o n

Data

The self in Infancy : theory and research I edlted by Ph111ppe Rochat. p, cm. -- (Advances In psychology ; 112) Includes blbllographical references and index. ISBN 0-444-81925-8 (alk. paper) I. Self In Infants. 2. Self-perception in infants. I. Rochat, Ph111ppe, 1950. II. Series: Advances In psychology (Amsterdam, Netherlands) ; 112. BF720.$44S45 1995 153,9--dc20 95-35838 CIP

ISBN: 0 444 81925 8 9 1995 Elsevier Science B.V. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the publisher, Elsevier Science B.V., Copyright & Permissions Department, EO. Box 521, 1000 AM Amsterdam, The Netherlands. Special regulations for readers in the U.S.A. - This publication has been registered with the Copyright Clearance Center Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the U.S.A. All other copyright questions, including photocopying outside of the U.S.A., should be referred to the copyright owner, Elsevier Science B.V., unless otherwise specified. No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. This book is printed on acid-free paper. Printed in The Netherlands

Preface This book is a collection of current theoretical views and research on the self in infancy, prior to self-identification and the well-documented emergence of mirror self-recognition. The focus is on the early sense of sell and the aim is to provide an account of recent research documenting the precursors of self-recognition and self-identification. By focusing on early infancy, this book provides an updated look at the origins of self-knowledge. The original idea behind the book was to ask prominent researchers in the field of infancy to discuss the origins of self-knowledge. In soliciting contributions, I gave authors the assignment to discuss recent empirical findings that bear on the issue of early self-knowledge, and to express their theoretical views on this problem. What is common to all contributors is that they are committed researchers, most of them with a long and outstanding experience in observing nonverbal behavior. Indeed, most of the theoretical assertions and positions contained in this book are based on recent, systematic empirical observations. The origins of knowledge about the self is arguably the most fundamental problem of psychology. It is a recurring theme in the classic works of James, Mead, Freud, and Piaget. As evidenced in current literature, today's developmental psychologists are clearly expressing a renewed interest in the topic. Recent progress in the study of infant behavior provides important and genuinely new insights regarding the origins of self-knowledge. The contributions contained in this book are the direct reflection of this novel trend. There are two parts to the book. The first is mainly theoretical, presenting a variety of current frameworks and conceptualizations of the self in infancy. The second part focuses more on research, reviewing experimental facts pertaining to the self in infancy as revealed in posture and action (Section 1); the perceptual origins of the self in infancy (Section 2); and the social origins of the self in infancy (Section 3). I would like to acknowledge the very professional help of MJ Wraga for the design, copy editing, and production of this book. Thanks also to Sue Hespos for her help. I dedicate this book to my children, Magali, Pablo, and C16o, as well as to Rana and my mother, Ren6e. Atlanta, June 1995.

This Page Intentionally Left Blank

vii

CONTENTS

Preface xi

List of Contributors

PART I

Theory

1

Are we automata? Eleanor J. Gibson

2

Criteria for an ecological self Ulric Neisser

17

The self as an object of consciousness in infancy George Butterworth

35

Early objectification of the self Philippe Rochat

53

A theory of the role of imitation in the emergence of self Andrew N. Meltzoff and M. Keith Moore

73

Aspects of self: From systems to ideas Michael Lewis

95

Relational narratives of the prelinguistic self Alan Fogel From direct to reflexive (self-) knowledge: A recursive model (self-produced) actions considered as transformations Pierre Mounoud The unduplicated self Daniel J. Povinelli

117

141

161

viii

10

The self as reference point: Can animals do without it? Emanuela Cenami Spada, Filippo Aureli, Peter Verbeek and Frans B.M. de Waal

PART II

Section 1

11

12

13

14

15

193

Research

The Self Revealed in Posture and Action

Self-knowledge of body position: Integration of perceptual and action system information Mark A. Schmuckler Using a computerized testing system to investigate the preconceptual self in nonhuman primates and humans Matthew J. Jorgensen, Stephen J. Suomi and William D. Hopkins

221

243

Move yourself, baby! Perceptuo-motor development from a continuous perspective Audrey L.H. Van der Meer and F. Ruud Van der Weel

257

Interactions between the vestibular and visual systems in the neonate Franqois Jouen and Olivier Gapenne

277

Two modes of perceiving the self Bennett I. Bertenthal and James L. Rose

Section 2

303

Perceptual Origins of the Self

16

The effect of blindness on the early development of the self Ann E. Bigelow

327

17

Intermodal origins of self-perception Lorraine E. Bahrick

349

ix

18

19

Self-orientation in early infancy: The general role of contingency and the specific case of reaching to the mouth John S. Watson The function and determinants of early self-exploration Philippe Rochat and Rachel Morgan

Section 3

20

375

395

Social Origins of the Self

Self/other differentiation in the domain of intimate socio-affective interaction: Some considerations Daniel N. Stern

419

21

Becoming a self Edward S. Reed

431

22

Understanding the self as social agent Michael Tomasello

449

Author Index

461

Subject Index

475

This Page Intentionally Left Blank

xi

Contributors Filippo Aureli

Yerkes Regional Primate Research Center and Department of Psychology Emory University Atlanta, GA 30322 USA

Lorraine E. Bahrick

Department of Psychology Florida International University University Park, DM 430 Miami, Florida 33199 USA

Bennett I. Bertenthal

Department of Psychology Gilmer Hall University of Virginia Charlottesville, Virginia 22903-2477 USA

Ann E. Bigelow

Department of Psychology St. Francis Xavier University Antigonish, Nova Scotia B2G 2W5 Canada

George Butterworth

Psychology Division University of Sussex Falmer, Brighton BN 1 9QN United Kingdom

Emanuela Cenami Spada

Yerkes Regional Primate Research Center and Department of Psychology Emory University Atlanta, Georgia 30322 USA

Frans B.M. de Waal

Yerkes Regional Primate Research Center and Department of Psychology Emory University Atlanta, Georgia 30322 USA

xii

Alan Fogel

Department of Psychology University of Utah Salt Lake City, Utah 84112 USA

Olivier Gapenne

LENA/EPBD Centre National de la Recherche Scientifique URA CNRS 654 Hopital de la Salp~.tri~re 47, Bd. de l'Hopital, Paris France F-75651 Cedex 13

Eleanor J. Gibson

Professor Emerita, Comell University RD1, Box 265A Middlebury, Vermont 05753 USA

William D. Hopkins

Yerkes Regional Primate Research Center and Emory University Atlanta, Georgia 30322 USA

Matthew J. Jorgensen

Division of Behavioral Biology Harvard Medical School New England Regional Primate Research Center Southborough, MA 01772-9102, USA

Franqois Jouen

LENA/EPBD Centre National de la Recherche Scientifique URA CNRS 654 Hopital de la Salp~.tri~re 47, Bd. de l'Hopital Paris France, F-75651 Cedex 13

Michael Lewis

Institute for the Study of Child Development Department of Pediatrics Robert Wood Johnson Medical School 97 Paterson Street New Brunswick, New Jersey 08903 USA

Andrew N. Meltzoff

Department of Psychology University of Washington Seattle, Washington 98195 USA

xiii

M. Keith Moore

Department of Psychology University of Washington Seattle, Washington 98195 USA

Rachel Morgan

Department of Psychology Emory University Atlanta, GA 30322 USA

Pierre Mounoud

Section de Psychologie Universit6 de Gentve Route de Drize 9 1227 Carouge Switzerland

Ulric Neisser

Department of Psychology Emory University Atlanta, Georgia 30322 USA

Daniel J. Povinelli

Laboratory of Comparative Behavioral Biology University of Southwestern Louisiana New Iberia Research Center 100 Avenue D New Iberia, Louisiana 00007-0560 USA

Edward S. Reed

Department of Psychology Franklin and Marshall University lancaster, Pennsylvania 17604 USA

Philippe Rochat

Department of Psychology Emory University Atlanta, GA 30322 USA

James L. Rose

Department of Psychology Gilmer Hall University of Virginia Charlottesville, Virginia 22903-2477 USA

xiv

Mark A. Schmuckler

Division of Life Sciences University of Toronto, Scarborough Campus Scarborough, Ontario M1C 1A4 Caoach

Daniel N. Stem

Section de Psychologie Universit6 de Gen6ve Route de Drize 9 1227 Carouge Switzerland

Stephen J. Suomi

National Institute of Health Animal Center 9000 Rockville Pike Bethesda, Maryland 20892 USA

Michael Tomasello

Department of Psychology Emory University Atlanta, GA 30322 USA

Audrey L.H. Van der Meer

Department of Psychology The University of Edinburgh 7, George Square Edinburgh EH8 9JZ Scotland

F. Ruud Van der Weel

Department of Psychology The University of Edinburgh 7, George Square Edinburgh EH8 9JZ Scotland

Peter Verbeek

Yerkes Regional Primate Research Center and Department of Psychology Emory University Atlanta, GA 30322 USA

John S. Watson

Department of Psychology Tolman Hall University of California Berkeley, California 94720 USA

PART I

Theory

This first part assembles current theoretical frameworks regarding the developmental origins of self-knowledge. Each chapter presents a particular view and conceptualization of the self in infancy, proposing different criteria and emphasizing particular aspects of early self-knowledge. The variety of approaches represented here include ecological/functional perspectives (Gibson; Neisser; Butterworth; Rochat), cognitivist views (Mounoud; Lewis; Meltzoff & Moore; Povinelli), dynamic systems/postmodern approaches (Fogel), and views from an animal research perspective (Cenami Spada, Aureli, Verbeek, & de Waal).

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

CHAPTER 1

Are We Automata? ELEANOR J. GIBSON

Cornell University

Issues in psychology seem always to be approachable from two perspectives: the structural view and the functional view. The issue of a "self' is no exception. It began its research life from a structuralist approach, as so much of classical experimental psychology did. But American functionalism offered a possible alternative way of addressing the question, and the two views have been clothed by their advocates and played against each other, enjoying swings of popularity. I think we are just now at a moment of confrontation, and I intend to take a stand. Having borrowed my title from William James (1879), it will of course be a functional stand, but I will not borrow his arguments, which were mainly philosophical. Better ones, in my (I hope) scientific way of thinking are now available. How should we think about a self, a person? As a concept based on a body image, a representation of oneself to oneself, with a face that can be presented to others? Or shall we think of ourselves in quite another way, as agents in control of our actions, in functional terms? I consider these two views, how they have unfolded in theory and research in the past half century, and then argue for my view. The first paper I remember reading on the subject of self-awareness in an infant was one published in 1948 in Enfance, entitled "Images du Corps et Conscience de Soi," by R. Zazzo. He observed his own child's responses to a mirror placed before him and to pictures of himself, through the child's first 3 years, and concluded: "By the way the child reacts to the image of his body, the mirror therefore reveals the origins of consciousness, the image of the body being essentially the consciousness of the self' (p. 43). He found self-recognition in the mirror at about one year, 7 months, and self-recognition of a photograph later, at 2 years, 9 months. When shown his photograph then, the child announced, "C'est moi." Predictably, with the advent of the cognitive revolution, research on selfrecognition grew, the favorite method continued to employ the mirror, and a test was evolved. A spot of rouge or paint was daubed on a child's nose, and a surprise

4

E L E A N O R J. G I B S O N

reaction at the mirror reflection was awaited. (Amsterdam, 1972; Lewis & BrooksGunn, 1979). Arguments have centered frequently on how early this reaction happened; most researchers suggest some time between 18 and 22 months. Observations of animals provided with mirrors reflecting themselves were not infrequent even before observations on human infants (Gallup, 1968). Most of the species observed responded socially, as if to another animal. But Gallup (1970) found that chimpanzees, after a few days provision with a mirror, spent a considerable amount of time in self-directed behavior (making faces, picking food from their teeth, etc.). After giving them access to a mirror for ten days, Gallup anesthetized the chimps and painted marks on their faces with a red dye. When they recovered from anesthesia, the chimps were again observed before the mirror. They apparently recognized their own facial features because they made many attempts to touch the marked area (unlike unmarked chimps). Similar procedures have been tried with other primates, but except for orangutans, no others seem to qualify for recognizing their own mirror images. Speaking of the failure of monkeys on this test, Gallup remarked, "Without an identity of your own it would be impossible to recognize yourself. And therein may lie the basic difference between monkeys and great apes. The monkey's inability to recognize himself may be due to the absence of a sufficiently well-integrated self-concept" (Gallup, 1977, p. 329). The linkage of mirror recognition and presence of a self-concept seems to be taken for granted. It is referred to again and again in a currently popular debate about whether animals have a "theory of mind." Are animals aware of what they do? Does self-recognition imply a "self-concept" and even "mental state attribution"? (See letters in the American Psychologist, August, 1994; Westergaard & Hopkins; Mitchell, Westergaard, Parker, & Boccia). There are those who want to feel that humans are special, as Descartes argued. But it seems that one can grudgingly admit a few of the great apes by this test, admitting that they may have a selfconcept and even intentions, but still keep out the animal hoi-polloi. Animals were seldom thought to have egos, even in the golden days of rat psychology (or maybe especially not then). Even the generous Tolman did not ascribe an ego to a rat, although he used the term ego. Body image theories of the self have had many defenders, such as psychoanalysts and psychiatrists. Missionaries, on the other hand, reported that "primitive men" did not recognize themselves in mirrors or in pictures. Stone Age man did not represent himself in pictures, although he gloriously represented deer, bison, and other animals on his cave walls. Is recognizing one's image not only the epitome of self-awareness, but also the crowning achievement of evolution and civilization? I have one anecdote that belies it. I once read the story of an elderly lady strolling down a boulevard and passing before a shop window. "Who is that old woman staring at me?" she asked herself. But as she took a step, touched the

ARE WE AUTOMATA?

5

ground with her cane and lifted it, she realized that it was her own reflection in the glass of the window. "But that isn't me," she thought. The superficial structures change, the wrinkles come, and the white hair. But the way we change ourselves, when we do, is in our intentions, our expectations, the choices we make, and the actions we perform. The core of our "selves" is not a representation of any kind, but the knowledge we have gained as participants in the world, the alternative ways to act that we have learned, and the way we select our actions and our hopes. I reject a static, representational concept of a self. I don't accept a structuralist definition because we need a functional theory with dynamics and the power of control incorporated in it. I have another anecdote to introduce this view. One First Day, at a Friend's Meeting that I sometimes attend, the silence was broken by a Friend who had apparently been impressed (and depressed) by recent findings and theories of astrophysicists about the universe. She said, "When I think about the immense cosmic forces and the awesome, endless stretch of space, I feel how infinitesimal, weak, and powerless we humans are." But I found myself thinking that this was an incorrect inference to draw, that it neglected an essential fact. Living creatures have the power to control their movements and actions, and no planet or meteor or force of the physical universe does. As humans we have selfcontrol, or agency, a far more remarkable kind of power than blind force. This is my candidate for how to think about a s e l f - not a structure, or image of a body or a face, but control of one's actions and interactions with the world and with others. This kind of self does not have to wait, either, until 18 months, or whenever facial features are recognized. Differentiation of distinctive features of objects, even such objects as faces, does not occur very early in perceptual development, but intentional activity does. I believe that knowledge of oneself begins with perception. Furthermore, as one who embraces an ecological approach to perception, I do not believe that perception begins with an image - - either retinal, mirror, photographic, or any other kind. Perception is an activity, the obtaining of information from a dynamic array in the environment surrounding the perceiver. This activity begins immediately at birth (and to some extent before). The obtainable information specifies events in both the surrounding environment and in the perceiver. It also specifies the relations between them, such as the fitness of the perceiver's action systems for using the affordances presented by the environment and the environmental consequences of any action that is performed or attempted. As actions are performed, information is generated about what the perceiver is doing and what he or she can expect to do - - in other words, about the self. There it is, in a nutshell w by your own actions shall you know yourself. Fortunately, we can now be more explicit about this statement because there is research available on how a self is specified in action and thus can be perceived.

6

ELEANOR J. GIBSON

A functional view of how a self is specified for infant perceivers can call upon evidence from several lines of research (E. Gibson, 1994). How do we know that an infant perceives information that specifies a self? To begin with, a self is differentiated from the external world of objects and events by detecting the difference between two kinds of events. There are movements of things in the world around me, and there are movements perpetrated by me, as an actor. I can perceive the difference between motion I have caused and motions caused by things or by someone else. There are at least two kinds of information available to the perceptual systems for detecting this difference; one is a kind of flow in the visual field that is caused by any movement of my head or body as I look around my surroundings. These flow patterns constitute visual feedback that is specific to my own movements and does not reflect the movements of anything else. J.J. Gibson's comment on this fundamental fact was, "To perceive the world is to coperceive oneself." As one moves forward (or is wheeled forward by someone else), there is a flowing array of continuous expansion. The focus of expansion specifies the exact direction in which one is heading m unique information for oneself. The whole optic array undergoes change, deforming in different ways depending on how a person is moving. But when objects in the world move, there is displacement relative to the observer, but no deformation or flow of the whole array. Research by Kellman, Gleitman, and Spelke (1987) on 4-month-old infants suggests that they are capable of detecting this difference. A second kind of information for differentiation of oneself from external things also derives from actions of the perceiver. An event such as action upon an object by the perceiver is specified by simultaneous multimodal information; that is, information accessible to more than one receptor system. Reaching toward a surface, for example, produces changes in the optic array, and also in proprioceptive, kinesthetic information. Information for the same event perceived simultaneously via two separate receptor systems is an unassailable argument for the existence of something external to one's own organism. Evidence exists that even neonates may perceive multimodal information as specifying the same object (Gibson & Walker, 1984). The arguments I have cited thus far implicate movement on the part of the perceiver as an essential source of information for a self distinct from others or things. I believe that for young infants, this is probably the case. But it can be argued that there is a kind of static information available, too. A perceiver always occupies a place. I am here; you are there. We have different perspectives for viewing what is happening around us, even when we are in company together. J.J. Gibson (1979, p. 112) pointed out that an observer stretched out quietly on a comfortable lounge chair can observe a unique scene, including his own legs, his nose, and even his mustache if he has one. Our noses are always with us, a kind of

ARE WE AUTOMATA ?

7

leading edge. I doubt that this information is very useful to an infant, but as babies grow older they learn that their own perspective is not the same as that of others. They even learn to anticipate what another's perspective might be (Piaget & Inhelder, 1956). There is definitely a social aspect for defining selves, though the uniqueness of a stationary perspective on the world is not apt to be detected early in infancy. Young children do not learn to employ the personal pronouns correctly until they have learned something about perspective differences (Loveland, 1984), although they may use many other words correctly. As an ecological psychologist who is particularly concerned with perceptual development, I have come to think that a major development in infancy is learning to perceive the affordances of the environment (Adolph, Eppler, & Gibson, 1993). Affordance is a term coined by J.J. Gibson (1979). It refers to a reciprocal relation between an individual and the environment in which he or she dwells. The environment offers opportunities that an individual organism may or may not be able to use. Animal species vary in the affordances that may be available and useful to them; a tree may afford shade for a human but be too small to shade an elephant. But it may afford food for the elephant and not the human. The elephant has a trunk to seize the tree and the strength to wrench it down. Developmental differences in affordances abound. Flat, solid, extended surfaces offer walkable places for human children and adults, but not for 3-month-old infants who have not yet acquired the trick of balancing and moving forward on two legs. But it is not only the anatomical equipment and power that makes an affordance usable for a creature. One must also learn, in many cases, what something affords. A spoon, for example, affords conveying food to the mouth. So do chopsticks. But a young child must learn to perceive that particular affordance of these tools and to wield them successfully. A tool must be designed so as to fit an animal's body scale and such appendages as can be maneuvered appropriately, but the animal must also learn to perceive whatever opportunity is offered. I introduce this concept because it offers a valuable insight into selfperception. To describe the information for perceiving an affordance any affordance - - it is not only necessary to describe the pertinent aspects of the environmental supports and offerings, but also the dimensional and dynamic capacities of the animal. Does the animal perceive the fit or nonfit of its own dimensions and powers to achieve some potential affordance? Whenever an infant learns about an affordance of the environment for itself, it must perceive (learn to perceive) its own dimensions and capacities (E. Gibson, 1994). These dimensions and powers change in scale with growth, so exploration of oneself and what one can do begins very early in life and continues through and beyond maturity. Information for body scale and body power is information for oneself, and with every activity that one engages in, it is being picked up. Perceiving the j.

8

ELEANOR J. GIBSON

acceptability of the width or height of an opening to be entered (Warren & Whang, 1987), the height of a step for comfortable climbing (Warren, 1984), and the surface of a floor or path for safe traversal (Gibson et al., 1987) are all achievements that provide information about oneself. Exploratory activity is natural in infancy and childhood, in particular, and there is no end to the selfrevelation it provides. I have yet, however, to approach the aspect of perceptual learning that led me to choose my title. Even more important for an infant's discovery of itself as a unique entity than learning about its dimensions and muscular strength is its discovery that it is possible for it to control changes in the environment. This, too, comes through activity and involves perceptual learning. It is the halhnark of "agency," the real heart of a self. A clever engineer could no doubt program a robot to choose a picture of itself from an array of mirror images or photos, but no engineer has been able to make a robot function spontaneously to perform an activity that effects a change in the world whenever the occasion is appropriate. As William Preyer (1890) observed long ago, the perception of a change produced by one's own activity is perhaps the most significant happening in the life of an infant. Perceiving oneself as a source of control, exerting causal effectiveness to make a perceptible change in the world, is the epitome of perceiving oneself. We speak of it as acting intentionally or even, in high-flown metaphorical terms, as exercising "free will." Have we any evidence about when and how infants come to perceive that they themselves are controlling an event in the world? Fortunately, we do. A baby is provided at birth with the ability to observe a good deal of what is going on around it. It can hear very well, and its visual system, though as yet far from optimal, is capable of discerning objects within its gaze limits, especially if they are moving. Most importantly, it is also sensitive to its own bodily postures and movements: somatic information about what it is doing. A baby moves spontaneously to the extent that its postural development allows, and spontaneously explores to the extent it can. Exploratory systems include mouthing, moving arms and legs, and moving the head. Exploring with any of these systems brings new information, as does contact with an object in the course of exploration. Any change in the world ensuing upon spontaneous exploratory actions is a perceptible consequence of the baby's own activity and is followed up with fresh efforts. Early research on such events resulted in a paradigm referred to as operant or instrumental conditioning. We should remember, however, that Preyer took note of such events, as did Piaget (1952) more recently. Piaget, in the course of daily observations of his own child, tied a string to the baby's wrist and attached the other end to a celluloid toy dangling from the hood of the baby's crib. As the child moved its arm, the toy jumped about, and very quickly the baby acted to make the interesting effect continue.

ARE WE AUTOMATA?

9

Research on operant conditioning with young infants took advantage of the spontaneous action systems available to them, and succeeded, I believe, because the systems were exploratory. Head turning, mouthing (referred to experimentally as nonnutritive sucking), and movements of the limbs (using the arms and kicking) are all exploratory and have all been used successfully. Consider some examples. Siqueland and Lipsitt (1966) established differential instrumental headturning with newborns. Head-turning was also used successfully by Papousek (1967), with a visual display as the environmental consequence. He emphasized the motivational effects on the infant of learning that a change in the environment could be effected by an activity of its own. Papousek and Papousek (1984) later pointed out that "crossmodal and serial processing of input from somatosensory organs, on the one hand, and extereoceptive organs, on the other" (p. 143), are basic for development of notions of causality and self-awareness. Head-turning has adaptive significance as an intentional act that can obtain a change of scene and new information. It has been used frequently, in the habituation paradigm, as a .means of studying infants' perceptual capacities. So also has nonnutritive sucking. An early paper by S iqueland and DeLucia (1969) was a real breakthrough, demonstrating that at 3 weeks, infants quickly learn to suck at a set amplitude for the opportunity to observe an interesting sight, such as a picture of a face or a cartoon figure, or to hear music or a human voice. They commented that making an infant's behavior "effective for producing changes in the extereoceptive environment provides a tool for studying the ontogeny of exploratory behavior" (p. 1144). Nonnutritive sucking combined with an habituation paradigm has been used in dozens of experiments for intensive study of the way infants hear phonemes and discriminate voices. The use of this activity intentionally to control a presentation is revealed most clearly in an experiment by Kalnins and Bruner (1973). In their experiment, high-amplitude sucking by 3month-old infants resulted in clearing the focus of a movie presented to them. Infants sucked at high amplitude to clear blurting, but ceased quickly when the focusing mechanism was disconnected. Furthermore, they achieved a rhythm of sucking and looking so as to maintain the focus smoothly. Studies of the control aspect of instrumental conditioning have particularly flourished with a method perfected by Rovee-Collier (Rovee & Rovee, 1969; Rovee-Collier & Gekoski, 1979). The method is similar to Piaget's observations described earlier. Rovee-Collier ties a ribbon around the baby's ankle, and fastens the other end to a mobile suspended above the crib. When the baby kicks, the mobile moves contingently with the kicks. Even very young infants (2 months, possibly less) learn to operate the mobile. Although both legs are kicked at first, the action becomes restricted to the operative one; and when the ribbon is shifted to the other leg, so are the baby's kicks (Rovee-Collier, Morrongiello, Aron, &

10

E L E A N O R J. G I B S O N

Kuperschmidt, 1978). How does this early accomplishment indicate perceived control and intention? I think in three ways, at least. First, the baby has discovered a predictable outcome of a specific spontaneous act and shows expectation of the consequences; second, the kicks are activated or deactivated in relation to context change (shift to the other leg, or disconnection of the mobile to make its motion noncontingent); and third, there are appropriate emotional accompaniments of both acquisition of control and loss of it when the mechanism of the mobile has been detached from the ribbon. The latter finding was demonstrated in experiments by Lewis, Sullivan, and Brooks-Gunn (1985) and Sullivan and Lewis (1989). Subjects who controlled the mobile expressed enjoyment and smiled, whereas subjects who could watch the mobile but could not control its motion fussed more and smiled less. Discovery of causal efficacy in other cases, too, has been reported to induce expressions of pleasure (Watson, 1972; Watson & Ramey, 1972). Although we seldom think of infant vocalizations as exploratory behavior, I believe they can be. They are spontaneous and they are sensitive to external social contingencies, which make them candidates for exercising control in a social situation. It seems that infants learn to use them very early to control social relationships (Mosier & Rogoff, 1994). Crying to get attention, and cooing for smiles and parental response are not only anecdotally supported, but have been the focus of considerable research, especially so-called "protoconversations" between mother and baby (Trevarthen, 1974; Brazelton, Koslowski, & Main, 1974). In fact, one of the ffLrStobservations of an infant learning about control was made in a social interaction with his own infant by Watson (1967), who referred to it as "contingency awareness." An experiment by Bloom (1977) studied the social situation of an adult-infant interchange. Infants 3 months old engaged in a social interaction with an experimenter, who vocalized, touched, and smiled at the infant. At specified intervals, the experimenter became unresponsive and silent for five seconds. Vocalization by the infant initiated the time-out period. The rate of infant vocalization was not lowered by this apparently negative contingency, but there was an effect on the patterned distribution of vocalizations. The infants paused between vocalizations, as if waiting for their turns, listening for resumption of vocalizing by the adult. Research by Murray and Trevarthen (1985) is strong evidence of the infant's awareness of the contingencies provided by this situation as well as awareness of its own role as potential controller. Infants were taped in live protoconversations with their mothers. Typically, there was give and take in turn: The infant responded with smiles and happy expressions to the mother's playful speech and gestures. But when an earlier tape of the mother was played back to the infant, her responses were of course not timed to meet the baby's expectations of response to his/her own communications. The baby no longer maintained control of her responses in the contacts, and the replay resulted in

ARE WE AUTOMATA?

11

withdrawal, protests, and puzzled looks. Other essays in this volume emphasize social origins of the self, but I believe it is valuable to point out here that social factors interact with means of learning about control. A self is attained through animal-environment interactions, both with other people and with physical objects and events. Does the phrase "animal-environment interaction" mean I consider that animals other than humans develop a self? Yes, I do, certainly insofar as they develop control over events in their surroundings, ones appropriate to their modes of life and needs. I am impressed with research by Mason (1978), which compares the behavior of wild-born rhesus monkeys and monkeys raised from birth in individual cages. The cage-reared monkeys were not only socially incompetent, but they also seemed to lack other skills and behaved in training situations as if "they had no strategy or plan." Monkeys raised with mobile companions coped better in problem-solving situations by exhibiting more exploratory looking and touching of objects. One project compared monkeys reared with either inanimate mother surrogates (fur-covered hobby horses) or with dogs. All the monkeys were given space to roam and a variety of playthings. The dogs did not behave maternally, but were very active companions. Mason reported big contrasts between the dog-raised monkeys and the monkeys with inanimate surrogate companions. The monkeys raised with dogs were more attentive to the environment, more responsive themselves, and more likely to "achieve an adaptive outcome by acting on the environment." The differences were interpreted as due to presence or absence of response-contingent stimulation. The active companions provided more opportunities for a developing monkey to "experience the fact that his behavior has effects on the environment and to learn that the events going on around him are amenable to his control" (p. 249). These comments remind us of the importance of developmental aspects of attainment of a self. The self in infancy (this book's concern) is far from a static structure. Indeed, it is the changes we observe during infancy that give us clues to the best way to conceive of a "self' and possibly to help promote a healthy one as childhood progresses. For a child to learn that his or her behavior has consequences that he or she can control is crucial. We may see such learning even as a tiny baby moves its arms, and discovers that it can bring a hand into view whenever that consequence may be desirable (Van der Meer, 1993). The perception of the intermodal contingency ~ the simultaneous somatic and external events resulting from the unitary act of ann-waving ~ is the kernel of this learning. Intermodal perception of their own actions by 5-month-old infants has been impressively demonstrated by Bahrick and Watson (1985). The infant subject was seated before two video screens, its legs free to kick, but screened from direct view. One screen displayed its own limb movement, which it could of course feel simultaneously.

12

E L E A N O R J. G I B S O N

The other screen displayed limb movements of another infant, identically dressed, or an asynchronous view of its own movement, previously recorded. The infants gazed preferentially at the novel display of the other infant or at their own asynchronous performance, thus showing awareness of their own movements as they were specified both kinesthetically and visually. This kind of proprioceptivevisual invariance is detected long before infants recognize a distinctive facial feature specified only visually in a mirror or a portrait of themselves. The information is multimodally, contingently specified, as it is in situations where we have seen that infants learn about their own powers of control. It seems likely that information for the self is first obtained in the discovery of control. Knowledge of control is gained in many situations, some of which I have described, and with development there must come more and more generalized expectations about situations that do and do not afford control by an infant's own actions. The concept of control is extended to others as infants learn about perspectives and how they differ from person to person, which leads to the expectation of intention in others as well as themselves (Tomasello, 1993). The learning of new affordances as new action systems (e.g., locomotion) become available extends the perceived self still further, as do the vast changes in body dimensions and power as growth continues. Developmental change is at the heart of an emerging self.

Conclusion To answer my original q u e s t i o n - Are we automata? Certainly not. From early infancy, we learn that we are agents; that our own actions can effect changes in the world, including the actions of other persons. Behavior is not automatic, robotic, mechanistic; it is intentional and flexible. I have been, in a sense, singing the praises of control and the discovery of ourselves as we learn about control. But finally, I append a caution. A young infant can exert control over very few situations. These situations will increase with development, and an infant's options will broaden as new action systems develop and the scope of the environment offers more numerous affordances. The power of selection thus increases, and as the repertoire increases, so does the power of control. But development does not bring complete freedom of control because the options are always limited by the environment of any given individual. Children born into African-American families or into Presbyterian ones, in wealth or in poverty have limits of control set by their cultural surroundings, although sometimes development brings opportunities to control that, too. Choices are made so

ARE WE AUTOMATA?

13

behavior is intentional, but freedom is only as complete as the alternatives available. In other words, we are not automata, but liberty is limited. I return finally to the "self," though I do not like the nominalization of the concept. Humans have a self in the sense that they control their own actions, and they discover this - - that they are agents, actors who create noticeable effects quite early in infancy. A "self' that is a representation, or a concept of an executive made of mental stuff that directs the lower bodily orders, is a fiction. We are ourselves because we function as individuals who interact intentionally with other individuals and the world in which we live. REFERENCES

Adolph, K. E., Eppler, M. A., & Gibson, E. J. (1993). Development of perception of affordances. In C. Rovee-Collier & L.P. Lipsitt (Eds.), Advances in infancy research: Vol. 8 (pp. 51-98). Norwood, NJ: Ablex. Amsterdam, B. (1972). Mirror self-image reactions before age two. Developmental Psychology, 5, 297-305. Bahrick, L. E., & Watson, J. S. (1985). Detection of intermodal proprioceptivevisual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21, 963-973. Bloom, K. (1977). Patterning of infant vocal behavior. Journal of Experimental Child Psychology, 23, 367-377. Brazelton, J. B., Koslowski, B., & Main, M. (1974). The origins of reciprocity: The early mother-infant interaction. In M. Lewis &. L. A. Rosenblum (Eds.), The effect of the infant on its caregiver (pp. 49-76). New York: Wiley. Gallup, G. G., Jr. (1968). Mirror-image stimulation. Psychological Bulletin, 70, 782793. Gallup, G. G., Jr. (1970). Chimpanzees: Self-recognition. Science, 167, 86-87. Gallup, G. G., Jr. (1977). Self-recognition in primates. American Psychologist, 32, 329-338. Gibson, E. J. (1993). Ontogenesis of the perceived self. In U. Neisser (Ed.), The

perceived self" Ecological and interpersonal sources of self-knowledge

(pp. 25-42). Cambridge, MA: Cambridge University Press. Gibson, E. J., Riccio, G., Schmuckler, M. A., Stoffregen, T. A., Rosenberg, D., & Taormina, J. (1987). Detection of the traversability of surfaces by crawling and walking infants. Journal of Experimental Psychology: Human Perception and Performance, 13, 533-544. Gibson, E. J., & Walker, A. S. (1984). Development of knowledge of visual-tactual affordances of substance. Child Development, 55, 453-460. Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. (Republished 1986, Hillsdale, NJ: Erlbaum) James, W. (1879). Are we automata? Mind, 4, 1-22. Kalnins, I. V., & Bruner, J. S. (1973). The coordination of visual observation and instrumental behavior in early infancy. Perception, 2, 307-314. Kellman, P. J., Gleitman, H., & Spelke, E. S. (1987). Object and observer motion in the perception of objects by infants. Journal of Experimental Psychology: Human Perception and Performance, 13, 586-593.

14

ELEANORJ. GIBSON

Lewis, M., Sullivan, M.W., & Brooks-Gunn, J. (1985). Emotional behavior during the learning of a contingency in early infancy. British Journal of Developmental Psychology, 3, 307-316. Lewis, M.L., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self New York: Plenum. Loveland, K.A. (1984). Learning about points of view: Spatial perspective and the acquisition of "I/you." Journal of Child Language, 11, 535-556. Mason, W. A. (1978). Social experience and primate cognitive development. In. G.M. Burghardt & M. Bekoff (Eds.), The development of behavior: Comparative and evolutionary aspects. New York: Garland Press. Mitchell, R.W., Westergaard, G.C., Parker, S.T., & Boccia, M.L. (1994). Mirror self-recognition and mental state attribution, 49, 761-762. Mosier, C.E., & Rogoff, B. (1994). Infants' instrumental use of their mothers to achieve their goals. Child Development, 65, 70-79. Murray, L., & Trevarthen, C. (1985). Emotional regulation of interactions between two-month-olds and their mothers. In T. Field & N. Fox (Eds.), Social perception in infants. Norwood, NJ: Ablex. Papousek, H. (1967). Experimental studies of appetitional behavior in human newborns and infants. In H.W. Stevenson, E.H. Hess, & H.L. Rheingold (Eds.), Early behavior: Comparative and developmental approaches (pp. 249-277). New York: Wiley. Papousek, H., & Papousek, M. (1984). Learning and cognition in the everyday life of human infants. In J.S. Rosenblatt, C. Beer, M. Busnel, & P.J.B. Slater (Eds.), Advances in the study of behavior: Vol. 14 (pp. 127-163). New York: Academic Press. Piaget, J. (1952, 1963). The origins of intelligence in children. New York: Norton. Piaget, J., & Inhelder, B. (1956). The child's conception of space. New York: Humanities Press. Preyer, W. (1890). The mind of the child. New York: Appleton. Rovee, C.K., & Rovee, D.T. (1969). Conjugate reinforcement of infant: Exploratory behavior. Journal of Experimental Child Psychology, 8, 33-39. Rovee-Collier, C. K., & Gekoski, M. J. (1979). The economics of infancy: A review of conjugate reinforcement. In H.W. Reese & L.P. Lipsitt (Eds.), Advances in child development and behavior: Vol. 13 (pp. 195-255). New York: Academic Press. Rovee-Collier, C. K., Morongiello, B. A., Aron, M., & Kuperschmidt, J. (1978). Topographical response differentiation and reversal in 3-month-old infants. Infant Behavior and Development, 1, 323-333. Siqueland, E.R., & DeLucia, C.A. (1969). Visual reinforcement of nonnutritive sucking in human infants. Science, 165, 1144-1146. Siqueland, E.R., & Lipsitt, L.P. (1966). Conditional head-turning in human newborns. Journal of Experimental Child Psychology, 3, 356-376. Sullivan, M.W., & Lewis, M.L. (1989). Emotion and cognition in infancy: Facial expressions during contingency learning. International Journal of Behavioral Development, 12, 221-237. Tomasello, M. (1993). Joint attention as social cognition (Rep. No. 25), Atlanta: Emory Cognition Project (pp. 1-36). Trevarthen, C. (1974). Conversations with a two-month-old. New Scientist, 2, 230235.

ARE WE AUTOMATA9

15

Van der Meer, A.L. (1993, August). Arm movements in the neonate: Establishing a frame of reference for reaching. Paper presented at the VIIth International Conference on Event Perception and Action, Vancouver, B.C. Warren, W.H. (1984). Perceiving affordances: Visual guidance of stair climbing.

Journal of Experimental Psychology: Human Perception and Performance, 10,

683-703. Warren, W.H., & Whang, S. (1987). Visual guidance of walking through apertures: Body-scaled information for affordances. Journal of Experimental Psychology: Human Perception and Performance, 13, 371-383. Watson, J.S. (1967). Memory and "contingency awareness" in infant learning. Merrill Palmer Quarterly, 13, 55-76. Watson, J.S. (1972). Smiling, cooing, and "the game." Merrill-Palmer Quarterly, 18, 323-339. Watson, J.S., & Ramey, C.T. (1972). Reactions to response-contingent stimulation in early infancy. Merrill-Palmer Quarterly, 18, 219-227. Westergaard, G.C., & Hopkins, W.D. (1994). Theories of mind and selfrecognition. American Psychologist, 49, 761-762. Zazzo, R. (1948). Images du corps et conscience de soi. Enfance, 1, 29-43.

This Page Intentionally Left Blank

The Self in Infancy- Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

17

CHAPTER 2

Criteria for an Ecological Self ULRIC NEISSER

Emory University

The reflexive pronoun self is often just a grammatical convenience; it applies to many things that are surely not "selves." Thus we say: -the hurricane blew itself out. -the scotch tape stuck to itself. -the computer turned itself off. -the comatose patient scratched himself. Such entities may do things to themselves, but they do not know what they are doing. Indeed, they do not know anything. The term self becomes psychologically interesting only when the activity in question is cognitive as well as reflexive, when there is self-awareness, self-consciousness, self-knowledge. Any analysis of self-knowledge (I am trying to choose the least controversial of the three terms above) must confront several obvious issues. What do individuals know about themselves, and how do they know it? When (developmentally speaking) do they know it? And how can we, as observers, be sure that they know it? In recent years I have been trying to address these questions in a more or less systematic way (Neisser, 1988, 1991, 1993; Neisser & Fivush, 1994; Neisser & Jopling, in press). My strategy has been to divide what people know about themselves into several distinct domains, based on the different forms of information that make self-knowledge possible. There are five such domains, different enough that for some purposes we can think of them as establishing different kinds of "selves." Two of these--the ecological self and the interpersonal s e l f - are based directly on perceptual information; hence they can and do appear in the preverbal infant. In speaking of different foci of self-knowledge as if they established different selves, I am following the lead of William James (1890). His classical chapter on

18

ULRIC NEISSER

"The Consciousness of Self' (note the cognitive emphasis of the title) includes such categories as the "material self," the "social self," and the "spiritual self." Unfortunately, James was neither a perceptionist nor a developmentalist. He offers no systematic informational basis for his categories, and makes no attempt to trace them to their roots. Nor is he equally interested in all of them: The material self, in particular, gets only two paragraphs in a chapter of 110 pages. Only three lines in those paragraphs concern the physical body. Our "material selves" are said to include not only our bodies but our clothes, our families, our home, our property, even our life work. The broad scope of this concept reflects James's preference for affective over cognitive criteria: he defines selves by what we care about, not by what perception gives us. While this may be a good way to emphasize the significance of affect (which indeed plays a key role, especially for the interpersonal self), I prefer to begin with cognition.

Five Kinds of Self-knowledge Although I will focus here on preconceptual aspects of the self, an overview of all the various "selves" may be useful first. Such a review may reassure the reader that other significant aspects have not been entirely forgotten. If self-concept, selfnarrative, and self-conscious introspection get short shrift here, it is only because they depend on language and thus appear somewhat later in development. -The ecological self is the individual situated in and acting upon the immediate physical environment. That situation and that activity are continuously specified by visual/acoustic/kinesthetic/vestibular information. As we shall see, infants perceive themselves to be ecological selves from a very early age. -The interpersonal self is the individual engaged in social interaction with another person. Such interactions are specified (and reciprocally controlled) by typically human signals of communication and emotional rapport: voice, eye contact, body contact, etc. This mode of self-knowledge, too, is available from earliest infancy. -The conceptual self, or self-concept, is a person's mental representation of his/her own (more or less permanent) characteristics. That representation, which varies from one culture to another as well as from one person to the next, is largely based on verbally acquired information. Hence, we can think of it as beginning in the second year of life. -The temporally extended self is the individual's own lifestory as he/she knows it, remembers it, tells it, projects it into the future. It cannot appear

CRITERIA FOR AN ECOLOGICAL SELF

19

until the child already has a conceptual self, a narratively organized episodic memory, and an explicit understanding of the continuity of persons over time say, until the fourth year. -The private self appears when the child comes to understand and value the privacy of conscious experience; when it becomes important that no one else has access to his/her thoughts, dreams, and interpretations of experience. I do not know whether this insight appears regularly enough to be counted as a developmental milestone, but it surely requires a temporally extended self on which to reflect.

Are Nonperceptual Forms of the Self Present in Infancy? Definitions like these raise operational questions. As observers, how can we tell whether a given individual m especially an infant m has one or another of these forms of self-knowledge? Let us take them one at a time. First, many recent experimental findings argue that infants are unlikely to have the kinds of experiences that establish a private self. Consider, for example, the problem that even 3-year-olds have in distinguishing reality from subjective appearance. They don't understand that an object which appears pink through tinted glass may in fact be white, or that what looks like a rock may just be a painted sponge (Flavell, Flavell, & Green, 1983). Children of this age also have trouble with the distinction between fact and belief, or between one person's belief and another's (Wimmer & Perner, 1983). It is not likely, then, that they--let alone much younger infants m could understand and value the subjective privacy of their own mental experience. We have much more direct evidence about the development of the remembered (temporally extended) self. Children as young as 2 years can recall past events if they are appropriately cued (Fivush, Gray, & Fromhoff, 1987), but they do not do so spontaneously. They show no interest in self-narrative and never sit around talking about old times. The core of the temporally extended s e l f - the idea of life as a continuing story, beginning with birth and extending through the present into the future m is entirely absent at that age. It develops only slowly in the third year or later (Hudson, 1990), in ways that depend on the surrounding culture (Rogoff & Mistry, 1990) as well as on parental encouragement. Thus, we can be reasonably sure that there is no temporally extended self in early infancy. The conceptual self is another matter. When does the self first "become an object to itself'? Like George Herbert Mead, I am sure this happens only as a result of social interaction. One "...becomes an object to himself only by taking the attitudes of other individuals toward himself within a social environment or

20

ULRIC NEISSER

context of experience and behavior in which both he and they are involved" (Mead, 1934 p. 138). By "object," Mead means an object of thought. This is by no means the same as being an object of perception. As we shall see, even very young infants perceive themselves to be real objects embodied and embedded in the environment. How do babies come to take themselves as objects of thought? Michael Tomasello (1993) has given an elegant (if speculative) account of this process. He begins by asking how a baby can come to take anything whatever as an object of thought N first, of someone else's thought. Starting at about 9 months of age, normal infants begin to exhibit the phenomena of "joint attention." They look where their mother is looking (or where she is pointing) and try to attract her attention to what they themselves are interested in. When they start to learn words, at about 1 year or so, it is always in contexts where they and their caretaker are attending to the same object or event (Tomasello & Farrar, 1986). It is only because Johnny already knows what his mother is thinking about (i.e., attending to) that he can know what her words mean and thus learn them for himself. On some occasions, however, mommy is attending not to some other object or event but to Johnny himself. It is on these occasions, when Johnny is an "object of thought" for his mother, that he can become one for himself as well. If we accept this scenario, we can date the onset of the conceptual self somewhere in the neighborhood of the baby's In'st birthday.

Origins of the Interpersonal Self Two forms of the self are given perceptually, distinguished by the forms of information that specify them. The ecological self the embodied individual purposefully engaged with the environment N is the principal topic of this chapter. Nevertheless, we cannot omit mention of the interpersonal self at this point. Babies engage in interpersonal communication from a very early age, long before they are conceptual selves and perhaps even before they can act effectively in the environment. The relevant evidence here is not mere social responsiveness. Responses like social imitation or preference for the mother's voice, now well established in neonates (DeCasper & Fifer, 1980; Meltzoff & Moore, 1989), easily could occur without the infant's being aware of its own role as an interpersonal agent. That attribution is justified only if the infant looks for--and findsmthe social consequences of its own social behavior. Just this was shown in a significant study by Lynne Murray and Colwyn Trevarthen (1985). Their experiment took advantage of the fact that 8-week-old infants enjoy participating in social dialogues "protoconversations" with

CRITERIA FOR AN ECOLOGICAL SELF

21

their mothers. During these exchanges they maintain attention, seek eye contact, and may smile and coo. Murray and Trevarthen set up a closed-circuit video system that made such exchanges possible even though mother and infant were in separate rooms. These televised protoconversations proceeded normally at first, as long as the baby could see and hear its mother "live," in real time. But when the subjects were shown videotapes of their mothers (tapes that had been made a few moments earlier during the live condition) they soon lost interest, showing signs of boredom if not distress. This means that their activities during the live condition must have been genuine interactions, not simply responses to stimuli. The infants were aware of (and enjoyed) their partner's responsiveness to their own social gestures m a responsiveness that was missing in the condition using videotape. In other words, they were aware of themselves as social agents. This demonstration shows that the interpersonal self, actively and purposefully responding to perceived social signals from a partner, is in place by 8 weeks of age. Much more could be said about this form of self-knowledge and its significance, and I have tried to say some of it elsewhere (Neisser, 1994). Here, however, it is time to turn to the ecological self.

The Ecological Self: Being in the Environment I described the ecological self above simply as "the embodied individual purposefully engaged with the environment." That will do for many purposes, but a more substantial definition is appropriate here. An ecological self is an individual who is, and perceives herself to be, located at a given place (or moving along a given path) in an extended environment of surfaces and objects. She has, and perceives herself to have, an extended body that is capable of interacting with the environment in a purposeful way. Those interactions are, and are perceived to be, relevant to her own needs and satisfactions m including the satisfaction that comes from purposive action itself. A first implication of this definition is that ecological selves are perceptually differentiated from their environments. The individual is in the environment but partly independent of it, moves through it, interacts with it, and consistently perceives this differentiated state of affairs. This achievement is only possible in species that are equipped with adequate perceptual systems, able to pick up the information that specifies the layout of the environment as well as the position and movement of the self. For humans, that information is primarily visual. (The role of other modalities will be addressed briefly below.) Moreover, it is primarily based on movement: on the visible motions of objects, of body parts, and of the individual him/herself in locomotion through the environment.

22

ULRIC NEISSER

The last of these, often called "egomotion," is the most informative of all. As the eye moves through the optic array (i.e., the complex, space-filling structure of reflected light that reaches it from every direction), it is exposed to distinct patterns of optic flow that specify both the path of locomotion and the layout of objects through which it is moving (Gibson, 1979). These patterns are of many kinds: Close approach to a surface produces rapid radial expansion of the corresponding sector of the array, a near object may occlude another that lies farther away, etc. Perceptual research has shown that adult human observers readily pick up and use these optic invariants to see how they are moving and where they are going. Although nonlocomotor infants do not yet have access to all of this information, they must get a great deal just from being carded around. Moreover, they have had controllable head and body movement from an early age. We can be confident that babies in their cribs see where the rails are, just as easily and accurately as pedestrians in the park see where the trees are. Both kinds of observers also see where they themselves are in the environment defined by those landmarks. They also see that the trees/rails stay fna~ly fixed in place, no matter how they themselves may move. Similar forms of information are produced by object motion even when the observer is stationary. The most familiar example is "looming": the explosive magnification of a sector of the array that specifies the approach of an object toward the eye. It has long been known that even very young infants - - and many species of a n i m a l s - will flinch away from such displays (Schiff, 1965). This is no mere reflex. Carroll and Gibson (1981; see also Gibson, 1982) showed that 3month-olds behave very differently when the looming object is an aperture (like the frame of a window) instead of a full textured surface. They do not flinch away, but lean forward to look through the opening. This behavior, which serves no purpose except visual exploration, strikingly illustrates the baby's awareness of the independently existing environment. Nonvisual forms of information can also be important. The impact of stumbling into a tree, the feel of running one's hand along a rail, the echo pattern reflected from a nearby wall all these are further information for the position and activity of the self in an independently existing environment. Normally, all of them are consistent with what is specified in the optic array. That consistency can be disrupted by experimental manipulations, most radically in studies where subjects wear prisms or lenses that invert the whole visual field. The results of such experiments show that vision, typically the most accurate perceptual system, tends to dominate and calibrate other modalities. In particular, proprioception and auditory localization soon shift to conform to what is visually given (Harris, 1965). Blind children, who must rely on other modalities in the absence of vision, are appreciably slower to develop an articulated ecological self (Bigelow, this

CRITERIA FOR AN ECOLOGICAL SELF

23

volume; Fraiberg, 1977; Hobson, 1993).

Agency I have defined the ecological self not only as situated in the environment but also as purposeful. Taken by itself, this is an easier criterion to meet. Babies are intentional agents almost from birth. Consider, for example, the so-called "circular reaction": Very young infants tend to repeat responses that have perceptible effects, especially if those effects are pleasant. Given the appropriate experimental arrangements (i.e., a properly wired pacifier), sucking is easily transformed into just such a response. Babies will suck persistently to bring pictures into view (Siqueland & DeLucia, 1969), to clarify a picture that was initially blurred (Kalnins & Bruner, 1973), or even just to hear tape-recorded spoken syllables (Eimas, Siqueland, Jusczyk, & Vigorito, 1971). Michael Lewis (1990a) has emphasized an important aspect of these behaviors: The infants exhibit obvious pleasure when their control of the effect in confirmed and equally obvious frustration when (during "extinction") it is disconfirmed. In Lewis's view, these emotional reactions are clear signs of intentionality. Although I agree with Lewis that such behaviors are evidence of purpose, they do not--by themselves m demonstrate the existence of an ecological self. They do not yet show that the infant is aware of its situation in the environment, of its own body and of the capabilities of that body. Intentionality may be a necessary condition of s e l f h o o d - no passive and purposeless entity is a s e l f - but it is not sufficient. More stringent criteria are needed. The most fundamental of these, I believe, is awareness of one's situation in an independent, spatially extended environment. The mere occurrence of intentional sucking (or of other conditioned operants) does yet show that this criterion has been met. Although sucking by itself does not demonstrate any spatial awareness, many other motivated behaviors obviously do. Bringing the hand to the mouth is an interesting early example: In newborns it becomes more frequent if the infant is given sucrose (Rochat, 1993; Rochat, Blass, & Hoffmeyer, 1988). Older infants, who can reach and grasp objects, often bring their hands to the mouth to facilitate oral exploration of objects they are holding. Rader and Vaughn (unpublished) have recently confirmed that this behavior is under intentional control. Objects that had been dipped in either a sweet or a bitter solution were repeatedly placed within reach of 4- and 5-month olds infants: After the first taste, subjects reached much more often for a sweet object than for a bitter one. Another obviously intentional spatial response that has been much studied is kicking. When one foot has been connected to an overhead mobile by means of a

24

ULRIC NEISSER

string, for example, even very young infants will keep on kicking almost indefinitely to make the mobile move (Rovee-Collier, 1989). In this case too, the baby is happy as long as its kicks are effective and becomes obviously frustrated when (with the string removed) they are not. Are those kicks just arbitrary motor responses that happen to have visual consequences, or is the baby aware of the spatial relations between itself and the mobile? I am inclined to take the latter view, but so far as I know there has been no research addressed to this question.

The Body The ecological self is not just located in the environment, but also embodied. Our ongoing experience almost always includes the awareness of being a mobile, coherent, effective, space-occupying body. Here too, although it is easy to emphasize the role of vision, we actually perceive the positions and motions of our bodies by means of several different perceptual systems: -vestibularly, via the organs of balance in the inner ear; -proprioceptively, through the vast system of neurons and receptors spread throughout the body itself and responsive to its motion; -tactually, when the body makes contact with external objects; -visually, not only on the basis of the optic flow patterns mentioned above but in a more concrete, objectlike way when parts of the body are actually in view; -through hearing, especially as in babbling or crying. The functioning of the vestibular system, which detects the direction of gravity as well as accelerations and decelerations of the head, seems to be largely determined by innate mechanisms (see Jouen & Gapenne, this volume). To some extent, this must also be true of the tactile/proprioceptive system. (A remarkable example: Melzack (1992) has described cases in which a person born without a forearm or lower leg nevertheless experiences a "phantom" of the missing limb.) On the other hand, we have already noted the results of studies with prism spectacles and other forms of visual rearrangement: These show that touch and proprioception are readily calibrated and recalibrated by vision. This is as it should be, given the changes in reach and stride and body size that occur during development. The dimensions and capabilities of the bodily self change over time, and bodily awareness must change with it. This principle applies even to the boundaries of the perceived self. The ecological self, as we experience it, includes anything that moves consistently

CRITERIA FOR AN ECOLOGICAL SELF

25

with us and under our control. Even things that are not parts of the biological body - - artificial limbs, clothes, cars, tools, and the like - - belong to the self as long as we are controlling, wearing, or using them. Many phenomenologists have commented on the compelling sense in which a tool can seem to be incorporated into the self (Merleau-Ponty, 1962, p. 143). These experiences produce what seems to be a paradox: I "know" that the hammer in my hand is not really a part of me, but it still feels as if it were. The paradox is resolved by distinguishing among different forms of information and the selves that they establish. The hammer is no part of my conceptual self, but while I am hammering it does become part of my ecological self. The opposite effect can also be observed, though less often. Under some conditions, genuine parts of the biological body are not perceived as belonging to the self. Our arms and legs, for example, are parts of our ecological selves only so long as they can be engaged in purposeful movement. A limb that cannot be deliberately moved, as in the "neglect" syndrome of clinical neuropsychology, is often experienced as a disturbing foreign object: What is this leg doing in my bed with me (Heilman, Watson, & Valenstein, 1985)? In his book A Leg to Stand On, Oliver Sacks (1984) describes how even a temporary, peripherally caused inability to move his leg produced the same dramatic and bizarre result.

Pain: A Digression The strong emphasis on activity in this discussion may seem one-sided to some readers. What about passive touch and bodily sensations? Especially, what about pain? Are we not most keenly self-aware when - - perhaps lying quietly in bed we suffer from a raging toothache? Not necessarily, I think. Although there can be no pain without an organism to experience it, there can surely be pain without self-knowledge (cf. Wittgenstein, 1958). Pain is a conscious experience, but consciousness can exist without self-consciousness. Suppose, for example, that an animal or an infant responds to noxious stimulation with obvious distress. As an observer I am very willing to take such distress as evidence of experienced pain. It does not follow, however, that there must be an experienced self. Even when it produces reflex responses or disrupts ongoing behavior, pain may dominate conscious experience in an unstructured, global way without being attributed to anything at all like a s e l f - - or indeed to anything at all. To be sure, this rarely happens to us. As adults, we usually incorporate our pains into one or another (or several) different aspects of the self: 1. Pains referred to body parts that can be purposefully moved are

26

ULRIC NEISSER

experienced as belonging to the ecological sell as when I find that my injured leg hurts if I walk. 2. Pain that is attended to in its own fight, as a salient aspect of inner experience, belongs to the private self. This aspect sometimes becomes focal, as for example when individuals are asked to rate the intensity of their pain. Nevertheless, it is not a necessary feature of pain experience. 3. When the individual focuses on what the pain may mean as a symptom of underlying illness, as a constraint on possible action, as a positive or negative aspect of the self-image m it becomes part of the conceptual self. Interestingly, this usually makes it worse. A plausible hypothesis is that pain control methods such as hypnosis and controlled relaxation work chiefly by reducing the involvement of the conceptual self in the ongoing experience. Because infants do not have a conceptual s e l f - let alone a private self options 2 or 3 above do not apply to them. In fact, they may not have option 1 either. Even when an initial ecological self has been established, the pain may be strong enough to disrupt its functioning. This is especially likely in early infancy. When an older child hurts his finger, he is likely to put it in his mouth: This means that the pain has been assigned to the appropriate part of the ecological self (i.e., the finger). Given their less robustly established ecological selves, I would not expect this behavior from babies. In any case, it is clear that responses to pain cannot be used to index the self in any simple way.

Affordances and Exploration So far, I have defined the environment independently of the self. To do so risks missing an important point: Almost every environment is rich in possibilities for action. Such affordances (Gibson, 1979) are real aspects of the environment, but defined with respect to given individuals at given moments. The floor affords walking to a toddler, crawling to a 10-month-old, and neither to a newborn. A particular object may afford reaching (if it is near enough) or throwing (if detachable and of the fight mass), but only to individuals with the appropriate capabilities. Many affordances can be perceived: We usually see what we might do before actually doing it. There can be misperceptions here (the floor may not actually afford walking because it contains a concealed trapdoor), but in normal environments they are relatively rare. A number of studies have shown that everyday body-scaled affordances m the climbability of stairs (Warren, 1984), the

CRITERIA FOR AN ECOLOGICAL SELF

27

sit-upon-ability of chairs (Mark, 1987), etc. - - are perceived quite accurately. Much of this research has used adult subjects, but there have been several studies of the perception of affordances by infants. Some of these concern reaching: For example, 5-month-olds easily distinguish what is and isn't within their reach (Field, 1976; Rochat & Goubet, 1993). The most systematic research program is that of Eleanor Gibson on infants' perception of surfaces of support. This work began with her classical study of the visual cliff (Gibson & Walk, 1960). The fact that most infants refuse to cross the "cliff" can best be understood in terms of the perception of affordances. No animal will locomote unless it perceives an affordance for locomotion, and no such affordance (i.e., no surface) is visible on the deep side of the cliff. More recently, Gibson and her associates have studied many other aspects of this problem (Gibson et al., 1987). In recent discussions of these issues, Gibson (1993, this volume) has suggested that the ability to perceive affordances is an especially good index of self-perception and self-awareness. All affordances are relations; they are never determined by the environment alone. The fact that an object "affords reaching" for a baby (is within her reach) says as much about the baby (e.g., the length of her ann) as it does about the position of the object. So the argument seems straightforward: If an infant sees a nearby object as within reach, she is ipso facto also seeing herself as an individual who can reach that far. Is this not a form of self-awareness? This argument requires somewhat more consideration. There are situations in which animals seem to act on perceived affordances, but are really just responding to stimuli. Consider, for example, how flies manage to make soft landings. It is now well established (Wagner, 1982) that the landing movements of the fly are controlled by the radially expanding optic flow field that results from approach to a surface. On detecting certain critical changes in that flow field, the fly slows down and extends its legs. One could easily describe this behavior as "perceiving the affordance of the surface for landing," but I would not like to conclude that flies are ecological selves. There is no reason to believe that the fly is aware of its real situation in an independently existing environment. More likely it is just a stimulus-response machine, one for which flow fields are among the effective stimuli. What might convince us that an animal or an infant is aware of that situation? One powerful piece of evidence, often emphasized by Gibson herself (1988, this volume) would be the appearance of exploratory behavior. Consider, for example, how an 8-month-old infant behaves at the top of a steeply sloping ramp when his mother calls to him from the bottom (Adolph, Eppler, & Gibson, 1993). At first he may misperceive the affordances, heading fight down the ramp and tumbling off (hopefully to be caught by the experimenter!). But after a minimum of experience,

28

ULRIC NEISSER

most babies stop at the top of the ramp and begin to explore. They hesitate, look around, feel the slope with their hands or their feet, change position. Only then do they decide whether or not to go, with their final decision depending on whether the slope is, in fact, shallow enough to afford crawling. Such a baby really is an ecological self, aware of itself as a choice-making individual with limited capabilities in an independently existing environment. There are many forms of exploration: looking around, touching, changing bodily attitudes, moving to new points of observation, etc. Many species of animals explore. Although I am no comparative psychologist, I cannot resist mentioning one form that was accepted and studied even in the darkest days of behaviorism. Psychologists of the 1940s were very interested in a phenomenon then called "VTE" (vicarious trial and error): the systematic side-to-side head movements of a rat at a choice-point in a maze. This is surely an example of exploration, by an animal with a substantial amount of ecological self-awareness. Although a rat engaged in VTE may indeed not know which way to turn, it apparently knows quite a lot about the situation as a whole: that there are two possible ways to go, that they lead to different places, that it can choose either of the two. VTE is a way of obtaining as much information as possble before making a choice. Infants, of course, have a much richer repertoire of exploratory behaviors. Another way in which an animal or an infant can demonstrate knowledge of environmental layout is by taking purposeful detours. The ability to move directly toward a goal (G) actually reveals very little about an individual's understanding. Like the fly in the flow field, it may just be responding to some simple stimulus. But taking a route that initially leads away from G - an indirect but eventually successful route, one not randomly chosen but also not previously explored m is clear evidence that something more is going on. An individual who does this evidently knows that it is in an extended environment, in which it cannot go directly to G from here but could possibly get to G from there.

Body Parts Being able to see parts of one's own body - - q s p e c i a l l y parts that can also be manipulated and controlled offers important opportunities for self-perception. Not all animals (or even all vertebrates) can do so. Considering where the eyes of fish are located, for example, it seems impossible for them ever to see any part of themselves. Animals like horses and cows, who have eyes on the sides of their head and few occasions to manipulate their limbs in the field of view, are also at some disadvantage in this respect. Primates, on the other hand, can visually

CRITERIA FOR AN ECOLOGICAL SELF

29

examine themselves just by inclining their heads. What's more, they can easily bring their h a n d s - and often also their f e e t - into the field of view. Once babies get the hang of this, they spend lots of time doing it. By observing its own hands and feet in action, the baby is able to detect relationships between the motions it initiates and the motions it sees. Piaget called this the mutual assimilation of kinesthetic and-visual schemata; Gibsonians prefer to call it the detection of amodal invariants. Until recently, most developmental psychologists believed that seeing one's hands in this way was an essential precursor of directed reaching and other manual skills. It now seems that this is not the case: Babies can successfully reach for sounding targets in the dark without seeing their hands at all (Clifton, Muir, Ashmead, & Clarkson, 1993). But although seeing one's hands is not essential for reaching, it may still be critical for developing a sense of being an extended ecological self. When babies watch the movements of their hands, it is (parts of) their own selves that they see. Several studies have shown that infants can distinguish between their own appendages (in these experiments, usually feet) and those of others. Bahrick and Watson (1985) presented 5-month-olds with two television screens side by side. One showed a real-time video of the subject's own legs; the other showed a videotape of the legs of another similarly clad baby, or of the subject's legs in a videotape made on an earlier occasion. An analysis of preferential looking showed that the subjects could easily make this distinction. They looked mostly at the legs of the stranger, perhaps searching for kinesthetic-visual invariants that were not actually there. Rochat and Morgan (1995) tested 3-month-olds in a more sophisticated version of the Bahrick/Watson paradigm and obtained an even more interesting result. By including a left-fight reversal of the televised image as one of their displays, they were able to show that the critical variable the information by which the infant identifies its own imaged legs in such experiments is the direction in which the legs are seen to move. Throughout the baby's previous experience, a directed movement like this (say, to the fight) has always produced a visual motion like that (also to the fight). Given a display where this is no longer the case, the infant does not recognize the visible legs as its own.

Mirrors

I am by no means the first to suggest that the sight of one's own body is an important factor in establishing a sense of self. Unfortunately, however, most researchers interested in this question have focused on what is really an advanced special case: the sight of oneself in a mirror. The well-known "rouge test"

30

ULRICNEISSER

(Amsterdam, 1972; Gallup, 1977) shows that infants do not properly identify mirror images of their own faces until late in the second year. It is sometimes suggested that this achievement marks the first appearance of self-awareness. As we have seen, however, this argument must be mistaken. Many other indices purposive movement, exploratory behavior, detours, pleasure in personal effectiveness, interest in one's own body parts m show that the ecological self is established within the first few months of life. What, then, is missing in a 1-year-old who fails the rouge test? There are several possibilities. The difficulty may just be a failure to understand the optics of mirrors, which systematically reverse many of the information structures that specify object position in normal viewing (Loveland, 1986). But this is probably not the whole story: Many primates who fail the rouge test will still use mirrors for other purposes (e.g., to see what is behind them). A more probable cause is that, at this age, the face is not yet an important component of the ecological self. Although 1-year-olds know a great deal about themselves and their bodies, they have had little reason to be interested in their faces. When this interest eventually develops, it manifests itself in several different ways: Two-year-olds not only pass the rouge test but tend to hide their faces when they are embarrassed, something they would never have done a year earlier (Lewis, 1990b). The same principles apply to the self-knowledge of animals. Chimpanzees and orangutans are the only nonhuman primates who pass the rouge test. Gorillas regularly fail it: They show no interest in the red painted spots on their foreheads that are so obviously visible in the mirror. Nevertheless, the very same gorillas do show a lively interest in similarly painted spots on their wrists (Suarez & Gallup, 1981)! Why would they do so, if they were not ecological selves? Again, I would like to suggest a more plausible interpretation. What such a gorilla lacks is not a sense of self, just a compelling interest in the appearance of its own face. Mirror-recognition is a phenomenon well worth studying, but it is deeply misleading as an index of early self-awareness.

Summary and Conclusions In short, there is plenty of evidence for the appearance of the ecological self at a very early age. Babies who watch their own hand movements in fascination and who distinguish their own body parts from those of others must know that they are independent agents. These are the same babies who clearly know where they are and how the environment extends around them, who explore that environment to determine what actions it may afford, who take pleasure in bodily movement as well as in being the agents of environmental effects. Table 1 lays out these forms

CRITERIAFOR AN ECOLOGICALSELF

31

of evidence m o r e systematically. The ecological self b e c o m e s manifest in purposive movement (especially when that movement involves deliberate detours or systematic exploration), in enjoyment of one's own powers of agency, and in various forms of self-inspection and self-exploration.

TABLE 1. Criteria for an Ecological Self. I. Awareness of the environment

a) S's movements are clearly directed toward particular environmental objects and adapted to the affordances of those objects. b) S monitors and adjusts his/her ongoing movements in progress, relying on specific and appropriate stimulus information (e.g., optic flow). c) S displays exploratory as well as goal-directed behavior. d) S demonstrates knowledge of the overall environmental layout, including regions not directly between self and goal (e.g., takes detours). II. Awareness of the body

a) S exhibits pleasure in (and tends to repeat) certain bodily actions, especially those that have consistent perceptible effects. b) S repeatedly examines and explores his/her own body and parts (e.g., self-touching, looking at hands and feet). c) S distinguishes his/her own body parts from those of conspecifics, and also from other objects (e.g., consistent preferential looking in a video display).

The range of phenomena in Table 1 illustrates a point that should be made more explicitly. Being an ecological self is not an all-or-none affair. Six-month old infants exhibit almost all of these criterial behaviors and are surely ecological selves, but they do not yet go on detours because they cannot m o v e around independently. Fetuses, in contrast, meet none of the criteria except perhaps IIa; they are certainly not such selves. There is a course of development in between, so it is easy to imagine infants at intermediate points who exhibit some of these characteristics without having them all. It would be a mistake to try to determine

32

ULRICNEISSER

the exact moment of onset of ecological selfhood, or indeed of any aspect of the self. In closing, I want to stress again that ecological awareness is not the only preconceptual form of self-knowledge. Babies are social as well as ecological creatures from the first. They soon come to know not only where they are and what they can do, but who they are with and what is going on. As interpersonal selves they anticipate the responses of others to their own social initiatives, cheerfully engaging in reciprocal play. These two forms of self-knowledge are equally fundamental for development. An awareness of one's real situation in both the physical and social worlds is the foundation on which other, later, more sophisticated aspects of the self are eventually built.

REFERENCES

Adolph, K. E., Eppler, M. A., & Gibson, E. J. (1993). Crawling vs. walking infants' perception of affordances for locomotion over sloping surfaces. Child Development, 64, 1158-1174. Bahrick, L. E., & Watson, J. S. (1985). Detection of intermodal proprioceptive-visual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21, 963-973. Carroll, J. J., & Gibson, E. J. (1981). Differentiation of an aperture from an obstacle under conditions of motion by three-month-old infants. Paper presented at the Meetings of the Society for Research in Child Development, Boston, MA. Clifton, R. K., Muir, D. W., Ashmead, D. H., & Clarkson, M. G. (1993). Is visually guided reaching in early infancy a myth? Child Development, 64, 1099-1110. DeCasper, A. J., & Fifer, W. P. (1980). Of human bonding: Newborns prefer their mothers' voices. Science, 208, 1174-1176. Eimas, P. D., Siqueland, E. R., Jusczyk, P. W., & Vigorito, J. (1971). Speech perception in infants. Science, 171, 303-306. Field, J. (1976). Relation of young infants' reaching behavior to stimulus distance and solidity. Developmental Psychology, 12, 444-448. Fivush, R., Gray, J. T., & Fromhoff, F. A. (1987). Two-year-olds talk about the past. Cognitive Development, 2, 393-409. Flavell, J. H., Flavell, E. R., & Green, F. L. (1983). Development of the appearancereality distinction. Cognitive Psychology, 15, 95-120. Fraiberg, S. (1977). Insights from the blind: Comparative studies of blind and sighted infants. New York: Basic Books. Gallup, G. G. J. (1977). Self-recognition in primates. American Psychologist, 32, 329-338. Gibson, E. J. (1982). The concept of affordances in development: The renascence of functionalism. In W. A. Collins (Eds.), The concept of development." Minnesota symposium on child psychology, Vol. 15 (pp. 55-81). Hillsdale, NJ: Erlbaum. Gibson, E.J. (1988). Exploratory behavior in the development of perceiving, acting, and the acquiring of knowledge. Annual Review of Psychology, 39, 1-41. Gibson, E. J., Riccio, G., Schmuckler, M. A., Stoffregen, T.A., Rosenberg, D., & Taormina, J. (1987). Detection of the traversability of surfaces by crawling and walking infant,s. Journal of Experimental Psychology: Human Perception and Performance, 13, 533-544.

CRITERIAFOR AN ECOLOGICALSELF

33

Gibson, E. J., & Walk, R. D. (1960). The "visual cliff." Scientific American, 202, 6471. Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. Harris, C. S. (1965). Perceptual adaptation to inverted, reversed, and displaced vision. Psychological Review, 72, 419-444. Heilman, K. M., Watson, R. T., & Valenstein, E. (1985). Neglect and related disorders (2nd Ed.). In K. M. Heilman & E. Valenstein (Eds.), Clinical neuropsychology (pp. 243-293). New York: Oxford University Press. Hobson, R. P. (1993). Through feeling and sight to self and symbol. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 254-279). New York: Cambridge University Press. Hudson, J. A. (1990). The emergence of autobiographical memory in mother-child conversation. In R. Fivush & J. A. Hudson (Eds.), Knowing and remembering in young children (pp. 166-196). New York: Cambridge University Press. James, W. (1890). Principles of psychology. New York: Holt. Kalnins, I. V., & Bruner, J. S. (1973). The coordination of visual observation and instrumental behavior in early infancy. Perception, 2, 307-314. Lewis, M. (1990a). The development of intentionality and the role of consciousness. Psychological Inquiry, 1, 231-247. Lewis, M. (1990b). Self-knowledge and social development in early life. In L. A. Pervin (Eds.), Handbook of personality (pp. 277-300). New York: Guilford. Loveland, K. A. (1986). Discovering the affordances of a reflecting surface. Developmental Review, 6, 1-24. Mark, L. S. (1987). Eyeheight-scaled information about affordances: A study of sitting and stair climbing. Journal of Experimental Psychology: Human Perception and Performance, 13, 361-370. Mead, G. H. (1934). Mind, self and society. Chicago: University of Chicago Press. Meltzoff, A. N., & Moore, M. K. (1989). Imitation in newborn infants: Exploring the range of gestures imitated and the underlying mechanisms. Developmental Psychology, 25, 954-962. Melzack, R. (1992). Phantom limbs. Scientific American, April, 120-126. Merleau-Ponty, M. (1962). Phenomenology of perception (translated by Colin Smith). London: Routledge & Kegan Paul. Murray, L., & Trevarthen, C. (1985). Emotional regulation of interactions between two-month-olds and their mothers. In T. M. Field & N. A. Fox (Eds.), Social perception in infants (pp. 177-197). Norwood, NJ: Ablex. Neisser, U. (1988). Five kinds of self-knowledge. Philosophical Psychology, 1, 3559. Neisser, U. (1991). Two perceptually given aspects of the self and their development. Developmental Review, 11, 197-209. Neisser, U. (Ed.). (1993). The perceived self" Ecological and interpersonal sources of self-knowledge. New York: Cambridge University Press. Neisser, U. (1994). Self-perception and self-knowledge. Psyke & Logos, 15, 392-407. Neisser, U., & Fivush, R. (Eds.). (1994). The remembering self" Construction and accuracy in the self-narrative. New York: Cambridge University Press. Neisser, U., & Jopling, D. (Ed.). (in press). The conceptual self in context." Culture, experience, self-understanding. New York: Cambridge University Press. Rader, N., & Vaughn, L. A. (unpublished). Intentional reaching in early infancy. Rochat, P. (1993). Hand-mouth coordination in the newborn: Morphology, determinants, and early development of a basic act. In G. J. P. Savelsbergh (Eds.), The development of coordination in infancy (pp. 265-288). Amsterdam: Elsevier.

34

ULRICNEISSER

Rochat, P., Blass, E. M., & Hoffmeyer, L. B. (1988). Oropharyngeal control of handmouth coordination in newborn infants. Developmental Psychology, 24, 459463. Rochat, P., & Goubet, N. (1993, March). Determinants of perceived reachability in infancy. Paper presented at the Meeting of the Society for Research in Child Development, New Orleans, LA. Rochat, P., & Morgan, R. (1995). Spatial determinants in the perception of selfproduced leg movements by 3- to 5-month-old infants. Developmental

Psychology. Rogoff, B., & Mistry, J. (1990). The social and functional context of children's remembering. In R. Fivush & J. A. Hudson (Eds.), Knowing and remembering in young children (pp. 197-222). New York: Cambridge University Press. Rovee-Collier, C. (1989). The joy of kicking: Memories, motives, and mobiles. In P. R. Solomon, G. R. Goethals, C. M. Kelley, & B. R. Stephens (Eds.), Memo~: Interdisciplinary approaches (pp. 151-180). New York: Springer-Verlag. Sacks, O. (1984). A leg to stand on. London: Duckworth. Schiff, W. (1965). Perception of impending collision. Psychological Monographs, 79, No. 604. Siqueland, E. R., & DeLucia, C. A. (1969). Visual reinforcement of non-nutritive sucking in human infants. Science, 165, 1144-1146. Suarez, S. D., & Gallup, G. G. (1981). Self-recognition in chimpanzees and orangutans, but not gorillas. Journal of Human Evolution, 10, 175-188. Tomasello, M. (1993). On the interpersonal origins of self-concept. In U. Neisser Ed.), The perceived self (pp. 174-184). New York: Cambridge University Press. Tomasello, M., & Farrar, J. (1986). Joint attention and early language. Child Development, 57, 1454-1463. Wagner, H. (1982). Flow-field variables trigger landing in flies. Nature, 297, 147-148. Warren, W. H. (1984). Perceiving affordances: Visual guidance of stair climbing.

Journal of Experimental Psychology: Human Perception and Performance, 10,

683-703. Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and constraining function of wrong beliefs in children's understanding of deception. Cognition, 13, 103-128. Wittgenstein, L. (1958). The blue and brown books. Oxford: Blackwell.

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

35

CHAPTER 3

The Self as an Object of Consciousness in Infancy GEORGE BU'IqqERWORTH

University of Sussex

Introduction: Consciousness of Self and Self-consciousness Traditional developmental psychology attributed little, if any, conscious awareness to the young infant. Over the past 30 years, such preconceptions have gradually been eroded: Demonstrations of selective attention even in newborn babies suggest that babies are not merely the hapless victims of a barrage of sensory stimulation. The traditional assumption that the newborn is merely a reflexive organism, all hungry body but with little mind and with only the most limited capacity to perceive self or the world, is no longer received wisdom. Yet elaborating a novel theory of the origins of self-knowledge from a new starting point has hardly begun. This paper will explore a distinction between "primary" consciousness of self, based on the facts of being embodied in social and physical reality, and "higherorder" reflective self-awareness, in early human development. It will first be necessary to say something about the history of consciousness in general before moving on to self-consciousness. As Humphrey (1992) points out, the meaning of the word has not only become narrower and narrower, it has turned around over historical time. The word derives from the Latin con, meaning "together with" and scire, meaning "to know." In the original Latin, the verb conscire (from which came the adjective conscius) meant literally "to share knowledge with other people." The circle with whom the knowledge was shared became tighter and tighter until, eventually, it included just a single person, the subject who was conscious. Most recently, there was a further shift of definition to having subjective knowledge, which by its very nature, no one else could have access to (i.e., knowledge of one's innermost thoughts and feelings). As Humphrey says, consciousness is rather like the word window, which has changed its meaning from "a hole where the wind comes in" to "a hole where the

36

GEORGE BUTFERWORTH

wind does not come in." It is as if consciousness has changed from "having shared knowledge" to "having intimate knowledge not shared with anyone except oneself." There has been erected in psychological theory a solipsistic barrier between developing self-knowledge and developing knowledge of the world. Consciousness has shifted from being a matter of public knowledge of public objects to private knowledge. Here we will show how it is possible to define consciousness transitively, in order to investigate how we can be conscious of self as an object, just as we can know about other physical objects. We will also consider how public, perceived objects themselves may constitute intermediaries between unrelated minds. Another distinction useful for a developmental account of the origins of self is that between "primary" consciousness, based on perception, and "higher-order" consciousness, or reflective self-awareness (Edelman, 1989). According to Edelman, primary consciousness is the state of being mentally aware of things in the world. Higher-order consciousness includes recognition by a thinking subject of his or her own acts or affections. It embodies a model of the personal and of the past, future, and present. In higher-order consciousness, there is direct awareness of mental episodes without involvement of sense organs. In essence then, the distinction between primary and higher-order consciousness is one between consciousness of the products of perception and consciousness of mental events in and of themselves. A simple-minded distinction between primary consciousness-ofself, which would occur when the self is the object of one's own perception, and higher-order self-consciousness, which would occur when the self is the object of one's own cognition, may help in unraveling the problem of self from a developmental perspective. The question toward which we now move is how the self can be the object of one's own perception.

Adualistic Confusion and the Origin of Self The assumption that there can be no psychological distance between the infant and the world has been widely assumed to characterize early experience. The newborn baby was traditionally considered to be undifferentiated from the world in her own self-awareness. In fact, Piaget (1954) actually spoke of the newborn as a visually perceiving two-dimensional tableaux that is completely undifferentiated as to components due to activity of self and components due to independent events in external reality. In his theory, vision only slowly acquires depth through sensorimotor coordination with touch and through the metric provided by the infant's own actions. According to this traditional view, muscular kinesthesis forms a primary space against which vision is calibrated. In Piaget's theory

SELF AS AN OBJECT OF CONSCIOUSNESS

37

(Piaget 1954, 1962) and other major theories such as the Freudian account, development proceeds from total lack of differentiation (adualistic confusion) to proximal sensitivity (awareness of the kinesthetic qualities of the infant, s own body) and finally to distal spatial sensitivity, as the infant gradually constructs planes of depth through her own activities. By about 18 months, largely as a result of the onset of independent locomotion, the infant becomes aware of herself as a totality, contained within an encompassing space. Profound adualistic confusion is a necessary theoretical consequence of the assumption that there can be no visual space perception at birth, but it is not an established empirical fact. An alternative view, "natural dualism," was first put forward by the philosopher Thomas Reid (1764). Natural dualism has been defined as "an immediate knowledge by mind of an object different from any modification of its own .... The ego and the nonego are thus given in an original synthesis, as conjoined in the unity of knowledge and in an original antithesis, as opposed in the contrariety of existence" (Baldwin, 1901, p. 134. Contrariety is defined as "the relationship between two contraries, an opposition between one thing and another"). James Gibson (1987) may have been influenced by Reid in formulating his theory of direct perception. Gibson's (1966) theory is that information about the environment, obtained through perceptual systems, is sufficient directly to inform the perceiver about her relationship with the world. Some implications of the Gibsonian approach to perception for theories of self-knowledge have been explored by Neisser (1988, 1993) and Butterworth (1988, 1992a, 1992b, 1995). Neisser distinguished between five kinds of self-knowledge, each of which may intersect with rather different periods of human development: 1) the ecological self, which is directly perceived with respect to the physical environment; 2) the interpersonal self, also directly perceived, which depends on emotional and other species-typical forms of communication; 3) the extended self, which is based on memory and anticipation and implies a representation of self; 4) the private self, which reflects knowledge that our conscious experiences are exclusively our own, also dependent on representation; and 5) the self-concept, defined as a theory of self based on sociocultural experience. This chapter will be concerned mainly with the first and second of Neisser's types of self-knowledge, i.e., with the level of direct awareness of self in relation to physical and social reality. Some characteristics of the ecological self that have been listed by Neisser (1988) are: a) it is specified by objective information; b) much of the information is kinetic and is available to several perceptual systems at once; and c) the ecological self is veridically perceived from infancy, but selfperception may still develop and become more adequate with increasing skill.

38

GEORGEBUTrERWORTH

The Self as the Object of the Infant's Perception The Gibsonian approach to the origins of self stands in contrast to more prevalent "indirect," adualistic theories, which are based on the idea of sensations that need to be embellished by inference, action, and memory. Direct perception means that information both about objects (including the self) and events in the environment, as they relate to behavior, is preserved in the energy patterns that impinge on perceptual systems. The ecological level of perception of reality is defined as follows by Gibson (1987): Awareness of the persisting and changing environment (perception) is concurrent with the persisting and changing self (proprioception in my extended use of the term). This includes the body and its parts and all its activities from locomotion to thought, without any distinction between the activities called "mental " and those called 'physical." Oneself and one's body exist along with the environment, they are coperceived (p. 418). Gibson departs from Sherrington (1906) in explaining awareness of the bodily self. According to Gibson (1987), proprioception is considered as a general function of perceptual systems, regardless of modality, rather than a specifically kinesthetic sense. In the traditional Sherringtonian account, proprioception is given internally by muscle and joint receptors. Gibson extended proprioception to include external feedback arising as a normal correlate of the exploratory activity of perceptual systems. Instead of kinesthesis being considered a special sense, proprioception becomes a general, self-specifying aspect of the informationseeking functions of perceptual systems. Thus, proprioception is a mechanism of self-sensitivity, common to all perceptual systems. Awareness of one's own movements can be obtained through vision or audition, as well as through the muscles and joints. Just as a bat may fly using echolocation to guide it, feedback from the visual environment may help the infant gain control of posture and hence be informative about the self. Proprioceptive information is given by the fact that we are embodied and in the world. Proprioceptive information is available even to babies because perceptual systems are simultaneously proprioceptive and exteroceptive in the coperception of self and environment. But, of course, this argument depends upon demonstrating that perception in very young babies is adequate to put them into a naturally dualist contact with reality. Contemporary research strongly puts into question the assumption of an initial adualism in early infant experience, in favor of Reid's natural dualism. Infants demonstrate at least a rudimentary differentiation between self and the physical environment as, for example, in neonatal reaching toward a visual target (Hofsten, 1989). They differentiate aspects of self, as in tactile self-exploration (Kravitz et al., 1978) or in the coordination between hand and mouth (Butterworth

SELF AS AN OBJECT OF CONSCIOUSNESS

39

& Hopkins, 1988) or in the visual control of the arms or legs (Van der Meer et al., 1995; Rochat & Bullinger, 1994). They also differentiate between self and the social environment, as evidenced by neonatal imitation (Meltzoff & Moore, 1977). They show previously unsuspected intersensory coordinations, e.g., between seeing and hearing (Castillo & Butterworth, 1981) and between vision and touch (Meltzoff & Borton, 1979), to cite just a few examples that show that even newborns can relate information obtained through different perceptual systems to the same external objects. Such observations put into question the assumption that early perceptual experience is uncoordinated with respect to the self or with respect to external referents. Perhaps the most compelling evidence against the adualistic confusion hypothesis comes from studies of visual proprioception that have shown that babies too young to walk or crawl nevertheless monitor their own postural stability with respect to the stability of the visual surroundings ( Lee & Aronson, 1974; Butterworth & Cicchetti, 1978; Jouen, 1990). Babies s ~ d or are seated in a "moving room" that has been suspended above the floor. The room swings in a parallel motion around the baby, so that the flow of optical information visually specifies a (nonexistent) loss of balance, even though the baby is actually in a stable posture. Babies compensate appropriately and sway or fall in the direction of instability specified by the visual flow field. The moving room studies demonstrate that babies invariably perceive the visual proprioceptive information as specifying a change in the spatial location of their own body. Much more evidence concerning the informative function of the visual flow field has been reviewed in earlier essays (Butterworth, 1992a, 1992b, 1995), and the fact that similar information is also available in auditory flow fields has been noted (Lee, 1993). Little purpose would be served by reiterating this evidence here. Instead, we may wish to consider in more detail how visual flow fields may serve to specify the self. What are the implications for the traditional concept of adualism? The infant's involuntary compensation to the misleading optic flow field can be thought of as an attempt to reverse a perceived change ofplace of her own body in order to maintain postural stability. By inverting the normal conditions of the ecology and thereby inducing a change of state of the surround, moving room studies reveal that the baby normally does make the distinction between a change of place of her body and a change of state of the environment. There is no question of adualistic confusion. Neisser (1988) argues that the ecological self does imply a form of unreflective consciousness, but, as he points out, this is not what is ordinarily termed "self-consciousness" because the ecological self is not all object of reflective thought. Nevertheless, self-perception does amount to an irreducible,

40

GEORGE B U T F E R W O R T H

basic form of self-awareness, which can be classified as a form of primary consciousness. Visual proprioceptive specification of self, long before there is independent locomotion, is consistent only with a theory that postulates "natural dualism" and "primary consciousness" as the original basis of experience. Later in development, at about 14 months, the ability partially to overrule discrepant optic flow patterns can be observed as the baby turns to see "what made the room move?" (Butterworth & Cicchetti, 1978). This later behavior does seem to reflect the kind of explicit self-consciousness where the event of room movement is understood to have occurred outside the infant's own agency. The difference between self-awareness and self-consciousness is a matter of degree. Adults in an unfamiliar posture can easily be unbalanced in the face of discrepant visual feedback (Lee & Lishman, 1975). This implies that in addition to specification of self through visual proprioception, there arises a form of selfknowledge that can, at least in part, overrule what is specified visually under nonecological conditions. There is one further issue that has been little explored in discussions of the origins of self, and this concerns the phenomenological basis for the information in the optic flow field. Although it is true that even adults are not normally conscious of optic flow per se, this may be because the flow carries information specific to the relationship between the body and the world. Under some conditions of optic flow, the body becomes the object of conscious awareness in the proprioceptive arena. This argument may be illustrated by the contrast between two types of flow field. In "looming," there is an explosive expansion of the optic array, with rapid occlusion of the part of the field corresponding to the background. The moment when the visual field becomes "all figure and no ground" coincides with a collision between the looming object and the observer. This information innately triggers head withdrawal and avoidance behavior (Ball & Tronick, 1971), presumably because optical looming carries life-threatening information. The opposite, "zooming," a flow pattern with explosive minification of the optic array, specifies the annihilation of the shrinking object at the moment when the visual field becomes "all ground and no figure." Zooming elicits close attention but no defensive behavior. Thus, looming and zooming may reveal that there is ultimately an existential basis for the information inherent in the dynamics of the optic array. In the case of loss of balance in the moving room, the phenomenological basis concerns that part of the optic array ordinarily specifying the stable background. An optic flow pattern derived from the background normally occurs only when the observer is in motion. Thus, the flow pattern in the moving room must specify that the observer has become unstable, and postural compensation

SELF AS AN OBJECT OF CONSCIOUSNESS

41

ensues. The phenomenological basis for phenomena such as these was first discussed by Michotte (1953).

Perception, Communication, and the Interpersonal Self Some aspects of direct perception, such as detecting the inforrnation that specifies the elasticity or rigidity of objects, may have primary application in social perception (Walker et al., 1980). Fogel (1993) lists 14 different ways that the dynamics of social interaction may be based on a rather small number of variant and invariant properties of the perceptual information that people display as a particular subclass of physical objects. These, coupled with the capacity for perception of emotional expressions, may constitute a large part of the necessary repertoire of abilities for social interaction in babies (see Table 1). Perhaps the best evidence of the infant's preadaptation for social experience comes from studies of imitation in early infancy. Precocious imitation was long ago called "participation" by Baldwin (1913). His terminology emphasizes imitation as a mechanism for realizing the socially constituted aspects of self. The importance of imitation for the origins of self is that consciousness of mutual, human relations provides the most direct feedback about one's own personhood. MacMurray (1933) put it as follows: "Complete objectivity depends on our being objectively related, in action as well as in reflection, to that in the world which is capable of calling into play all the capacities of consciousness at once. It is only the personal aspect of the world that can do this" (p. 134). Baldwin (1913) agreed that imitation plays a central role in the development of self-knowledge: " My sense of myself grows by my imitation of you and my sense of yourself grows in terms of myself' (p. 185). Notwithstanding its controversial status in contemporary psychology, there is now extensive evidence for neonatal imitation. Imitation in human newborns has been shown for tongue protrusion, mouth opening, lip pursing, sequential finger movements, blinking, vocalization of vowel sounds, and emotional expressions (Maratos, 1982; Meltzoff & Moore, 1983; Field et al., 1982; Kugiumutzakis 1985; Reissland, 1988). Vinter (1986) showed that newborn infants imitate the dynamics (not the statics) of the acts they observe; they need to see the act in progress in order to imitate. By the end of the first year of life, however, it is sufficient for the infant to see the end state (e.g., tongue protruded) in order to imitate. By 1 year of age, imitation can also take on symbolic properties; it is no longer merely participation in the literal act, as in the neonate. Neonatal imitation is just the first level of a developing system of interpersonal relatedness that may contribute in important ways to acquiring self-knowledge.

42

GEORGE BU'ITERWORTH

T A B L E 1. Varieties of Perceptual Information Available to Support Social Interaction. lnvariant

Information

Action

Distance dispLacement, expansion and contraction, increasing and decreasing intensity.

1. Expansion from a point and magnification or increasing ofintensitw 2. Contraction toward a point and minification or diminishing of intensity, 3. Maintaining constancy in size of elements or. intensity

I. Approaching a partner. becoming louder, surging, crescendo, explosive 2. Avoiding, leaving or withdrawing, fading away, trailing off, becoming softer 3. Maintaining a constant distance, leveling off', framing

Lateral displacement and rotation, movement a.gamst a background

1. Deletion of background texture on one side of an object, and addition of texture on the other side 2. Shearing of texture

I. Partners body moves across perceptual field. hiding and revealing

aga.i.nst a

Elasticity and rigidity, shape and surfixce deformation

const.'JLnt

background

1. Deformation of shape

2. Deformation of surface

3. Rigidity. of form D~solution and emergence of form

I. Dissolution of perceptual texture 2. Emergence or" perceptual texture

Color and texture

I. Changes itl color

2. Changes ha texture Frequency and regularity

1. Changes in spatial density or temporal ti'equency 2. Changes in regularity. of time or space betv.'cetl eveIl r.$

At~cr Fogel 1993.

2.Turning toward or away, rubbing 1. Changes in body posture, stance, gait 2. Changes in facLal expression, dimpling, wounds, swelling, muscle contractions and flexions, changes in flow 3. lmmobiliq,/, stiffness, inse ns itivit3.', un pass iv i v,' I. Disappearance. silence pausing, leave taking, ending a topic 2. Appearance, action following pause, greettng, growth, eruption, aggregation, beginning a new topic 1. Blushing, tantling, becoming pale, reddening (anger, exertion) 2. Wrinkles, creasing of skin I. Number of textural elements, beats (claps, head nods, vocal sounds), pitch of voice, synchrony in time 2. Rhythms, regularity, vs. irregularity of time be~,.~:en beats or points, uniform vs. notluniform distribution of spati:d elemeuu

SELF AS AN OBJECT OF CONSCIOUSNESS

43

How is neonatal imitation possible when it involves parts of the body the infant cannot see? The Gibsonian argument would be that the mechanism depends on proprioceptive aspects of the visual perception of tongue protrusion. Perception carries information for self and for the external environment. Perception of tongue protrusion can be considered as if it were a phase of the action of protruding the tongue. In the newborn, the information for tongue protrusion can be directly perceived, and although memory for the observed action is, in some sense, involved, the response is not primarily reconstructed through memory. Rather, it is elicited through direct perception, just as postural compensations are elicited by the dynamics of the flow field in the moving room situation. In this case, however, the information for imitation is not proprioceptive, but "alteroceptive." Alteroceptive perception seeks "'awareness of and the potentiality for interaction with another psychological being" (Trevarthen, 1993, p. 127). Although imitation of remembered events also develops, as shown by the capacity for deferred imitation that increases with age (Meltzoff, 1988; Vinter, 1986), newborn imitation can be taken as evidence for direct, primary consciousness of an interpersonal self. There is the question of the function of imitation in such young human beings. Kugiumutzakis (1992) draws on evidence from neonatal preference for the sound and affective tone of the voice to suggest that imitation reflects an innate motive for communication. He argues, with Trevarthen (1993), that newborns show an "innate intersubjectivity" and that they distinguish between self and others from the outset. Kugiumutzakis argues that this is true hetero-imitation derived from a relationship between self and other. Imitation cannot be dismissed as simply the cyclic repetition of reflexive activity conceived as a mechanistic response released by the adult. Although the evidence suggests that infants do recognize the correspondence between aspects of their own bodily self and those of other people, this may be a necessary but insufficient condition for perception of an interpersonal self because the relationship lacks any emotional significance when stated so baldly. The infant's capacity for emotional expression may also play an important part in early communication. Neisser (1988) defines the interpersonal self as the self engaged in immediate, unreflective social interaction with another person. He argues that the essential information for the interpersonal self comes into existence only when people are engaged in interpersonal behaviors that are synchronized. There arises a mutuality of experience (or intersubjectivity) that is confirmed by the reciprocal effects of gestures, emotions, and expressions from the partner. The interpersonal aspect of self is brought into existence through the information created by these forms of early communication.

44

GEORGE BUTI'ERWORTH

Evidence for such finely attuned behavior, and for the sharing of affect, between 2-month-old infants and their mothers has been obtained by Murray and Trevarthen (1985). They have studied the mutual correlation of activity in motherinfant pairs: the fine synchrony of movements, vocalizations, and expressions of pleasure as the behaviors of partners who form an indissociable whole. Stern (1985, 1993) has described a similar process that he calls "affect attunement," whereby the mother matches the infant's feelings with her own. If the baby is expressing joy, the mother does so, too, perhaps by a different form of nonverbal expression but in a manner precisely synchronized with the infant's emotional expression. Stern argues that through experiences such as these, the baby comes to be aware of the variants and invariants of the emotional relationship with the partner and of the organization and manifestation of each species' typical form of emotional expression. The emotions form an invariant constellation of feeling qualities that are experienced as belonging to the self, while interaction with others is the eliciting condition for such self-specifying experience. Trevarthen distinguishes innate primary intersubjectivity (infant's conscious awareness of mother, especially in relation to emotion) from secondary intersubjectivity (infant's conscious joint awareness with mother of the world of objects). Primary intersubjectivity consists of the exchange of feelings, a common code of cooing noises, facial and hand movements, concentration, pleasure, and surprise that manifest even in very early social interactions between the 2-monthold infant and the mother. "The universal emotions are the natural bridge between minds at any age" (Trevarthen, 1991, p. 3). Primary intersubjectivity, Trevarthen argues, can be thought of as a directly perceived, conversational consciousness where communication occurs through the dynamic, transient shifts of emotion, as revealed in emotional expression of infant and adult alike. The innate perceptual abilities of the baby make possible the comprehension of such complex exchanges. Toward the end of the first year of life, the infant achieves secondary intersubjectivity, which is based on jointly constructed meaning, the negotiation of conventional knowledge and common purposes, and communication through symbols. ( Sperry & Trevarthen, 1990; Trevarthen, 1991, 1992).

Joint Visual Attention The final part of this paper will briefly introduce phenomena of joint visual attention between infants and adults. Stated simply, the research concerns how an infant knows where someone else is looking, how a baby knows where someone else is pointing, and how babies produce pointing for other people. In studying the

SELF AS AN OBJECT OF CONSCIOUSNESS

45

foundations of referential communication (how babies share objects with other people), the definition of self-consciousness that might apply is the transitive one of joint awareness of public objects. That is, the object in the world offers an opportunity for minds to meet in joint attention to its properties. There is no doubt that babies as young as 6 months are able to change their own line of sight to follow a change in the attention of another person. Contrary to the traditional assumption that infants are totally egocentric (i.e., lost in an undifferentiated sell) and therefore unaware of other minds, babies will take a change in the focus of attention of their social partner as indicating a potentially interesting sight. In our carefully controlled studies, an adult turns slowly and deliberately to look at one of several targets positioned around the room. Babies can find the target the adult is looking at, and we have described the ecological, geometric, and representational mechanisms of joint attention, which arise during the first 18 months of life (Butterworth & Jarrett, 1991). One of the most striking phenomena we have discovered is that the ability to look where someone else is looking in the first year of life is circumscribed by the boundaries of the infant's own visual field. With the potential targets being located behind the infant, babies do not search behind themselves when the mother changes her line of gaze (Butterworth & Cochran, 1980). Instead, on the adult's signal, the baby turns through about 40 degrees within his or her own visual field and, failing to encounter a target, gives up. Only at 18 months does the baby succeed in searching in the invisible space behind him- or herself when the adult looks there. These data suggest that the infant takes her own visual field to be held in common with others. Of course, if objects are noisy, then the baby can turn behind to locate them but this merely illustrates the different properties of the auditory and visual systems with respect to the ecological processes of spatial orienting. Visual perception necessarily originates at a particular viewpoint, but the infant behaves as if others have a perspective on a common visual space. The boundaries of joint attention are defined by the periphery of the visual field. Even after 12 months, when the baby begins to comprehend manual pointing, this still does not extend the boundaries of joint attention. Grover (1988; Butterworth & Grover, 1988; 1989) showed that babies fail to search beyond the boundaries of the visual field even when the mother looks and points behind the baby. Infants produce pointing at about 14 months, and we know that this is an important, species-specific gesture that bridges nonverbal and verbal communication. Pointing, with the typical extended index finger posture of the hand, is species-specific to humans and may reflect the adaptation to communication of the specialized morphology of the hand (Butterworth, 1991). In our most recent studies, we have examined the production of pointing in babies.

46

GEORGE BU'Iq'ERWORTH

We use automated toys: either a toy truck that can move from place to place or a remote-controlled doll figure that moves its arms and legs. Babies find these objects very interesting; they will point at them and they will often check visually that the adult has taken notice. Checking reveals a concern for the effectiveness of communication. We were particularly interested to find recently that when placed in pairs, 14-month-old babies also point for each other and check with each other that their infant partner is attending to the object that has been pointed out. (Franco, Perrucchini, & Butterworth, 1992). William James (1946) argued that joint visual attention depends on expressive movements that lead unrelated minds to terminate in the same perception. Objects, he said, are coterminous, mutual aspects of experience. Indeed, he argued that other minds are known only by virtue of the body's expressive movements and their effects on one's own perception. A change in another person's visual orientation, or manual pointing, signals to the infant the possibility of an object, just as the changes in emotional expression we have been discussing may refer transparently to the feeling states that accompany them. James argues that minds have space in common as a receptacle for experience and that both perception and emotion have their perceptible objects. By the second year, the baby is entering the linguistic community and is beginning also to show evidence of reflective self-awareness, or higher-order consciousness. Recognizing the self in a mirror is revealed when the infant removes a surreptitiously placed dab of odorless rouge from the nose, using the mirror reflection as a guide. Rouge removal has been taken as a particular index of self-consciousness. Human infants solve the task at about 15 months, which suggests that the infant by this age has a visual image of her own appearance. Table 2 summarizes the generally agreed sequence of stages in self-recognition in mirrors (plus more recent studies using video feedback). We may note, however, that contemporary research shows that some aspects of this sequence may need revision. For example, the assumption that early responsiveness to the mirror merely depends on detecting a contingency between one's own movement and that of the reflection may need to be modified in the light of studies such as that by Bahrick and Watson (1985), who showed, using delayed video feedback, that apparent preference for contingency may actually depend on detecting the invariant relationship between proprioceptive information for self-movement and visual feedback from the motion. Evidence for an extremely early beginning for such intermodal specification of self has recently been obtained by Van der Meer & Lee (1995), who showed that newborns will keep a weighted limb within the visual field so long as they receive contingent video-recorded visual feedback. When the feedback is noncontingent (because the infant sees the nonweighted limb over the

SELF AS AN OBJECT OF CONSCIOUSNESS

47

video), the weighted limb drops out of view. Thus, contingency may be only one aspect of the complex of proprioceptive and visual feedback serving to specify the self in perception. TABLE 2. Summary of Main Stages in Mirror and Video Self-recognition Tasks During Infancy. Developmental Stage

Age

Characteristics

Unlearned attraction to images of others

3-8 months

Interest in mirror, touches, smiles, behaves "socially" to reflection.

Self as a permanent object

8-12 months

Aware of stable categorical features of self, locates objects attached to body using mirror image, differentiates contingent from non-contingent videorecordings of self*

Self-other differentiation

12-15 months

Uses mirror to locate others in space. Differentiates own video-image from that of others*.

Facial feature detection

15-24 months

Recognition based on specific features. Success on "rouge removal" tasks.

* These are actually very conservative measures of infants' social awareness. Using contingent and non-contigent video-feedback Trevarthen 1992 has found that infants as young as 3 months show signs of distress and disengagement when shown non-contingent video-feedback of their own interaction with their mothers.

The mirror task may also serve to illustrate the distinction between primary and higher-order consciousness of self. Gallup (1970) showed that chimpanzees and orangutans are the only primates capable of solving the rouge-removal task. Recognizing one's self in a mirror is actually a complex intellectual problem, and interest in the mirror image may be based on different mechanisms at different ages (Butterworth, 1992a). Rouge removal may require self-identification by means of memorized distinctive features, plus comprehension of the identity of the reflected image with self and attribution of the reflected image to the self. It is not clear whether self-recognition through distinctive features depends on a visual memory of one's own facial features or whether it may be based on accumulated social experience that informs the individual that in general, human faces do not have red marks on them and therefore there is something wrong with the face when the

48

GEORGE BUTTERWORTH

mark is seen in the mirror. This latter, more general social mechanism requires the individual to recognize the self in the mirror but to refer the reflected red mark to a general model of what faces should look like. Whichever interpretation proves correct, on the evidence of untrained performance on the mirror task, such abilities are restricted to higher primates (however, see Itakura, 1991 for the effects of training on the mirror task with lesser primates). It seems possible that mirror self-recognition marks both an ontogenetic and phylogenetic boundary between primary and higher-order consciousness. Gallup (1992) has suggested that performance on the rouge-removal task predicts performance of monkeys and apes on introspectively based social strategies and on tasks that require the imputation of mental states to others. Such a "theory of mind" interpretation would also link rouge removal to higher-order consciousness. Edelman (1988) reminds us that through language, we may simultaneously experience in interaction primary and higher-order consciousness. The fact that we each have unique characteristics, both in the forms of our bodies and in our autobiographies, will lead to the development of self-awareness and human selfconsciousness throughout life.

Conclusion

The evidence from infancy suggests three major conclusions. First, there is evidence for primary consciousness of self in human infants. Second, human social communication may be of central importance to the development of specifically human consciousness, both in its primary and higher-order forms, and third, the most recent research on perception in babies suggests that the perceptual abilities of the infant play an important part in primary consciousness of self. The ecological approach to perception has revealed many unsuspected aspects of the original abilities of babies. The weight of the evidence is such that aspects of early development, previously thought to be independent of perception, now have to be reconsidered. There is emerging a view of the origins of self both as a process within perceiving and as a product of perception, and this offers a new foundation for theories of the development of higher-order self-consciousness. It seems likely, however, that the ecological, embodied self will remain as the situated core of the self in the world. The ecological self acts as a personal frame of reference for the cognitive, linguistic, and culturally defined aspects of self that constitute the mechanisms for and content of higher-order self-consciousness.

SELF AS AN OBJECT OF CONSCIOUSNESS

49

REFERENCES

Bahrick, L., & Watson, J. S. (1985). Detection of intermodal proprioceptive-visual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21 (6), 963-973. Baldwin, J. M. (1901). Dictionary of philosophy and psychology." Vol. 1. New York and London: Macmillan. Baldwin, J. M. (1913). Social and ethical interpretations in mental development." A study in social psychology. (5th Ed.) New York: Macmillan. Ball, W., & Tronick, E. (1971). Infant responses to impending collision: Optical and real. Science, 171, 818-820. Butterworth, G. E. (1992a). Origins of self-perception in infancy. Psychological Inquiry, 3(2), 103-111. See also 10 peer commentaries pp. 112-133. Butterworth, G. E. (1992b). Self-perception as a foundation for self-knowledge. Psychological Inquiry, 3(2), 134-136. Butterworth, G. E. (1995). An ecological perspective on the origins of self. In J. Bermudez, N. Eilan, & A. Marcel (Eds.), The body and the self Cambridge, MA: MIT Press. Butterworth, G. E., & Cicchetti, D. (1978). Visual calibration of posture in normal and motor retarded Down's syndrome infants. Perception, 7, 513-525. Butterworth, G. E., & Cochran, E. (1980). Towards a mechanism of joint visual attention in human infancy. International Journal of Behavioral Development, 3, 253-272. Butterworth, G. E., & Hicks, L. (1977). Visual proprioception and postural stability in infancy" A developmental study. Perception, 6, 255-262. Butterworth, G. E., & Grover, L. (1988). The origins of referential communication in human infancy. In L. Weiskrantz (Ed.), Thought without language (pp. 5-25), New York: Oxford University Press. Butterworth, G. E., & Hopkins, B. (1988). Hand-mouth coordination in the newborn human infant. British Journal of Developmental Psychology, 6, 255-262. Butterworth, G. E., & Jarrett, N. L. M. (1991). What minds have in common is space: Spatial mechanisms serving joint visual attention in infancy. British Journal of Developmental Psychology, 9, 55-72. Castillo, M., & Butterworth, G. E. (1981). Neonatal localization of a sound in visual space. Perception, 10, 331-338. Edelman, G. M. (1989). The remembered present. New York: Basic Books. Field, T. M., Woodson, R., Greenberg, R., & Cohen, D. (1982). Discrimination and imitation of facial expressions in neonates. Science, 218, 179-181. Fogel, A. (1993). Developing through relationships: Origins of communication, self, and culture. Hemel Hempstead: Harvester Press. Franco, F., Perrucchini, P., & Butterworth, G. E. (1992, September). Referential communication between babies. Paper presented at the Vth European Conference on Developmental Psychology, Seville, Spain. Gallup, G. G. Jr. (1970). Chimpanzees' Self recognition. Science, 167, 86-87. Gallup, G. G. Jr. (1992). Levels, limits, and precursors to self recognition: Does ontogeny recapitulate phylogeny? Psychological Inquiry, 3(2), 117-118. Gibson, J. J. (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin. Gibson, J. J. (1987a) The uses of proprioception and the detection of propriospecific information. In E. Reed & R. Jones (Eds.), Reasons for Realism" Selected essays of James J. Gibson (pp. 164-170). Hillsdale, NJ: Erlbaum.

50

GEORGEBUTFERWORTH

Gibson, J. J. (1987b). A note on what exists at the ecological level of reality. Unpublished paper, 1978. In E. Reed & R. Jones (Eds.), Reasons for Realism: Selected essays of J. J. Gibson (pp. 416-418). Hillsdale, NJ: Erlbaum. Grover, L. (1988). Comprehension of the manual pointing gesture in human infants. Unpublished doctoral dissertation, University of Southampton, England. Hofsten, C. von (1989). Transition mechanisms in sensorimotor development. In A. de Ribaupierre (Ed.), Transition mechanisms in child development. (pp. 233-258). Cambridge: Cambridge University Press. Humphrey, N. (1992). A history of the mind. London: Chatto and Windus. James, W. (1947). Essays in radical empiricism: A pluralistic universe. London: Longmans. (original work published 1912) Jouen, F. (1990). Early visual-vestibular interactions and postural development. In H. Bloch & B.I. Bertenthal (Eds.), Sensory-motor organization and development in infancy and early childhood (pp. 199-216). Dordrecht: Kluwer. Kaitz, M., Meschulach-Sarfaty, O., & Auerbach, J. (1988). A re-examination of newborns' ability to imitate facial expressions. Developmental Psychology, 24(1), 3-7. Kugiumutzakis, G. (1985). The origin, development, and function of early infant imitation. Unpublished doctoral dissertation, University of Uppsala, Sweden. Kugiumutzakis, G. (1992). Intersubjective vocal imitation in early mother-infant interaction. In J. N adel & L. Camioni (Eds.), New perspectives in early communicative development. (pp. 23-47) London: Routledge. Kravitz, H., Goldenberg, D., & Neyhus, A. (1978). Tactual exploration by normal infants. Developmental Medicine and Child Neurology, 20, 720-726. Lee, D. (1993). Body environment coupling. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self knowledge. (pp. 43-67) Cambridge, MA: Cambridge University Press. Lee, D., & Aronson, E. (1974). Visual proprioceptive control of standing in human infants. Perception and Psychophysics, 15, 529-532. Lee, D., & Lishman, J. R. (1975). Visual proprioceptive control of stance. Journal of Human Movement Studies, 1, 87-95. MacMurray, S. (1933). Interpreting the universe. London: Faber. Maratos, O. (1973). The origin and development of imitation during the first 6 months of life. Unpublished doctoral dissertation, University of Geneva. Meltzoff, A. N. (1988). Infant imitation and memory: Nine month olds in immediate and deferred tests. Child Development, 59, 217-225. Meltzoff, A., & Borton, R. W. (1979), Intermodal matching by human neonates. Nature, 282, 403-404. Meltzoff, A. N., & Moore, M. K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 75-78. Michotte, A. (1953). The emotional involvement of the spectator in the action represented in a film: Toward a theory. Revue Internationale de Filmologie, 4, 8796. Translated by A. Costall in: Thines, G., Costall, A., & Butterworth, G. E. (1991). Michotte' s Experimental Phenomenology of Perception. Hillsdale, NJ: Lawrence Erlbaum. Murray, L., & Trevarthen, C. (1985). Emotional regulation of interactions between 2month-olds and their mothers. In T. M. Field & N. A. Fox (Eds.), Social perception in infants. Norwood, NJ: Ablex. Neisser, U. (1988). Five kinds of self knowledge. Philosophical psychology, 1(1), 3559. Neisser, U. (Ed.) (1993). The perceived self" Ecological and interpersonal sources of self-knowledge. Cambridge, MA: Cambridge University Press.

SELF AS AN OBJECI' OF CONSCIOUSNESS

51

Piaget, J. (1954). The construction of reality in the child. New York: Basic Books. Piaget, J. (1962). Play Dreams and Imitation in the child. New York: Norton. Reid, T. (1990). An Inquiry into the human mind: On the principles of common sense. Bristol: Thames Antiquarian Books. (Original work published 1764) Reissland, N. (1988). Neonatal imitation in the first hour of life: Observations in rural Nepal. Developmental Psychology, 24, 464-469. Rochat, P., & Bullinger, A. (1994). Posture and functional action in infancy. In A. Vyt, H. Bloch, & M. Bornstein (Eds.), Francophone perspectives on structure and process in mental development. (pp. 15-34), New Jersey: Erlbaum. Sherrington, C.S. (1906). On the proprioceptive system, especially in its reflex aspect. Brain, 29, 467-482. Sperry, R., & Trevarthen, C. (1991, October). Paper presented at a meeting on Molecules and the mind: The mind-body problem in epistemology and in the history of science. Cortina-Ulisse European Award (27 Edition August 1991) with Giorgio-Cini Foundation and the Sigma-Tau Foundation, Venice. Stern, D. (1985). The interpersonal world of the infant. New York: Basic Books. Trevarthen, C. (1991, October). Consciousness in infancy: Its origins, motives, and causal potency. Paper presented at a meeting on Molecules and the Mind: The Mind-Body problem in Epistemology and in the history of science. Cortina-Ulisse European Award (27 Edition August 1991) with Giorgio-Cini Foundation and the Sigma-Tau Foundation,Venice. Trevarthen, C. (1992). The functions of emotions in early infant communication and development. In J. Nadel & L. Camioni (Eds.), New perspectives in early communicative development. (pp. 48-81) London: Routledge. Van der Meer, A., Van der Weel, F. R., & Lee, D. (1995). Lifting weights in neonates: Body building in progress. Science, 267, 693-695. Vynter, A. (1986). The role of movement in eliciting early imitation. Child Development, 57, 66-71. Walker, A., Owsley, C. J., Megaw-Nyce, J., Gibson, E. J., & Bahrick, L.E. (1980). Detection of elasticity as an invariant property of objects by young infants. Perception, 9, 713-718.

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

53

CHAPTER 4

Early Objectification of the Self PHILIPPE ROCHAT

Emory University

On what basis do we assume that animals, young or old, human or nonhuman, are endowed with a sense of self? Does a sense of self, as an entity situated in and differentiated from the environment, imply self-awareness? These questions run through the book, and this chapter is an attempt to address them within a functional theoretical framework. The primary assumption of this framework is that young infants are actors in a meaningful environment. From their precocious and rapidly developing propensity to explore and act in relation to functional goals stems a sense of self as an agentive, differentiated, and situated entity in the environment (the "ecological self," Neisser, 1991). This sense of self does not imply any selfconsciousness or self-awareness, 1 and does not emerge with the development of particular representational systems as it is sometimes suggested. Also, the ecological self is not specific to humans, but is an emergent property of any biological system that perceives and acts in relation to functional goals. However, at least in humans and possibly in other nonhuman primates (see chapters by Povinelli; and Spada, Aurelli, Verbeek, & de Waal, this volume), there is obviously more to self-knowledge than the sense of the ecological self. Along with the direct, early sense of being an agent in the world develops the indirect idea of "me," or the sense of self as both subject of action and object of reflection (i.e., self-awareness). Where does this aspect of self-knowledge come from, and how? In this chapter, I propose that this question needs to be considered in light of the fact that from birth, infants are not merely passive reactors to nonspecific stimulation. In the context of goal-oriented action systems, infants develop an ability to perceive the effectivities of their own body, long before they are capable of recognizing themselves in a mirror. Furthermore, starting at least by the end of the first month, and possibly before, infants engage in exploratory activities that support and are probably at the origin of the process of an objectification of the

54

PHILIPPE ROCHAT

self. In repeating actions and when apparently compelled to observe systematically their consequences on objects and people, young infants come to specify both what they are capable of doing (the ecological self), and who they are as sentient, intentional, and emotional entities. However, I will argue that the developmental origins of self-awareness are primarily social, emerging in the interaction of the infant with others. Others provide infants with a constructive mirror (the social mirror) that supports from birth an objectification of their own feelings and emotions. The reciprocity characterizing social interactions is viewed as instrumental in scaffolding self-awareness. But first, let us consider two precocious facts of life from which stems an early sense of self: goal orientation and exploration.

Goal Orientation and Exploration in the N e w b o r n Until fairly recently, and in the footsteps of William James, many authors insisted on the chaotic, blooming, buzzing confusion of neonates. Ren6 Spitz presented newborn behavior as "... random, unstructured, and ... inconsistent" (Spitz, 1965, p. 54). At first sight, many reasons exist to consider neonates as highly immature entities, whose behavior is on the whole erratic, manifesting vegetative (organismic) rather than psychological functioning. From this perspective and in relation to the environment, early behavior was interpreted as the expression of an initial state of fusion or undifferentiation between young infants and their surroundings (Wallon, 1942/1970; 1981; Piaget, 1952). Within the psychoanalytic tradition, the initial stage of behavioral development is sometimes described as "normal autism" (Mahler, Pine, & Bergman 1975). Accordingly, young infants behave in independence of, not in attunment to the environment: "...the reaction to any stimulus that surpasses the threshold of reception in the weeks of normal autism (first two months) is global, diffuse, syncretic - - reminiscent of fetal life. This means that there is only a minimal degree of differentiation, and that various organismic functions are interchangeable" (Mahler et al., 1975, p. 43). Before dismissing this view, it is important to note that its adoption is understandable in light of the rapid development occurring in the course of the first months, which brings about increasing organization, predictability, and apparent intentionality to infant behavior. If this development is apparent to any observer, this does not preclude the existence of an initial organization of behavioral systems that are highly predictable and remarkably attuned to particular features of the environment. In fact, the impetus of such development is to be found in the dynamics of this initial organization of behavioral or action systems, which

EARLY OBJECI'IFICATION OF THE SELF

55

defines the avenues of behavioral change (Rochat & Senders, 1991; Rochat, in press). In a somewhat contradictory statement, Mahler et al. note a few paragraphs below the above quotation that in the rare and fleeting state of alert inactivity (when they are not fussy, sleeping, or crying), young infants and even neonates demonstrate responsivity to external stimulation. The authors suggest that "...this fleeting responsivity to external stimuli makes for the continuity between the normal autistic phase and later phases." (Mahler et al., 1975, p. 43). It appears that the initial autism is not absolute and that there are moments in which neonates are receptive to their environment. But these fleeting moments are in fact what import to any understanding of newborn psychology. It is during these moments, which become increasingly protracted over time, that young infants reveal their actual state of mind in relation to the outside world: how they function in the environment, what they perceive of it, and how they act upon it. When crying or sleeping, infants as well as adults are indeed self-absorbed, in a momentary state of unrelatedness to the environment that can be interpreted as normal autism, or undifferentiation. Such is not the case in the state of alert inactivity when young organisms appear to tune into, and be receptive to, the resources of their environment. Relative to the fleeting state of alertness of young infants, recent research provides much evidence of precocious plasticity, goal orientation, organized action toward functional goals, and exploratory activities. This research supports the idea of an early differentiation and puts seriously into question the interpretation of an initial autism. Infants from birth appear to behave as actors in a meaningful environment. But what does that mean? It means that newborn behavior cannot be merely reduced to a collection of discrete reflexes that progressively become oriented and intentional. Aside from the inventory of responses normally observed by pediatricians in their routine neurobehavioral assessment of neonates, infants manifest from birth actions that are oriented toward particular environmental resources, such as people or food. They are not merely responsive to nonspecific stimulation, as in the case of intense light or an air puff (e.g., blinking response), abrupt vestibular stimulation (e.g., Moro response), or tactile strokes from any external object whether it be the finger of the examiner, a pencil, or a pacifier (e.g., rooting, Babkin, or Babinsky responses). Recent infancy research demonstrates that infant behavior shows selectivity and attunement to particular features of the environment. It indicates that early behavior is more meaningful than mere reflexes, which are by definition, and in the original sense of the physiological reflex arcs described by Sherrington (1947/1961), automatic and the expression of rigid stimulus-response links. For example, newborns orient toward sound sources but show more propensity to

56

PHILIPPE ROCHAT

orient toward the source of a rattle sounds than a pure tone (Clarkson & Clifton, 1991). When presented with cloth pads carrying either the body odor of a stranger or the body odor of their own mother, newborns tend to orient more toward the latter (MacFarlane, 1975). From 1 month, infants show enhanced visual attention to particular facial features, such as eyes and mouth, that are most relevant for the reading of emotional expressions (Maurer & Salapatek, 1976). These types of observations demonstrate that young infants are not merely responsive in the reflex (physiological) sense discussed above, but are functionally oriented toward meaningful features of the environment: sounds, odors, and visual configurations that are functionally relevant for the maintenance of their transactions with the environment, and ultimately their survival among people and objects. Although this selectivity is based upon the functioning of preadapted action systems such as sucking or rooting, such functioning entails selectivity, hence basic functional values that guide groping and exploration. The claim that newborn infants are not merely responsive is best supported by evidence of exploratory activities at the onset of development. Newborns appear to be not only selective with regard to environmental features, but also to engage in active probing of their environment. Two sets of empirical observations on early oral activity illustrate this point. In a recent article on imitation of various types of tongue protrusion by 6week-old infants, Meltzoff & Moore (1994) report striking data demonstrating that these infants engage in an active exploration and approximation of the model. This spontaneous approximation leads to more accurate matches of the model (e.g., tongue protrusion to one side of the mouth) over few successive trials. Aside from the active exploration involved in the imitation of tongue protrusion, there is an exploratory component to the oral behavior of young infants when a suckable object is introduced into their mouth. Neonates engage in more than sucking in response to intraoral stimulation. When various intraoral objects (pacifiers) varying in shape, texture, and consistency are introduced, neonates selectively manifest particular movements of the tongue, gums, and lips (Rochat, 1983, 1987). The amount of such oral-haptic activity depends on the degree of eccentricity of the pacifier compared to the normal physical characteristics of the biological nipple. The more physically eccentric the pacifier, the less infants engage in activities serving a nutritional function (sucking), and the more they engage in activities serving a perceptual function (oral-haptic exploration; Rochat, 1983, 1987). From birth, infants appear to engage in scanning activities of intraoral objects that are not readily assimilable to sucking and the functional goal it serves (i.e., the eventual ingestion of food). The exploratory component of infant behavior as in the case of early imitation and newborns' oral activity suggests that early on, young infants are not merely

EARLY OBJECTIFICATION OF THE SELF

57

responsive, but show orientation toward functional goals. This orientation is first determined by action systems that are functional at birth (e.g., imitating or feeding) and are readily organized to tap into environmental resources that are essential for the neonate's survival (e.g., people and nutritional objects).

Early Sense of Self and the Perception of Body Effectivities by Young Infants If we accept that at no point in development infants can be reduced to automata, that their behavior is not merely a collection of S-R links, and if we consider that from the beginning behavior is guided by functional goals and survival values that are an integral part of the newbom's action systems, then the claim of an early sense of self is reasonable. If a biological system does not simply respond automatically to meaningless stimulation, and if, on the contrary, it shows exploration, plasticity (discovery of new solutions), and orientation towards functional goals, it implies that it knows something about itself: It perceives itself as an agent, differentiated, and situated in the environment. Accordingly, any biological system that expresses exploration, plasticity, and goal-oriented actions is endowed with a sense of self. This means that not only humans and human infants possess a sense of self, but all mammals, birds, and insects. When an animal adjusts its behavior to achieve a functional goal and is not merely responding to a stimulation, it needs to be capable of at least three things: to distinguish between itself (the agent) and the goal to be achieved; to situate itself in relation to the goal; and to perceive its own potential effectiveness to achieve this goal. For example, in order to plan a successful action, a foraging bird that detects a worm on the branch of a neighboring tree will be required at minimum: to differentiate itself from its prey; to situate itself in relation to the prey; and to perceive that it can fly in order to reach the prey. These are three basic ingredients of a sense of self that Neisser (1988, 1991) first coined as the ecological self. Like birds and other animals, young infants do manifest a sense of the ecological self: a sense of a differentiated, situated, and agentive entity. Because infants are actors and explorers in relation to functional goals, and not mere automata reacting in rigid ways to stimuli, they have a sense of the ecological self. An illustration of this argument can be made on the basis of some empirical observations on early reaching behavior. From an early age, infants show the propensity to bring their hands in contact with objects in the environment, first for short contact with no grasping, then to grasp and bring them to the mouth, and eventually to engage in fine manual exploration to transform them or use them as tools. Reaching is probably the

58

PHILIPPE ROCHAT

earliest clear expression of a goal-oriented action, the goal being at minimum to have manual contact with the object. It is a good index of the early sense of self. Hofsten (1984) demonstrated rudiments of reaching behavior within the first week following birth. Newborns were shown to wave their arms in the direction of an object-target translating in front of them. They showed significantly more arm movements when the object was present in their field of view. Coordination and anticipation in reaching develop rapidly in the course of the first 6 months, to include fine arm and hand adjustments in relation not only to the spatial location of the object, but also to its size and shape (Lockman, Ashmead, & Bushnell, 1984; Hofsten & Ronnqvist, 1988; Clifton, Rochat, Litovsky, & Perris, 1991). This anticipation indicates, with no ambiguity, that the infant perceives the affordance of the object in terms of its reachability. In reaching the way young infants reach, what is implied is that they situate themselves as actors in relation to an object, which is differentiated from themselves, is out there in the environment, has particular characteristics, and moves in a certain way relative to themselves. The study of early reaching also demonstrates that infants perceive their own effectivities as actors in the environment, this perception being an integral part of the early sense of serf. As adults, we constantly gauge what is safe or unsafe to do. If we stand on the ledge of a skyscraper, we perceive the danger of our situation and resist the temptation of leaning too far to have a better view of what is happening below. We understand our situation and what it affords. In less extreme situations, we perceive what we can do successfully with a minimum of energy expenditure. We adjust our posture in the anticipation of lifting a heavy object, and we stretch out in order to catch a fly ball. This tight link between perception and action is based on the perception of what objects afford for our actions, hence the perception of our own body effectivities. Knowing what we can do, without harming ourselves or wasting too much energy, is crucial for our survival and is a basic aspect of self-knowledge. Recently, a colleague and I collected data suggesting that the perception of body effectivities determines reaching behavior by young infants when they start successfully to grasp objects around them (by 4 months). These data demonstrate young infants' accuracy in perceiving the effectivities of their own body in reaching. Here is a sample of what we observed. In two different experiments (Rochat & Goubet, 1993), infants were presented with an object for reaching, which was presented successively at four different distances. Infants were placed in an uptight infant seat with the object centered at their shoulder height. The nearest distance placed the object about 30 cm from the infant's torso, in alignment with the toes. The other three distances expanded from this referential distance by 5 inches. At Distances 1 and 2, the object was within

EARLY OBJECTIFICATION OF THE SELF

59

reach of the infant. At Distance 3, it was at the limit of prehensile space; the infant could eventually touch it, but only with intense stretching forward of the trunk and upper limbs. At Distance 4, the object was out of the infant's reach. During 30-second presentations, frequency and duration of gaze at the object, latency to reach, and reach attempts were scored. In the first study, three groups of infants aged 5-6 months were compared based on their relative ability to control independent sitting (i.e., their ability to coordinate reaching of the hand(s) and leaning of the trunk). Results showed that at distances 3 and 4, gazing activities and the frequency of reach attempts increased, depending on the infant's relative sitting ability. With increased control over self-sitting, infants demonstrated an expansion of the perceived limits of prehensile space. In the second study, 5- to 7-month-olds were analyzed with either light (2g) or heavy (200g) bracelets attached to their wrists. Reaching with heavy bracelets moved forward the infant's center of mass when reaching, and reduced the limits of maximum reachability without losing balance. The rationale was that if infants were sensitive to this change, they should reach less with the heavy bracelets. Results indicated that at distances 3 and 4 only, the frequency of reach attempts decreased when infants wore the heavy bracelets compared to when they wore the light ones. These results suggest that infants as young as 4 months are sensitive to what their body affords for action, by detecting and adjusting with remarkable accuracy the perception of their own body's effectivities. They adjust the planning of their activity (reaching) by perceiving sudden experimental changes in their bodily characteristics (i.e., weighted limbs causing forward displacement of their body's center of mass). These results indicate that when starting to reach, 4-monthold infants plan their reach based on both the perception of their situation in the environment and particular postural constraints. They detect visual and proprioceptive information specifying their relation as actors to the object and to the region of maximum extension of the body without losing balance (i.e., the "region of postural reversibility"). Note that the region of postural reversibility determines the perceived limits of prehensile space in adults as well (Rochat & Wraga, 1994). Thus, infants, like adults, demonstrate an ability to perceive the limits of what they can do without losing balance. This detection is the direct expression of a sense of the body's effectivities. As for any other self-generated and goal-oriented actions, the planning of reaching behavior is based on the perception of body effectivities. In relation to a functional goal (e.g., bringing hand(s) in contact with an object), this perception integrates the actor's sense of its own capability and its particular situation in the environment. The argument proposed here is that this perceptual ability is a primary aspect of self-knowledge. It corresponds to the sense of the ecological self, which is an emergent property of any biological system that does not merely

60

PHILIPPE ROCHAT

respond to stimuli, but acts, explores, and invents new means to achieve functional goals. In humans, this ability is expressed at, and develops from, birth. However, at least in our species, the development of self-knowledge implies much more than the development of an early sense of self construed as the progressive discovery of body effectivities. Infants also develop self-awareness that is not implied by the early sense of self discussed so far. What are the origins of an ability to perceive oneself, not only as a subject of action (the ecological self), but also as an object of reflection and recognition (the conceptual self)? How does the idea and specification of "me" emerge in development, and what are the mechanisms underlying the progressive objectification of the self? These questions are fundamental from both a developmental and an evolutionary perspective. Indeed, the emergence of selfconsciousness potentially indexes a quantum leap in both ontogeny and primate evolution (e.g., see Gallup, 1982; Povinelli, 1987).

Object Exploration and the Early Objectification of the Self Parallel to the development of goal-oriented action systems from which emerges an early sense of self (e.g., the perception of body effectivities in relation to functional goals), infants also develop a propensity to contemplate and analyze the results of their own actions on objects. When infants explore and act in the environment, they learn as much about the object with which they are interacting as they learn about themselves. To explore objects is indeed to coexplore oneself, to paraphrase Gibson's (1979) original quote about the inseparable process of perceiving the environment and the self. I propose that from the perceptual analysis of their own actions and their consequences on physical objects, infants gather more than information specifying the ecological self and what objects afford for action. They also gather information about their own vitality and the dynamics of their own emotions. In general, infants do not only perceive and act in order to achieve utilitarian goals, such as the search for food or social comfort. This fact was already emphasized by Werner & Kaplan (1963), who distinguished two basic modes infants engage in that become rapidly differentiated in the course of the first months: the action on or with things versus the contemplation of things. From an early age, and aside from acting toward basic utilitarian goals, infants do engage in playful monologues that are linked to a contemplation of the self. These monologues appear to have the main function of acting out current feelings and emotions that are experienced internally by the infant (e.g., when happy infants from the second month vocalize, babble, coo, and get motorically excited with no

EARLY OBJECq~ICATION OF THE SELF

61

other apparent purpose than to enjoy these activities and experience themselves as actors). This process is tentatively viewed here as a primary objectification of the self by which feelings and emotions become objects of multimodal exploration. Aside from the playful monologues infants appear to engage in, by exploring and acting in the environment infants also develop a sense of what they are at an emotional level. The idea proposed here is that as young infants explore and act on objects and on their own body, they are simultaneously gathering information about their own internal states by acting them out. I borrow here from the concept of "vitality affect" introduced by Stern (1985) to capture a certain quality of feelings and emotions experienced by young infants that are traditionally not accounted for in the literature. Following Stem, vitality affects correspond to: "...these elusive qualities (that) are better captured by dynamic, kinetic terms, such as 'surging,' 'fading away,' 'fleeting,' 'explosive,' 'crescendo,' 'decrescendo,' 'bursting,' 'drawn out,' and so on. These qualities of experience are most certainly sensible to infants and of great daily, even momentary, importance. It is these feelings that will be elicited by changes in motivational states, appetites, and tensions (Stern, 1985; p. 54). These feelings are important aspects of the perceived self, and from birth on infants are actively involved in investigating them. Early on, there is an objectification of feelings and emotions in the process of perceiving and acting on physical objects, which include the body. While interacting with objects and exploring the perceptual consequences of their own actions on their own body, infants are actually externalizing their own feelings of vitality, which become public and accessible to multiple perceptual modalities, in addition to being felt internally or subjectively. This process can be construed as an early objectification of the self (self as object of reflection; hence, object of exploration) that develops in parallel to the ecological self (self as subject of action). Let me try to articulate this idea with an example: Consider an infant systematically kicking a mobile hanging low from above her crib. What kind of perceptual information is available to her? First, one can assume that the infant detects (as indexed by the systematicity of her kicking ) the temporal contingency between her own movement and the effect it has on the mobile. This information specifies a causal link from which certain expectations can be made: When the feet contact the mobile, it happens to move contingently. By exploring the results of her own action, the infant detects what Watson labels a perfect algorithm, or perfect contingency: "...a temporal pattern between two events that potentially reflects the causal dependency between them" (Watson, 1994, p. 134). The detection of this temporal invariant is based on multimodal information (visual,

62

PHILIPPE ROCHAT

proprioceptive, tactile, and auditory). It specifies both the infant as an agent, who is differentiated, and situated in the environment (the ecological self), as well as the affordances of the activated object (that it is kickable, moveable, and noisy). But is that all? When infants engage in exploratory activities within the context of such playful soliloquies, they are actually detecting much more than a temporal contingency, the affordances of an object, and their own efficacy as agent. By exploring the visual consequences of their own actions, they are also experiencing their own force and the dynamics of their own emotional tone. Their own vitality is reflected to them via the externalized movements they cause (e.g., the movements of the mobile, its noise, the felt and seen movements of their own legs, their impact on the object, etc.). In a sense, the way the object moves reflects to the infants not only what they are causing, but also how they are causing it (with more or less force, for example). This process represents an early objectification of the self in the sense that it is an exploration of the dynamics of internal properties of the self, which are accessed via the exploration of an external event. Literally, the infant acts out the experience of subjective, internal feelings. It is the physical object, animated by the infant, that externalizes qualitative features of the self: its vitality and the intrinsic dynamics of felt emotions. This information goes beyond the specification of the ecological self because it specifies intrinsic aspects of the self as agent. These aspects pertain to the dynamics of motivational and emotional forces that animate the infant as she perceives and acts in the physical environment. In terms of the process involved, the developing objectification of the self (the idea of "me") might originate in the perception of these dynamic aspects of the self that are externalized or "acted out," in addition to being felt internally. The idea is that by animating objects and exploring how they move, infants actually externalize and gain further perceptual (e.g., visual, auditory) access to the dynamics of their own feelings. It allows them to detect the invariants of their unique vitality. From an early age, infants appear to be actively involved in exploring their rich emotional experiences while acting on their own body or interacting with objects in the environment. Piaget (1952), and Baldwin (1906) before him, noted that young infants engage in playful repetitions of action schemes, first on their own body (primary circular reactions) and eventually on external objects (secondary and tertiary circular reactions). This propensity for repetition favors selfexploration and the discovery of one's own effectivity and vitality. In particular, the repetition of actions, such as the transport of the hand to the mouth, thumb sucking, or the kicking of a mobile, enable young infants to calibrate the effectivity of their own actions, as well as to specify their own force and vitality. Here is one among many other examples of what Piaget (1952) observed regarding the early propensity for self-examination, and what I consider to be active

EARLY OBJECTIFICATION OF THE SELF

63

emotional monologues in the context of circular reactions. The first is in relation to primary circular reactions which, once again, are body oriented: Observation 12: At 1 month and 3 days, Laurent puts out his tongue several times in succession. He is wide awake, motionless, hardly moves his arms, and makes no sucking-like movements; his mouth is partly open and he keeps passing his tongue over the lower lip. At 1 month and 5 day.s, Laurent begins sucking-like movements and then the sucking is gradually replaced by the preceding behavior. At 1 month and 6 days, he plays with his tongue, sometimes by licking his lower lip, sometimes by sliding his tongue between his lips and gums. The following days this behavior is repeated and always with the same expression of satisfaction (Piaget, 1952, p. 50). This observation demonstrates a 1-month-old infant's propensity to repeat a newly discovered action over protracted periods of time. This action is generated for the apparent pleasure of its repetition. What is interesting is that this action of the tongue is not stereotypical, but rather is modulated and associated with an unmistaken emotional expression of satisfaction. The following observation pertains to self-examination and a playful monologue in the context of secondary circular reactions (repeated actions that are object oriented): Observation 94: At 3 months, 5 days, Lucienne shakes her bassinet by moving her legs violently (bending and unbending them, etc.), which makes the cloth dolls swing from the hood. Lucienne looks at them, smiling, and recommences at once. These m o v e m e n t s are simply the concomitants of joy. When she experiences great pleasure Lucienne externalizes it in a total reaction including leg movements. As she often smiles at her knick-knacks she caused them to swing (Piaget, 1952, p. 158).

Again, this observation demonstrates the infant's propensity to become selfabsorbed, this time by observing traces of her vitality in the visual consequences of her own action on the object. This exploration is accompanied by positive affect (smiling), which suggests that concomitant to the exploration of her own agency, the infant is also immersed in the exploration of her own pleasure and emotional experience, an obviously important aspect of the self. Lewis, Sullivan, & BrooksGunn (1985) (cited by Lewis, 1990)also observe the systematic expression of joy by infants as young as 2 months who are learning the contingency between a pulling movement of the arm and a visual stimulation. In this instrumental learning situation, an arm pull by the infant triggered the short appearance on a screen of an image showing an infant's smiling face accompanied by the sound of a singing child. Interestingly, the authors report that during an extinction phase in which the contingency was surreptitiously suppressed, infants switched their emotional expression from joy to anger. They expressed joy again when the contingency was reestablished. These observations once again point to the fact that

64

PHILIPPE ROCHAT

the exploration of the self as agent is inseparable from the exploration of concomitant vitality and emotions. The specification of experienced feelings and emotions is an important aspect of the early development of self-knowledge. As suggested above, it is inseparable from the development of the ecological self and from the discovery by young infants of their own efficacy in the physical world. Again, in perceiving the results of their own actions, infants do not only gain information about whether or not they are successful in achieving a goal. They also gain information about their idiosyncratic way of doing things and about the emotional aspects of themselves as actors. They learn about their unique vitality, strength, and impact on the environment. What I have proposed so far is that by acting on objects, infants cast their own specific way of doing things, and it is based on the detection of this information that infants probably start specifying themselves as objects of reflection. Accordingly, the perceptual analysis of self-generated actions and their consequences on physical objects might be an important factor in the development of the self as both subject of action and object of reflection. However, this contribution is probably minimal in comparison to the objectification of the self that stems from social interactions and the constructive reciprocity provided by others.

Early Objectification of Self in Others In perceiving and acting in the social environment, infants develop a special set of expertise that they do not acquire in interacting with inanimate, nonintentional objects. I will argue that before recognizing themselves in mirrors, pictures, or films, infants start recognizing themselves in others, via imitation and the reciprocity of social interactions. In reproducing emotional expressions, perceiving emotions in others, monitoring their own, and probing how they impact on others, infants discover aspects of themselves that they could not discover otherwise. It is essentially from the experience of the social mirror that they develop self-awareness, and in particular the awareness of themselves as objects of reflection. Based on the work of Eckman, Levenson, & Friesen (1983), which demonstrated that particular facial movements influence one's emotion-specific physiological changes, Meltzoff (1993) suggests that in imitating the facial expression of others, young infants might directly experience the emotional state of the person they are imitating. In other words, imitation may be the process that underlies the infant's emotional empathy with others. In relation to self-

EARLY OBJECTIFICATION OF THE SELF

65

knowledge, the intriguing idea proposed by Meltzoff means that infants from birth might engage in the process of an objectification of their own emotions via the imitation of body movements and the facial expressions of others. Note that infants from birth are invariably positioned by caretakers and engaged in particular games that provide them with opportunities to imitate and act in reciprocity. Some 50 years ago, Spitz (1965) provided striking observations of infants in a crowded orphanage displaying the syndrome of infantile hospitalism. When deprived of frequent exposure to the constructive "mirror" afforded by others, infants tended to withdraw from the world, apparently losing their basic sense of self as active participants in both the physical and human environment. Although we have seen in the preceding section that opportunities for an early objectification of the self occur via the interaction between infants and physical objects (i.e., the monologue of circular reactions), this process is obviously not sufficient to account for the development of self-consciousness. Self-consciousness is essentially coconstructed and needs to be conceived of as a byproduct of social interactions (i.e., the reciprocity characterizing the dialogue with others). In his compelling book, Kaye (1982) suggests that self-consciousness corresponds to the child's realization of other persons' intentional agency. In this process, the child views herself as another person equally endowed with intentionality. Kaye describes multiple frames by which parents and caretakers might scaffold infants' knowledge in general, and self-knowledge in particular, in terms of an awareness of being an intentional entity among others. For example, Kaye describes how adults carry out what they detect as the infant's intention (instrumental flame). Parents manifest exaggerated consequences to the infant's action (feedback frame), or provide the infant with actions to be imitated within appropriately timed turn~ taking patterns (modeling frame). These frames are expressed to the infant from birth and determine the developing ability to conceive of themselves as intentional agents among others. Self-awareness depends on social interactions, and in particular on the systematic scaffolding of empathic, exquisitely attentive, and more expert others. The parental frames described by Kaye (1982) are pervasive in the sense that any empathic adult expresses the ability to foster and scaffold the infant as they interact together. In fact, it is remarkable how systematic and robust adults are at modulating their behavior as a function of the infant's age, and how parents in general tune in their infant's progress by constantly adjusting the frames of their interaction. This ability is indeed instrumental in fostering development and is probably at the origins of competencies, including self-awareness, that might be exclusive to the human species and perhaps few of its close relatives (see Tomasello, Kruger, & Rather, 1993, for a discussion of this interesting issue).

66

PHILIPPE ROCHAT

Infants from birth show a great inclination to explore and interact with others, developing precocious signs of social skills. This inclination is at the origins of what Neisser (1991) calls the "interpersonal self," and what Trevarthen (1974) interprets as the "primary intersubjectivity" expressed by young infants in the course of the first months. Research demonstrates the social attunement of young infants in the context of face-to-face interactions (Stern, 1985; Tronick, 1980; Murray & Trevarthen, 1985). Features that pertain to others are shown to have a strong attraction value for neonates, who orient preferentially toward the social resources of their environment. For example, they demonstrate increased visual attention toward facial features, in comparison to other, nonfacial configurations (Fantz, 1963; Johnson & Morton, 1991). This early inclination contributes to the development of self-knowledge. In general, by perceiving and acting in a social context, infants gain a deep, private knowledge about themselves. The human environment for young infants, as for older individuals for that matter, is a "deep mirror "2 that scaffolds further objectification of the self as a sentient, emotional, and intentional entity. Others, construed as social objects, afford much more than physical objects: In addition to specifying our situation in the environment as other objects do (where we are and what we can do, which correspond to the early sense of self), they afford a deep reflection about our worth and what we feel. In interacting with others, infants engage in both a "doing in the world" mode from which the early sense of self emerges, but also and simultaneously, in a "reflecting" mode from which selfawareness emerges. According to this view, others are privileged objects in the environment that afford the idea of "me"; hence, self-consciousness. It is true that when infants encounter resistance with physical objects in the environment that, for example, a particular object is not easily suckable and requires much effort to provide nutrients (think of a clogged rubber nipple) it informs them both about the quality of this object and about the relative efficacy of their own actions. Physical objects do sometimes provide "deep reflection" about what we are (e.g., weak, strong, cowardly, or brave), but only interactions with others invariably reflect to us what we are and how we feel. As an illustration of the deep reflection provided by others to young infants, and of a typical parental frame at work from the earliest age, here is a short, reconstructed excerpt of a mother interacting with her 2-month-old infant via a close-circuit video system as partof an ongoing research conducted in collaboration with Ulric Neisser and Viorica Marian at the Emory Infant Laboratory: The mother with a high pitch, motherese voice: "...Hi baby...hi my baby...are you happy to see me?...Are you going to give me a smile? .... Are you? .... Are you going to give a smile to mommy? .... (the baby starts to look away from the mother)...

EARLY OBJECTIFICATION OF THE SELF

67

The mother: "...look here baby...what's happening my baby? .... Are you sad?... Are you unhappy?... Aren't you going to give a smile to mom? .... C'mon, give a smile to mom..." etc. This dialogue illustrates the typical affective scaffolding provided by the mother, who encourages a certain type of hyper-positive, highly charged exchange where positive affects are displayed, exaggerated, and requested with insistence. The mother reflects to the infant a certain affective tone, and by the modulation and timing of her voice (not the content of her words, of course), she is commenting literally on-line about the feelings of her baby in an obvious attempt to control and make them more positive. The mother in this example is an affective broadcaster of her infant's affective state changes not unlike the running description of a s p o r t s c a s t e r - except that she is commenting in order to impact positively on her infant. She is a commentator as a way of being an active participant, and in particular an initiator of particular affects in the ongoing interaction. This affective scaffolding is unique to the interaction with others and provides infants with a deep reflection about the dynamics of their own emotions, and ultimately about themselves as sentient and intentional entities. Again, the idea of "me," or self-consciousness, is mainly social in its origins, via the process of the "deep reflection" afforded by the reciprocal and constructive social mirror. As infants interact from birth with both physical and social objects, they simultaneously develop a sense of themselves as subject of action (ecological sell) and as object of reflection (self-awareness). Although self-awareness manifests itself unambiguously by the second year when infants become explicit about their idea of "me" (see Lewis, this volume), self-awareness does develop in the course of the first months. Much research is needed to unveil the early ontogeny of selfawareness in relation to the objectification of the self afforded by the systematic reciprocity and scaffolding of others.

Conclusion: The Paradox of the Physical Mirror As a conclusion, I would like to convey a final idea regarding reactions to specular images or mirror reflections, which often have been used by comparative and developmental psychologists as a mean to assess the emergence of self-awareness (e..g., see the influential work of Gallup, 1970; Lewis & Brooks-Gunn, 1979). This idea is that reactions to specular images are interesting, not because they denote the presence or absence of self-awareness, but because they can potentially denote subjects' awareness of a fundamental paradox attached to these images: the

68

PHILIPPEROCHAT

simultaneous experience of self and nonself (others). I propose that sensitivity to such a paradox underlies particular reactions of young children and adults confronted for the first time with their specular image. These reactions are revealing of a basic perceptual differentiation that takes place in the course of the first months. This differentiation is construed as the perceptual prerequisite of selfawareness. In general, it is erroneous to assume that self-consciousness emerges suddenly by the second year of life, as numerous researchers have proposed using mirror self-recognition as a paradigm. Mirrors are ambiguous at best and cannot be used as the crucial test of self-consciousness. When infants face their own image and start showing emotional embarrassment in the mark test, it is not because they become conscious of themselves, but because they come to terms with the ambiguity of physical mirrors as objects in the environment. What they experience, in my view, is the ambiguity reflected by the specular image, which specifies two things that are normally viewed as one: the sense of self, and the sense of another individual that is by definition distinct from the self. On one hand, the specular image specifies what the infant is (the ecological self) via the perfect temporal contingency of visual-proprioceptive information. On the other hand, it specifies simultaneously someone else for the mere reason that it is something that looks like another person, i.e., an animated and differentiated entity that is externalized (a nonself entity by definition). Children's embarrassment probably corresponds to a deep perceptual and emotional puzzlement associated with the highly unusual experience afforded by the mirror. Interestingly, this puzzlement seems to be much more dramatic in mature individuals who have never experienced the ambiguity of the specular image and who are confronted for the first time with their own reflection in a large mirror. The anthropologist Edmund Carpenter (1975) reported striking observations collected from adult individuals of an isolated tribe (the Biami) living in the Papuan plateau where neither slate or metallic surfaces exist, and where rivers do not provide clear physical reflections. Here is what Carpenter reported regarding these individuals' initial reaction to a large mirror, a wholly new experience for them: They were paralyzed: after their first startled response - - covering their mouths and ducking their heads they stood transfixed, staring at their images, only their stomach muscles betraying great tension. Like Narcissus, they were left numb, totally fascinated by their own reflections; indeed, the myth of Narcissus may refer to this phenomenon (Carpenter, 1975, pp. 452-453).

In conclusion, let us remember that the experience of the mirror violates something fundamental that infants experience from birth: the perception of

EARLY OBJECTIFICATIONOF THE SELF 69 themselves and of objects (including people) as differentiated and situated entities in the environment. If the determinants of self-awareness are mainly social, this awareness is fundamentally based on the detection of perceptual information that specifies the self as an entity differentiated from others. There is an element of fascination in mirrors, not because they are a key to self-knowledge, but because they afford an illusion. They are a source of ambiguous information that conflicts with what infants learn from birth as actors in a meaningful environment.

NOTES

1. The terms self-consciousness and self-awareness are considered as equivalent, both linked to a reflexive process whereby the self is both subject of action and object of reflection. In contrast, the sense of the ecological self is viewed as direct, pertaining exclusively to the self as subject of action.

2. Deep is used here to provide a contrast between the reflecting process of our person provided by people interacting with us, and the specular image reflected by the physical mirror that is a superficial, bidimensional reflection of our physical envelope. REFERENCES

Baldwin, J.M. (1906). Mental development in the child and the race (3rd ed.). New York: Augustus M. Kelley. Carpenter, E. (1975). The tribal terror of self-awareness. In P. Hockins (Ed.), Principles of Visual Anthropology. The Hague, Netherlands: Mouton. Clarkson, M.G., & Clifton, R.K. (1991). Acoustic determinants of newborn orienting. In M. Weiss & P.R. Zelazo. (Eds.), Newborn attention: Biological constraints and the influence of experience (pp. 99-119). Norwood, NJ: Ablex. Clifton, R.K., Rochat, P., Litovsky, R.Y., & Perris, E.E. (1991). Object representation guides infant reaching in the dark. Journal of Experimental Psychology: Human Perception and Performance, 17(2), 323-329. Eckman, P., Levenson, R.W., & Friesen, W.V. (1983). Autonomic nervous system activity distinguishes among emotions. Science, 221, 1208-1210. Fantz, R.L. (1963). Pattern vision in newborn infants. Science, 218, 179-181. Gallup, G.G. (1970). Chimpanzees: Self-recognition. Science, 167, 86-87. Gallup, G.G. (1982). Self-awareness and the emergence of mind in primates. American Journal of Primatology, 2, 237-248. Gibson, J.J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. Hofsten, C. von (1982). Eye-hand coordination in newborns. Developmental Psychology, 18(3). 450-461. Hofsten C. von,& Ronnqvist, L. (1988). Preparation for grasping an object: A development study. Journal of Experimental Psychology: Human Perception and Performance, 14, 610-621. James, W. (1890) The principles of psychology. New York: Henry Holt. Johnson, M.H., & Morton, J. (1991). Biology and cognitive development: The case of face recognition. Oxford: Blackwell.

70

PHILIPPEROCHAT

Kaye, K. (1982). The mental and social life of babies: How parents create persons. Chicago: University of Chicago Press. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. New York: Plenum. Lewis, M., Sullivan, M., & Brooks-Gunn, J. (1985). Emotional behavior during the learning contingency in early infancy. British Journal of Developmental Psychology, 3, 307-316. Lockman, J., Ashmead, D., & Bushnell, E. (1984). The development of anticipatory hand orientation during infancy. Journal of Experimental Child Psychology, 37, 176-186. MacFarlane, A. (1975). Olfaction and the development of social preferences in the human neonate. In R. Porter & M. O'Connor (Eds.), Parent-infant interaction CIBA Foundation Symposium 33 (pp. 103-117). Amsterdam: Elsevier-North Holland. Mahler, M.S., Pine, F., & Bergman, A. (1975). The Psychological birth of the human infant: Symbiosis and individuation. New York: Basic Books. Maurer, D.,& Salapatek, P. (1976). Developmental changes in the scanning of faces by young infants. Child Development, 47, 523-527. Meltzoff, A.N. (1993). The centrality of motor coordination and proprioception in social and cognitive development: from shared actions to shared minds. In G. Savelsbergh (Ed.), The development of coordination in infancy (pp. 463-496). Amsterdam: Elsevier Science Publisher. Meltzoff, A.N., & Moore, M.K. (1994). Imitation, memory, and the representation of persons. Infant Behavior and Development, 17, 83-99. Murray, L., & Trevarthen, C. (1985). Emotional regulation of interaction between twomonth-olds and their mothers. In T.M. Field & N.A. Fox (Eds.), Social perception in infants. Norwood, NJ: Ablex. Neisser, U. (1991). Two perceptually given aspects of the self and their development. Developmental Review, 11, 197-209. Neisser, U. (1993). The self perceived. In U. Neisser, (Ed.), The perceived self." Ecological and interpersonal sources of self-knowledge. (pp.3-24). Cambridge, MA: Cambridge University Press. Piaget, J. (1952). The origins of intelligence in children. New York: International Universities Press. Povinelli, D.J. (1987). Monkeys, apes, mirrors, and minds: The evolution of selfawareness in primates. Human Evolution, 2, 493-507. Rochat, P.(1983). Oral touch in young infants: Responses to variations of nipple characteristics in the first months of life. International Journal of Behavioral Development, 6, 123-133. Rochat, P. (1987). Mouthing and grasping in neonates: Evidence for the early detection of what hard and soft substances afford for action. Infant Behavior and Development, 10, 435-449. Rochat, P. (in press). Early development of the ecological self. In C. Dent-Read & P. Zukow-Goldring. (Eds.). Changing ecological approaches to development: Organism-environment mutualities. Washington, D.C.: APA Press. Rochat, P., & Senders, S. J., (1991). Active touch in infancy: Action systems in development. In M.J. Weiss & P.R. Zelazo (Eds.), Infant attention: Biological constraints and the influence of experience (pp. 412-442). Norwood, NJ: Ablex. Rochat, P., & Goubet, N. (1993, March). Determinants of infants' perceived reachability. Poster presented at the 60th Meeting of the Society for Research in Child Development, New Orleans, LA.

EARLY OBJECTIFICATIONOF THE SELF 71 Rochat, P., & Wraga, M. (1994). Perceiving what is reachable: Systematic errors in the perception of an affordance. (Rep. no. 29) Emory Cognition Project, Emory University. Sherrington, C. (1961). The integrative action of the nervous system. New Haven: Yale University Press. (original work published 1947) Spitz, R.A. (1965). The first year of life : A psychoanalytic study of normal and deviant development of object relations. New York: International Universities Press, Inc. Stern, D. (1985). The interpersonal world of the infant. New York: Basic Books. Tomasello, M., Kruger, A., & Ratner, (1993). Cultural learning. Behavioral and Brain Sciences. 16, 495-552. Trevarthen, C. (1974). The psychobiology of speech development. In E.H. Lenneberg (Ed.), Language and brain: Developmental aspects. Neurosciences Research Program Bulletin, 12, 570-585. Tronick, E. (1980). On the primacy of social skills. In D.B. Sawin, L. O. Walder, & J.H. Penticuff (Eds.), The exceptional infant: Psychosocial risks in infant environment transactions. New York: Bruner & Mazel. Wallon, H. (1970) De l'acte it la pensde: Essai de psychologie comparde (From act to thought: Essay in comparative psychology). Collection Champs Flammarion. (original work published 1942) Wallon, H. (1981). Comment se d6veloppe chez renfant la notion du corps propre. (How does the child develop the notion of own body). La reconnaissance de son image chez l'enfant et l'animal. Collection Textes de Base en Psychologie (P. Mounoud & A.Vinter, Eds.). Neuch~tel, Switzerland: Delachaux et Niestl6. Werner, H., & Kaplan, H. (1963). Symbolic formation: An organismic-developmental approach to language and the expression of thought. New York: Wiley. Watson, J.S. (1994). Detection of self: The perfect algorithm. In S.T. Parker, R.W. Mitchell, & M.L. Boccia (Eds.), Self-awareness in animals and humans." Developmental perspectives (pp. 131-148). Cambridge: Cambridge University Press.

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

73

CHAPTER 5

A Theory of the Role of Imitation in the Emergence of Self ANDREW N. MELTZOFF and M. KEITH MOORE

University of Washington

In this essay, we will provide data and arguments substantiating three propositions: 1) infants differentiate self from nonself and also recognize the commonality between themselves and other people, starting from birth; 2) infants' development of the notion of self is inextricably intertwined with their notion of other individuals; and 3) body imitation plays a role in infants' development of a notion of self. We suggest that imitation is a psychological bridge between self and other that bears two-way traffic starting from the neonatal period. This is the aboriginal state from which the development of the self proceeds. The newborn brings innate structure to his or her first interactions with people, and yet interactions with other individuals profoundly alter the notion of self. The challenge for a theory of selfdevelopment is to specify this innate structure and the way that it is subsequently reorganized. A theory of development that mischaracterizes the newborn is flawed; a model of the innate that ignores development misses the human capacity for reconceiving things, even our self. We are proposing a perspective on the status of the self in infancy that is different from two historically influential views. The first derives from Piaget (1952, 1954) and holds that there is no initial bridge between self and other because the infant is "adualistic" at birth. Piaget denies the existence of a bridge, along with any primitive grasp of the other. Piaget's task is to show how the infant bootstraps him- or herself out of an adualistic universe, how the infant eventually discovers other persons as separate entities. The second perspective ha~s social psychological roots deriving from Cooley (1902) and Mead (1934). In this view, infants are initially selfless and then molded into the adult state through social interaction.

74

A .N. MELTZOFF 8z M .K. MOORE

Philosophers, too, have pondered the self. Descartes was impressed with the asymmetry between knowledge of self and other. We seem to have direct firstperson experience of aspects of the s e l f - our thoughts, feelings, and intentions. However, we cannot experience the inner world of others. Even if we have an initial sense of self, what connects us to others? How do we come to believe that others are more sentient than the tables, trees, or automata around us - - that others have minds like our minds? In sum, psychologists such as Piaget have focused on the initial problem of distinguishing self and other, and philosophers such as Descartes have asked how self and other can be brought together. One has primarily inquired about differentiation; the other about commonality. The developmental struggle, according to both psychologists and philosophers, is how we gain objectivity about the self and conversely gain an understanding of the other as a center of subjectivity. Both recognize that a mature notion of self requires a parallel construction of self and other. The commonsense adult notion of "self' seems to entail both body and mind. Students of infancy interested in the self have most often focused on the perception of the body. Substantial progress has been made in this area. However, the adult notion of self goes beyond the body, and entails psychological dimensions such as intentions, memories, and thoughts about the self (e. g., Bermddez, Marcel, & Eilan, 1995; Campbell, 1994, 1995; Goldman, 1993; Gopnik, 1993). Less progress has been made in exploring the origins of these psychological aspects of self. In investigating them, one expects to see a lengthy developmental course (beyond infancy), although the earliest progenitors may still be discerned in the prelinguistic period. In this essay, we shall be concerned not only with the origins of the self as a perceived body, but also with the progenitors of the properly psychological aspects of self. First, we will show that neonates recognize the equivalence between self and other because they imitate, but that imitation is not solely confined to pure body perception. Rather, it appears to be mediated by a representation of themselves and other people. Although very young infants imitate specific behaviors, there is no indication that they can stand outside of this interaction and conceive of the more abstract notion of a "matching relationship." Second, we will show that older infants ~ infants in the second year of life m are not focused on specific behaviors, but construe imitation as a generative game that transcends specific gestures. This notion of mutuality transforms and extends the primitive grasp of self-other relations that was available to the neonate. Third, we will report an experiment with older infants showing that they ascribe goals to the behavior of others even when those goals are not attained. Moreover, the findings indicate that these older infants do not ascribe goals to inanimate objects. The possibility is raised that attributing goals and intentions to the self is intertwined

IMITATION AND THE EMERGENCE OF SELF

75

with an understanding of the other's goals and intentions. All three of these research programs, but particularly the last two, take us beyond the issue of recognizing one's body and illuminate progenitors of the notion of self as a bearer of psychological properties.

Ways of Measuring Preverbal Self Developmental psychologists have devised clever techniques for investigating the notion of the nonverbal self in infants. These include visual self-recognition in mirrors (Amsterdam, 1972; Lewis & Brooks-Gunn, 1979) and recognition and intercoordination of one's own movement patterns (Bahrick & Watson, 1985; Butterworth, 1995; Butterworth & Hopkins, 1988; Rochat, 1993; Rochat & Morgan, 1995; Van der Meer, Van der Weel, & Lee, 1995; Watson, 1994; see also chapters by these authors, this volume). These measures inform us about primitive "levels" of self (Butterworth, 1990; Lewis, 1994; Neisser, 1988, 1991, 1994; Stem, 1985; see also chapters by these authors, this volume) but are only seen as the first steps toward the mature adult notion. These nonverbal measures of self-recognition do not provide an engine for development of the self. The fact that infants rub rouge off their faces after seeing themselves in a mirror is a measure of visual self-recognition, yet only an extreme view would actually suppose that such mirror-guided self-inspection plays a critical role in the development of a "self." Moreover, most of these measures focus on the isolated infant. A child would have a very disturbed notion of self if he did not identify with other persons, even if he could recognize his own image in a mirror. A fuller notion of self would seem to involve how one understands oneself in relation to other persons. We suggest that there is a naturally occurring behavior that can also be used to measure the preverbal notion of self. This naturally occurring behavior is imitation. The human infant is highly imitative. In De Poetica, Aristotle proposed that imitation was a distinguishing mark of humans: "Imitation is natural to man from childhood, one of his advantages over the lower animals being this, that he is the most imitative creature in the world, and learns at first by imitation." Imitation provides a visible readout of the infant's notion of bodily self (Baldwin, 1906). We poke out tongues to infants, essentially asking, "Do you have one of these?" If the infant selectively responds with a tongue protrusion, this is a nonverbal indicator that the infant has a primitive body schema that includes the tongue as a differentiated aspect. We can purse our lips or wiggle our fingers to see if the infant identifies these body parts as well. Imitation provides more than a cataloging of the self's body parts. It provides a measure of interpersonal

76

A.N. MELTZOFF •

M.K. MOORE

correspondences. Imitation entails not simply intracorporeal recognition and coordination (as in mirror/TV self-recognition tasks), but also intercorporeal mapping. By imitating, the infant is showing that a specific body part of the other can be mapped to a specific organ of the self. Moreover, imitation is a bidirectional bridge. The infant can imitate the adult, and the adult can also imitate the infant. Parents spend many hours imitating their infants, shaking rattles after infants shake, blowing raspberries when infants blow raspberries, and making all sorts of exaggerated facial movements when the infants do so (Bruner, 1975; Stem, 1985; Tomasello, Kruger, & Ratner, 1993; Trevarthen, 1979). We suggest that such imitative exchanges are naturally occurring explorations of self-other correspondence during which infants come to know, among other things, what the self looks like from the outside: the purely visual manifestations of felt body movements. Thus, the infant uses the other to learn about the self, just as surely as using the sells experiences helps to interpret the behavior of others. It is all too easy to focus on only one direction, but we will show the power of considering imitation as a bidirectional interpersonal bridge. The human infant is highly imitative, but at what age does imitation begin? This is not an idle question about "competent infants," for if imitation indicates what we have suggested, any complete theory of the development of the notion of self will include the ontogenesis of imitation. It is universally accepted that 18month-olds imitate adults. Meltzoff and Moore (1977, 1983, 1989) reported a series of studies showing that imitation begins at birth. At first, this finding was considered surprising because it was not anticipated by traditional stagedevelopmental theories such as Piaget's. The phenomenon of early imitation has now been replicated and extended in more than 20 studies in a dozen independent laboratories. Successful imitation of a variety of different facial and manual gestures has been demonstrated. Studies have also documented the crosscultural universality of early imitation, demonstrating the phenomenon in the U.S. (Abravanel & Sigafoos, 1984; Field, Woodson, Greenberg, & Cohen, 1982), Canada (Legerstee, 1991), France (Fontaine, 1984), Greece (Kugiumutzakis, 1985), Switzerland (Maratos, 1982; Vinter, 1986), Sweden (Heimann, 1989; Heimann & Schaller, 1985; Heimann, Nelson, & Schaller, 1989), Israel (Kaitz, Meschulach-Sarfaty, Auerbach, & Eidelman, 1988), and rural Nepal (Reissland, 1988). The question that immediately arises is whether early imitation is "real" imitation. This is a fuzzy question, because the word real is undefined. (Is Weiskrantz's, 1966, blindsight "real" sight? Are 6-month-olds' smiles expressions of "real" joy?) The motivation behind the question is, however, an insightful one. The deeper question is whether the early imitative behavior involves an intentional mapping between self and other, or is merely reflexive and automatic. If early

IMITATION AND THE EMERGENCE OF SELF

77

imitation were merely reflexive, it would be an interesting biological adaptation but would not necessarily bear on issues of self. If infants are intentionally matching the adult's act, this would have larger implications.

Early Imitation as Intentional Matching of Self to Other Several experiments were done to distinguish reflexive from intentional imitation. Meltzoff and Moore (1977) showed that 12- to 21-day-old infants could imitate four different adult gestures: lip protrusion, mouth opening, tongue protrusion, and finger movement. These gestures were carefully selected to test whether the imitative response was a specific match or merely a global reaction. The results favored specificity inasmuch as infants did not appear to confuse actions or organs. They differentially responded to tongue protrusion with tongue protrusion and not lip protrusion, thus showing that the specific organ could be identified. They also differentially responded to two different actions produced by the same organ (lips protruding versus lips opening), thus showing that the movement pattern was .being extracted. The range of gestures demonstrated and the specificity of the imitative acts suggested that a generative matching mechanism had been uncovered. Another study tested whether infants could imitate even if there was a temporal gap imposed between perception and production. We wanted to prevent infants from initiating the response while the adult gesture was demonstrated. Reflexes do not jump such gaps. The experimental technique was to put a pacifier in the infant's mouth during the stimulus presentation. Infants engaged in competing motor activity (sucking on the pacifier) during the presentation. The adult then stopped gesturing, assumed a neutral face pose, and only then removed the pacifier. The results showed that infants were able to imitate, in contrast to what might be expected by the reflexive account. These studies seemed to indicate that a straightforward reflexive model could not account for early imitation. Further studies suggested that early imitation is a goal-directed, intentional activity. In one study, 6-week-old infants were shown a novel oral movement: a large tongue protrusion to the side (Meltzoff & Moore, 1994). The prediction from a reflexive model is either: a) no response, if the unusual adult gesture was not innately specified as a "triggering stimulus"; or b) persistence in a preset motor pattern of ordinary tongue protrusion without modification by the adult act. In fact, infants imitated and gradually corrected their imitative attempts to achieve a more faithful matching of the target. For most babies, the initial response was not an exact copy of the adult. Instead, they made mistakes. The early attempts focused on the lateral components: The tongue either

78

A .N. MELTZOFF 8~ M.K. M O O R E

went into the cheek or was thrust slightly forward and then slid laterally during retraction. A microanalysis of the response profile documented that infants corrected these initial attempts over successive efforts. This pattern of correcting one's own behavior over time would seem to suggest that, even at this age, infants can monitor their own acts w an early indicator of "self' at the level of body and perhaps the earliest progenitor of self-awareness. These findings of infants homing in on the presented target are in line with the mechanism for early imitation suggested by Meltzoff and Moore (1977, 1983, 1989, 1994). We proposed that infant imitation depends on a process of active intermodal mapping (AIM). The central notion is that imitation, even early imitation, is a matching-to-target process. The goal or behavioral target is specified visually. Infants' self-produced movements provide proprioceptive information that is compared to the visually specified target. Thus, AIM hypothesizes that the perception and production of hmnan acts can be represented within a common supramodal framework, that infants are not limited to modalityspecific information about body movements in space. The supraanodal fraanework is the foundation on which imitation is built. Although the AIM hypothesis highlights error detection and correction, it does not rule out visual-motor mapping of elementary acts on a "first try." The crux of the hypothesis is that the adult act serves as a genuine target for the infants' behavior. There may be a delimited set of primitive acts (e.g., midline tongue protrusion) achieved with little need of feedback. Other, more complex acts involving the computation of transformations on these primitives (tongue to the side) may require proportionately more proprioceptive monitoring. In fact, the results demonstrated that young infants did not immediately trigger accurate imitations of the novel tongue-to-the-side behavior; they needed to correct their behavior to achieve it. Some infants responded in an even more revealing way to the tongue-to-theside display. They poked out their tongues and simultaneously turned their heads to the side. This head movement was not in the stimulus, but was the infants' version of how to get their bodies to do a novel act involving both tongue protrusion and an off-midline direction. Tongue protrusion + head turn was not the work of a mindless reflex. It was a creative error in which infants extracted the "goal" of the act and strived to get their bodies to duplicate it. We also came across an "experiment of nature" that further illuminated early imitation. Some neonates have an anatomical malformation that prevents them from protruding their tongues because of an attached freniculum. The freniculmn is the piece of skin attaching the tongue to the bottom of the mouth. The malformation extends the freniculum to the front tip of the tongue, preventing tongue protrusion. It is corrected by simple surgery. When shown no gestures or

IMITATION AND THE EMERGENCE OF SELF

79

when shown mouth openings, infants with an attached freniculum acted normally. Indeed, their mouth-opening imitations were indistinguishable from other infants'. The case of theoretical import was their reaction to the tongue protrusion gesture, which, of course, they could not duplicate. The infants attempted to poke out the tongue and then became frustrated and cried. This suggests that even for young babies, their intention is differentiated from the actual motor movements produced. The correction of imitative responses, the commission of creative errors in reaching a goal, and the frustration of the physically handicapped infants all suggest a common story. In all of these cases, the stimulus did not simply trigger a fixed response. In the first and third cases, infants made repeated attempts, and their intention was not satisfied by the initial motor performance stemming from it. This suggests that there is a differentiation between the representation of the target act that was derived from the external world and the representation of the infant's own body acts. The intention is apparently to bring these two into congruence. The second case shows that the response was not fully programmed, but was actively constructed by the infants; they responded with their best interpretation of what they saw.

A Function of Imitation: Identification and C o m m u n i c a t i o n w i t h Others Given the notion of representation discussed above, we can better understand the function that early imitation plays in self-other relations. We will show that infants can bring to mind and reinitiate imitative exchanges from the past. These memories of others' acts can serve to identify individuals, and reinitiating the imitative exchanges plays an identificatory and communicative function. One relevant study involved a substantial delay (Meltzoff & Moore, 1994). We showed 6-week-old infants a gesture on one day and brought them back after a 24-hour delay. Infants were shown the same person with a passive-face pose. Thus, the target gesture was not in the perceptual field. If the motor system was tightly coupled to what was seen, the infants should also display a passive face. That is not what happened. Infants remembered and imitated the gesture the adult did the day before. Infants who had seen the person demonstrate mouth opening 24 hours earlier stared at the adult and then performed that gesture; infants who had seen tongue protrusion produced tongue protrusions. They were imitating based on memory. It is as if the infants were asking, "Didn't we play this? .... Isn't this our game?" Simpler notions of perceptual-motor couplings or resonances do not seem to capture the full richness of the behavior because infants were not resonating to what was in view (which was a passive face).

80

A.N. MELTZOFF & M.K. MOORE

Another experiment led to similar conclusions. In this study, 6-week-old infants were presented with two different adults: the mother and a male experimenter (Meltzoff & Moore, 1992). We found that the way these people appeared and disappeared was critically important. When, for example, the adults changed places without the infant tracking the exchange, the infant paused to inspect the new person and then performed a burst of the actions shown by the previous person. We might have an infant who had been watching her mother showing the mouth-opening gesture. The mother then left, and the stranger appeared and showed tongue protrusion. When the infant saw the new adult, the infant stopped acting, looked at the face, and then performed an intensive bout of the previous person's gesture. What can account for this? Why should the infant produce the old person's gesture and not be driven by the gesture in current view? (We have also shown that infants can switch their behavior to match the adult under certain conditions; see Meltzoff & Moore, 1992.) We suggest two interrelated ideas to account for this pattern of behavior: First, young infants do not have a fully developed system for maintaining the unique identity of people over breaks in perceptual contact (Bower, 1982; Meltzoff & Moore, 1995; Moore & Meltzoff, 1978; Moore, Borton, & Darby, 1978). By what rules do they reidentify a particular person as being the same one again? Featural similarity alone is no guarantee that this is the same one. Mother may wear a kerchief, put her hair in curlers, or lean over the babies' crib with her hair falling in her face. Is the mother to become a series of different people as she is featurally altered? Does the young baby have a multitude of mothers, one for each appearance? Spatiotemporal information such as that provided by visually tracking a person as he or she moves in the environment (regardless of feature changes) is also important for weighing the identity of a person. It is thus understandable that infants can be confused about the unique identity of persons in a multiperson situation involving appearance and disappearance, especially if they have not completely tracked the exchange. Second, we propose that in addition to spatiotemporal and featural criteria for identity, infants also use functional criteria to sort out issues of identity, especially for other people. It is not only how a person looks, but how a person acts and what games they afford (to adapt a Gibsonian, 1979, term) that helps to verify their identity. We are proposing that one function early imitation serves for infants is to clarify who is in front of them when the identity of the person has been put into doubt (as it may be when a person leaves and a person takes the same place). We thus have two studies in which infants do not imitate what is presently in the perceptual field. In one, infants imitate the remembered actions from yesterday when not being shown those gestures. In the other, the infants imitate the

IMITATION AND THE EMERGENCE OF SELF

81

previous person's behavior instead of the one in front of them. Neither case is reducible to reflexes alone because the infant is n o t directly reflecting the body movements given in perception. Rather, infants seem to deploy imitative behaviors as a tool for probing the identity of the other, as if the earlier episode is a kind of shared memory or shared experience with the other that helps to identify the person. This suggests that infants remember individual others and their own interactions with them. Instead of having generalized reactions to all others (such as in smiling, cooing, and greeting of "humans in general"), the uniqueness of interpersonal relations is born. Others become distinct individuals, and relations with them become special and differentiated from each other for the infant. One wonders whether a foundation for the earliest notion of self as an entity emerges, in part, as the invariant abstracted from the distinctly different relationships in which it participates.

T h e O t h e r as a R e f l e c t i o n of S e l f

Thus far, we have offered some suggestions about the mechanism underlying early imitation, its intentional character, and the function it serves in the world of the young infant. Once we begin to take early imitation seriously, it puts a new light on the traditional measures of self, such as the mirror self-recognition studies. Mirror self-recognition tests only assess a narrow dimension of self, the recognition of one's visual appearance. From everything we have argued to this point, it should be clear that we think a prior, and more fundamental step in the development of self-understanding arises from awareness of one's own movements and body postures. You may need mirror experience to learn that your face does not normally have a red dot on it or that your eyes are blue. However, if our arguments about facial imitation are sound, you don't need mirror experience in order to visually recognize your own body movements. In principle, this intermodal equivalence has been detectable from birth. Visual instantiations of your normally unseen body movements can be directly related to the movements that are felt. Visual self-recognition based on unfelt/unseen visual appearances (such as often tapped in mirror/photograph recognition studies) is distinguishable from selfrecognition based on bodily movement patterns, and we believe that the latter provides the ontogenetic foundation for acquiring the former (Meltzoff, 1990). How can we begin to investigate infants' ability to recognize that seen human movements are "like me" or "like the movements that are felt"? Several approaches are possible; some involve presenting young infants with TV recordings of their leg movements (Bahrick & Watson, 1985; Rochat & Morgan, 1995, this volume; Watson, 1994, this volume). The question is whether they prefer visual images of

82

A.N. MELTZOFF & M.K. MOORE

their own movements to other control images. These are fascinating studies, and the results are compatible with the hypothesizing and findings so far offered. The findings suggest that infants in the first half year of life are monitoring their own body movements through proprioception because they can recognize visual instantiations of their unseen leg kicks. This interpretation is consonant with our own view of early imitation, which we based on the fact that infants use proprioceptive monitoring to correct their unseen imitative movements to bring them into line with the visual target. Recently, we extended this work in a series of studies in which an adult experimenter acted as a kind of "social mirror" to the infant, reflecting everything the baby did. We wanted to know if infants could recognize this self-other correspondence despite the absence of featural identity because the adult did not look like the infant. Two experiments were conducted with 14-month-olds (Meltzoff, 1990). The first study investigated whether or not infants at this age showed more interest in their own behavior being reflected to them by another person than they did when the adult imitated another baby who was not the self. The infants sat at a table, across from two adults who sat side-by-side. All three participants were provided with replicas of the same toys. Everything the infant (A) did with his toy was directly mimicked by one of the adults (A'). If A banged the toy three times on the table, A' banged his toy three times on the table. The second adult (B) actively manipulated the toys. Furthermore, we wanted this adult not only to be active, but to do "babylike" things with the toys so that no preference for A' could be based solely on a differentiation of adult versus infantile actions. In our yokedcontrol procedure, there were two TV monitors situated behind the infants and in view of the adults. One monitor displayed the actions of the current infant, live, and the other displayed the video recording of the immediately preceding infant. The job of each adult was to mimic one of the infants. Both adults performed in perfectly infantile ways, but only one matched the perceiving infant. Could the infants recognize which adult was acting like the self and which like another baby? We thought that if infants could detect that their actions were being matched, they would prefer to look at A' and also smile at him more. We also thought that infants would tend to test the relationship between the self and the imitating other by experimenting with it. For example, infants might modulate their acts by performing sudden and unexpected movements to check if A'was still shadowing them. Adults do this when they unexpectedly catch sight of themselves in a store video camera; they wave their arms or make a sudden movement to check whether the image on the screen follows suit. The results showed that infants looked longer at A', smiled more often at him, and directed more testing behavior toward him. These results suggest that the subjects are recognizing the relationship between the actions of the self and the actions of the imitating other. How did the

IMITATION AND THE EMERGENCE OF SELF

83

babies detect this relationship? Broadly speaking, two kinds of information are available. The first is information based solely on temporal contingency. According to this alternative, the infant need only detect that whenever he does x the adult does y. The infant need not detect that x and y are structurally congruent in any way, only that they are temporally linked. A second alternative is that the infant can do more than recognize the temporal contingency. In particular, the infant may be able to recognize that the actions of the self and other have the same form: They are structurally congruent. To differentiate these alternatives, a study was designed in which the purely temporal aspects of the contingency were controlled by having both experimenters act at the same time. This was achieved by having three predetermined pairs of target actions. Both experimenters sat passively until the infant performed one of the target actions on this list. If and only if the infant exhibited one of these target actions, both experimenters began to act. The imitating adult performed the infant's act, and the control adult performed a different behavior that was paired with it from the predetermined target list. For example, whenever an infant shook a toy, the imitating adult also shook his toy, carefully shadowing the infant. The behavior of the other adult was also under the temporal control of the infant, but this adult performed a different type of action. Whenever the infant shook his toy, the control adult would slide his matched toy, also carefully shadowing the speed and duration of the infant's act. This design achieves the goal of having the acts of both adults' actions contingent on the infant's. What differentiates the two experimenters is not the purely temporal relations with the acting subject, but the structure of their actions vis-hvis the subject. The results showed that the infants looked, smiled, and most importantly, directed more testing behavior at the matching actor. Even with temporal contingency information controlled, infants can recognize the structural congruence between the acts they see others perform and the acts they do themselves. The data illuminate infants' perception of their own bodies and movement patterns. Infants do not just recognize that another moves w h e n they move (temporal synchrony), but recognize that another moves in the same m a n n e r as they do (structural congruence). In that sense, they can recognize that their own acts as felt are like the acts seen in others. It is important to put this work on adults' imitation of infants in relation to the work on early imitation. First, the fact that infants can imitate from birth does not in itself prove that they would be sensitive to or interested in being imitated. The work on early imitation showed that they can map from the other to the self. The work on being imitated shows that they can turn this around. They can recognize being matched, thereby mapping from self to other. The fact that infants

84

A.N. MELTZOFF & M.K. MOORE

can map in both directions is why we consider imitation a bidirectional bridge between self and other. Second, this change of direction has some interesting implications. There is a reversal of roles inasmuch as the infant becomes the model in these episodes. The infant acts first, and the other follows. Thus, the infants' acts become the focus of attention and determine the interaction. The question immediately arises as to how younger infants would respond to being imitated. We have investigated this question. We found that 6-week-olds are particularly attentive to being imitated and clearly differentiate it from the situation of being shown gestures that are not in their control. However, there are two h~portant differences between the younger and the older infants. First, while the younger infants do increase the frequency of the particular gesture being imitated, there is no evidence that they switch to doing mismatching gestures to see if they will be copied. For example, if the adult is imitating infant mouth opening, infants will increase that behavior; however, they do not switch to tongue protrusion to test this relationship. Second, it appears that younger infants interpret being copied as a causal relation; their acts cause the adult's acts. The older infants seem to go beyond this interpretation and treat the interaction as a shared game that is being played with a social other. Older infants understand being imitated in a different, more advanced way than the younger ones. For older infants, the testing response progresses from shnple modification of ongoing behavior, to initiating highly unusual behaviors, and then to manifesting joy when they see the adult respond in kind. Infants are happy to engage in these imitative exchanges for long periods of time (more than 20 minutes without break). Part of what seems to be giving the older infant such joy in the case of being imitated is that a "matching game" is being played. By matching game, we mean a matching relationship abstractly considered the notion that "you will do what I do" where the particular behaviors are infinitely substitutable. It is not the notion that tongue protrusion causes tongue protrusion x leads to x and y leads to y but rather that the social other is doing "the same as" I do. By 14 months, infants undoubtedly know that adults are not under their total control, and another part of the joy of this exchange is the realization that although the infant doesn't actually control the other (the probable illusion of the neonate), nonetheless the other is still willing to do just what I do. These two factors together help to explain why infants will engage in a matching game for such extended periods of time m for far longer periods and with far greater joy than watching themselves in a mirror.

IMITATION AND THE EMERGENCE OF SELF

85

H u m a n Acts Are Understood as Purposive; Machine M o v e m e n t s Are Not We have argued that imitation in very young infants is constrained to matching specific acts, and older infants operate at a more abstract level. For older infants, the focus is not on specific acts but on the matching relationship itself. We have suggested that older infants can recognize that an adult has taken on the goal of matching the infants' behavior. This implies that infants see the goals of behavior and perhaps even attribute intentions to others. Adults clearly construe people in this way: We think of most human acts as purposive, as aimed toward achieving some goal. We know that our own behavior is intentional and suppose that the behavior of others is also. Furthermore, we don't suppose that inanimate objects have such things as intentions and goals. Do infants interpret human acts in this way? Do they make this distinction between humans and inanimate objects? Might their understanding of their own goals and intentions be linked in some way to their understanding that o t h e r s - entities who are "like me" m also have goals and intentions? A recent study suggests that by 18 months of age, infants go beyond imitating the visible surface behavior of the adult and take into account the purposiveness of human acts (Meltzoff, 1995). In the critical test situation, infants saw an adult who demonstrated an "intention" to act in a certain way. hnportantly, the adult never fulfilled this intention; he tried but failed to perform the act, so the end-state was never reached. The goal toward which the adult was striving therefore remained unobserved by the infant. To an adult, it was easy to see the actor's intention. The experimental question was whether infants registered this behavior in purely physical terms, or whether they too could read through the surface behavior to the underlying goal, which remained unseen. The subjects, who were too young to give verbal reports, revealed how they interpreted the event by what they chose to reenact. In the experiment, each infant was assigned to one of four groups that varied according to what was shown. For one group, the adult demonstrated a target act for each of the five tasks. For another group, the adult tried to perform the target acts, but none of the goal-states was successfully achieved. Control groups were shown neither the target acts nor the attempt to produce them. The results showed that infants who saw the unsuccessful attempt or the full target act produced target acts at a significantly higher rate than the controls. It was also striking that infants were as likely to perform the target after seeing the adult "trying" as they were after seeing the actual demonstration of the target behavior itself. We interpret this pattern of data as showing that 18-month-olds can infer the goal toward which a sequence of actions is aimed, even though the end-state is never attained.

86

A.N. MELTZOFF & M.K. MOORE

How would infants respond to a mechanical device that mimicked the same movements performed by the actor in the failed-attempt condition? One of the stimuli used in the first study was the attempt to pull apart a dumbbell-shaped object. The adult failed to achieve this end because her hand "accidentally" slipped off the end of the cube as she tried to pull it apart. The inanimate device was constructed so that it too slipped off the end of the dumbbell. The device did not look human, but it had poles for arms and mechanical pincers for hands (see Figure 1). The pincers "grasped" the dumbbell at the two ends just as the human hands did. One mechanical arm was then moved outward Oust as in the human case), and. its pincer slipped off the end of the dumbbell (just as the human hand did). As shown in Figure 1, the movement patterns in space closely mimicked those of the human's.

FIGURE 1. Schematic of the presentations used in Meltzoff (1995). Top panel shows the human hand sliding off the end of the dumbbell. Bottom panel shows the inanimate device mimicking this movement. The results showed that infants did not attribute a goal or intention to the movements of the inanimate device. Although they looked at the device as long as at the human display, and even smiled at the movements about as much (showing no signs of fear or avoidance), they simply did not see the sequence of actions as implying a goal. Infants were six times more likely to pull the dumbbell apart after seeing the adult perform the failed attempt as they were when the failure was performed by the inanimate device. Indeed, they were no more likely to pull the dumbbell apart in the inanimate condition than they were in a baseline condition in which no demonstration at all was modeled. Taken together, these findings indicate that infants interpreted human acts within a purposive framework; moreover, equivalent movements of inanimate objects are not interpreted in this way.

IMITATION AND THE EMERGENCE OF SELF

87

A Developmental Theory: Imitation as an Interpersonal Bridge We have considered three aspects of infant imitation and what they tell us about the development of the preverbal self. During the neonatal period, infants imitate the specific behaviors they see. By toddlerhood, infants appreciate imitation at the level of a matching relationship expressing a mutuality that transcends particular behaviors. Another experiment showed that these older infants will imitate acts they do not see; they will enact what the experimenter tried to do, recognizing that the self and other can attain a common goal even if the surface behavior differs. Our view is that development of the self and other is intertwined, with imitation being a bridge that carries interpersonal information between self and other, as will be elaborated below.

Supramodal Representation: InterpersonalDifferentiation with Equivalence To account for early imitation, we used the notion of a supramodal framework. This notion of supramodality (which is a modification of ideas introduced by J. J. Gibson, 1966; E. J. Gibson, 1969; and Bower,1982) has deep implications for the infants' notion of self in relation to others. In our view, early imitation shows that infants can represent human movement patterns they see and ones they perform using the same mental code (Meltzoff & Moore, 1977, 1983, 1994). The perception of the adult's act is registered in such a way that it can be used directly for the execution of a motor plan. There is thus something like an act space or primitive body scheme that allows the infant to unify the visual and motor/proprioceptive information into one common supramodal framework. (The recognition of self-movements on TV is also interpretable within this view: Bahrick & Watson, 1985; Rochat & Morgan, 1995; van der Meer et al., 1995; Watson, 1994). This way of describing early imitation puts emphasis on the commonality between self and other. As soon as one focuses on commonality, it immediately raises the psychologist's question about differentiation. Perhaps the supramodal code means that there are no grounds for distinguishing self from other, for distinguishing proprioceptive information from exteroceptive information. Thought of in this way, the supramodal system is simply a translation device for turning visual perceptions into motor output, a perception-production traa~sducer. In our view, there are three reasons to think that a more differentiated system is in play in early infancy. First, the infant need not produce what is given to perception. In the pacifier study (Meltzoff & Moore, 1977), the infant responded after the adult stopped, when the actual stimulus was a passive face. This same effect was shown even more forcefully in the Meltzoff and Moore (1994) study in which infants imitated after a 24-hour delay. Thus, the information picked up by

88

A.N. MELTZOFF & M.K. MOORE

vision can be accessed at a later time. This seems to require a stored representation of the adult's act and not simply a transduction or perceptual-motor coupling. Second, as we have seen, the imitative acts can be corrected to achieve a more faithful match. Information about one's own acts has to be available for comparison to the representation of the adult's act. Third, infants show special interest in being imitated themselves, which indicates the capacity to recognize when their unseen facial behavior is being copied. Such recognition implies that there is a representation of their own body. These pieces of evidence go beyond the simple transducer account. They suggest that, at minimum, the infant's representational system performs three functions: 1) preserving information about the external world; 2) preserving information about one's own body; and 3) a means of comparison. This suggests a differentiation in the supramodal system whereby the representation of the other's body is separate from the representation of one's body. Although they both use supramodal language, they are not confused. The cognitive act is to compare these two representations N in one case, to match one's own acts to the other (imitative correction), and in the other case, to detect being matched oneself (recognizing being imitated). Thus, in our view, the mental code may use supramodal "language," but the system is not just one undifferentiated supramodal whole.

Imitation as a Catalyst for Developing a Notion of Self Although the initial supramodal framework we are postulating is a powerful system for relating self to other, it still has profound limitations. Because early imitative matching is done on the basis of supramodal equivalence, modalityspecific information would not be preserved. For example, the infant can imitate without yet knowing what his acts look like in a purely visual sense, as it were from the outside looking in. Mirrors may be one avenue for such development. Another is by being copied in exchanges involving mutual imitation. To the extent that the imitation is a faithful one, infants would gain a sense of what their felt acts look like, and this would be especially true for unseen body parts and movements (faces, etc.). There are aspects of self that can only be known by seeing reflections of yourself as others see you. Mutual imitation also affords a critical step forward in the emergence of self in a second, perhaps more profound, way. Let's examine the case of a toddler's mutual imitation in more detail. At first, the infant performs her own actions without regard to their effects on the adult (because the infant doesn't yet know that she is in a mutual imitation experiment). When the infant notices the adult movement, she shifts her attention to that behavior and then begins to vary her own behavior. This, we are supposing, is done to determine whether the adult match was a chance or repeatable event. When the infant sees that the adult continues to match, the

IMITATION AND THE EMERGENCE OF SELF

89

infant produces highly novel behavioral sequences and looks on gleefully as the adult follows suit. This interaction carries a whole array of information: a) the adult's behavior matches her own; b) it is not a random congruence; c) the specific behaviors don't matter because the invariant in this situation is "to match"; and d) from the infant's viewpoint, the infant's own novel behaviors were intended acts. All of these ingredients provide grounds for an important realization by the infant: "I intend to produce these acts, the adult performs these same acts, they are not chance events; therefore, the adult intends his acts." In other words, the similarity transcends the surface behavior. This is a matching of intended acts. There are again bidirectional implications from considering these interactions at the level of mutually intended acts, instead of only shared behaviors. Going in one direction, the infant now knows more about the other. The adult is an intending other. But going in the other direction, because the other is also like me, the infant could realize that he or she is also an intender. This would constitute an advance in the infant's notion of self. We have already shown that the infant has had intentions from an early age, but having intentions is different from being aware that one has intentions. The development of a notion of self as one who intends is a step forward in the infant's "level" of understanding self. Thus, in our developmental model, mutual imitation provides a mechanism for moving beyond the recognition of one's body. To use the Neisser/Butterworth typology, it provides a developmental mechanism for the infant transcending the "ecological" self. The model offers a partial account of how humans might begin to develop the notion that self and other are commensurate bearers of psychological properties like goals and intentions. If our developmental theory is correct, it delimits the class of entities to which the infant ascribes psychological properties. Toddlers should see other people not as things, but as purposive beings because people can be imitated, are like them, and engage in mutual imitation. The experimental findings showed that 18-montholds do see others in purposive terms. They interpret even the unsuccessful attempts of others as aimed at a goal. Indeed, infants performed the goal acts just as readily after seeing failed attempts as after seeing the successful act by the adult. It was as if they "saw through" the surface behavior to the underlying goal of the act. The infants were not limited to duplicating what was seen; they did what the adult "intended" to do. Just as importantly for our theoretical model, the infants did not respond in this way to the movements of the inanimate object. When presented with the same sequence of movements traced by a mechanical device instead of by a human, they responded at baseline levels. Apparently, the inanimate movements were not seen in purposive terms.

90

A.N. MELTZOFF & M.K. MOORE

Conclusions

W e have argued that imitation is both a measure of preverbal self and also a mechanism for the development of the notion of self. Imitation serves in the development of self because it is a two-way bridge between the self and other. On the one hand, it is a naturally occurring measure of body perception and awareness through proprioception, i.e., Neisser's (1991) ecological and early interpersonal selves. On the other hand, we have argued that imitation affords a means for going beyond body perception to an awareness that one has goals and intentions just as others do. The adult notion of self encompasses one's body, but is more than an awareness of body because it encompasses psychological properties. This adult notion does not develop in isolation; one's notion of self is inextricably intertwined with one's notion of others.

ACKNOWLEDGMENTS

Funding was provided by NIH (HD-22514). We thank Calle Fisher and Craig Harris for assistance on the research; Pat Kuhl and Alison Gopnik for helpful discussions on the problems addressed; and Philippe Rochat for prompting us to bring this writing project to fruition. The order of authorship is alphabetical; the work was thoroughly collaborative. Requests for reprints should be sent to Andrew N. Meltzoff, Department of Psychology, University of Washington (Box 357920), Seattle, WA 98195-7920. REFERENCES

Abravanel, E., & Sigafoos, A.D. (1984). Exploring the presence of imitation during early infancy. Child Development, 55, 381-392. Amsterdam, B. (1972). Mirror self-image reactions before age two. Developmental Psychobiology, 5, 297-305. Bahrick, L. E., & Watson, J.S. (1985). Detection of intermodal proprioceptive-visual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21, 963-973. Baldwin, J.M. (1906). Mental development in the child and the race (3rd ed.). New York: Augustus M. Kelley. Bermt~dez, J., Marcel, A.J., & Eilan, N. (1995). The body and the self Cambridge, MA: MIT Press. Bower, T.G.R. (1982). Development in infancy (2nd ed.). San Francisco: W.H. Freeman. Bruner, J.S. (1975). From communication to language: A psychological perspective. Cognition, 3, 255-287. Butterworth, G. (1992). Origins in self-perception in infancy. Psychological Inquiry, 3, pp. 103-111. Butterworth, G. (1995). An ecological perspective on the origins of self. In J. Berm?adez, A.J. Marcel, & N. Eilan (Eds.), The body and the self (pp. 87-107). Cambridge, MA: MIT Press.

IMITATION AND THE EMERGENCEOF SELF 91 Butterworth, G., & Hopkins, B. (1988). Hand-mouth coordination in the newborn baby. British Journal of Developmental Psychology, 6, 303-314. Campbell, J. (1994). Past, space, and self. Cambridge, MA: MIT Press. Campbell, J. (1995). The body image and self "consciousness."In J. Berm~dez, A.J. Marcel, & N. Eilan (Eds.), The body and the self (pp. 29-42). Cambridge, MA: MIT Press. Cooley, C.H. (1902). Human nature and the social order. New York: Charles Scribners. Field, T.M., Woodson, R., Greenberg, R., & Cohen, D. (1982). Discrimination and imitation of facial expressions by neonates. Science, 218, 179-181. Fontaine, R. (1984). Imitative skills between birth and six months. Infant Behavior and Development, 7, 323-333. Gibson, E.J. (1969). Principles of perceptual learning and development. New York: Appleton-Century-Crofts. Gibson, J.J. (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin. Gibson, J.J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. Goldman, A.I. (1993). The psychology of folk psychology. Behavioral and Brain Sciences, 16, 15-28. Gopnik, A. (1993). How we know our minds: The illusion of first-person knowledge of intentionality. Behavioral and Brain Sciences, 16, 1-14. Heimann, M. (1989). Neonatal imitation, gaze aversion, and mother-infant interaction. Infant Behavior and Development, 12, 495-505. Heimann, M., Nelson, K.E., & Schaller, J. (1989). Neonatal imitation of tongue protrusion and mouth opening: Methodological aspects and evidence of early individual differences. Scandinavian Journal of Psychology, 30, 90-101. Heimann, M., & Schaller, J. (1985). Imitative reactions among 14- to 21-day-old infants. Infant Mental Health Journal, 6, 31-39. Kaitz, M., Meschulach-Sarfaty, O., Auerbach, J., & Eidelman, A. (1988). A reexamination of newborn's ability to imitate facial expressions. Developmental Psychology, 24, 3-7. Kugiumutzakis, J. (1985). Development of imitation during the first six months of life (Uppsala Psychological Rep. No. 377). Uppsala, Sweden: Uppsala University. Legerstee, M. (1991). The role of person and object in eliciting early imitation. Journal of Experimental Child Psychology, 51, 423-433. Lewis, M. (1994). Myself and me. In S.T. Parker, R.W. Mitchell, & M.L. Boccia (Eds.), Self-awareness in animals and humans (pp. 20-34). Cambridge: Cambridge University Press. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. New York: Plenum Press. Maratos, O. (1982). Trends in the development of imitation in early infancy. In T. G. Bever (Ed.), Regressions in mental development: Basic phenomena and theories (pp. 81-101). Hillsdale, NJ: Erlbaum. Mead, G. H. (1934). Mind, self, and society from the standpoint of a social behaviorist. Chicago: University of Chicago Press. Meltzoff, A. N. (1990). Foundations for developing a concept of self: The role of imitation in relating self to other and the value of social mirroring, social modeling, and self-practice in infancy. In D. Cicchetti & M. Beeghly (Eds.), The self in transition: Infancy to childhood (pp. 139-164). Chicago: University of Chicago Press. Meltzoff, A.N. (1995). Understanding the intentions of others: Re-enactment of intended acts by 18-month-old children. Developmental Psychology, 31(5).

92

A.N. MELTZOFF • M.K. MOORE

Meltzoff, A.N., & Gopnik, A. (1993). The role of imitation in understanding persons and developing a theory of mind. In S. Baron-Cohen, H. Tager-Flusberg, & D. Cohen (Eds.), Understanding other minds: Perspectives from autism (pp. 335366). New York: Oxford University Press. Meltzoff, A.N., & Moore, M.K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 75-78. Meltzoff, A.N., & Moore, M.K. (1983). Newborn infants imitate adult facial gestures. Child Development, 54, 702-709. Meltzoff, A.N., & Moore, M.K. (1989). Imitation in newborn infants: Exploring the range of gestures imitated and the underlying mechanisms. Developmental Psychology, 25, 954-962. Meltzoff, A.N., & Moore, M.K. (1992). Early imitation within a functional framework: The importance of person identity, movement, and development. Infant Behavior and Development, 15, 479-505. Meltzoff, A.N., & Moore, M.K. (1994). Imitation, memory, and the representation of persons. Infant Behavior and Development, 17, 83-99. Meltzoff, A.N., & Moore, M.K. (1995). Infants' understanding of people and things: From body imitation to folk psychology. In J. Berm~dez, A.J. Marcel, & N. Eilan (Eds.), The body and the self (pp. 43-69). Cambridge, MA: MIT Press. Moore, M.K., Borton, R., & Darby, B.L. (1978). Visual tracking in young infants: Evidence for object identity or object permanence? Journal of Experimental Child Psychology, 25, 183-198. Moore, M.K., & Meltzoff, A.N. (1978). Object permanence, imitation, and language development in infancy: Toward a neo-Piagetian perspective on communicative and cognitive development. In F.D. Minifie & L.L. Lloyd (Eds.), Communicative and cognitive abilities: Early behavioral assessment (pp. 151-184). Baltimore, MD: University Park Press. Neisser, U. (1988). Five kinds of self-knowledge. Philosophical Psychology, 1, 3559. Neisser, U. (1991). Two perceptually given aspects of the self and their development. Developmental Review, 11, 197-209. Neisser, U. (1994). The perceived self" Ecological and interpersonal sources of selfknowledge. Cambridge, MA: Cambridge University Press. Piaget, J. (1952). The origins of intelligence in children. New York: International Universities Press. Piaget, J. (1954). The construction of reality in the child. New York: Basic Books. Reissland, N. (1988). Neonatal imitation in the first hour of life: Observations in rural Nepal. Developmental Psychology, 24, 464-469. Rochat, P. (1993). Hand-mouth coordination in the newborn: Morphology, determinants, and early development of a basic act. In G. Savelsbergh (Ed.), The development of coordination in infancy (pp. 265-288). Amsterdam: Elsevier Publisher. Rochat, P., & Morgan, R. (1995). Spatial determinants in the perception of selfproduced leg movements by 3- to 5-month-old infants. Developmental Psychology, 31 (4). Stem, D. N. (1985). The interpersonal world of the infant. New York: Basic Books. Tomasello, M., Kruger, A.C., & Ratner, H.H. (1993). Cultural learning. Behavioral and Brain Sciences, 16, 495-552. Trevarthen, C. (1979). Communication and cooperation in early infancy: A description of primary intersubjectivity. In M. Bullowa (Ed.), Before speech (pp. 321-347). New York: Cambridge University Press.

IMITATION AND THE EMERGENCE OF SELF

93

Van der Meer, A.L.H., Van der Weel, F.R., & Lee, D.N. (1995). The functional significance of arm movements in neonates. Science, 267, 693-695. Vinter, A. (1986). The role of movement in eliciting early imitations. Child Development, 57, 66-71. Watson, J.S. (1994). Detection of self: The perfect algorithm. In S.T. Parker, R.W. Mitchell, & M.L. Boccia (Eds.), Self-awareness in animals and humans (pp. 131148). Cambridge: Cambridge University Press. Weiskrantz, L. (1986). Blindsight: A case study and implications. Oxford: Clarendon Press.

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

95

CHAPTER 6

Aspects of Self: From Systems to Ideas MICHAEL LEWIS

Institute for the Study of Child Development, Robert Wood Johnson Medical School

The nature of the mental states of shyness, shame, and modesty have as their emotional element self-attention. It is not the simple act of reflecting on our own appearance, but the thinking what others think of us, which excites a blush (p. 325.) I have received authentic accounts of two little girls blushing at the ages of between 2 and 3 years ....It appears that the mental powers of infants are not as yet sufficiently developed to allow of their blushing (p. 310). - Charles Darwin (1872)

The Expression of the Emotions in Man and Animals

My daughter, a medical student, enjoys pointing out to me articles about biological topics that use the self word. Not too long ago, she sent me an article by Von Boehmer and Kisielow (1990) entitled, "Self and Nonself-discrimination by T-Cells," and one by Harding, Gray, McClure, Anderson, and Clarke (1990) entitled, "Self Incompatibility: A Self-recognition System in Plants." We use the terms self and nonself in reference to plants and to cells as well as to humans. If the term self used here is confusing, consider several examples from the human literature. We use the term self-regulation when we talk about newborn infants (Kopp, 1982) and intersubjectivity in 6-month-olds (Stem, 1985). Understanding the term self does not get much better when we consider adult humans; for example, the Western view of self as an "I-self' versus an Eastern view of self as a "we-self" (see Geertz, 1984; Roland,

1988). The p h e n o m e n a of multiple

personality disorder (Ross, C.A., 1989) - - the idea of multiple selves rather than a single self recently has caught our attention, although Prince (1905/1978) detailed this disorder nearly a century ago.

96

MICHAEL LEWIS

Even in our everyday lives, we are confronted with explaining selves, our own as well as others. For example, much of my motor action the very act of writing these words m although initially planned, is carried out by the machinery of my body, which includes by definition, self-regulation and self-other differentiation. How could it be that I can write or even speak, almost effortlessly, complicated phrases and thoughts without giving much attention to the processes that give rise to them? I certainly know, as I sit here writing, that I have a plan to write this chapter and an outline, which I have made to help formulate my thoughts. It is clear that I have intentions and desires and presumably the ability to carry out the task of thinking and writing. Yet the very acts themselves seem to emerge from me almost effortlessly. Indeed, if I focus my attention on them, I find that doing so interrupts the very act that I am performing. It is clear, then, that this self of mine the body and the mind - - that carries out this task does not need and, in fact, may be hindered by my paying attention to myself. A self is necessary to formulate at least sometimes what it is that I wish to think about, but does not appear to be involved in the process that actually carries out the task of thinking. Consider this example: We give a subject the problem of adding a 7 to a sum of 7's that preceded (e.g., 7 + 7 = 14 + 7 = 21 + 7 - 28, etc.). It is clear that as we carry out this task, we cannot watch ourselves do the arithmetic. It would seem that one aspect of the self has set up the problem, another will solve it; and it is likely that the first will evaluate the result of what the second did. These diverse examples from plants, cells, and adults all address the single topic of what it could mean when we use the term self. Our problem is made no more easy when we consider such topics as self- deception and akrasia, a Greek term used to mean "lack of will or self-control." Let us consider self-deception. How is it possible for a self to deceive its self? It would appear to be a logical impossibility, but only if we believe that a self is a single thing. A self as a single thing could not deceive its self. If, however, we conceive of a self in the manner that Freud (1959) did, one that consists of several aspects or features, then we would be able to argue that one part of the self can deceive another part. The problem of akrasia is still another issue. Aristotle (1953) makes reference to the very well-known phenomena where we have an intention to do X but find ourselves unable to do X, or doing Y instead. This situation is familiar to all of us. For example, the other night after a large meal in a restaurant, the dessert tray was brought to the table. As it approached, my intention was not to order a dessert and, thus, to spare myself the calories. Yet by the time the others at the table had made their choices, I was ready to order and did so. How was it that my intention to not X became X?

ASPECTS OF SELF

97

The idea of intentions failed, or akrasia, suggests that the one way to understand the self is to assume the position that there are multiple aspects to the self and that different features have different intentions. Thus, such everyday experiences of self-deception and akrasia give rise to the idea of the self as a modular system, an idea applied to brain structure and process (Gazzaniga, 1988). It is clear that whatever the self may be, it is a complex multi-aspect sort of "thing" or "process." This multi-aspect self has been considered in many different ways. In early writing, I have referred to it in terms of "subjective" versus "objective" self-awareness (Lewis & Brooks-Gunn, 1979; see also Duval & Wicklund, 1972) or the machinery of self versus the idea of "me" (Lewis, 1994). It is clear that the adult self is made up of a variety of different aspects, functions, and structures, which only occasionally work in harmony. In fact, in the psychoanalytic construction of personality and selfhood, it is the intrapsychic struggle between aspects of the self that accounts for the dynamic as well as the psychopathological features of human behavior. Indeed, the idea of multiple features or aspects of the self is perhaps the most widely held belief for all who are interested in the self (see Wylie, 1961, for a more historical review of this problem). Given this idea of a multi-aspect adult self, how are we to treat the idea of the development of self?. From a developmental perspective, not all these aspects exist at birth or even develop together at the same time. If they did, there would be little that develops. Thus, it is essential when studying the development of the self that we first agree to the general principle that the term self in and of itself imparts little meaning because it does not specify particular aspects of a self. If investigators talk about the existence of a self at birth or even at 3 months (Gopnik & Meltzoff, 1994; Watson, 1994), they mean something very different than what others might mean when they talk about the self as evolving in the middle of the second year of life (Lewis, 1992; Lewis & Brooks-Gunn, 1979). What I should like to discuss in this essay first is the need to make a distinction between different aspects of the self and to argue that the aspect that we adult humans refer to as ourselves is, in fact, a rather unique aspect of self; one that we share with few other species (the exceptions being great apes and, perhaps, porpoises and whales). This aspect of the self develops somewhere toward the middle of the second year of life (Lewis & Brooks-Gunn, 1979; Lewis, 1994). It may grow out of other aspects of the self that appear earlier, or it may have little connection to them ~ being related only as part of a developmental function of emerging skills associated with maturational processes. More important, however, is the need to make clear, both in our conceptions and language, that the functions of this late-maturing aspect of self not be assigned to earlier aspects of the self. Unfortunately, this confusion appears throughout the infancy literature. This

98

MICHAEL LEWIS

problem arises in many areas and has to do with the inclination to attribute to the very young infant elaborate mental states. Take, for example, the distinction between intentionality and intention. Intentionality refers to goals aa~d processes built into a system that may not require any elaborate mental state. Intentions are mental states. For example, a flower turning toward the sun a tropism might be said to have intentionality, but the flower does not have intention as a mental state. Likewise, an infant's mimicking of a tongue protrusion of its mother, or an infant's smiling after its mother smiles, may reflect intentionality m something goal-directed and organized m but not the mental state of intention. Thus, we need to distinguish between two types of intention; those without any "mental state about its goals" and those with (Roitblat, 1990). In order to try to understand the term self, I will first consider it from a systems perspective. This may enable us to differentiate some of the functions of the early self from those of a later one, for if the self can be thought of as a complex system with interrelated aspects and functions, the assumption underlying early functions may not require us to invoke mental concepts or states, such as intentions, or complex schemas (also mental states), such as a loving mother. Having achieved the idea of articulating a self-system, I will try to show how some of our conjectures of the early self rely too much on assuming mental states that can only be found in the later self. This attempt to distinguish between various aspects of the self will enable us to explore the developmental course that leads from the machinery of our early self-system (something I have called the "subjective self') to the idea of "me" (something I have called the "objective self').

The Self-System When we consider the idea of a self-system, we need to focus on the general features of systems. By considering the self as an exemplar of a general system, we may be able to shed some light on some of the conundrums that face us as we try to understand what selves mean. Following Von Bartalanffy (1967), systems consist of: 1) elements; 2) their interactions; 3) the effect on the system of changes in any element, the effect of one element on another; 4) the assumption that the system is greater than the sum of its parts; 5) the assumption that every system has goals; and 6) the assumption that systems adjust to change. 1. Elements. The self is made up of various aspects or elements. It is only a unity in the sense that it is a system. Exactly what these aspects may include is open to question. For example, we could look at such aspects as consciousness and unconsciousness, or following Freud, the id, ego, and superego. In addition, we could focus on functions of self, including the ability to discriminate self from

ASPECTS OF SELF

99

other, the ability to recognize the self, the ability to evaluate the self, and the knowledge the self has about its self and other selves. Other aspects of the self could be added: James' religious self, the self of beliefs and ideas, for example (James, 1890). Aspects or elements can be viewed from a biological-structural perspective, something that is necessary if we are to include in the self-system the activity of the body itself; for example, the regulation of temperature, the metabolism of food, and oxygen adjustments. Increasingly, we accept the idea that the human brain is made of aspects that can act separately, yet exist within the same body.. Thus, for example, the activity of the amygdala and what it can or cannot learn in some cases is independent of what the neocortex itself knows (LeDoux, 1990). Or the case of the split-brain patient who is aware haptically of the number that he feels, but verbally and consciously cannot tell us what it is (Gazzaniga, 1985). Finally, the example of the idiot savant also suggests the nonunity of the self and brain function. The hyperlexic child of 2 years who can read like an adult, but who can understand nothing of what she reads is an example of this (Lewis, 1985). The idea of a modular brain has to support the idea that the self is made up of elements rather than a single unity or thing. Moreover, these aspects or elements of the self have different developmental courses; some may exist from birth, and some will appear over the course of the early years of life. Some elements even continue to develop across the whole life span. 2. Elements Interact. It is the nature of all systems that the elements that make them up are in communication with each other. The form of that communication may not be known, but they are in communication. Within any living system there needs to be communication between the elements of that system. This can include a unit as small as a cell, a plant or animal, or even a more complex organism. This means that the elements are aware of each other. This awareness must be differentiated from the awareness adult humans have about themselves. It is a communication awareness in much the same way that my thermostat is aware of the temperature in the room, or my T-cells are aware of the foreign protein they rush to attack. Thus, to confuse us further, the idea of selfawareness needs to be broken down into at least the awareness that the various aspects of the self have about each other, including body awareness, and the awareness we attribute to an adult when we hear her say that she is aware that she is wearing a red dress. It is also the case that the various elements of the system have to be able to identify each other as belonging to the same system, much like the T-cell has to distinguish between foreign and nonforeign protein. Self-other differentiation has to be a part of how all systems work. This capacity may be built directly into the system, needing no experience for its development. It is likely to be built into the

1O0

MICHAELLEWIS

operating rules through some, as yet undefined, biological principle. On the other hand, the operating rules may be built into the system, but need experience to elicit their function. The work of Watson (1994) suggests that the functional selfother distinction is the consequence of experience; in particular, the contingency between the actions of the infant and the infant's environment. This analysis opens the possibility that at least some aspects that make up the system are organized through experience, and others may need little experience to operate. As I sit here writing, the elements of the system I call " m y s e l f ' are regulating: My body shivers as the room cools, and my blood sugar level changes, giving information to me (thus forming a mental state) about the necessity to put on a sweater or to have dinner. Regulation m the interaction of the elements in a system m is a property of all living matter. The notion of self-regulation in the infant, therefore, need make no assumption about mental acts or intentions, although it is intentional behavior. This notion of self-regulation is another problem about the use of the self word. When we use the phrase self-regulate, do we gain any better meaning than if we use the term regulate? Implied in selfregulation is the idea of intention or a mental act. Do we need such a term or belief about mental acts in so young an organism? And if we do, where would such mental acts come from? They might be built into the system, but then there would be little work to be found in development because they exist from birth. But systems regulate and have intentionality, without necessarily having intentions or mental states. Thus, to say self-regulation may trap us into believing that the system has intentions. Infants regulate and have intentionality; they may not have intentions! Thus, one could say that all systems self-regulate, thus implying nothing at all about a self-that-intends (mental states), or we can mean that at the systems level intention exists, but it is not the same kind of intention that exists in the human adult. The intention implied in self-regulation is more like that of a green leaf turning toward the sunlight. If we choose to call it an intention, it is the lowest level of intention, but it is not the same intention as that of an adult who has an intention to write a paper or finish a sentence (Lewis, 1990b).

3. Mutual Affects. Elements of any system must affect one another. What happens to one element of the self-system impacts on others. The impact may be quite direct, or the impact may be indirect. The shivering of my body in a cold room sets off mental states, including knowledge of why my body is shivering and an intention to put on a sweater. An indirect effect may be found where one aspect of the self informs another aspect without the latter knowing anything about the former. For example, the amygdala conditioned to a loud noise stimulus may produce a state of fear, while at the same time the organism does not know why it is fearful (LeDoux, 1990). Nevertheless, that element of the system that is fearful must affect the element of the system that forms a mental state such as found in

ASPECTS OF SELF

101

the verbal expression I am fearful. The idea of the unconscious influencing one's actions or thoughts is too well known to need elaboration here. It is the unfortunate case that we know too little about both the elements and how it is that these elements affect one another. Our task, in fact, is nothing less than the articulation of the elements and the exploration of how they might interact and affect one another. It would be an exercise not unlike Freud's attempt to explain how the unconscious affects conscious behavior or how conscious behavior affects the unconscious (although I might add that Freud had little use for the idea that the conscious could affect the unconscious). 4. The sum is greater than the parts. It seems clear that if we wish to argue that there are multiple elements, then it follows that understanding one element will not result in our understanding the self-system itself. To understand the self is to understand all of its aspects and how they work together. 5. Systems have goals, and goals require action. In order to act, it is necessary for the system to be able to distinguish between its self and other selves. Whether this ability is learned (see Watson, 1994) or, as others have suggested, part of the very process of a c t i o n - including perceiving, feeling, and thinking is still debatable (see Butterworth, 1990). What appears to be necessary if we take a systems approach is that organisms cannot act without being able to distinguish between self and other. Thus, one feature of the goal structure of a system is its necessity to distinguish itself from other. On a biological level, this may be between cells; on a human level, it may be a function of preformed categories, or it may be learned. The self-other distinction is an essential and necessary feature of systems in that all systems need to reorganize their boundaries. In the human self-system, what def'mes the boundary of the system itself is in part determined by human biology and in part by the cultural rules. The "I-self' of the West, a concept of the self as a single bounded entity in opposition to all other such entities, or the "we-self' of the East, where the self is seen only in a nexus; is determined by the rules of the particular culture (Geertz, 1984). Another goal of any system is to preserve itself. Preservation of self at a biological level includes the goals associated with regulation. Self-system goals operate at other levels as well. Nozick's (1981) row boat analogy is an example of the attempt to preserve the idea of ourselves. In the Nozick example, a board of the boat is replaced each year. After 50 years, all the boards have been changed, yet the boat still remains the same. In an analogous fashion, one might say that even though I look, feel, and think differently at different times, it is always still me. There is ample evidence from the social psychological literature to convince us that we tend to reconstruct our past so that the narrative that we construct is consistent with what we are now (see M. Ross, 1989). Identity, the idea of myself

102

MICHAELLEWIS

as myself, in spite of the changing sell is part of the psychological requirement of a human self to maintain the self. 6. Changing systems. All living systems change, yet they also require constancy. Aspects of a system, especially a developing one, must change, yet maintain themselves as a system. The self-system that exists in infancy has a limited number of elements, yet, with development, the self-system becomes more complex. Through a multitude of processes, including retaining old aspects, transforming others, and adding new ones, the immature system becomes mature and, at the same time, maintains its self. The processes of change result in disequilibrium and equilibrium, through which the system is able to maintain itself. Such an analysis of the general properties of a system as applied to the self points out that complex systems may contain such elements as awareness, communication, and intentionality, without necessarily involving any mental states. That the self-system eventually develops mental states and, in particular, the mental state of the idea of "me," which then leads to the idea of others also having a melike-me, does not mean that the infant from the beginning has such states. The distinction between the machinery of myself (the system properties) and the idea of "me" (a mental state) needs to be made.

Myself and Me We may best be able to approach the topic of what a self may be by considering a distinction that I find useful. What may be needed is to clarify the distinction between what I will call the "system properties" of the self: something that many, if not all, living organism possess, including young infants; and the idea of "me": a mental state that emerges slowly in the human young, possibly as a function of frontal lobe maturation, and which is likely to exist (although not in a human form) in the great apes. As I sit here in my room, I have no trouble recognizing myself. I know where I am and why I am here. I can tell the way I smell and when I speak; I can hear my voice. The sun's warmth through the window is comfortable. Sitting here, I can think about myself when I stop writing, and I can wonder whether I will find my way to a friend's house tonight. I wonder about my appearance. Is my hair combed properly? As I get up and go to the mirror, I see myself. I see the reflected surface of my being. "Yes, that is me," I say, fixing my messy hair. I know a great deal about me. One of the things I know is how I look: for example, that there is a scar above my left eyebrow. I look familiar to myself, even though I have changed considerably with age. Pictures taken of me 20 or 30 years ago still look like me

ASPECTS OF SELF

103

to me, even though I know that when I look at myself in the mirror, I will not look as I did then. This concept of self is an idea, a mental state that is a particularly powerful one for me; it is an idea with which I cannot part or forget. It is one around which a good portion of the network of many of my mental states center. This is not to say that what I know about myself is all I know. In fact, this idea of myself is only one part of myself; there are many other parts of which I do not know. There are the activities of my body. There are other parts beside my body of which I do not know. True, I may know them as ideas, but I cannot find them or locate them in myself. I have no knowledge of a large number of my motives: organized coherent thoughts and ideas that we sometimes call unconsciousness, and that control large segments of my life. I have no knowledge of how my thoughts occur or why I feel one way or another. Nevertheless, I know that I think and feel, even without this knowledge. The claim has been made that it is possible to know of all things related to the self. Take, for example, the yoga's belief in the control of much of our autonomic nervous system functioning. Although it may be true that I could know more of some parts of myself if I chose to focus my attention on them, it is nonetheless the case that what is known by myself is greater than what I can state I know. If such facts are true, it is fair to suggest the metaphor of my self. I imagine myself to be a biological machine that is an evolutionarily fit complex of processes: regulating, growing, feeling, thinking, planning, and learning. One aspect of this machine is the idea of "me." An aspect of this machine that knows is that it knows itself, but also knows that it does not know all of itself. The me that recognizes the me in the mirror, is part of that machinery and is likely located in the frontal lobes (Weiskrantz, 1986). The self, then, is greater than the me, the me being only a small portion of myself. This idea can be best understood from an epistemological point of view. The idea that I know is not the same as the idea that I know I know. The "me" aspect of the self that I refer to is that which knows it knows. The failure to make the distinction between these features of self can be blamed for much confusion when studying the issue of development. The distinction between self and me or between knowing and knowing I know involves two aspects of me. If we do not confuse knowing with knowing I know, then the argument around the issue of the developmental sequence in self becomes clearer. As I have already suggested, many features of the self exist early and exist as part of the system from birth or soon after; the idea of "me" - - the knower who knows - - is not developed until somewhere in the middle of the second year of life (Darwin, 1872; Lewis, 1992). We have already given some attention to two early features of the self. These are self-other differentiation and self-regulation. They are likely to be part of the

104

MICHAELLEWIS

machinery of myself and not related to the idea of "me." Certainly, by 3 months, and most likely from birth, the infant can differentiate itself from an other. Selfother differentiation also has associated with it a type of self-awareness. It is the self-awareness of elements of a system in communication with one another. This type of recognition and the self-other differentiation are part of the hardware of any complex system. T-cells do recognize and differentiate themselves from foreign proteins. The newborn infant recognizes and responds appropriately to intersensory information. Therefore, we should not expect that these aspects of self are the differentiating features when we compare wildly different organisms. All organisms, as systems, should have these capacities. What may distinguish organisms in regard to their system organization is the complexity of the machinery of these systems. What may distinguish humans from most other living organisms is not the functions of the human system, but the ability to have mental states and, more specifically, the mental states related to the idea of "me." The ontogenetic and phylogenetic coherence found to date supports the idea that in order to understand the concept of self, we need to disentangle the common term self into at least two aspects. These I call the machinery of the self and the idea of "me." They have been referred to by other terms; for example, objective self-awareness, which reflects the idea of "me"; and subjective self-awareness, which reflects the machinery of self (Lewis, 1990b, 1991, 1992). The same objective-subjective distinction has been considered by Duval and Wicklund (1972). In any consideration of the concept of self, especially in regard to adult humans, it is important to keep in mind that both biological aspects exist. There is, unbeknownst to us most of the time, an elaborate complex of machinery that controls much of our behavior, learns from experience, has states and affects, and affects our bodies, most likely including what and how we think. These processes are, for the most part, unavailable to us. What is available is the idea of "me," a mental state. What is particularly impressive is the recent research on brain function, whose findings point to the possibility that different areas of the brain may be associated with different functioning. Thus, both the machinery of the self and the mental state involving the idea of "me" appear to be the consequences of different biological processes and locations. For example, the recent work by LeDoux (1990) points to specific brain regions that may be responsible for different kinds of self-processes. Working with rats, LeDoux found that even after the removal of the auditory cortex, the animals were able to learn to associate an auditory signal with a shock. After a few trials, the rats showed a negative emotional response to the sound, even though their auditory cortex had been removed. These findings indicate that the production of a fear state is likely to be mediated by subcortical

ASPECTS OF SELF

105

regions, probably the thalamic-amygdala sensory pathways. Similar findings have been reported in humans, which suggests that states can exist without one part of the self experiencing them. Weiskrantz (1986), among others, has reported on a phenomena called "blindsightedness." Patients have been found who lack the visual cortex, at least in one hemisphere. When they are asked if they can see an object placed in their blind spot, they report that they cannot see it; that is, they do not have the experience of the visual event. The self reflecting on itself, my recognition of what I know, the "me" B the mental state - - in fact, does not see. When, however, they are asked to reach for it, they show the ability to reach, at least some of the time, for the object. Thus, they can "see" the event, but cannot experience their sight. These findings, as well as Gazzaniga's work (1988) on split, brain patients, suggest that separate brain regions are responsible for the production and maintenance of both the machinery of self-processes and the mental state of the idea of "me." A similar analysis involving memory has been suggested by Tulving (1985). This idea of the machinery of the self, or subjective awareness, versus the mental state, or the idea of "me," an objective awareness, can best be seen in emotional life. I have tried, in the past, to distinguish between emotional states and experiences and have argued that adults can have emotional states and yet have no experience of them (Lewis, 1990a; Lewis & Michalson, 1983). This distinction is especially true for the infant if, by experience, we refer to a mental state about the self. Thus, for example, if I say, "I am happy," I mean by that statement that I am in an emotional state of happiness and I can experience that state. The young infant can be in a state, but may not have an experience of that state. Emotional states, therefore, refer to objective self-awareness, or the machinery of ore system. This machinery can have goals, can learn and profit from experience, can control functions, and can react to events, including people. The experiences of our emotional states refer to objective self-awareness. The idea of different aspects of the self are clearly necessary, given the data of our adult lives. The question is whether such distinctions make sense in considering the development of self. An interesting example from the adult literature is given by Pribram (1984), who describes a patient in whom the medial part of the temporal lobe, including the amygdala, had been bilaterally removed: I once had the opportunity to examine some patients in whom the medial part of the temporal l o b e - including the a m y g d a l a - had been removed bilaterally. These patients, just as their monkey counterparts, typically ate considerably more than normal and gained up to 100 pounds in weight. At last I could ask the subject how it felt to be so hungry. But much to my surprise, the expected answer was not forthcoming. One patient who had gained more than 100 pounds in the several years since surgery was examined at lunchtime. 'Was she hungry?' She answered, 'No.' 'Would she like a piece of rare, juicy steak?' "No.' 'Would she like

106

MICHAELLEWIS a piece of chocolate candy?' She answered, 'um-hum,' but when no candy was offered she did not pursue the matter. A few minutes later when the examination was completed, the doors to the common room were opened, and she saw the other patients already seated at a long table eating lunch. She rushed to the table, pushed the others aside, and began to stuff food into her mouth with both hands. She was immediately recalled to the examining room, and questions about food were repeated. The same negative answers were obtained again, even after they were pointedly contrasted with her recent behavior at the table. Somehow the lesion had impaired the patient's feelings of hunger and satiety, and this impairment was accompanied by excessive eating! (Pribram, 1984, p. 25)

Here we can see a clear distinction between the subject's objective self-experience and her subjective state of hunger.

Fallacies and Confusion in Distinguishing Between Mental States and the Machinery of Self In 1960, when research in infancy first started, the general proposition underlying the beliefs in an infant's ability rested on the assumption (following William James, 1890) that the infant was a blooming mass of confusion; that, in fact, the infant was an incompetent, immature adult. In the course of the next 35 years, ample discoveries have led us to believe that the infant is a highly capable organism. This is predicated on the discoveries that infants' perceptual abilities, although immature, are quite competent and that infants are capable of gathering information from their environments. Moreover, ample research led to the belief that infants can learn, have memory, and profit from their interactions in the world. Finally, it was shown that complex abilities (Bower, 1974), rather than simple schemata, as originally proposed by Piaget (1955), did exist in infants, and so they were capable of intersensory integration (Spelke, 1976), imitation (Meltzoff & Moore, 1977), and complex learning (Rovee-Collier, 1987). Moreover, the belief in deep structures, first from Chomsky's idea of innate grammar (1965) and, more recently, to Leslie's, Frith's, and other's belief in a grammar of meaning (Baron-Cohen, Leslie, & Frith, 1985; Leslie & Frith, 1988, 1990), have led us to assume incorrectly that there exists in infants from the beginning an elaborate and complex set of mental states and accompanying actions. Thus, in the last 35 years, we have moved from the belief of the very young infant as an incompetent organism to a stance where the infant is perceived as a miniature adult. The reasons for such a shift are unclear, but the evidence can be found everywhere. For example, consider a recent paper by Acquarone (1992), discussing a 3-month-old infant with cerebral palsy:

ASPECTS OF SELF 107 The baby came in hanging on mother's left arm, looking at the therapist very briefly but intently and intelligently, even though her body was lifeless, thus giving hints of a potential to develop and a wish to link with the therapist. We sat on the floor. Baby, with a twisted, half-fallen head, looked briefly at the therapist, half-smiled, and then looked startled. The therapist took it as some faint wish for contact in a very uncoordinated way. Thereafter, the therapist held both stiff little hands and helped her clench her fist, which produced another twisted look, and half-smile. Holding the baby's hands, the therapist talked gently about how nice it feels to be touched and firmly held and the baby jerked uncoordinatedly and stiffened. The therapist noticed the baby's overall joy and that the infant wanted her legs touched and massaged, even though the baby growth prevented direct contact . Mother asked the therapist what she was seeing in the child. The therapist observed that the baby had some determination in linking with people, with her, but she needed to go through stages such as finding her own body boundaries, her identity, her mother's. Mother asked whether she could help. Mother's request for advice on how to help the baby was answered by observations about the baby being looked at, talked to, being held, and closing her hands, having her legs and body touched and massaged (p. 47; italics added).

Here we see in the therapeutic situation the writer suggesting complex mental states that this h a n d i c a p p e d 3 - m o n t h - o l d infant possessed. The child's m o t o r behavior is interpreted so as to reflect mental states, unlikely to occur in a child of this age. Similar difficulties can be found elsewhere. For example, Crittenden (1994), in talking about the organization of infant behavior, writes: For example, an infant might conclude 'When I signal how I feel, my mother behaves in ways that are comforting,' rather than, 'When I cry, my mother picks me up.' Under less auspicious conditions, an infant might encode the more complex and discouraging information that, 'When I signal, my mother rejects me,' or 'When I am quiet, she does not bother me.' These examples show how behaviors can be perceived and encoded in procedural models as a function of classes rather than reflexes or links between specific self and other behaviors. These inferred mental states would seem to exist as basic properties of a self, even as the self develops and moves toward mental state capacities, such as bound in the experience of emotion, or as found in the idea of me (p. 82; italics added).

Also consider these mental states taken from the same paper: "For example, a securely attached child might have the paired generalizations that, 'My mother loves and understands me,' and 'I am lovable and capable o f being understood'" (p. 83; italics added). Crittenden, like m a n y others, clearly is willing to attribute these complex mental states to young infants, "even as the self develops." Thus, before the idea of "me" occurs, the infant has mental states that need the idea of "me." This type of logical error is found throughout the literature. Stem's idea about the development of self (1985) contains this same difficulty. He states that, "There is no confusion between self and other in the beginning, or

108

MICHAELLEWIS

at any point during infancy" (p. 10). This is not a problem because, as I have tried to indicate, this may be a property of all systems; thus, self-other differentiation could exist from the beginning. However, this systems idea is not what Stern is addressing. In fact, he is willing to give the newborn complex mental states; the ability to experience itself as a mental state: "I am suggesting that the infant can experience the process of emerging organization as well as the result, and it is this experience of emerging organization that I call the emergent sense of self. It is the experience of a process as well as a product." (p. 45). For Stern, the self, even from the beginning, has extraordinary mental capacities. It can view itself coming into existence! I think that there are some difficulties with such a view of selfdevelopment. First, it must rest on the belief that infants are not only highly capable of actions, including perceptions, thoughts, and learning, but of complex mental states concerning themselves and their existence. Although the infant has been shown to be highly capable in terms of some early capacities, these appear to be more reflexive in nature than cognitively based. There is no direct evidence of such mental states, nor are we ever informed as to how we might show there is. Moreover, this view of the self does not allow for much self-development, certainly not the development of the mental states of self. If, by birth, the child is capable of experiencing the self, then this self is a self capable of objective selfawareness at the same time that the self is coming into existence. It experiences itself being created. Such a phenomena is hard to understand, and we are offered little support for it. This difficulty, especially as it relates to self, has historically appeared before. Because of this lack of objective awareness, it is difficult to see how Stem can ascribe to the infant mental states dependent on self-awareness. He explicitly sees the infant capable of experiencing its own emergent properties. This problem has been viewed by others (e.g., Kemberg, 1976; Lacan, 1968). They all claim that the young infant is capable of experiencing its emergent self and, therefore, of experiencing anxiety over its nonexistence. This is similar to Otto Rank's (1929/1952) notion of birth anxiety. Freud, in his critique of this view, rightly points out that anxiety is a signal, and as such, has to be experienced. Only the ego can experience: "The id cannot be afraid as the ego can; it is not an organization and cannot estimate situations of danger." (Freud, 1936/1963, p. 80). Because the ego emerges only slowly, certainly not at birth, there can be no objective experience. The problem resides in the fact that any anxiety over nonexistence or experiencing the emerging self, as an adult might experience it, cannot occur to an organism that has no objective self-awareness or mental state of its own existence. It is not possible for an organism to experience itself or to be anxious about its existence prior to the capacity to think about itself as existing; that is, prior to being able to experience itself and prior to its being able to imagine its nonexistence.

ASPECTS OF SELF

109

The core self and the issue of intersubjectivity, also as discussed by Stern (1984), present us again with the same type of problem. Intersubjectivity, as he and others have defined it, is related to objective self-awareness; that is, I have mental states of myself and the other self and the connection between them. However, it might be possible to have intersubjectivity without a mental state if we think of intersubjectivity as contagion; for example, smiling when another smile appears. This social contagion, like when one bird flies off, all others on the telephone wire do so, or when laughter produces laughter, does not rest on mental states: none need be evoked. Under such a definition, intersubjectivity becomes a set of complex behavior patterns that are triggered by other behaviors. Intersubjectivity thus may be less controlled by mental states, and more by simple rules, such as circular reactions, as proposed by Piaget. If it is not based on intentions in terms of complex means and representations, it may be more like automatic social responses. Thus, intersubjectivity between a mother and her 6month-old can take place as a function of complex patterns systems, one that may be present in any species. However, intersubjectivity as discussed by Stem, which involves a mental states based on the ideas of "me," "you," and "our relationship," should not be possible at this age. Certainly, there is no support for it, and the confusion between mental states and contagion is not often even tested. Part of the problem is, as I have already indicated, one of what we might mean when we use such terms as awareness, intention, or intersubjectivity. It is the confusion between the machinery of the self leading to complex patterns of social behavior and the development of mental states. Self-awareness can be used simply to reflect the machine's capacity to monitor aspects of itself. Such capacity does not need a mental state. Unfortunately, often self-awareness is meant to imply a mental state. It is this confusion in our terms and in our conceptions that leads us to assume that the behaviors in early infancy reflect analogous, if not identical, processes and functions, which we usually assign to adults. This problem of confusing machine self with the idea of "me," a mental state, leads to many difficulties. While the interactive behavior between a 3- to 6month-old infant and its mother might be viewed as an early example of intersubjectivity (i.e., the ability of the family members to share experiences to match, align, or attune their behavior to each other), it might, on the other hand, reflect much simpler processes. These processes may not involve mental states, but instead, may indicate simple rules of contagion or attention-getting and holding. For example, it is more efficient to stop moving or talking when someone else is talking. Paying attention requires no mental state, but only the interactions between a dyad, one of whom possesses complex mental states and the other, biological capacities.

110

MICHAELLEWIS

Thoughts About Selves I have tried to argue that our language when we think of the self often fails us. Terms like awareness, intention, and regulation, and ideas about self-other impart important information because each of these aspects (or if one prefers, functions or skills) can occur in very different ways. However, if we differentiate them, a true developmental model is possible. I will restate some basic principles articulated throughout the early part of the paper. 1. All living systems self-regulate. By this we mean that within any living system there needs to be communication between parts of that system. It we can call awareness, but not the mental state of awareness. This can include a unit as small as a cell, a plant or animal, or even more complex organisms. As I write, my systems are regulating my temperature or regulating my blood sugar level. Regulation is a property of living matter. Regulation makes no assumptions about objective self-awareness or intentions, although there is intentionality in the process. 2. In order to act, it is necessary for organisms to be able to distinguish between self and other. Whether this ability is learned, as Watson suggests (1994) or, as others have suggested, part of the process of action N including perceiving, feeling, and thinking is unknown (Butterworth, 1990); what appears to be so is that no organism can act without being able to distinguish between self and other. The ability to regulate or to distinguish self from other is part of the machinery of all living systems (Von Bartalanffy, 1967). 3. Even higher-order functions such as perception and mental states, and complex actions, such as driving a car, can be carried out by adult humans without objective self-awareness; that is, without their being able to look in and observe the processes that allow these behaviors to be carried out. I cannot watch myself think. I can only look at the product of my thinking. 4. A unique aspect of some self-systems is objective self-awareness. By objective self-awareness I mean the capacity of a self to know it knows or to remember it remembers. It is this mental state that we refer to when we say self-awareness. The capacity of objective self-awareness may be uniquely human (although the great apes, porpoises, and whales appear capable of this, as well; Lewis, 1994). 5. Specific developmental processes of the self follow the general principles of development. Earlier capacities, such as the machinery of the self, may give rise to later capacities, such as mental states (e.g., the idea of "me"), but are not transformed. Furthermore, both capacities exist once the latter emerges. Thus, unlike a more classical genetic epistemological

ASPECTS OF SELF

1 11

approach, I see the retention of earlier structures or functions as not only possible, but as a necessary aspect of development. In some sense then, old structures in interaction with the environment and/or as a function of maturation give rise to new structures. These new structures do not replace the old ones, but coexist with them. Under certain conditions, individuals will utilize the most mature aspect they have achieved. However, this does not mean that other aspects are not utilized. In some sense then, mature adults possess within their repertoire all aspects, whereas younger children possess only those aspects already achieved. The end result of any developmental trajectory is the existence of all aspects. In fact, once the various aspects have emerged, they are likely to be elaborated over the entire life course; not only do they not replace one another, but they are likely to continue to develop (Fischer, 1980). Finally, similar types of behavior which may or may not be related to the same aspect - - exist in adults as in infants. For older children - - perhaps past 2 to 3 years of age - - and adults, both aspects of self exist and are used. The aspect used by adults may be valuable in understanding their development. We recognize, then, that there are different aspects of self. Objective selfawareness, an occurrence that we have marked as taking place in the middle of the second year of life (Lewis & Brooks-Gunn, 1979), joins the subjective self, which either exists at birth or develops soon after. Objective self-awareness does not replace subjective self-awareness, but coexists with it. Thus, adults are capable of functions that involve both objective and subjective self-awareness. In fact, as the chapters in this volume will attest, the objective self can even think about the subjective self.

The Intentional Stance When I started studying psychology over 35 years ago, human behavior was explained without the evocation of mental states or actions. The dominant theories concerning knowing were based on learning, either classical Pavlovian conditioning or operant conditioning. For the most part, both of these theories had little use for mental states or acts and preferred instead the idea that the complex behavior of humans could be explained by such ideas as associational learning, combined in some fashion to create complex behaviors. Skinner, for example, showed that pigeons could be taught to play ping-pong through the conditioning of very small behaviors, rather than through mental states such as intentions, plans, memories, or schemas.

112

MICHAEL LEWIS

Somewhere around the middle of the 1960s, there developed in psychology a paradigm shift. At this point, mental states, cognitions, and schemata reentered into our explanations. The old learning models, which did not rely on mental states or even cognitions, disappeared, and in their place, cognitive science developed. Mental states, cognitions, and the idea of mind have now ahnost totally replaced the learning paradigms that have all but disappeared. Thus, from a historical point of view, we can see in the last 30 years a shift in our world view, a movement from mechanistic theories to theories of mind. As in all world views, this new paradigm replaced the old, not because the old view was found wrong or the new view could be proven correct, but because of the paradigmatic shift (Kuhn, 1962). Given that we now accept the possibility of mental states, we are faced with the perplexing question of how to assign them. For example, let us take the mental state of intentionality. We can all agree that nonliving matter does not have intentions. So far, there is agreement. There also is agreement that plantlike living things do not have mental states and therefore do not have intentions. Although agreement is great, there are people willing to assign some mental states to plants. Not surprisingly, these people tend to like plants and have many of them. Some believe that the plant "knows" if you love it, and they believe that the plant will thrive if it is loved by a person. From plantlike life on, we begin to have serious disagreement. If we are prepared to give mental states to animal life, then to which form of animal life are we prepared to ascribe them? Single cells like T-cells? Amoebae? Do insects have mental states like intentionality? Bees? As we can readily see, the assignment of mental states to any living organism varies with the world view taken. It is clear that some of us are prepared to assign mental states to newborns while others are not. It is not possible to prove either view incorrect because each rests upon a different world view. What is possible to do is to question the developmental models derived from such world views and to make sure that each model is logical within its own system. From a logical point of view, we need to question any system that assigns mental states that can "observe" the same mental state developing. The idea that the self can watch the self develop is an example of the illogic often underlying some notions of the development of mental states. Furthermore, from a developmental point of view, if we give the newborn or very young infant mental states like those of adults we are forced to ask ourselves, what then develops? The world view that assigns mental states or their analogy to the newborn allows for little development because the adult form of a mental state or something similar is there from the beginning. Such a world view must stand in opposition to the idea of development.

ASPECTS OF SELF

1 13

One of the problems that I believe we all experience in thinking about the young infant's self, and one that often leads many to believe that the self can experience itself come into being, is that we do not remember what the self not thinking about the self is like. I believe we can approximate what this might be like by viewing ourselves in a slightly different manner than we usually do. Imagine, if we can, what the self is like when we are completely engrossed in some activity, like watching an exciting movie, listening to a concert, or engaged in finishing a chapter. At such times, we lose the idea of "me" because our attention is tumed away and because it is focused elsewhere. At these times, we are in a state, perhaps, as Csikszentmihalyi describes, of flow (1990). We have no sense of ourselves, and we lose where we are, what time it is, and even who we are. Even so, we are still able to carry out complex tasks, learn, and solve problems. Thus, the idea of "me" is not required for us to behave. This state is what I imagine exists for the infant prior to the idea of "me" emerging. Finally, I return to Darwin's (1872/1969) observations made over 100 years ago. He, too, saw that mental powers and states were necessary for the idea of "me." He, too, saw that these did not exist at birth or even soon after, but developed, and appear in the second year of life.

REFERENCES

Acquarone, S. (1992). What shall I do to stop him crying? Psychoanalytic thinking about the treatment of excessively crying infants and their mothers/parents. Journal of Child Psychotherapy, 18(1), 33-56. Aristotle (1953). Ethics. (translated by J. A. K. Thomson). London: Penguin Books. Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Mechanical, behavioural, and intentional understanding of picture stories in autistic children. British Journal of Developmental Psychology, 4, 113-125. Bower, T. R. (1974). Development in infancy. San Francisco: Freeman. Butterworth, G. (1990). Origins of self-perception in infancy. In D. Cicchetti & M. Beeghly (Eds.), The self in transition: Infancy to childhood (pp. 119-137). Chicago: University of Chicago Press. Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press. Crittenden, P.M. (1994). Peering into the black box: An exploratory treatise on the development of self in young children. In D. Cicchetti & S. Toth (Eds.), Rochester symposium on developmental psychopathology, Vol. 5: The self and its disorders (pp. 79-148). Rochester, NY: University of Rochester Press. Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience. New York: Harper & Collins. Darwin, C. (1969). The expression of the emotions in man and animals. Chicago: University of Illinois Press. (Original work published 1872) Duval, S., & Wicklund, R. A. (1972). A theory of objective self-awareness. New York: Academic Press. Fischer, K.W. (1980). A theory of cognitive development: The control and construction of hierarchies and skills. Psychological Review, 87, 477-531.

114

MICHAELLEWIS

Freud, S. (1963). The problem of anxiety. New York: Norton. (Original work published 1936) Freud, S. (1959). The ego and the id. In J. Strachey (Ed. and Trans.), The complete psychological works of Sigmund Freud: Vol. 19 (pp. 3-66). London: Hogarth Press. (Original work published 1923) Gazzaniga, M. S. (1985). The social brain: Discovering the networks of the mind. New York: Basic Books. Gazzaniga, M. S. (1988). Brain modularity: Towards a philosophy of consciousness experience. In A. J. Marcel & E. Beseach (Eds.), Consciousness in contemporary science (pp. 218-256). Oxford: Clarendon. Geertz, C. (1984). "From the native's point of view": On the nature of anthropological understanding. In R. A. Shweder & R. A. LeVine (Eds.), Culture theopy." Essays on mind, self, and emotion. (pp. 123-136) Cambridge, MA: Cambridge University Press. Gopnik, A., & Meltzoff, A. N. (1994). Minds, bodies and persons: Young children's understanding of the self and others as reflected in imitation and theory of mind research. In S. T. Parker, R. W. Mitchell, & M. L. Bocchia (Eds.), Self-awareness in animals and humans: Developmental perspectives (pp. 166-186). Cambridge, MA: Cambridge University Press. Harding, V., Gray, J. E., McClure, B. A., Anderson, M. A., & Clarke, A. E. (1990). Self-incompatibility: A self-recognition system in plants. Science, 250, 937941. James, W. (1890). The principles of psychology. New York: Holt. Kernberg, O. F. (1976). Object relations theory and clinical psychoanalysis. New York: Aronson. Kopp, C. B. (1982). Antecedent of self-regulation: A developmental perspective. Developmental Psychology, 18, 199-214. Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago: University of Chicago Press. Lacan, J. (1968). Language of the self Baltimore: Johns Hopkins University Press. LeDoux, J. (1990). Cognitive and emotional interactions in the brain. Cognition and Emotions, 3(4), 265-289. Leslie, A.M., & Frith, U. (1988). Autistic children's understanding of seeing, knowing and believing. British Journal of Developmental Psychology, 4, 315-324. Leslie, A.M., & Frith, U. (1990). Prospects for a cognitive neuropsychology of autism: Hobson's choice. Psychological Review, 97, 122-131. Lewis, M. (1985).Gifted or dysfunctional: The child savant. Pediatric Annals, 14, 733744. Lewis, M. (1990a). Thinking and feeling m The elephant's tail. In C.A. Maher, M. Schwebel, & N. S. Fagley (Eds.), Thinking and problem solving in the developmental process: International perspectives (the WORK) (pp. 89-110). Hillsdale, NJ: Lawrence Erlbaum. Lewis, M. (1990b). Social knowledge and social development. Merrill-Palmer Quarterly, 36(1), 93-116. Lewis, M. (1991). Ways of knowing: Objective self-awareness or consciousness. Developmental Review, 11, 231-243, Lewis, M. (1992). Shame, the exposed self. Zero to Three, 7(4), 6-10. Lewis, M. (1994). Myself and me. In S. T. Parker, R. W. Mitchell, & M. L. Boccia (Eds.), Self-Awareness in animals and humans: Developmental perspectives (pp. 20-34). New York: Cambridge University Press. Lewis, M. & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self New York: Plenum.

ASPECTS OF SELF

1 15

Lewis, M., & Michalson, L. (1983). Children's emotions and moods: Developmental theory and measurement. New York: Plenum. Meltzoff, A. N., & Moore, M. K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 75-78. Nozick, N. (1981). Philosophical explanation. Cambridge, MA: Belkings Press. Piaget, J. (1955). Language and thought of the child. (M. Gabain, Trans.). New York: Harcourt Brace. (Original work published 1926) Pribram, K. H. (1984). Emotion: A neurobehavioral analysis. In. K. R. Scherer & P. Ekman (Eds.), Approaches to emotion (pp. 13-38). Hillsdale, NJ: Erlbaum. Prince, M. (1978). The disassociation of a trersonality. New York: Oxford University Press. (Original work published 1905) Rank, O. (1952). The trauma of birth. London: Kegan Paul. (Original work published 1929) Roland, A. (1988). In search of self in India and Japan. Princeton, NJ: Princeton University Press. Ross, C.A. (1989). Multiple personality disorder. New York: Wiley. Ross, M. (1989). Relation of implicit theories to the construction of personal histories. Psychological Review, 96(2), 341-357. Roitblat, H. L. (1990). Causation, intentionality, and cognitive action theory. Psychological Inquiry, 1(3), 263-265. Rovee-Collier, C. (1987). Learning and memory in infancy. In J. D. Osofsky (Ed.), Handbook of infant development (pp. 98-148) New York: Wiley. Spelke, E. S. (1976). Infant's intermodal perception of events. Cognitive Psychology, 8, 533-560. Stem, D. N. (1985). The interpersonal world of the infant. New York: Basic Books. Tulving, E. (1985). How many memory systems are there? American Psychologist, 40, 385-398. Von Bartalanffy, L. (1967). Robots, men, and mind. New York: Brazilles. Von Boehmer, H., & Kisielow, P. (1990). Self-nonself discrimination by T-Cells. Science, 248, 1369-1372. Watson, J. S. (1994). Detection of self: The perfect algorithm. In S. T. Parker, R. W. Mitchell, & M. L. Boccia (Eds.), Self-awareness in animals and humans: Developmental perspectives (pp. 131-148). New York: Cambridge University Press. Weiskrantz, L. (1986). Blindsight: A case study and implications. Oxford: Oxford University Press. Wylie, R. C. (1961). The self concept. Lincoln, NB: University of Nebraska Press.

This Page Intentionally Left Blank

The Self in Infancy" Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

117

CHAPTER 7

Relational Narratives of the Prelinguistic Self ALAN FOGEL

University of Utah

Introduction and World V i e w

During the past ten years, there has been a radical change in me conceptualization of the self in infancy. There is a growing consensus that a sense of self exists long before infants recognize their own image in a mirror, before the acquisition of language or symbolic abilities. The prelinguistic self is believed to be based on the direct perception of the self as a part of a relationship with the physical and social environment. In this paper, I offer a theoretical explanation for the early emergence of self in infancy. This theory is based on a postmodem perspective that+ views the self as a dynamic relational system (Jencks, 1992; Porush, 1991). Rather than borrowing metaphors from the physical sciences to explain psychological phenomena, postmodernism suggests that metaphors from humanistic disciplines may be more appropriate. The self and intrapsychic experiences have been described, from a postmodern perspective, in terms of metaphors such as dialogue (Bruner, 1990; Hennans & Kempen, 1993); drama, narrative, or rhetoric (Goffman, 1974; Sarbin, 1986); life history (Harre, 1988; Sampson, 1991; Shotter, 1984); and voice (Bahktin, 1988; Mitchell, 1988). Dialogue, drama, and history evoke dynamic living processes, imbued with emotion and meaning. Drama and history do not have fixed outcomes or fixed features. Characters who remain rigid under the weight of changing circumstance are seen as tragic and flawed; these characters become fully human only if they have the opportunity to struggle with those flaws and open themselves to change and revelation. The actions of historical figures and dramatic characters are complex, transformational, and surprising. The modernist world view - - based on deterministic and positivistic paradigms m describes the self in terms of mechanistic metaphors derived from the physical sciences. Mechanistic metaphors of the self include a reified notion of individual separateness and the idea of a localized and solid core, the result of a

118

ALANFOGEL

linear accretion of experiences. Mechanistic metaphors include: internalization, introjection, social construction, organization, intersubjectivity, identity, and agency. All of these metaphors evoke a physical process of separating or connecting, building or assembling. One imagines, through these metaphors, a linear string of displacements of action or information, things that collide or spin, lock or unlock, match or mismatch. In this brief account of world views, I am not suggesting that those who take a postmodern or humanistic perspective on the self are superior to those who do not. The essence of the postmodern agenda is the encompassing of multiple world views and multiple perspectives: It is the liberation of many voices into an emergent dialogue of discovery (Gergen, 1994; Jencks, 1992). Indeed, recent accounts of the development of self in infancy have taken to mixing metaphors: at times mechanistic, at times humanistic. Both Emde (1994) and Stern (1985), for example, rely on mechanistic models of the core self while at the same time recognizing the role of dialogue and narrative. I am also not suggesting that mixing metaphors is, in itself, a guarantee of a reasonable theory of the self. Indeed, both of these authors have been criliqued for inconsistencies in their approaches to self and for failures to recognize the clash of world views taking place in their own writing (Cushman, 1991; Fogel, 1993). A theory of the self in the late 20th century requires, at a minimum, an explicit recognition of a multiplicity of world views: Retreats into scientific positivism may be comforting but ultimately unsatisfying as a way to understand the richness of the psyche (Gergen, 1994). Any choice of humanistic metaphors may give an initial appearance of broadening the scope of inquiry into the self, but if it is done without an understanding of the history and intellectual status of such terms within the humanistic disciplines, it loses its scientific potency as a form of explanation. Metaphors can be used as scientific terms subject to verification and systematic research (Fogel, 1993). That research, however, may not be based on mechanistic experimental methods, but on more qualitative approaches to discourse and rhetorical analysis used in humanistic disciplines. A corollary position is that scientific terms m including mathematical models m within positivistic accounts are never more than metaphors: They are not the "real thing"; they are meaningful only within a linguistic-cultural system of scientific discourse, and therefore they can only be analyzed with respect to their effects within the language community of scientists (Fogel, 1993; Gergen, 1994; Oyama, 1993; Shotter, 1984). From a postmodern perspective, therefore, the self is defined in social relational terms and is never fully individuated. The self has multiple voices and themes, and is always part of cultural dialogues with people and objects. I propose that the human self is inherently relational at all ages, and its origins can be found

RELATIONALNARRATIVES 119 at the beginning of life. The self is a living and dynamic dialogical process, a set of loosely connected, nonverbal narrative themes and variations created from the history of communication with people and with things. The Western self objective, detached, stable, independent m can be explained as a narrative form, but it is only one of many narratives available to all selves as part of the dynamics of growing up in a complex social and cultural environment.

Relationships and the Self Dialogical Processes and the Creation of Narrative Before discussing a relational perspective on self, I will review current thinking regarding interpersonal relationship processes. In the view taken here, interpersonal relationships are living systems mutually created through a dynamic process of dialogue between participants. In order for relationships to develop as living systems, the dialogue must be conceptualized as a dynamic process that generates both stability and novelty (Fogel & Branco, in press). Traditional views of interpersonal communication have relied on discrete state models. In these views, communication is parsed into discrete messages that are transmitted across channels to receivers who process the information and produce messages in response. Although this view of communication is useful for particular kinds of research applications, it fails to meet a fundamental requirement for any developing system: There is no way to generate novelty. Discrete state systems are rule-bound, and there are no theoretical means for changing the rules (Fogel, 1993). An alternative model of communication that meets the requirement of explaining both stability and novelty is based on a continuous process model (Fogel, 1993). In this view, communicative action, like all action, is dynamic and creative, able to adjust continuously to changing circumstances. Communication is a continuous coregulation of joint activity, rather than a series of discrete messages. Even when one is in a receptive or listening mode one is still acting, and those actions affect the ongoing actions of the partner. Communicative dialogue has been conceptualized as a three-step process. A acts and B responds. But as B composes the response, A is simultaneously changing so that the observed changes in A affect the response of B at the very moment it is being performed (Markova, 1987). Valsiner (1994) proposes that dialogical processes can be understood as a "temporal triplet...where the first and third parts account for the emergence and dissipation (respectively) of the middle component" (p.15), and in which the boundaries of each component are not clearly demarcated.

120

ALANFOGEL

Narratives are coherent stories m having a beginning, middle, and end m but they are not scripted in advance. Narratives unfold as part of the interpersonal process of telling the story. Narratives are creative activities in which the speaker can alter the story in collaboration with the listener, or in which groups creatively coconstruct joint narratives (Goffman, 1974; Harre, 1988; Hennans & Kempen, 1993; Knoespel, 1991; Sarbin, 1986). Using the concept of coregulation, I (Fogel, 1993, in press) have suggested that novelty also emerges from nonverbal and preverbal communication in both human and nonhuman species. Coregulation can be distinguished from simple synchrony, matching, or attunement because coregulation is defined by the same patterns of emergent novelty and mutual creativity that have been recognized in dialogical and narrative approaches to interpersonal communication. Examples are the use of postural coorientation, kissing, and other courtship behaviors; breaking and establishing mutual gaze; game playing, and fighting. Infants and young children are capable of entering into these forms of creative nonverbal communication (Fogel, 1993; Grammer, 1989; Heath, 1984; Kendon, 1975; Scherer, 1982). Through coregulated dialogical processes, stable patterns calledframes emerge as the historically derived set of repeating patterns of discourse (Bateson, 1955; Fogel, 1993; Goffman, 1974). Frames are coregulated consensual agreements about the scope, setting, and focus of dialogue. Frames are rituals, plots, or routines, regularities in the social process to which participants return, keeping the same overall pattern of coaction against a background of variability. Examples of frames include parent-infant games, peer play rituals, unresolved disagreements, and role relationships. Frames provide the context for the meaning of actions, thoughts, and feelings of the participants. The character or quality of a relationship is evident in part from the types of frames and the ways in which they are organized. It is also evident from the dynamics of frames: their flexibility in adapting to novel situations, and the flexibility of the system to shift between frames when necessary (Fogel & Branco, in press). Frames that have little flexibility and variability are experienced as rigid, boring, or distressing, as in perfunctory greetings or standing arguments. Relationships with relatively few frames, and with difficulty changing to or inventing new frames, feel stagnant and confining. From both a dynamic systems perspective and a dialectical perspective, coregulation and framing are complementary processes. The former is the dynamic and creative aspect of communication that generates novelty and meaning. The latter is the stabilization of coregulated routines. Once frames become established, they can change in a number of ways. They can become elaborated and extended, as in the deepening of intimacy between romantic partners. Frames can become

RELATIONALNARRATIVES 121 abbreviated, as when a mere glance between romantic partners symbolizes prior intimate frames. Finally, interpersonal frames can destabilize and even disappear, as when romantic partners end their intimate relationship. One can think of frames as living systems. They need continued maintenance to stay alive, rejuvenation when they are ailing, and they are dynamic processes that only exist as they are creatively enacted through coregulation (Baxter, 1994; Fogel, 1993; Lyra & Rossetti-Ferreira, 1994; Wilmot, 1994). In the following section, I propose that the self, like interpersonal relationships, is best understood using the dynamic concepts of coregulation and framing. I suggest that selves are relational rather than unitary; that the identity of the self derives from its set of self frames and the dynamic links between and within them; and that selves are living systems that require continuous activity, maintenance, and rejuvenation in much the same way as interpersonal relationships.

The Relational Self Within the literature on the human self, a number of scholars have proposed that the dynamics of self-experience resemble social dialogues. William James and George Herbert Mead, early proponents of this view, distinguished between the "I" as the thinking agent, and the "me" as the object of self-thought (James, 1890; Mead, 1934). More recently, this distinction has been broadened to consider the self as the emergent narrative between two or perhaps many more imagined psychological positions (Bahktin, 1988). Position is the dynamic counterpart to the older (dramatistic) concept of role, used to refer to the different perspectives that constitute self-dialogues (Harre & van Langenhove, 1983): 9. . we conceptualize the self in terms of a dynamic multiplicity of relatively autonomous I positions in an imaginal landscape .... The I has the possibility to move, as in a space, from one position to the other in accordance with changes in situation and time ... the I has the capacity to imaginatively endow each position with a voice so that dialogical relations between positions can be established .... As different voices, these characters exchange information about their respective me's and their worlds, resulting in a complex, narratively structured self .... The I in one position can agree, disagree, understand, misunderstand, oppose, contradict, question, and even ridicule the I in another position (Hermans et al. 1992, pp. 28-29). One could discard the dichotomy between "I" and "me" and refer instead to the self as a dialectical process between multiple imagined positions. A list of hypothetical dialectical processes that create self-experience is given in Table 1. These dialectical positions are not conceptualized as discretely different poles, but rather as ends of a continuum. Nor should one think about these dimensions as discretely different from each other: The items in the table merely list a set of

122

ALANFOGEL

metaphors that may be potentially useful in understanding a relational self. In this paper, I explore two of these dialectical dimensions as applied to the infant self: perception-action and self-other. Like an interpersonal relationship, the relational self cannot be completely determined or defined because it is always constituting itself creatively through dialogue. The form of self dialogue is similar to the form of social discourse. The self is not "in" the individual because it manifests the positions of others and can imagine itself in times and places other than the present. The self is not a unitary locus of agency because it is composed of multiple positions, none of which is more important than any other (Hermans & Kempen, 1993). Furthermore, the self is not consistent across time because there are continuous dialogues between the perceived actual self, past selves, and expected future selves. Self-dialogue creates a constantly emerging view of self as multiple aspects of the self are compared and evaluated (Halen & Bosma, 1994; Werff, 1990). Bruner's notion of the distributed self, known only through a lifelong interpretive procedure, is similar. Bruner found that in autobiographical writing, ostensibly about one's past, almost one half of the narrative units are in the present tense. The writers do not merely recount the past but are "deciding what to make of the past narratively at the moment of telling" (Bruner, 1990, p. 122).

Identity as the Relational History of Self-frames If the self is relational, how can we explain the emergence of an identity: a coherent sense of the self through time and the self as having particular characteristics? I suggest that the feelings of a cohesive and unified identify can be explained by the same process that creates stable consensual frames within a social relationship. Self-frames are stable dialogical patterns that emerge in the discourse between self-positions (Scheibe, 1986). Over time, these patterns take on a coherent form. Self-frames, like relationship frames, are dynamically stable. They are dynamic to the extent that they must be recreated through introspective selfexperience. They are stable to the extent that the history of self-dialogue tends to constrain the boundaries of the self-frames into a small and repeating set of themes (Fogel, 1993). Self-frames have a number of experiential forms. They may be narrative forms, stories by which one recognizes and maintains the self. Self-stories have a beginning, middle, and end. They always contain an evaluative component (e.g., self as hero or victim) (Hermans & Kempen, 1993; Sarbin, 1986). Self-frames may be preferred pattems of cognitive-emotional processes (Demos, 1988; Lewis, in press). Such patterns may include preference for the avoidance or approach of challenge and risk, or a tendency to feel certain kinds of emotions (e.g., fear vs.

RELATIONAL NARRATIVES

123

TABLE 1. Forms of self-experience through dialogue. Dialectical dimensions are extreme positions along various dimensions of dialectical opposition. Selfexperiences arise in the dialogues between dialectical positions. A sense of coherence emerges as these patterns of discourse stabilize into self-frames.

Dialectical Dimension

Self-Experiential Processes

Self-other

Private-public, attachment-separation, autonomy-interdependence, I-Thou

Action-perception

Doing-being, creating-consolidating, flowinterruption, expressing-hiding

Emotional processes

Motivations: approach-withdraw, happy-sad, interest-boredom, anger, fear, etc.

States of consciousness

Sleep-wake, conscious-unconscious, realimaginary, material-spiritual

Sensory systems

Vision, hearing, taste, touch, smell, and their intersensory comparisons

Before-after

Self through time, self-sameness-disjunction, synchrony-asynchrony

Stability-change

Certainty-Uncertainty, sameness-novelty, safety-risk, security-insecurity, opennessdefensiveness

Coherence-chaos

Attunement-detachment, illusion-disillusion, identity-diffusion

Convergence-divergence

Cooperation-conflict, comparing relationships, comparing actions and effects

Tools and objects

Comparisons between different types of tools that extend the self into the environment

Control-abandon

Thinking-feeling, cold-warm, reason-passion

C ategodcal-contin uous

Verbal-nonverbal, linear-nonlinear, sequentialsimultaneous

124

ALANFOGEL

excitement) under particular contextual conditions. Self-frames also may be experienced as patterns of value and commitment to particular forms of relationship (Halen & Bosma, 1994). Because the self is composed of many different frames, we still have to explain the existence of a sense of coherence and a unified identity. I suggest that this may arise, in cultures where such patterns are valued, by the convergence of common themes across different self-frames. We do not need a "super" self m a self-homunculus - - to organize these separate frames within the self. The different self-frames are not separated from each other as in a multiple personality disorder. Instead, they often overlap by virtue of their participation in similar forms of dialogue (Fogel, 1993; Tomkins, 1980). As a developmental psychologist, to take myself as an example, my professional endeavors may overlap with my parenting. There is a set of selfframes that comprise my personal dialogues about work, and another set of frames that are about my thinking and questioning regarding parenting. At times, parental self-dialogues can overlap with professional self-dialogues. To the extent that common themes emerge and stabilize in these complex dialogues, it gives me a feeling of internal coherence. There is no "higher" self that records, represents, and stores these dialogues: There is only the dynamic stability of emergent frames. A self in which one position and one type of frame dominates all the others is not dynamically cohesive. This form of self-unity is rigid and exclusionary, containing a disjunction between each of its real and imagined self-frames. Developmental change of the self, therefore, is the process of creating, comparing, consolidating, integrating, and differentiating relational self-frames (Fogel, 1993). In the following section, I turn to an analysis of the self in infancy. I suggest that from the beginning of life, the infant self is relational: It unfolds dynamically with respect to discourse within the self and between self and other.

Nonrelational Theories of Self in Infancy

Representation Without Relationship Although the idea of a prelinguistic or preconceptual sense of self is not new, scholars of infancy have not conceptualized that self as relational. The infant self they construct has an objective existence as a stored representation and a unity of purpose, control, and organization. One source of evidence used to argue for selfcoherence in infancy is the assumption that experience coming from different sensory systems is convergent, or crossmodally invariant (cf. Stern, 1985). In addition to crossmodal invariance, there is presumed to be a clear link between perception and action leading to a sense of self as agent. The experience, for

RELATIONAL NARRATIVES

125

example, of seeing the hands and feeling them move and act at the same time is presumed responsible for establishing a sense of body coherence and unity. These scholars propose that the infant sense of self is developed over time as a generalized representation of each of these coherences between perception and control: The so-called perceptual or ecological self is the experienced central locus of control. According to Emde and colleagues, "a sense of coherence and of agency are the cardinal features of the self-system along with a fundamental sense of control (i.e., ownership) of body and action" (Emde, B iringen, Clyman, & Oppenheim, 1991, p. 252). Neisser's (1991) concept of an ecological self has a similar origin. He views the ecological self as the core of self-experience, present from birth, to which is added more cognitively complex forms of coherence, making, successively, a remembered self, a private self, and a conceptual self. These different selves are not the same as the dialogical positions of a relational self. Rather, each type of self is a coherent stored representation and each has an independent developmental course. Stern (1985) also relates the perceptual detection of invariants to self-produced action. Like Emde, Stern argues that the infant self is the central stored representation of coherence across the various perceptual and motor domains. The "sense of a core self results [because] the infant has ... the ability to integrate all of these self-invariants into a single subjective perspective" (Stem, 1985, pp. 71-72).

Critique of Nonrelational Self Theories What is the problem with defining the infant self simply as the represented coherence between different perceptual, motor, and affective modalities? One problem is that if the self-experience of the older child and adult is conceptualized as relational, based on narrative and dialogue, then one has to account for its developmental origins. According to Stem and Neisser, for example, the relational self is a later occurrence, made possible by the acquisition of language and a system of conceptual thought that emerges in the third year of life. Although both of these authors posit an interpersonal self early in the first year, this self is based on representations rather than on narrative processes. This is not a reasonable explanation because it does not account for how the self transforms from coherence to dialogue, from unity to multiplicity. How can a coherent self, based in invariance and singularity, change into a narrative, open, and questioning selff Language, with its linear structure and its categorical lexicon, does not seem like a good tool for creating a dynamic relational self. Language, as a mode of discourse, is inherently limited. Indeed, people have to work very hard to capture in language the depth and quality of their personal experiences.

126

ALANFOGEL

A second issue arises with respect to the social origins of self. A coherent representation is inherently solipsistic. How can a coherent unitary representation be related to anything? How can a coherent representational self change due to social experience? Most theories that appeal to unified self-representations assume that the infant acquires learned associations between their stored representations of themselves and their stored representations of others. Stern (1985), for example, believes that the absent other exists as a separate coherent mental representation, an "evoked companion" that becomes associated with the mental representmion of the self. A final problem with a nonrelational approach to infant self is that a singular, coherent, concretely represented adult self would be pathological. Healthy adult selves require flexibility, are embedded in shifting and jointly constructed narrative coconstructions, and are open to conflicts and their resolution. Perhaps a self that is pathological for an adult is normal for an infant? Indeed, many aspects of infant behavior would not be considered normal if displayed by an adult: extremes of emotion, excessive dependence on a single person, and poor motor control when eating. Nevertheless, a more parsimonious model of the infant self would suggest that, like the adult sell it is open, dynamic, and dialogical. Therefore, the task of self-theory in infancy is to find ways to illuminate how a prelinguistic and preconceptual individual can participate in relational modes of experience and can create personal narratives based on relational processes.

A Relational Theory of Infant Self I propose that the infant self, like the adult sell is fundamentally relational. In this section, I discuss how a relational self in infancy might be conceptualized along similar dialectical dimensions as those discussed for adults (see Table 1). Some of these forms of self-experience are in the realm of perception, action, affect, and participation in social relationships and are therefore available to infants. Here, I review literature in the areas of perception-action relationships and self-other relationships. The goal is not to prove that the infant self is relational rather than coherent. Rather, the purpose of this section is to make a plausible argument that a relational metaphor is a potentially more productive approach to the study of early self-development than the existing coherence metaphor. The reader should not expect hard data or empirical verification. At this stage in theory development, my goal is to develop the narrative tools to begin constructing a relational world view. Further work is needed to build a set of empirical models from a relational perspective.

RELATIONAL NARRATIVES 127

Dialectical Positions Related to Perception and Action Nonrelational theories presume that the self is located in the body, as a coherent and centrally stored representation. A relational perspective assumes that the self is distributed as a relationship between the body and the environment. According to Gibson (1966, 1979) individuals perceive themselves at the same time that they perceive the environment. Individuals perceive visual flow in relation to their own location, and movement and information that is generated is both dynamic and relational: It specifies the link between person and environment (Michaels & Carello, 1981; Reed, 1987). Humans see their own nose, at a minimum, at all times in the visual field. Thus, what is out "there" is never independent of what is "here." The body is more than a reference point in the relationship between perception and action. As a position in a dialogue, the body imparts a specific quality to the experience of perception (Fogel, in press; Hermans & Kempen, 1993; Johnson, 1987; Shotter, 1984). Action and perception are situated activities that are meaningful only with respect to the total individual-environment system, including the body (Bruner, 1990; Ginsburg, 1985; Heft, 1989; Hermans & Kempen, 1993; Kolers & Roediger, 1984; Sarbin, 1986). Infants in the first months of life can make a mobile move by repeating a pattern of kicking that they were taught weeks earlier (Rovee-Collier, Enright, Lucas, Fagan, & Gekoski, 1981). However, if they were crying during training, they do not remember the correct kicking pattern later, when they are not crying. Infants do not remember the procedure if the crib decorations are changed between training and testing (Butler & Rovee-Collier, 1989; Fagen, Ohr, Singer, & Klein, 1989; Rovee-Collier et al., 1981). Let us return to the example of seeing the hands moving in the visual field, which nonrelational theories assume is a source of self-coherence. It is certainly the case that the production of force in the arms is precisely coupled with changes in the visual array for the hands, giving the infant the opportunity to perceive the hands as part of the self. For nonrelational theorists, the important feature of this link is the detection of invariance between the visual and motor perceptual fields. The details of the fields, the experience of moving and seeing as unique sensorimotor realms, is less important. From the detection of invariance, the subject retains a representation of the self as the locus of control. This representation is amodal, that is, it is an abstract, nonqualitative, superordinate scheme that is believed to regulate later encounters with the environment. I suggest that there is no such superordinate abstraction. There is only the dynamically stable frame that sustains and renews the discourse between eye and hand each time it recurs. Denying the existence of an abstract super-self does not imply that the self is without affect and cognition. The phenomenological

128

ALANFOGEL

experience associated with this frame may vary from elation to frustration, and it is always imbued with qualities specific to those embodied modalities of perception: muscular exertion, the uniqueness of arm (e.g., compared to leg) movements, the depth and brilliance of visual (e.g., compared to tactile) experience, and the like. Furthermore, these frames don't exist independent of larger patterns of activity (Fogel, in press; Ginsburg, 1985; Heft, 1989; Varela, 1983). Nonrelational theories assume that the abstract representation is some kind of distillation of experience, the kernel of invariance after shedding the husk of everyday variability of action. A relational theory assumes, on the contrary, that the frames emerging from the hand-eye discourse are more than the parts taken separately. Novelty emerging in the hand-eye dialogue is not predictable knowing only the characteristics of hand or eye taken alone. This is a fundamental principle of dynamic systems thinking (emergent properties of self-organization) and of dialectical theory (synthesis is emergent from the dialogue). There is also another way in which the hand-eye coordination is relational: It is embedded within larger individual-environment relational processes. The visual aspect of identifying the hand as belonging to self would not be possible without the ability to compare the action of the hand to the background visual field. Thus, the perception of the hands as part of the self makes sense only with respect to the perception of stationary and moving things that are perceived as not part of the self (Gibson, 1979). The self is not simply the abstract agent of observation, not simply the invariant convergence between action and perception. It is the dialogic relationship between the situated and intentional agent of observation, the body part perceived and proprioceived, and the background perceptual flow field in which the body is located. Data on infant perception and action can be reinterpreted according to this relational model. We know, for example, that knowledge of spatial relations and absent objects is related to self-motion (Acredolo et al., 1984; Campos & Bertenthal, 1988). Kittens who have acquired visually guided locomotion are more skilled at reaching at objects with their paws than those who cannot locomote (Hein & Diamond, 1972). Thus, locomotion produces a sense of self as agent, but also establishes an embodied relationship between self and environment. What we perceive and know is not "about" the environment, but rather is "about" the experiential and dynamic relationship between individual and enviromnent. Butterworth (1992) described infant dialoguelike self-directed actions that have a quality similar to the James-Mead 'T' that acts and the "me" that is the focus of that action. In the first few hours of life, newborns touch their own heads in an ordered sequence beginning with the mouth, then moving to the face, the head, the ear, the nose, and the eyes. This occurs only when the infant is awake (Kravitz et al., 1978, as reported by Butterworth). Just prior to hand-to-mouth contact,

RELATIONAL NARRATIVES

129

newborns will open their mouthes in anticipation. Such patterns of hand-to-mouth and hand-to-head contact occur in fetal development in both humans and other species. As infants acquire visually guided reaching, they self-explore by touching different parts of the body beginning with the fingers and ending months later with the toes (Kravitz et al. 1978, as reported by Butterworth). Rochat (in press) also has found similar patterns of self-exploration in young infants, including both visual and tactile self-exploration. These activities have the quality of a dialogue between the position taken by the act of touching and the position taken by that act of perceiving the area that is touched. This so-called "double touch" (Rochat, in press) could be interpreted as action required to detect and represent invariances about the self. On the other hand, it could also be the case that the double touch is a form of self-dialogue: The sequences of infant double touch that unfold over time are a form of nonverbal relational narrative, leaving each infant to create uniquely different and dynamically evolving self-frames. Perceiving and experiencing the sell even in infants, is not merely abstracting and representing. The dialogical self-frames that result from perception-action relationships have similar properties as verbal narratives. First of all, they are situated activities that derive their meaning from the relational context in which they are embedded. Second, perception-action frames repeat themselves, with variations, in similar contexts, therefore giving them the flavor of remembered action or the reconstruction of the past in the present. Finally, perception-action frames have a beginning, middle, and end. Infants explore themselves and their environments according to preferred sequences. Although these sequences may be structured by the situation (one must look before reaching, reach before grasping, grasp before exploring), the experience for the infant is likely to be partly related to these obligatory temporal patterns. The sense of self for the infant, therefore, must be related to knowing how to create a good perception-action story line and how to manipulate the elements of the story to highlight different aspects of this dialectical dimension. Self is never independent of sequence, story, and situation. I now turn to an examination of another dialectical process that constitutes self: the dialogue between self and other people.

The Self-other Dialectic If we assume that the infant self is fundamentally dynamic and relational, able to construct nonverbal narrative frames across repeated encounters, then it is relatively easy to understand how other persons contribute to the development of infant self-frames. The self is perceived as the relationship of the infant to the environment because the self is inherently part of the perception of the nonself. This nonself may include both inanimate and animate objects, nonhuman and

130

ALANFOGEL

human individuals. The importance of these social relationships is that they highlight particular affordances for the infant, and in so doing, emphasize particular modes of self-other relationship. During coregulated communication with another person, infants can create information about their own actions in relation to another person whose actions are adjusting to the individual at the same time that the individual is adjusting to those actions (Fogel, 1993). The concept of mutual coactive creation of the self through communication is similar to Kaye's (1980) concept of shared activities and intentions; to Trevarthen's (Trevarthen & Hubley, 1978) concept of primary intersubjectivity, in which the self emerges as part of a mutually regulated exchange between infant and caregiver in early infancy; to Sander's concept of mutual regulation in the mother-infant system; and to Mead's (1934) concept of the mutually significant gesture. Because the acts to which one is adjusting become adjusted to one's own adjustive actions, the process gives immediate and directly perceived meaning to one's own actions. A common example is infant face-to-face play with the parent during the first six months of life. Parents and infants become increasingly coordinated in their mutual smiling, gazing, and postural adjustments over several months of mutual play. Infants come to know themselves as participants in this discourse. How do they come to know themselves? Some suggest that infants detect contingencies between their own and the parmer's actions, and develop a cognitive sense of self that is based on expectation of a particular pattern of contingent responding (it may be consistent or inconsistent; e.g., Gianino & Tronick, 1988). Others suggest that the infant builds up a representation of self, other, and self with other that is the abstracted invariance from the history of interaction. In this view, the infant perceives that experience belongs to the self because these invariances converge on the self via the matching of motivation and proprioception (Stem, 1985). Stem (1993), for example, uses the metaphor of a "microplot," or a dynamic narrative memory that includes a series of events that repeatedly occur over time. Similar to notions of script or episodic memory, however, this metaphorical narrative is stored in memory as a whole sequence that can be replayed: not as a relational process or as relational positions (Fogel, 1993). My approach, outlined earlier, is that self-frames are not stored or represented. They are emergent creations from dialogue between relational positions. Whenever the conditions for creating the same or similar narrative process are present, the infant is able to enter again into similar dynamic forms of discourse. Narrative remembering occurs not by replaying a memory tape, not by acting out expectations, and not by calling up a represented narrative script. Rather, remembering is a recreative process in which the entire constellation of infant action differs according to the circumstances. As the positions and conditions of

RELATIONALNARRATIVES 131 discourse change, the infant enters into different patterns of self-organization with the person or object. These patterns cannot be understood as merely contingent responses: They involve the active participation of the infant's whole body and psyche. These patterns of self-organization include hand and limb movements, postural shifts, gaze, facial expression, autonomic changes, and experiential alteration in felt affect (Brazelton, Koslowski, & Main, 1974; Demos, 1988; Fogel & Thelen, 1987; Trevarthen, 1993). Affective experience plays a central role in the self-other dialectic (Demos, 1992; Stem, 1985; Trevarthen, 1993). One view is that infants can perceive their o w n affective states as different from another's because affect is inherently motivational. According to Tomkins (1962) and Demos (1992), the affective system amplifies perceptual information into a form that motivates the infant to act in specific ways vis-h-vis the situation. A functional perspective on emotion (Barrett & Campos, 1987) suggests a similar motivating role of families of discrete affects leading to "action tendencies" that are adaptive in specific circumstances. The infant, as the locus of these feelings and actions, can experience the self in relation to others. These and other views of the role of emotion in the development of the self suggest the need for a fixed set of discrete emotions in the newborn that form the seeds for the germination of a core self (Demos, 1992; Emde, Biringen, Clyman, & Oppenheim, 1991; Tomkins, 1962; Trevarthen, 1993). Such a perspective is not, however, theoretically necessary for the assertion that emotional experiences amplify and facilitate the relational dialogues of the self in relation to another person. Several recent perspectives suggest that infant emotions are not discrete, but rather are dynamic and continuous. Although stable regularities may emerge from these emotion dynamics, a diverse array of affective motivational processes may be available to the infant. Out of the self-other dialogue, particular constellations of cognitive and affective experience are likely to consolidate as part of the relational narratives of self with other. These constellations emerge as affective regularities that add emotional color and depth to the developing relational plot (Fogel et al., 1992; Lewis, in press; Sarbin, 1986). In his later work, Tomkins (1980) retreats from discrete emotions as the unit of individual experiences and uses the metaphor of the "scene" (or script) as the basic unit of self-experience. In a scene, characters of self and other are preserved in their dialogical positions. Scenes having similar themes are related to each other by the infant. The infant psyche, rather than a simple additive compilation of represented scripts, is dynamic as events within scenes "magnify" or attenuate each other to create emergent forms of affective experience connected by self-other dialectics. Furthermore, the originally remembered scenes change dynamically as

132

ALANFOGEL

they are combined with other scenes leading to a transformation of self-experience: What is remembered is emergent from the historical process and is assembled nonlinearly, much like the emergence of topics in a dialogue (Fogel & Branco, in press). According to Tomkins, "what is important from the point of view of script theory is that the effect of any set of scenes is indeterminate until the future happens and either further magnifies or attenuates experience.., the consequence of any experience is not singular but plural" (Tomkins, 1980, p. 219). Self-other dialogues, like perception-action dialogues, have the same characteristics of verbal narratives. This is demonstrated by the differential patterning of infant-other interpersonal frames as a function of person and situation. Infants as young as several months show very different organized pattems of action in relation to different social partners, such as between a peer and an adult (Fogel, 1979a), between mother and father (Parke, 1979), and between mother and an unfamiliar female (Mizukami et al., 1990). Infants as young as 6 weeks will repeat similar patterns of action, particularly imitative responses, when encountering the same person on repeated occasions (Meltzoff & Moore, 1994). Different patterns of organization appear in relation to interactive disturbances such as simulated maternal depression (Cohn & Tronick, 1983), inten'uption of maternal action during the "still-face" procedures (Cohn & Elmore, 1988; Fogel et al., 1983; Tronick et al., 1978), and changes in the animacy of the partner (Field, 1979; Legerstee, Corter, & Kienapple, 1990), changes in behavior following separation from the mother (Field et al., 1986; Fogel, 1979b). These differences in responding typically cannot be ascribed to a single action such as crying, gaze aversion, or smiling. Rather, they include systemic changes in the coaction of the infant's hands, arms and legs, facial expression, and gaze direction. These changes are not simply contingent upon a change in the socialcontextual situation. Rather, they involve concurrent coactivity within each different context, and the patterns of infant organization are cocreated through interaction in that context. Because these patterns of action involve the whole infant, and because they are always taken in relation to specific contexts, each pattern of organization reflects a different dialogically constructed narrative of selfin-relation. One can observe similar relational narratives arising from self-other dialectics in the second year of life. Compliance and self-monitoring activities, for example, are enhanced in relationships that are coregulated, that is, those in which parental demands are not unilaterally imposed on infants, but developed as part of ongoing self-other discourse (Lutkenhaus, Bullock, & Geppert, 1987; Parpal & Maccoby, 1985; Schaffer & Crook, 1980). Similarly, the sells relationship to family members arises in mother-child and sibling-child discourse during the second year,

RELATIONALNARRATIVES 133 as teasing, support, and prohibition are coregulated in the family (Dunn & Munn, 1985). During peer interactions in the second year, children begin to establish coregulated communication without the intervention of an adult. They are able to create play dialogues, first involving simple imitations and later using complementary actions in response to the partner. These forms of play are cooperative activities in which frames are coregulated. Participation in these activities enhances self-monitoring and self-regulation, as well as the child's conceptual understanding of self and other (Brownell & Carriger, 1990; Dunham et al., 1989; Eckerman, 1994; Wolf, 1982). During the preschool period, children begin to define themselves linguistically with respect to other people. Talk about the self contains themes of similarity or difference from other people, or with respect to actions performed with others ("We went way down by the pool"), to others ("I taught my little brother how to color"), or by others ("Jimmy hit me"). Often, children's descriptions of events that occur when alone are cast in terms of relationships with others ("No one was holding my hand"; Miller, Mintz, Hoogstra, Fung, & Potts, 1992, pp. 53-54). When children first use the word mine at 2 years in the context of social play, they are defining ownership (a relational concept) rather than taking possession of something. They will say mine if another child takes a toy, but may not complain as the other child uses the toy (Levine, 1983). The emergence of self-conscious emotions (pride, shame, and guilt) can be conceptualized as particular types of narrative forms. Shame, for example, occurs within narratives that focus on falling short of standards and the inadequacy of the self in relation to those standards (Barrett, 1984; Mascolo & Fischer, 1984). These findings suggest that linguistic self-narratives are a developmental continuation of a relational self that begins in the first months of life. Observers who are focused on language as the only means of establishing self-other relationships, or as the only means of constructing self-other narratives, suggest that the relational self begins around 18 months when infants recognize themselves in a mirror and can linguistically label the self (Harter, 1983; Lewis & BrooksGunn, 1979; Pipp et al., 1987; Stipek, Gralinski, & Kopp, 1990). However, other research shows that this achievement may be due in part to the child's early exposure to mirrors in the social context and to using mirrors in that context for self-recognition (Mitchell, 1993). Conclusion: Complex Dialectics I have suggested that the fabric of self-experience is woven from different types of dialectics, many of which are available to young infants. I reviewed research in two

134

ALANFOGEL

dialectical modalities: perception-action and self-other dialogue. The self is not a represented compilation of these experiences, but rather is a relational process that is continuously evolving. Within any one of the dialectical systems mentioned in Table 1, indeterministic emergent processes are likely to occur as the narrative unfolds. Complexity in the self is also generated because these dialectical processes do not occur in isolation. Indeed, they are often concurrent with each other, and their overlap affords additional sources of creativity in self-narratives. I have already suggested how the dialectic between different emotions may enter into the relational narratives between the infant and another person. The emotions can amplify or magnify the self-other dialogues, predisposing the dyad to form relational frames that are mediated by the emergent motivational features of the emotion dialectic. In a similar manner, emotion dialectics may interface with perception-action dialectics. Increasing complexity occurs as these dialectics interface with all the others listed in Table 1. The relational narratives of the prelinguistic self are patterned self-frames that are emergent from the complexity of the overlapping dialectical processes. These narratives symbolize the individual's ways of being-in-relation. At a higher level, an emergent sense of self-coherence or self-identity coalesces from comparisons between the pluralistic relational narratives. This sense of coherence is not an immobile or fixed core, but rather is a flexible and dynamically stable coalition of many voices playing many parts in the self-drama. From the perspective of relational narratives, therefore, the self is an open system of emerging possibilities and creative processes. The self is a dynamic system whose dialogues mnplify and attenuate the voices, and out of which new voices and themes will emerge. Although the infant self is simpler in its dialogical themes and frames compared to later, one can however conceptualize a fully relational infant self. Experiencing the self is a profoundly engaging experience, composed of selfrenewing and self-organizing relational self-frames. Nonrelational theories seem to assume that without language, the infant is not privy to deeply rewarding selfreflective experiences. At least such theories do not allude to these processes. These nonverbal narratives have a similarly creative, similarly engaging quality for the infant self as for the adult. This does not imply that the infant has the same ability to reflect upon its own life and identity as the adult. Rather, the infant self is a living system, it is unfinished like all selves, and finds its motivation for selfrenewal in the continuing elaboration of the dialogue within and between selfframes.

RELATIONAL NARRATIVES 135 ACKNOWLEGMENTS

The author acknowledges the helpful comments of Hui-chin Hsu, Christy Nelson, and Andrea Pantoja. This work was supported in part by a grant from the National Institute of Mental Health (R01MH48680). REFERENCES

Acredolo, L. P., Adams, A., & Goodwin, S. W. (1984). The role of self-produced and visual tracking in infant spatial orientation. Journal of Experimental Child Psychology, 38, 312-327. Bahktin, M. M. (1988). Discourse in the novel. In N. Mercer (Ed.), Language and literacy from an educational perspective Vol. 1 (pp. 47-57). Philadelphia, PA: Open University Press. Barrett, K. C. (1995). A functionalist approach to shame and guilt. In J. P. Tangney & K. W. Fischer (Eds.), Self- conscious emotions: The psychology of shame, guilt, embarrassment, and pride (pp. 25-63). NY: The Guilford Press. Barrett, K. C., & Campos, J. J. (1987). Perspectives on emotional development II: A functionalist approach to emotions. In J. D. Osofsky (Ed.), Handbook of infant development (2nd Ed.) (pp. 555-578). NY: John WHey & Sons. Bateson, G. (1955). The message: "This is play." In B. Schaffner (Ed.), Group processes Vol. 2. Madison, NJ: Madison Printing Co. Baxter, L. A. (1994). A dialogic approach to relationship maintenance. In D. J. Canary & L. Stafford (Eds.), Communication and relational maintenance (pp. 233254). San Diego, CA: Academic Press. Brazelton, T. B., Koslowski, B., & Main, M. (1974). The origins of reciprocity. In M. Lewis & L. Rosenblum (Eds.), The effect of the infant on its caregiver (pp. 49-76). NY: Wiley. Brownell, C. A., & Carriger, M. S. (1990). Changes in cooperation and self-other differentiation during the second year. Child Development, 61, 1164-1174. Bruner, J. (1990). Acts of meaning. Cambridge, MA: Harvard University Press. Buffer, J., & Rovee-Collier, C. (1989). Contextual gating of memory retrieval. Developmental Psychology, 22, 533-552. Butterworth, G. (1992). Origins of self-perception in infancy. Psychological Inqui~, 3, 103-111. Campos, J. J., & Bertenthal, B. I. (1988). Locomotion and psychological development. In F. Morrison, K. Lord, & D. Keating (Eds.), Social interaction in individual development. Academic Press: New York. Cohn, J. F., & Elmore, M. (1988). Effect of contingent changes in mothers' affective expression on the organization of behavior in 3-month-old infants. Infant Behavior and Development, 11, 493-505. Cohn, J. F., & Tronick, E. Z. (1983). Three-month old infants' reaction to simulated maternal depression. Child Development, 54, 185-193. Cushman, P. (1991). Ideology obscured. American Psychologist, 46, 206-219. Demos, E. V. (1992). The early organization of the psyche. In J. W. Barron, M. N. Eagle, & D. L. Wolitzky (Eds.), Interface of psychoanalysis and psychology (pp. 200-232). American Psychological Association. Dunham, P., Dunham, F., Hurshman, A., & Alexander, T. (1989). Social contingency effects on subsequent perceptual-cognitive tasks in young infants. Child Development, 60, 1486-1496.

136

ALANFOGEL

Dunn, J., & Munn, P. (1985). Becoming a family member: Family conflict and the development of social understanding in the second year. Child Development, 56, 480-492. Eckerman, C. O. (1994). Toddler's achievement of coordinated action with conspecifics: A dynamic systems perspective. In L. B. Smith & E. Thelen (Eds.), Dynamical systems in development: Applications (pp. 333-357). Cambridge, MA: MIT Press. Emde, R. N. f1994). Individuality, context, and the search for meaning. Child Development, 65, 719-737. Emde, R. N., Biringen, Z., Clyman, R. B., & Oppenheim, D. (1991). The moral self of infancy: Affective core and procedural knowledge. Developmental Review, 11, 251-270. Fagen, J. W., Ohr, P. S., Singer, J. M., & Klein, S. J. (1989). Crying and retrograde amnesia in young infants. Infant Behavior and Development, 12, 13-24. Field, T. (1979). Infant behaviors directed toward peers and adults in the presence and absence of mother. Infant Behavior and Development, 2, 47-54. Field, T., Bega-Lahr, N., Scafidi, F., & Goldstein, S. (1986). Effects of maternal unavailability on mother-infant interactions. Infant Behavior and Development, 9, 473-478. Fogel, A. (1979). Peer vs. mother directed behavior in 1- to 3-month-old infants. Infant Behavior and Development, 2, 215-226. Fogel, A. (1980). The effect of brief separation on 2-month-old infants. Infant Behavior and Development, 3, 315-330. Fogel, A. (1993). Developing through relationships. London: Harvester Wheatsheaf and University of Chicago Press. Fogel, A. (in press). Information, creativity, and culture. In C. Dent-Reed & P. ZukowGoldring (Eds.) Changing ecological approaches to development." OrganismEnvironment Mutualities. APA Publications. Fogel, A., & Branco, A. U. (in press). Meta-communication as a source of indeterminism in relationship development. In A. Fogel, M. Lyra, & J. Valsiner (Eds.), Dynamics and indeterminism in developmental and social processes. Hillsdale, NJ: Erlbaum. Fogel, A., Diamond, G. R., Langhorst, B. H., & Demos, V. (1982). Affective and cognitive aspects of the two-month-old's participation in face-to-face interaction with its mother. In E. Tronick (Ed.), Social interchange in infancy: Affect, cognition, and communication, (pp. 37-57) Baltimore: University Park Press. Fogel, A., Nwokah, E., Dedo, J., Messinger, D., Dickson, K. L., Matusov, E., & Holt, S. A. (1992). Social process theory of emotion: A dynamic systems perspective. Social Development, 1, 122-142. Fogel, A., & Thelen, E. (1987). Development of early expressive and communicative action: Reinterpreting the evidence from a dynamic systems perspective. Developmental Psychology, 23, 747-761. Gergen, K. J. (1994). Exploring the postmodern: Perils or potentials? American Psychologist, 49, 412-416. Gianino, A., & Tronick, E. (1988). The mutual regulation model: Infant self and interactive regulation, coping and defense. In T. Field, P. McCabe, & N. Schneiderman (Eds.), Stress and coping across development (pp. 47-68). Hillsdale, NJ: Erlbaum. Gibson, J. J. (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin. Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin.

RELATIONAL NARRATIVES 137 Ginsburg, G. P. (1985). The analysis of human action: Current status and furore potential. In G. P. Ginsburg, M. L. Brenner, & M. von Cranach, (Eds.), Discovery strategies in the psychology of action (pp. 255-279). London: Academic Press. Goffman, E. (1974). Frame analysis: An essay on the organization of experience. Cambridge, MA: Harvard University Press. Grammer, K. (1989). Human courtship behavior: Biological basis and cognitive processing. In A. E. Rasa, C. Vogel, & E. Voland (Eds.), The sociobiology of sexual and reproductive strategies (pp. 147-169). New York: Chapman and Hall. Halen C. van, & Bosma, H. A. (1994). Problems of self-conception during the life span. Poster presented at the XIIth biennial meetings of the ISSBD, Amsterdam. Harre, R. (1988). The social construction of emotions. New York: Basil Blackwell. Harre, R., & Langenhove, L. van (1983). Varieties of positioning. Journal of the Theory of Social Behavior, 21, 393-406. Harter, S. (1983). Developmental perspectives on the self-system. In E. M. Hetherington (Ed.), Handbook of child psychology: Socialization, personality, and sound development: Vol. 4, (pp. 275-385). NY: Wiley. (P. H. Mussen, General Editor). Heath, C. (1984). Talk and recipiency: Sequential organization in speech and body movement. In J. M. Atkinson & J. Heritage (Eds.), Structures of social action: studies in conversation analysis (pp. 247-265). NY: Cambridge University Press. Heft, H. (1989). Affordances and the body: An intentional analysis of Gibson's ecological approach to visual perception. Journal for the Theory of Social Behavior, 19, 1-30. Hein, A., & Diamond, R. M. (1972). Locomotory space as a prerequisite for acquiring visually guided reaching in kittens. Journal of Comparative and Physiological Psychology, 81, 394-398. Hermans, H. J., Kempen, H. J. G., & Loon, R. J. P. van (1992). The dialogical self: Beyond individualism and rationalism. American Psychologist, 47, 23-33. Hermans, H. J. M., & Kempen, H. J. G. (1993). The dialogical self" Meaning as movement. San Diego: Academic Press. James, W. (1890). The principles of psychology. New York: Holt. Jencks, C. (1992). The postmodern agenda. In C. Jencks (Ed.), The postmodern reader (pp. 10-39). NY: St. Martin's Press. Johnson, M. (1987). The body in the mind: The bodily basis of meaning, imagination, and reason. Chicago: The University of Chicago Press. Kaye, K. (1980). Why we don't talk 'baby talk' to babies. Journal of Child Language, 7, 489-507. Kendon, A. (1975). Some functions of the face in a kissing round. Journal of the International Association for Semiotic Studies, 15 (4), 99-334. Knoespel, K. J. (1991). The emplotment of chaos: Instability and narrative order. In N. K. Hayles (Ed.), Chaos and order: Complex dynamics in literature and science (pp. 100-122). Chicago: University of Chicago Press. Kolers, P. A., & Roediger, H. L. (1984). Procedures of mind. Journal of Verbal Learning and Verbal Behavior, 23, 425-449. Kravitz, H., Goldenberg, D., & Neyhus, A. (1978). Tactual exploration by normal infants. Developmental Medicine and Child, 20, 720-726. Legerstee, M., Corter, C., & Kienapple, K. (1990). Hand, arm, and facial actions of young infants to a social and nonsocial stimulus. Child Development, 61, 774784. Levine, L. E. (1983). Mine: Self-definition in 2-year-old-boys. Developmental Psychology, 19, 544--549.

138

ALANFOGEL

Lewis, M. D. (in press). Cognition-emotion feedback and the self-organization of developmental paths. Human Development. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. New York: Plenum. Lutkenhaus, P., Bullock, M., & Geppert, U. (1987). Toddlers action: Knowledge, control, and the self. In F. Halisch & J. Kuhl (Eds.), Motivation, intention, and volition. Berlin: Springer-Verlag. Lyra, M., & Rossetti-Ferreira, M. C. (1994). Transformation and construction in social interaction: A new perspective on analysis of the mother-infant dyad. In J. Valsiner (Ed.), Child development within culturally structured environments, Vol. 3: Comparative-cultural and co-constructionist perspectives. Norwood, NJ: Ablex. Markova, I. (1987). On the interaction of opposites in psychological processes. Journal of the Theory of Social Behavior, 17, 279-299. Mascolo, M. F., & Fischer, K. W. (1995). Developmental transformations in appraisals for pride, shame, and guilt. In J. P. Tangney, & K. W. Fisher (Eds.),

Self-conscious emotions: The psychology of shame, guilt, embarrassment, and pride (pp. 64-113). NY: The Guilford Press. Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago Press. Meltzoff, A. N., & Moore, M. K. (1994). Imitation, memory, and the representation of persons. Infant Behavior and Development, 17, 83-99. Michaels, C. F., & Carello, C. (1981). Direct perception. Englewood Cliffs, NJ: Prentice-Hall. Miller, P. J., Mintz, J., Hoogstra, L., Fung, H., & Potts, R. (1992). The narrated self: Young children's construction of self in relation to others. Merrill-Palmer, 38, 4567. Mitchell, R. W. (1993). Mental models of mirror-self-recognition: Two theories. New Ideas in Psychology, 11, 295-325. Mitchell, S. A. (1988). Relational concepts in psychoanalysis: An integration. Cambridge, MA: Harvard University Press. Mizukami, K., Kobayashi, N., Ishii, T., & Iwata, H. (1990). First selective attachment begins in early infancy: A study using telethermography. Infant Behavior and Development, 13, 257-271. Neisser, U. (1991). Two perceptually given aspects of the self and their development. Developmental Review, 11, 197-209. Oyama, S. (1993). How shall I name thee? The construction of natural selves. Theory & Psychology, 3, 471-496. Parke, R. (1979). Perspectives on father-infant interaction. In J. Osofsky (Ed.), Handbook on infant development (pp. 549-590). New York: Wiley. Parpal, M., & Maccoby, E. E. (1985). Maternal responsiveness and subsequent child compliance. Child Development, 5, 1326-1334. Pipp, S., Fischer, K. W., & Jennings, S. (1987). Acquisition of self- and motherknowledge in infancy. Developmental Psychology, 23, 86-96. Porush, D. (1991). Fictions as dissipative structures: Prigonine's theory and postmodemism's roadshow. In N. K. Hayles (Ed.), Chaos and order: Complex dynamics in literature and science, (pp. 54-84). Chicago: The University of Chicago Press. Reed, E. S. (1987). Why do things look as they do? The implications of J. J. Gibson's: The ecological approach to visual perception. In A. Costall & A. Still (Eds.), Cognitive psychology in question, (pp. 90-114). New York: St. Martin's Press.

RELATIONAL NARRATIVES 139 Rochat, P. (in press). Early development of the ecological self. In Changing ecological approaches to development: Organism-environment mutualities. Eds. APA Publications. Rovee-Collier, C. K., Enright, M., Lucas, D., Fagan, J., & Gekoski, M. J. (1981). The forgetting of newly acquired and reactivated memories of 3-month-old infants. Infant Behavior and Development, 4, 317-331. Sampson, E. E. (1991). Social worlds, personal lives. San Diego: Harcourt Brace Jovanovich. Sarbin, T. R. (1986). Narrative psychology: The storied nature of human conduct. New York: Praeger. Schaffer, H. R., & Crook, C. K. (1980). Child compliance and maternal control techniques. Developmental Psychology, 16, 54-56. Scheibe, K. E. (1986). Self-narratives and adventure. In T. R. Sarbin (Ed.), Narrative psychology: The storied nature of human conduct (pp. 129-151). New York: Praeger. Scherer, K. R. (1982). Methods of research on vocal communication: Paradigms and parameters. In K. Scherer & P. Ekman (Eds.), Handbook of methods in nonverbal behavior research. Cambridge, MA: Cambridge University Press. Shotter, J. (1984). Social accountability and selfhood. Oxford: Blackwell. Stem, D. N. (1983). The early development of schemas of self, other, and "self and other." In J. D. Lichtenberg & S. Kaplan, (Eds.) Reflections on self psychology (pp. 49-104). Hillsdale, NJ: Lawrence Erlbaum. Stem, D. N. (1985). The interpersonal world of the infant. New York: Basic Books. Stipek, D. J., Gralinski, J. H., & Kopp, C. B. (1990). Self-concept development in the toddler years. Developmental Psychology, 26, 972-977. Tomkins, S.S. (1962). Affect, imagery, and consciousness: Vol.1. New York: Springer. Tomkins, S.S. (1980). Affect as amplification: Some modification in theory. In R. Plutchik & H. Kellerman (Eds.), Emotion, theory research and experience: Vol. L Theories of emotion. New York: Academic. Trevarthen, C. (1993). The function of emotions in early infant communication and development. In J. Nadel & L. Camaioni (Eds.), New perspectives in early communicative development (pp. 48-81). New York: Routledge Press. Trevarthen, C., & Hubley, P. (1978). Secondary intersubjectivity: Confidence, confiding, and acts of meaning in the first year. In A. Lock (Ed.), Action, gesture, and symbol: The emergence of language (pp. 183-227). New York: Academic Press. Tronick, E., Als, H., Adamson, L., Wise, S., & Brazelton, T. B. (1978). The infant's response to entrapment between contradictory messages in face-to-face interaction. Journal of the American Academy of Child Psychiatry, 17, 1-13. Valsiner, J. (1994). Culture and human development: A co-constructionist perspective. In P. van Geert & L. Mos (Eds.), Annals of theoretical psychology (Vol. 10). New York: Plenum. Varela, F. J. (1983). Principles of biological autonomy. New York: North-Holland. Werff, J. J., van der (1990). The problem of self-conceiving. In H. Bosma & S. Jackson (Eds.), Coping and self-concept in adolescence (pp. 14-33). Berlin: Springer. Wilmot, W. W. (1994). Relationship rejuvenation. In D. J. Canary & L. Stafford (Eds.), Communication and relational maintenance (pp. 255-273). San Diego: Academic Press. Wolf, D. (1982). Understanding others: A longitudinal case study of the concept of independent agency. In G. Foreman (Ed.), Action and thought (pp. 297-327). New York: Academic Press.

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

141

CHAPTER 8

From Direct to Reflexive (Self-) Knowledge" A Recursive Model (Self-produced) Actions Considered as Transformations PIERRE MOUNOUD

Universit~ de Gen~ve

This book is based on three major assumptions: one of them I fully subscribe to, the other two are on the contrary problematic. The first assumption to which I subscribe states that object knowledge and self-knowledge are inseparable. It refers to Gibson's theory (1966, 1979), which considers that all perception implies a coperception of the object and of the perceiver him/herself and that all perceptual systems are self-referential. This assumption could be completed by the following statement inspired by Piaget's theory: All knowledge about objects implies knowledge about (subject's) potential actions related to these objects. As a consequence, all that we know about the development of object knowledge could be transposed to the development of selfknowledge, including featural knowledge as well as motion knowledge. Expressed differently, object knowledge includes self-knowledge, particularly as it is correlated to actions. The second assumption asserts that during the second year of life, a conceptual or categorical self related to self-recognition in the mirror appears, as assessed for example by the Gallup (1970) mirror test. One can wonder, what does the emergence of this level of self-recognition in the general context of self-knowledge development mean more precisely? Does this behavior constitute a final state, a basis for further development or, as I suspect, just a step in the elaboration of selfknowledge as a complex phenomenon that cannot be restricted to self-recognition and even less to a particular form of it? More precisely, self-knowledge cannot be limited to featural knowledge and must include knowledge about actions, which is usually referred to as motion knowledge.

142

PIERREMOUNOUD

I prefer to use the term knowledge about action for knowledge related to (self-) actions that are considered as transformations connecting configurational or featural knowledge. Action knowledge is related not only to self-produced actions, but also to perceived actions. In a similar way, the body schema cannot be considered exlusively as visual identification of the various parts of the body (perceptual aspect) (Pick, 1922) or as a system of postural transformations (motoric aspect) (Head, 1920; Schilder, 1968), but as an amalgam of both points of view (Ajuriaguerra, 1976; Hecaen & Ajuriaguerra, 1952). As Wallon (1959, p. 253) wrote, "the problem of body schema is not only related to its constitutive images, but also to the relationships between gestural space (self-motion) and object space (object motion)." I will try in this chapter to develop and explicate the aspects of self-knowledge related to actions that are considered as transformations. The third assumption asserts that before the emergence of a conceptual or categorical self, a preconceptual self can exist. Neisser (1993) refers to this as the "ecological self." Now, such a formulation raises delicate theoretical and terminological problems. In particular, what differentiates the conceptual forms of self-knowledge from the preconceptual ones? In fact this is related to (the more general problem concerning) the definition of different knowledge systems and the relationships they maintain among one another. A major part of my chapter will be devoted to this issue. I will base my discussion on the ideas developed by Mandler and Piaget on that topic. But first I will clarify the way I use the terms direct and reflexive in order to qualify different functioning modes of any knowledge systems.

The Difference Between Direct and Reflexive Knowledge I will qualify knowledge as "direct" when the subject's processing capacities (structures or networks) are adequate or adapted to certain dimensions of the environment and their variations or when the structures or networks are adapted to some categories of problems encountered by the subject or when the patterns of information (static or dynamic) are in correspondence with action patterns. In such cases, the processing is automatized. For newborns, this direct knowledge results from phylogenesis. When knowledge is direct (or when the processing is automatized), it is as if the subject has no need to "think," to "reflect on," or to "mediate" before acting. There is a direct coupling between subject and environment (homeostasis). It is possible to say that the subject is under a simple stimulus response control, following Neisser (this volume) comments about flies' landing movements in response to optical flow. He concludes that flies do not

FROM DIRECT TO REFLEXIVE SELF-KNOWLEDGE

143

necessarily perceive themselves as distinct from the environment. One could ask to what extent humans behave in a similar way when we speak of direct knowledge. On the other hand, confronted with new problems, subjects can be considered to be in a disequilibrium state or as inadapted (homeorhesis). In such cases, they have to modify their structures or networks or to elaborate new ones. This is usually the case in the course of development. During the elaboration of new structures to the new category of problems (or to similar problems but processed by new structures), subjects are in a state of disequilibrium. In humans, these phases of disequilibrium manifest themselves by searching behaviors or exploratory activities that are associated with (what is usually called) "thought," "reflection," or "explicitation processes," as well as various states of consciousness and intentionality. The relationship between subject and environment becomes (in a certain way) "indirect" or "mediated." In these cases, I suggest that such knowledge should be qualified as "reflexive." From my point of view, reflexive knowledge is a transitory phenomenon. It is necessary as long as the subject is elaborating new structures or networks. Reflexive knowledge can be related to the executive or integrative functions attributed to the prefrontal cortex (Dubois, Pillon, & Sirigu, 1994). When new structures have been constructed and automatized for new categories of problems, then knowledge manifested by the subject's behavior should again be qualified as direct. Take, for exmnple, when the infant (around 12 months of age) succeeds without difficulties (in an automatized way) to retrieve an object located behind an obstacle, or when the child (around 3 years) succeeds without difficulties (i.e., in a systematic way) to nest cups of different sizes. Because I consider developmental processes as recursive, cognitive development can be characterized by a succession of levels of direct and reflexive knowledge. Consequently, it is no longer possible to consider direct knowledge as more "primitive" than reflexive knowledge. In such a perspective, direct knowledge as manifested in the newborn must be considered as resulting from previous reflexive knowledge in the course of phylogenesis. The origin of disequilibrium or disadaptation could be internal or external to the subject. Nevertheless, during ontogeny it is reasonable to consider the internal transformations as predominant and responsible for the major restructuring of cognitive systems. If we consider, for example, the setting up of inter- and intrahemispheric connections (the coupling of connected neural networks) as studied by Thatcher (1994), one can figure out their consequences on the equilibrium in the relationship between the subject and its environment. Considered in relation to these distinctions, my position can be summarized in the following way:

144

PIERREMOUNOUD 9At birth there is a first knowledge system already constituted, which I call "sensorial" or "sensorimotor," which includes a variety of direct selfknowledge based upon encapsulated subsystems.These various subsystems are integrated into a whole system. 9 On the other hand, there is also from birth on a second knowledge system in elaboration called "perceptual" or "perceptuomotor," which makes possible the elaboration of new (self-)knowledge by the different specific subsystems. During the elaboration phases of the new system and its subsystems, knowledge is reflexive and eventually becomes direct when it is automatized. I estimate the achievement of the whole perceptuomotor knowledge system to be at around 3 1/2 to 4 years. 9Knowledge of the first sensorimotor system is "direct" and nonexplicit. At the achievement of the perceptuomotor system (and subsystems), knowledge is again direct but could be explicit if necessary (i.e., if necessitated by the encountered situations). 9From 3 1/2 to 4 years on, another knowledge system (called "concrete") starts in a recursive way. The perceptual system now elaborated will take the turn of the constituted system.

In order to discuss the problem of the existence of different knowledge systems and their possible relationships, I will first examine recent articles by Mandler (1988, 1992). Then I will present Piaget's theory related to knowledge about action-transformations and reflexive abstraction in order to demonstrate how selfknowledge related to action is one of the major components of knowledge systems. Finally I will describe my own conception of the construction of new knowledge systems.

On the Existence of Different Knowledge Systems Mandler's Point of View Recently, Mandler wrote two articles entitled "How to built a baby" (1988, 1992). In her first article (1988), which I have discussed extensively elsewhere (Mounoud, 1993a), Mandler defines what she calls a "dual representational system." On the one hand, there is a "sensorimotor knowledge system" (or sensorimotor procedures), based on sensorimotor, nonsymbolic representations, and, on the other hand, there is a "conceptual knowledge system" (or declarative knowledge), based on conceptual and symbolic representations. The existence of this second system is due to the human infant's innate capacity to symbolize.

FROM DIRECT TO REFLEXIVE SELF-KNOWLEDGE

145

These two systems differ in the following way: Sensorimotor knowledge is not accessible to consciousness, and its acquisition does not require conscious accompaniment, whereas conceptual knowledge is accessible to consciousness, for purposes of recall or thinking. They differ with regard to their respective origins as well. Sensorimotor knowledge is derived from perceptual input, based on what objects look like, without adding something "above or beyond what the object looks like" (Mandler, 1988, p. 118). Conceptual knowledge is based on a process of elaboration of perceptual input, resulting from perceptual analysis, and is equivalent to a mental comparison process. In her second article, Mandler (1992) specifies what the process of perceptual analysis in the elaboration of conceptual knowledge is and defines what she calls "conceptual primitives" (constructed). Before being constituted of conceptual representations, the conceptual knowledge system would initially be based on another type of representation called "image schemas" (conceptual primitives), which are derived from perceptual structures. I have to point out that in her 1988 article, Mandler states that conceptual knowledge is not due to a transformation of procedural knowledge. In her 1992 article, conceptual knowledge results from a (representative) redescription of perceptual schemas following a model borrowed from Karmiloff-Smith (1991) and from Slobin (1985), who in turn were inspired by Talmy (1983), Johnson (1987), and Lakoff (1987). Consequently, the process of redescription corresponds to the process of perceptual analysis. Perceptual schemas, before being redescribed by means of language in a propositional form (as conceptual schemas), are redescribed as image schemas. Image schemas are defined as declarative, analogical and nonpropositional knowledge. These representations are rather global in character and are also quite crude. They do not require detailed featural analysis. The major difference between perceptual and image schemas lies in the fact that image schemas contain only fragments of the information originally processed by the perceptual schemas (Mandler, 1992, p. 602). Perceptual analysis (or representative redescription) takes place "on aspects of the input not previously analyzed" (p. 592), "on a new kind of information" (p. 589). Nevertheless, these aspects have been necessarily processed by perceptual schemas, if one takes into account that perceptual analysis is directly based on them and selects only fragments. There results a new kind of information according to Mandler, only in the sense that "a piece of perceptual information is recoded into nonperceptual forms that represent a different format: a vocabulary of meanings" (p. 589). In addition, she also states that the vocabulary of image schemas is a set of elementary meanings (p. 590). Finally, Mandler declares that a brief structural description of a percept (a perceptual schema) can be done by means of an image schema (p. 601). Insofar as perceptual schema can be described by an

146

PIERREMOUNOUD

image schema, I have difficulty understanding how the processing realized by perceptual and image schemas differs, except for the filtering process that takes place between the two schemas. How could the major differences between perceptual and conceptual knowledge be grounded? Although captivating, Mandler's attempt to explain how a nonperceptual understanding of objects develops in opposition to empiricist theories seems to lead to a dead end. She herself considers that Piaget's theory fails to explain in a satisfactory way how sensorimotor schemes are transformed into concepts. (For Piaget, as we will see later, sensorimotor schemas are concepts). She also considers that for Piaget, the emergence of mental images at around 18 months corresponds to the appearance of object concept. Inasmuch as conceptual knowledge is described as a redescription of perceptual schemas, it is difficult to figure out how Mandler could escape to empiricism, unless she considers perception as potential action, as Arbib (1980) has, for example. But nothing similar is found in her articles. On the contrary, she seems to be averse to any attempt to locate the basis of the understanding in action or in what she calls "motor processes," "felt movement of the self," or "bodily experience," without mentioning "manipulating objects" or "physical interactions": concepts she uses in reference to Piaget's theory. Like many other developmental psychologists, she does not conceive of how knowledge related to self-actions could possibly play a major role in cognitive development. We come back to the first assumption outlined in the introduction, which states that self-knowledge is inseparable from object knowledge or includes it. I must confess that if action (knowledge) is limited to manipulation or physical interactions, I have the same aversion as Mandler. But I am convinced that Piaget's ideas related to action are very different from such a depiction. For that reason I will (briefly) present what I consider to be the essence of Piaget's point of view.

Piaget's Point of View As an introduction, it is necessary to specify the meaning of the terms concept and conceptual for both Mandler and Piaget in order to eliminate some misunderstandings. What Mandler tries to explain is the emergence of conceptual knowledge, whose mature form would be expressed through language ("accessible for purposes of recall," "potentially expressible verbally"). Consequently, the emergence of concept corresponds to the emergence of spoken language. For Mandler, a concept is in a way the verbal redescription of previous imagery (analogical) knowledge, which is itself a redescription of perceptual knowledge. In this manner, image schemas (conceptual primitives) constitute an inte~xnediary (representational) level between perceptual schemas and conceptual schemas.

FROM DIRECT TO REFLEXIVE SELF-KNOWLEDGE

147

For Piaget, the existence of concepts is neither bounded to the emergence of speech nor to the emergence of mental images, which only provide the configurational knowledge of the states. Concepts are basically dependent upon the coordinations of actions (material or mental) that supply the knowledge of the transformations (production rules). For Piaget, an image or a word representing an object never constitutes a concept. An image or a .word become concepts only when they are integrated into a transformational system that defines their relations with other images or other words (or even other percepts). In this manner, a sensorimotor scheme is already the practical expression of a concept. A scheme is a concept in the sense that it allows meaning to be conferred to objects. Consequently, for Piaget the problem is not to explain the transition from sensorimotor schemes to concepts, but rather to explain the transition from concepts expressed by practical activities based upon sensorimotor schemes (coordination of actions) to concepts expressed by reasoning based upon logical operations (coordination of internalized actions) or formal operations. The transition from one level of conceptualization to another is due to a process of interiorization as reminded by Mandler but in the interiorization of action schemes (coordination) and not of imitative activities. To conclude, I would like to suggest that all knowledge systems are conceptual systems in a broad sense and that the conceptual knowledge system defined by Mandler is a particular case attuned to the emergence of language. With regard to the structures of the newborn, the problem is somehow more complex. For Piaget, these structures could not be called conceptual because they are considered as "biological," and for Mandler they are nonsymbolic and nonaccessible to consciousness. However, for many authors such as Jackendoff (1992), they have to be called conceptual. Personally I agree with this last statement. Let us now consider how Piaget explains the transition from one knowledge system to another. As already mentioned, he calls upon the process of "interiorization" of the action coordinations (also called general coordinations of action), which gives rise to mental operations precisely defined as interiorized or mental actions. This process has been described by Piaget under the name of reflexive abstraction (or convergent reconstruction with overtaking), which is a recursive process (Piaget, 1967). Karmiloff-Smith's (1991) process of representative redescription has some similarities with the reflexive abstraction process. According to Piaget (1967/1971), this process can be defined in the following way: Reflexive abstraction consists first of becoming conscious of the existence of one of the actions or operations previously made by the subject himself, that is to say, noting its possible interest, having neglected it so far .... Second, the action notes

148

PIERRE MOUNOUD has to be "reflected" (in the physical sense of the term) by being projected onto another plane m for example the plane of thought as opposed to that of practical action ....Third, it has to be integrated into a new structure, which means that a new structure has to be set up, but this is only possible it two conditions are fulfilled: (a) the new structure must first of all be a reconstruction of the preceding one if it is not to lack coherence and continuity... (b) it must also, however, widen the scope of the preceding one, making it more general by combining it with the elements proper to the new plane of thought; otherwise there will be nothing new about it. These, then, are the characteristics of a "reflection," but now we are taking the term in the psychological sense, to mean a rearrangement, by means of thought, of some matter previously presented to the subject in a rough or immediate form (Piaget, 1967, p. 366; 1971, p. 320). It is obvious that when Piaget wrote this definition, he had in mind the

transition from sensorimotor intelligence to concrete thought or from concrete to abstract thought. Nevertheless, taking into consideration the recursive character of the process and the general background of his theory, the definition could be generalized to any transition, in particular to the transition from reflex to sensorimotor schemes. According to Piaget's perspective, "concepts" at the level of representative intelligence only appear in conjunction with the emergence of logical operations (as internalized actions), at around 6 to 7 years of age. In a similar way, concepts at the level of sensorimotor intelligence only appear in connection with the emergence of the general coordinations of actions, at around 16 to 18 months of age. However, there are prior to concepts stricto sensu, preconcepts, which are characterized by the insufficient regulation between their intension and extension. At this point, I must introduce a distinction made by Piaget (1961) between two categories of knowledge instruments, respectively called "operative" and "figurative" instruments. Operative instruments are those that provide knowledge about transformations (mainly the schemes or the operations); in contrast, figurative instruments supply knowledge about states of reality or the results of transformations. The figurative instruments correspond to three types of signifiers as defined by Piaget: perceptual indices, mental images, and abstract symbols. For him, these two categories of instruments are inseparable, but they are dissociated in order to facilitate the analytic description. Nevertheless, he has treated them as if they could exist independently of each other. The distinction between k n o w l e d g e about states and k n o w l e d g e about transformations is essential because it makes it possible to understand correctly the notions of action schemes and mental operations, as well as the importance attributed to k n o w l e d g e about actions. By k n o w l e d g e about actions as transformation, Piaget refers to the ability of understanding the connection between two successive states of a situation or of an event. It could also refer to the

FROM DIRECT TO REFLEXIVE SELF-KNOWLEDGE

149

knowledge about production rules (generative rules, i.e., grammar), logical rules, physical laws, or statistic laws. It may be useful to give some examples in order to illustrate this distinction between knowledge about states and transformations and their connections. The scheme or concept of support (prototype of means-end coordination mastered by the infant at around 8 months of age),, which consists of pulling a support in order to reach for an object laying down on it, corresponds to the mental structure that makes possible the connection of the action-transformation "pushpull" to the states "being out of reach" and "being reachable." The scheme integrates knowledge about the states "being on" and "being beside," in a transformational system. At a more analytical level of explanation, it is also possible to describe the various states "being on" and "being beside" as linked by more elementary coordinated action-transformations such as "laying down" or "lifting up," etc. The scheme of face recognition, acquired at around 2 or 3 months of age (which succeeds a scheme already present at birth), corresponds to the mental structure that connects the various states of a face defined by configurations of perceptual indices (front view, side view, etc.) to action-transformations (head rotations, subject's or object's rotation). The scheme of (shape or) size constancy is the insertion of the various sizes of an object related to its distance from the perceiver in a transformational system (system of transformations) governing the moves of the object. Present at birth, it can be reconstructed during the first months of life. The scheme of object permanence (the "objective" form), achieved at around 16 to 18 months of age, is the mental structure that connects the various successive states of a set of objects (their different localizations or relative positions) to their successive displacements (transformations) when these displacements are organized or structured by the subject into a system. I would like to point out that for Piaget, there is already at birth a "practical" form of object permanence (as opposed to the "objective" one), revealed in particular by the capacity of the newborn to recuperate the nipple when lost, i.e., to adequately rotate his/her face (transformations) in order to modify the "state" of the nipple from "out of the mouth" to "in the mouth" (Piaget spoke of the sucking reflex scheme, but as a matter of fact he was referring to the rooting reflex). This permanence presupposes, as stated by Rochat & Morgan (this volume), some implicit knowledge about the mouth. This was for Piaget a practical form of permanence inherent to functioning, as opposed to the "objective" permanence produced by functioning. I hope these examples clarify the nature and importance given by Piaget's theory to knowledge about actions (about the general coordinations of actions) as

150

PIERRE MOUNOUD

systems of transformations. It is the structure of action coordination or the coordinate actions that connect the successive states of a situation and define its invariants. It is more the whole organization or structure than the actions themselves. It is also in that sense that we are dealing with something more abstract than manipulations or physical interactions, as referred to by Mandler and many other developmental psychologists. For Piaget's theory, the states of the world (the successive states of a given reality) are understandable only as far as they are connected or linked with the transformations that generated them (the production rules). For him, this knowledge about transformations could not be directly derived or extracted from perception. If the 8-month-old baby searches for an object placed under or behind another one, it is not because s/he has perceived the occlusion of one by the other (the information is not incorporated in the structure of the visual flow), but because s/he is able to organize in a system the respective displacements between the two objeets. Similarly, if the newborn again finds the nipple, it is not because s/he perceived it escaping from her/his mouth, but only because s/he possesses a structure (called the "sucking reflex" by Piaget) that coordinates her/his actions with her/his perceptions. The newborn can compensate a displacement (head rotation) by its inverse. In addition, these coordinated sucking activities allow the baby to understand the states of "being inside" or "being outside" the mouth and the relationships that connect them. This partly corresponds to the image schema of containment as defined by Mandler, following Johnson (1987) and Lakoff (1987). Next, I shall analyze in detail the genesis of this concept of containment as described by Mandler.

The Concept of Containment: A Comparative Analysis of Mandler's and Piaget's Viewpoints In order to be more concrete, I will present in a critical way the origin of the concept of containment as described by Mandler (1992) and then contrast her view with Piaget's. The concept of containment is the capacity to understand that a given object (the container) can contain another one. The container may be a part of the body such as the mouth or the hand, or an object. According to Mandler, the concept of containment appears in the infant at around 5 months of age as an image schema (conceptual primitive). First, for Mandler this concept is bounded to notions such as "going in" or "going out," "opening" and "closing." These notions are precisely what Piaget calls knowledge about action as transformation. Thus, the infant must have as a prerequisite some knowledge related to these actions and to their meanings. This corresponds to the first step of the reflexive abstraction process: The subject has to

FROM DIRECT TO REFLEXIVE SELF-KNOWLEDGE

151

notice or to become conscious of the existence of and possible interest of some actions produced by him or her. Second, according to Mandler (in reference to Lakoff, 1987), the image schema of containment has three structural elements: interior, boundary, and exterior. These elements are based on two different capacities: the capacity to consider objects as having boundaries and an inside separated from an outside (Spelke, 1988) and the capacity to consider objects as being in or out of a container. For Piaget, this is related to the knowledge of three distinct "states" that can be understood (processed and stored) independently from the knowledge related to the actions that produced them. Knowledge about states results from perceptual activities that constitute configurations of perceptual indices (as configurations of haptic and tactile indices corresponding to the perception of an object in the mouth, out of the mouth, or only in contact with the mouth; or configurations of visual indices corresponding to the perception of an object inside, outside, or in contact with another one). In addition, Mandler mentions that according to Johnson (1987), bodily experience can be the basis of the understanding of containment. Nevertheless, she is not convinced that bodily experience is a necessary condition for perceptual analysis. She considers that it is "easier to analyze the sight of the milk going into and out of a cup than milk going into and out of one's mouth." But she does not explain the reasons of this relative ease. However, she concedes that "food as something taken into the mouth" could be an early conceptualization of containment (Mandler, 1992, p. 597). Third and finally, the concept of containment would result from the cluster of related image schemas. For Piaget, the sensorimotor scheme or concept of containment results from the coordination (the cluster!) of elementary (noticed) actions (going in, going out, opening, closing, etc.) themselves, connected with the corresponding resulting states of reality (being inside, being outside, being opened, being closed, etc.). The coordination of actions gives access to the transformation rules, to what organizes the successive changes of states, to what gives them a meaning. From my point of view, there is already at birth schemes or concepts of containment belonging to the first knowledge system (sensorial). Other concepts elaborated by means of the second knowledge system (perceptual) would follow the first ones. Taking into consideration (self-)knowledge about actions in the construction of new systems seems to me the only way to locate the origin of knowledge "above or beyond what the object looks like" (i.e., beyond the perceptual similarities and differences, Mandler, 1988, p. 118; 1992, p. 595). This is the reason why Piaget has always been opposed to empiricism. Knowledge is not mainly derived from perception, but from the understanding of (self-)actions as transformations (which I consider essential to the construction of "conceptual

152

PIERRE MOUNOUD

primitives," in spite of Mandler's reluctance). Actions as transformations cannot be reduced to the concepts of motion, manipulation, or interaction. Knowledge about actions is more directly connected with concepts such as plans, programs, templates, potential actions, or more generally to the whole organization of actions, the coordinative structures as labeled by Bernstein (1967). The concept of body effectivities suggested by Rochat (this volume) is another attempt to specifiy this complex notion.

Limits of the Piagetian Approach The preceding pages constitute an argument in favor of Piaget's ideas on the knowledge about action coordination or logico-mathematical experience. Nevertheless, this conception has several gaps, which explains at least in part, why it has been rejected and distorted. The major problems of Piaget's theory are due to his radical structuralist approach. When Piaget defined the structures underlying behaviors, he did so by taking exclusively into consideration actions (physical or mental) independently of the object categories (the various contents) to which they apply. In other words, he defined the knowledge about actions (operative knowledge) independently of the knowledge about states (figurative knowledge) or more precisely in relation to abstract, nonspecified states. Actiontransformations have also been characterized in a very abstract way and have been qualified for example as "direct," "inverse," "reciprocal," and "correlative." These transformations have lost all of their functional dimensions and above all their meanings, and have been reduced to their logical aspects. Whole structures have been defined in order to explain very general competencies without any specificity and functional values. The difficulties and criticisms generated by this approach are well known. By rejecting whole structures, researchers also rejected or lost the ideas related to action-transformations as a possible base for the production rules at the origin of our behavior and also of a possible basis to understanding cognitive development. The central concept of actions as transformations has been denatured and reduced to the trivial idea of manipulations and physical interactions that correspond to what Piaget has called the "physical experience" or the empirical abstraction (also called simple or Aristotelian abstraction), to which he conceded very limited credit. The essence of cognitive development for Piaget is based on the logico-mathematical experience as related to actions and corresponding to the reflexive abstraction process. Thanks to his radical structuralism, Piaget was able to formalize very different types of behaviors by means of the same formal structure. Thus, at the sensorimotor level, he defined the behavior of the newborn and the 18-month-old

FROM DIRECT TO REFLEXIVE SELF-KNOWLEDGE

153

with the same structure (Piaget, 1937). Because his goal was to demonstrate the emergence of new formal structures, he was forced to qualify them as "practical" and "objective" in order to differentiate them. The structure related to the reflex schemes was called "practical" and "biological." It was described as inherent to the action, as opposed to the "objective" one related to the sensorimotor schemes that are considered to be produced by the action and qualified as mental or psychological. A similar opposition has been suggested more recently by Karmiloff-Smith (1991), who compares knowledge in the humain brain, not accessible to consciousness, biologically specified (knowledge in the cognitive systems, embedded in procedures) to knowledge accessible to other parts of the brain as data structure. From my point of view, it is possible to say that any knowledge system is biologically determined and inherent to actions. The knowledge system manifested by the newborn results from a phylogenetic construction and does not radically

differ from the other systems. To increase the confusion, Piaget subsequently relabeled the "objective" structures as "practical" without giving any explanation for this major change (e.g., Piaget, 1947). I have discussed this problem in various articles (see Mounoud, 1979, 1993a; Mounoud & Hauert, 1982; Mounoud & Vinter, 1981; see also Hauert, 1980, 1990; Vinter, 1985, 1990). Finally, I have to mention that the majority of the general coordinations of action that Piaget described in the course of the sensorimotor stage is already part of the newborn's repertoire, such as for example hand and mouth coordination, visual and manual coordination, means-end coordination, the actions of adding and substracting, etc. These are some of the reasons that led me to consider the structures defined by Piaget as preformed (Mounoud, 1979). From my point of view, it is obvious that knowledge about actiontransformations cannot be defined separately from knowledge about the states of objects and that both cannot be characterized in such an abstract way as Piaget did. Furthermore, meanings only result from the connection between knowledge about states and transformations expressed by the functional properties of actions. The concept of body effectivities suggested by Rochat (this volume) integrates these two kinds of knowledge. Apart from their formal dimensions, actions have functional properties at the origin of the meanings manifested in the behavior or attributed to the objects. Consequently, the origin of the concept of containment is not to be exclusively found either in the perception of the object (in the perceptual analysis of the object) or in the general coordination of the actions of opening and closing the hands or the mouth, but in the discovery of the functional properties of these actions (their meanings) such as grasping or releasing objects, containing or

154

PIERRE MOUNOUD

being contained, and of the values attached to these functions. This is done by means of implicative relations, declarative representations, or predicative functions. Conceptualization or thought precedes action in cognitive development. To conclude, it is possible to say that if Mandler gave an exclusive role to perceptual activities in the elaboration of knowledge systems, Piaget gave an exclusive role to the formal aspects of the action coordination. Below I will suggest a point of view that tries to combine these two aspects and also introduces the functional aspects of actions.

Do Knowledge Systems Include Knowledge About Actions? At this point, we must discuss whether the knowledge systems or the conceptual structures include only perceptual data or if they integrate other data, such as knowledge related to transformations (production or generative rules) as manifested in actions. Do representations incorporate potential actions, action plans, or knowledge about the effectivities of the body? According to Piaget, transformations (or production rules) are initially only accessible to the subject by means of his/her own actions. Consequently, it is important to specify if knowledge about self-actions is exclusively based on perceptual information or the perceptual analysis that accompanies them, or if the organization itself of the action, its planning, could be at least partly available. If knowledge about actions is only accessible through perception, then it still has exclusively a perceptual origin. On the contrary, if knowledge about actions is represented in the form of rules, plans, programs, strategies, recall schemas, etc., then there could be some good reasons to search in that direction for the origin of an essential part of our knowledge, which precisely concerns transformations. To conclude this point, I will briefly refer to an idea recently suggested by Jackendoff (1992) concerning conceptual structures. For him, one of the components of innate conceptual structures could be body representation, which encodes internal states of muscles and joints and the locus and character of body sensations. With such a definition, the body is exclusively considered on the perceptual side or in Piaget's terminology, with regard to knowledge about states without any connection to knowledge about transformations. As I mentioned in the introduction, body schema is not only related to perceptual configurations (featural knowledge), but includes a system of transformations as well that connects the configurations usually called postural schema. For Jackendoff, body representation is a way station between the intention to act (part of the conceptual structure) and the form of motor commands. I consider the intention to act as a concept very close to knowledge about actions, and I am very much in

FROM DIRECT TO REFLEXIVE SELF-KNOWLEDGE

155

favor of introducing it in the conceptual structure. One could say that the reflexes and sensorimotor schemes are basically intentions to act or potential actions. In my opinion, the newborn has a preformed conceptual structure (called "perceptual"), which includes a body representation (knowledge about states and transformation; Mounoud, 1979, 1981). This perspective diverges from Piaget's and Mandler's points of view. Now we still have to deal with the crucial problem related to the construction of new conceptual systems. The problem will not be settled in the same way according to the position regarding the initial state of the newborn, but this issue will not be discussed here.

The Construction of New Knowledge Systems According to Mandler, what are the necessary ingredients needed in order to explain the development of new knowledge systems? As we have already seen, the redescription of perceptual schemas into image schemas should be sufficient. However, image schemas provide only knowledge about states or configurations (which can include motion perception as changes of states), but this knowledge excludes the understanding of transformations inherent in action coordinations. All of the knowledge related to the "intention to act," "potential actions," the "effectivities of the body," and action plans is missing. Consequently, in order to (re)construct a (new) knowledge (or conceptual) system, it is necessary to postulate, in addition to the construction of image schemas, the construction of abstract schemas (propositional in nature), which provide knowledge about actiontransformations. In such a perspective, the infant becomes conscious not only of fragments of the information originally processed by the perceptual schemas recoded into image schemas (featural knowledge of featural self) as stated by Mandler, but also of some intentions to act or action planning (motion knowledge or motion self-knowledge). In that manner, s/he will discover not only the successive states of an object that "goes away" or "comes near" her/him (variations in the apparent size) or that turns around (variations of shape), but also transformations (production rules) that connect and explain them (extension and flexion of the elbow, rotation of the wrist). (By bending or stretching my elbow, I can bring near or move away my hand or an object held and can vary its apparent size. By twisting my wrist, I can rotate my hand and modify its apparent shape). Such representations must be declarative in nature and could be expressed or formulated in a more abstract way. The new representations (image and abstract schemas) would be initially partial and elementary. Their coupling would form the connection between the knowledge of states and transformations (elementary) at the origin of meaning (the

156

PIERRE MOUNOUD

effectivities of the body). Thus, the infant discovers (becomes conscious of, notices) some aspects among the whole set of aspects processed by his/her first constituted knowledge system (sensorial), as for example his/her hand as one part of his/her body that could be in various states (opened or closed, prone or supine, near or far away) as well as in the action-transformations that connect them (opening and closing, rotating, bending and stretching). In a similar way, the infant discovers his/her milk bottle as an object that could be in various states (stand up or reverse, fully or partially visible, etc.) and the transformations that connect them (turning up, turning down, turning back, moving behind or in front, etc.). To summarize, it is by the construction of abstract schemas (representations) propositional in nature and of image schemas (representations) analogical in nature that the infant reconceptualizes the states and transformations of objects, of others, and of her/himself (her/his body), which define the events to which s/he is confronted. For many years, I have thought that these two types of schemas (analogical and propositional) followed one another within a stage (Mounoud, 1986b). More recently (Mounoud, 1990; 1993a), I have begun to consider that they may coexist initially in a partially dissociated way and may later on operate in a combined, integrated way. The conceptual (sensorial) system of the newborn is also composed of image schemas like the 3D model structures of Marr (1982), including representations of the various states of the body; and of abstract schemas, including representations of the action-transformations of the body.

Conclusion

The exceptional abilities of humans to modify behavioral determinants during development can be explained by the emergence at various periods and in particular at birth of new coding capacities. These new capacities force the organism to retranslate, redefine, reinterpret, and reformulate some of the information accessed, that is to say, to construct new representations, new frames of reference, and new categories. The construction of these representations is made through a relatively slow and complex process requiring a few years. I have described several times and again recently this construction process, so it will not be presented here (see Mounoud, 1984, 1986a and 1986b, 1988, 1992, 1993a, 1993b, in press; see also Vinter, 1990). According to this perspective, the newborn's exceptional competencies are explained by preformed representations qualified as sensorial or sensorimotor. They account for the intersensorimotor coordinations that characterize the newborn's behavior. During their first weeks, infants behave in certain situations as if the

FROM DIRECT TO REFLEXIVE SELF-KNOWLEDGE

157

surrounding world and their body are meaningful: Numerous stimuli constitute for them organized patterns of information in response to which they produce organized action patterns (e.g., early prehension, imitation, direct knowledge). But more or less simultaneously, infants behave as if the situations they confront constitute "polymorphous sets" or a "confusing and ambiguous universe" without precise functional meaning (in other words, with problematical situations), as, for example, in their awkward attempts to reach for an object between the 2nd and 5th months (from approximately the 6th to the 20th week) or in their unskillful attempts to retrieve a hidden object (the A-not-B error) between the 8th and 10th months (all of these situations can be characterized by a disequilibrium state). Thus, infants need several months to be able to recategorize situations and reorganize or replan their actions. It is not before 6 months that infants are able to grasp in a partly adapted manner a visually perceived object; not before 1 year that they succeed in regulating or in accurately planning in advance the orientation and the shaping of the hand as a function of the size and orientation of the object; not before the age of 16 to 18 months that their grasp begins to be regulated as a function of the object weight inferred from its size and/or texture; not before 20 to 24 months that their prehension adjusts to reciprocal orientation between two objects; and not before 36 months that they fit together five cups of different sizes (i.e, Greenfield et al., 1972; Hofsten, 1989; Lockman, 1990; Mounoud, 1983). It seems as if the infant possesses at birth action procedures (or procedural sensorimotor representations) adapted to a set of situations (direct knowledge). These representations are by nature unconscious, or they relate to a nonreflexive consciousness (cf. Marcel, 1983), as all automatic or automatized behaviors can be considered unconscious in nature. The emergence of new coding capacities causes the infant to elaborate new representations that I have called "perceptive" and that go along with reflexive knowledge. Another way of expressing the same story is to say that infants, in the course of their development, construct knowledge (or concepts) that must lead them to construct new know-how. Development is therefore a matter of shifting not only from direct know-how to reflexive knowledge (as argued by Piaget), but also, and in an equally large extent, shifting from reflexive knowledge to new, unconscious know-how. Rey (1934) spoke of the withdrawal of active intelligence during automatization processes. It is in this way that new know-how new skills are learned and automatized (prehension, walking, imitation, localization, etc). To conclude, it is possible to consider development as an alternmion between: a) periods of adaptation (adaptation in the different domains is more o1" less optimal according to the experiences realized); and b) periods of reorganization. Periods of adaptation are characterized by automatized behaviors that can be described as reactive or interactive (direct knowledge); periods of reorganization are characterized

158

PIERRE MOUNOUD

by transitional conscious activities that provide to the infant's behavior an active and intentional character (reflexive knowledge). These functioning modes depend on infants' planning abilities and vary as a function of infants' developmental level and the situations confronting them.

ACKNOWLEDGMENTS

I thank Frangoise Schmitt for her valuable secretarial assistance and C. A. Hauert, D. Page, and P. Zesiger for their very helpful comments. REFERENCES

Ajuriaguerra, J. de. (1976). Manuel de Psychiatrie de l'enfant. Paris: Masson. Arbib, M.A. (1980). Perceptual structures and distributed motor control. In V.B. Brooks (Ed.), Handbook of physiology: Vol. III. Motor control. Bethesda, MD: The American Physiological Society. Bernstein, N.A. (1967). The coordination and regulation of movements. Oxford: Pergamon Press. Dubois, B., Pillon, B., & Sirigu, A. (1994). Fonctions int6gratices et cortex pr61frontal chez l'homme. In X. Seron & M. Jeannerod (Eds.), Neuro-psychologie Humaine (pp. 453-469). Liege: Mardaga. Gallup, G.G. (1970). Chimpanzees: Self-recognition. Science, 167, 86-87. Gibson, J.J. (1966). The senses considered as perceptual systems. Boston: HoughtonMifflin. Gibson, J.J. (1979). The ecological approach to visual perception. Boston: HoughtonMifflin. Greenfield, P.M., Nelson, K., & Saltzman, E. (1972). The development of rulebound strategies for manipulating seriated cups: A parallel between action and grammar. Cognitive Psychology, 3, 291-310. Hauert, C.A. (1980). Propri6t6s des objets et propri6t6s des actions chez l'enfant de 2 5 ans. Archives de Psychologie, 48, 95-168. Hauert, C.A. (Ed.). (1990). Developmental psychology: Cognitive, perceptuo-motor, and neuropsychological perspectives. Amsterdam: North Holland. Head, H. (1920). Studies in neurology. Oxford: London. Hecaen, H., & Ajuriaguerra, J. de. (1952). M~connaissances et hallucinations corporelles. Vol 1. Paris: Masson. Hofsten, C. von. (1989). Transition mechanisms in sensorimotor development. In A. de Ribaupierre (Ed.), Transition mechanisms in child development: The longitudinal perspective (pp. 233-258), Cambridge: Cambridge University Press. Jackendoff, R. (1992). Languages of the mind: Essays on mental representation. Cambridge, MA: The MIT Press. Johnson, M. (1987). The body in the mind: The bodily basis of meaning, imagination, and reasoning. Chicago: University of Chicago Press. Karmiloff-Smith, A. (1991). Beyond modularity: Innate constraints and developmental change. In S. Carey & R. Gelman (Eds.), The Epigenesis of Mind: Essays on Biology and Cognition (pp. 171-197). Hillsdale, NJ: Erlbaum. Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind. Chicago: University of Chicago Press.

FROM DIRECTTO REFLEXIVESELF-KNOWLEDGE 159 Lockman, J.J. (1990). Perceptuo-motor coordination in infancy. In C.A. Hauert (Ed.),

Developmental psychology:Cognitive, perceptuo-motor, and neuropsychological perspectives (pp. 85-111), Amsterdam: North Holland. Mandler, J.M. (1988). How to build a baby: On the development of an accessible representational system. Cognitive Development, 3, 113-136. Mandler, J.M. (1992). How to build a baby: II. Conceptual primitives. Psychological Review, 99(4), 587-604. Marcel, A.J. (1983). Conscious and unconscious perception: An approach to the relation between phenomenal experience and perceptual processes. Cognitive Psychology, 15, 238-300. Marr, D. (1982). Vision. San Francisco: Freemam. Mounoud, P. (1979). D6veloppement cognitif: Construction de structures nouvelles ou construction d'organisations internes. Bulletin de Psychologie, 33, 343, 107118. Mounoud, P. (1981). Cognitive development: Construction of new structures or construction of internal organizations. (Author, Trans.). In I.E. Sigel, D.M. Brodzinsky, & R.M. Golinkoff (Eds.), New directions in Piagetian theory and practice (pp. 99-114). Hillsdale, NJ: Erlbaum. (Original work published 1979) Mounoud, P. (1983). L'6volution des conduites de pr6hension comme illustration d'un module du d6veloppement [Evolution of reaching behaviors as an illustration of a developmental model]. In S. de Sch6nen (Ed.), Les d~buts du d~vetoppement (pp. 75-106). Paris: Presses Universitaires de France. Mounoud, P. (1984). A point of view on ontogeny. Human Development, 27, 329334. Mounoud, P. (1986a). Action and cognition: Cognitive and motor skills in a developmental perspective. In M.G. Wade & H.T.A. Whiting (Eds.), Motor Development in Children (pp. 373-390). Dordrecht: M. Nijhoff. Mounoud, P. (1986b). Similarities between developmental sequences at different age periods. In I. Levin (Ed.), Stage and structure (pp. 40-58). Norwood, NJ: Ablex. Mounoud, P. (1988). The ontogenesis of different types of thought. In L. Weiskrantz (Ed.), Thought without language (pp. 2545). Oxford: Oxford University Press. Mounoud, P. (1990). Cognitive development: Enrichment or impoverishment? In C.A. Hauert (Ed.), Developmental psychology: Cognitive, perceptuo-motor, and neuropsychological perspectives (pp. 389-414). Amsterdam: North Holland. Mounoud, P. (1992). Les concepts d'6quilibration et de structure chez Piaget dans "La Naissance de rlntelligence" (1936) et "La Construction du R6el" (1937). In D. Maurice & J. Montangero (Eds.), Equilibre et dquilibration dan,s l'oeuvre de Jean Piaget et au regard de courants actuels (Cahier No. 12, pp. 31-43). Gen~ve: Fondation Archives Jean Piaget. Mounoud, P. (1993a). The emergence of new skills: Dialectic relations between knowledge systems. In G.J.P. Savelsbergh (Ed.), The development of coordination in infancy (pp. 13--46). Amsterdam: North Holland. Mounoud, P. (1993b). Les r61es non sp6cifiques et sp6cifiques des milieux dans le d6veloppement cognitif. In J. Wassmann & P. Dasen (Eds.), Les savoirs

quotitdiens: Les approches cognitives dans le dialogue interdisciplinaire.

Fribourg: Presses Universitaires. Mounoud, P. (in press). A recursive transformation of central cognitive mechanisms: The shift from partial to whole representation. In A. Sameroff & M. Haith (Eds.), Reason and Responsability: The pcL~sage through childhood. Chicago: University of Chicago Press.

160

PIERRE MOUNOUD

Mounoud, P., & Hauert, C.A. (1982). Development of sensorimotor organization in young children: Grasping and lifting objects. In G.E. Forman (Ed.), Action and thought: From sensorimotor schemes to symbolic operations (pp. 3-35). New York: Academic Press. Mounoud, P., & Vinter, A. (1981). Representation and sensorimotor development. In G. Butterworth (Ed.), Infancy and epistemology: An evaluation of Piaget's theo~ (pp. 200-235). Brighton: Harvester Press. Neisser, U. (1993). The self perceived. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 3-21). Cambridge, MA: Cambridge University Press. Piaget, J. (1937). La construction du rdel chez t'enfant. Neuchfitel: Delachaux & Niestl6. Piaget, J. (1947). La psychologie de l'intelligence. Paris: A. Colin. Piaget, J. (1961). Les mOcanismes perceptifs. Paris: Presses Universitaires de France. Piaget, J. (1967). Biologie et connaissance. Paris: Gallimard. [Biology and knowledge: An essay on the relations between organic regulations and cognitive processes] (B. Walsh, Trans.). Chicago: University of Chicago Press, 1971. Pick, A. (1922). St/3rung der Orientierung am eigenen K~rper. Psychol. Forsch., 1, 303. Rey, A. (1934). L'intelligence pratique chez l'enfant. Paris: Alcan. Schilder, P. (1968). L'image du corps. Paris: Gallimard. (Edition originale 1935) Slobin, D. (1985). Crosslinguistic evidence for the language-making capacity. In D. I. Slobin (Ed.), The crosslinguistic study of language acquisition: Vot. 2. Theoretical Issues (pp. 1157-1256). Hillsdale, NJ: Erlbaum. Spelke, E.S. (1988). Where perceiving ends and thinking begins: The apprehension of objects in infancy. In A. Yonas (Ed.), Perceptual development in infancy (pp. 197-234). Hillsdale, NJ: Erlbaum. Talmy, L. (1983). How language structures space. In H.L. Pick, Jr. & L.P. Acredolo (Eds.), Spatial orientation: Theory, research, and application. New York: Plenum Press. Thatcher, R. W. (1994). Cyclic cortical reorganisation. Origins of human cognitive development. In G. Dawson, & K.W. Fischer (Eds.), Human behavior and the developing brain (pp. 232-266). New-York: Guilford. Vinter, A. (1985). L'imitation chez le nouveau-nd. Paris: Delachaux & Niestl6. Vinter, A. (1990). Sensory and perceptual control of action in early human development. In O. Neuman & W. Prinz (Eds.), Relationships between perception and action." Current approaches (pp. 305-324), Berlin: Springer. Wallon, H. (1959). Kinesth6sie et image visuelle du corps propre chez l'enfant. Enfance, 3-4, 252-263.

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

161

CHAPTER 9

The Unduplicated Self DANIEL J. POVINELLI

University of Southwestern Louisiana New Iberia Research Center

"The present...is really a part of the past n a recent p a s t - delusively given as being a time that intervenes between the past and the future. Let it be named the specious present, and let the past...be known as the obvious past." - E.R. Clay

If pressed, developmental psychologists offer 18-24 months as the period in which humans infants develop a self-concept, citing this as the age at which infants are able to recognize themselves in mirrors (Amsterdam, 1972; Lewis & BrooksGunn, 1979; for other achievements suggesting this as the onset of selfconception, see Kagan, 1981). Although contributors to this volume may not completely disagree with this view, most are united in the belief that long before 18 months, human infants process and store many kinds of information about themselves m a belief that highlights the need for understanding the preconceptual origins of the self in early infancy. In their view, the kinds of self-conceptual behaviors that 18- to 24-month-olds display (such as self-recognition in mirrors) are seen as the expression of an already sophisticated self-representational system. In this chapter, I shall steer a narrow path between these two caricatures of the emergence of the self-concept. On the one hand, I will acknowledge the sophistication of the 18- to 24-month-old's self-knowledge by contrasting it with preconceptual forms of knowledge present much earlier in infancy. Indeed, I will defend the proposition that by 18-24 months, most children have developed a qualitatively new system for encoding information about the self. On the other hand, I will emphasize the immaturity of the 2-year-old's self-knowledge by offering the hypothesis that a striking difference exists between the 18- to 24month-old's self-conceptual system and that of older preschool children. I outline a theoretical model which argues that once the young child's conception of self is in place at 18-24 months, it continues to develop from an initial system of self-

162

DANIELJ. POVINELLI

representations that is largely restricted to the here and now (the "present self") to a later system that is capable of temporal differentiation (the "proper self'). I propose that these two hypothetical self-constructs emerge asynchronously during the course of development. It is the latter construct that allows the child to knit together historical instances of him- or herself into a unique, unduplicated self. It is what, in William James' (1890/1950) words, allows our consciousness to say, "I am the same self that I was yesterday" (p. 332; italics in original). Throughout this essay, I will provide empirical grounding for the theoretical discussion by examining the reactions of human infants and children, as well as chimpanzees and other nonhuman primates, to visual presentations (and representations) of themselves (e.g., seeing themselves in mirrors). I take this approach not because I believe that mirrors are the only methodological tool for understanding the development of the self m they most certainly are not. Instead, I focus on mirrors because the different kinds of reactions that live versus delayed visual feedback of the self provoke from human infants and children of various ages, as well as other species, provide an important window into the architecture of their knowledge about the self. In addition, the self-recognition paradigm lends itself easily to our comparative program, which seeks to identify homologous cognitive developmental pathways among the great apes and humans. Furthermore, I shall maintain that self-recognition in mirrors reflects a much deeper understanding of the self than simply knowing what one looks like.

Representational Development The foundation of my theoretical model of the development of the self-concept is that its emergence at 18-24 months is but one manifestation of a more general capacity for representation. In other words, although the self-concept is an important domain of inquiry for developmental psychologists, its development is by definition not possible until the capacity for representation proper emerges. In order to sustain such a view, I need to provide a framework for thinking about the development of representation in human infancy. The formal account of transformations in the infant's representational abilities that I adopt is derived largely from the theoretical work of Olson & Campbell (1993), although the particular application of this to self-representation is my own. Representations versus Schemata During infancy, humans develop the capacity to construct action schemata. Schemata can be thought of as neural control programs that are triggered by stimuli in the external world. These programs control motor output, and hence

THE UNDUPLICATEDSELF 163 they warrant a description as causal structures. Throughout the first 18 months of life, the human infant advances from the use of relatively simple action schemata (reaching, grasping, head-turning) to the later elaboration and complex deployment of such schemata (one hand reaches to open a box, the other reaches inside the box, grasps an object, pulls it toward the mouth). Gradually, such schemata become automatized and can be deployed readily and in rapid succession in the presence of the relevant stimuli. According to sophisticated rules of generalization, these schemata can be triggered in appropriate "novel" contexts. The progressive construction of these behaviors requires an increasingly sophisticated distinction between self and the environment, presumably mediated by proprioceptive feedback. But despite their eventual sophistication, these internal neural structures are not representations of the external world: Schemata are activated by a present object or event; they are causally connected to that object or event, but they do not represent that object or event. Their activity is tied to stimulating conditions; consequently, there is no need for the activating condition to be held in mind. It is present, and therefore present to the mind (Olson & Campbell, 1993, p. 15). Thus, despite the complexity of the behavior that these early schemata can execute, they are not representations p r o p e r - they are not held in mind in the absence of the objects of perception that trigger them in such a way that they can be intrinsically connected (see below) to other schemata. Olson and Campbell (1993) conclude that it is not until 18-24 months that mental representation the ability to create an intentional connection between a schema (in the absence of its stimulating condition) and another object or event in perception becomes possible. On this view, earlier emerging demonstrations of object permanence using habituation-dishabituation techniques (e.g., Baillargeon, 1987) would not be interpreted as evidence of an active ability to create an intentional relation between two schemata, although it may be evidence of early implicit connections among schemata (see Olson, 1993). Although I recognize this point to be debatable, I shall stipulate this implicit/explicit distinction without further justifying it. A wide range of behaviors blossom in the child at this age, including symbolic play, mirror self-recognition, simple acts of altruism, self-descriptive utterances, mastery smiles, statements about desires, sophisticated gaze-following, linguistic markers of the failure to self-generated plans, acts of intentional cooperation, development of the self-conscious emotions, an explosion in naming skills, second-order classification skills, and an understanding of referential focus, just to name a few (Leslie, 1987; Amsterdam, 1972; Lewis & Brooks-Gunn, 1979; Zahn-Waxler & Radke-Yarrow, 1982; Kagan, 1981; Bartsch & Wellman, 1995; Bischof-KOhler, 1988; Butterworth & Jarrett, 1991; Gopnik, 1982; Brownell & Carriger, 1990; Lewis, Sullivan, Stanger, & Weiss, 1989;

164

DANIELJ. POVINELLI

Macnamara, 1982; Langer, 1986; Baldwin, 1993). In contrast to schemata, mental representations are a means of "maintaining a relation [an intentional connection] with an object or event in its absence" (Olson & Campbell, 1993, p. 14; italics in original). (Note that this meaning of representation captures its linguistic origins as the "re-presentation" of an object or event in its absence.) One defining feature of the capacity for representation is the ability to actively create relations between things directly perceived and things only conceived. Such relations are called propositions. Propositions can embody either linguistic or imaginal relations between separate schemata. Thus, with the advent of representational ability, the infant now has the capacity to construct many types of propositions. For example, Olson (1993) speculates that the explosion in naming skills displayed by 18- to 24-month-olds is the direct result of their new-found ability to form propositions, expressed as instance or category relations: "This [object of perception] is a ball [held in mind]." Similarly, classification becomes possible on the basis of hierarchical categories: "This ball [object of perception] is a toy [held in mind]." Other relations become possible as well: "The ball [held in mind] is under that cup [object of perception]." Also, and critical to my later hypothesis, causal attribution should become possible: "This event [object of perception] was caused by that action [held in mind]." Representing Propositions Olson (1993) has argued that there is a further elaboration of the child's capacity for representation between 3 and 5 years (see also Olson & Campbell, 1993). He argues that, whereas by 18-24 months infants have the resources to hold in mind one schema (independent of the environmental context that stimulates it) while their perceptual system attends to something else and creates an intentional relation between the two, by 4 years of age children develop the ability to represent the propositions themselves. One consequence of this is that it allows propositions to be evaluated (true or false) in relation to perceived states of affairs in the world. In this view, these representations are not limited to mental states, but in fact are part of a broader developmental transition. Olson (1993) has used this model to explain how once a young child has the conceptual resources to hold in mind more than one representation at a time, he or she is able to pass most standard theory-of-mind tasks. Indeed, I should be careful in committing to a particular version of how the child becomes capable of embedding representations within representations. In recent years, a number of proposals have been put forward to account for changes during the preschool years in young children's ability to cope with representations (Flavell, 1988; Forguson & Gopnik, 1988; Pemer, 1991; Lillard, 1993). As a case in point, consider Flavell's (1988)

THE UNDUPLICATED SELF

165

interpretation of his research on the development of young children's understanding of the appearance-reality distinction: Children of this age [3-year-olds] also believe, as we generally do, that each object or event in the world has only one nature m one "way it is" - - at any given point in time. It cannot be two or more very different, mutually contradictory, and i n c o m p a t i b l e things at the s a m e time; rather, it can only be one thing. Consequently, it makes no sense to them to hear something described as being radically different than the single way it "is" (with "is" not differentiated from "seems to them at the moment") (Flavell, 1988, p. 245).

Olson's (1993) theoretical position offers one account of the underlying reason for this inability of children younger than about 3 or 4 years of age to cope with such situations. But regardless of the particular view to which one subscribes, once the child achieves the functional ability to understand that individual things can have multiple states, profound changes quickly follow in their understanding of objects, events, and mental states. I shall argue that this transition has an equally profound impact on the child's conception of self as an entity with a personal history and future.

The Present Self Based on these foundations, let us assume that from birth forward (the period covered by most contributors to this volume) the developing infant elaborates upon the construction and deployment of schemata. By 18-24 months, however, the infant has the capacity to hold in mind a representation of the self (a selfconcept) while its perceptual system is directly engaged with objects or events in the world. This initial system continuously updates and replaces its selfrepresentations, but because it cannot hold in mind two representations at the same time, it is unable to store former representations of the self in. r e l a t i o n to these new ones. Borrowing from William James (1890/1950), I shall refer to this representation of the self as the "present self." According to my model, the most primitive manifestation of this present self is the child's ability to form mental declaratives that amount to self-descriptions of their physical or mental states (especially such as agency and desire) (e.g., "I am building a house," "I am hungry," "I cause(d) this," "I want this"; Kagan, 1982; see Bartsch & Welhnan, 1995, for extended analysis of these kinds of spontaneous utterances in 2-year-old children). For now, I define this representation of the sells immediate state as (1)

Si

,

166

DANIELJ. POVINELLI

where S denotes representational-based knowledge of the self and the subscript i fixes the location at the immediate location in time. This initial self-conceptual system is familiar to adults (James, 1890/1950). Indeed, part of the current hypothesis is that this present self is not replaced by a different conceptual system later, but rather remains in place throughout later developmental elaborations (see below). Despite this similarity to our adult understanding of self, my assumption that simultaneous comparisons of different representations are impossible implies that self-representations are largely "on-line"; what is represented is only a single representation of the child's physical and mental states. Thus, by definition this system carries with it no capacity to integrate previous mental or physical states with current ones. This is not to say that previous self-representations play no role here; some subset of them may well be stored in memory and may even provide default inputs into the child's Si 9Thus, as the child's current perceptual information about itself changes, its self-representation is updated to match those changes. Some of these successive self-representations may be relatively stable, or at least may experience strong continuity from one to the next, and hence they may not be updated very much or very often. On the other hand, some of these self-representations (such as the child's desires) may be updated relatively frequently. This allows us to formally define S i as a temporally localized, integrated set of self-experiences represented by the child:

(2)

Si = ({ si, physical }, {si, psychological })

Several important clarifications about this view of the initial self-concept are needed. First, the present is not a "knife-edge," but instead carries with it some temporal degradation (James, 1890/1950). In other words, our conception of the here and now can never truly be the here and now because by the time we direct our attention to that instant, it has already vanished. This is what E.R. Clay (cited by James, 1890) meant by his reference to the "specious present." Thus, any description of the present self must recognize a time corridor into the immediate past and into the immediate future. It is difficult at this point to specify the exact time dimensions along which the child (or the adult) carves up the present, but ultimately it will be critical to do so. Indeed, in a later portion of this essay I offer some suggestions as to how we might experimentally investigate this issue. The second clarification concerns the fact that memories of the relatively distant past clearly exist in very young infants, and children who are 2 to 3 years of age have verbal access to such memories and can recall details of events that occurred at remote points in the past (see Nelson, 1989, 1991, 1993). Thus, this model must grant the 18- to 24-month-old child access to at least some of those

THE UNDUPLICATED SELF

167

memories. Any attempt to define a present self must explain how children can have access to these past representations if S i is temporally truncated in the manner I have stipulated. I speculate that from the 2- to 3-year-old's perspective, these stored self-representations are "atemporal" in the sense that they have no temporal or causal relation to S i 9Thus, many (but not all) of these previous states can be readily recalled (e.g., Nelson, 1989; Gopnik & Slaughter, 1991), but they need have no relation to S i 9I will explore the implications of this issue later, but for now I acknowledge this access and incorporate it into the present self. Finally, by 18-24 months of age, young children have also developed the capacity for imagination and can also talk about events likely to happen in the future. Again, however, I suspect that these imagined states are atemporal in the same sense as representations of previous states.

The Proper Self Next, I consider how the emergence of the capacity to hold in mind several representations simultaneously (which develops in the later preschool years) may interact with the on-line self-conceptual system described above. Just as the advent of the shift from schemata-based to representation-based knowledge allows for the emergence of the present self, I hypothesize that an additional capacity to embed propositions within representations has an important impact upon the child's selfconceptual system. Such domain-general changes, which allow children to simultaneously compare multiple representations, ought to have important ramifications for the child's self-representations. Using Olson's (1993) terminology, I speculate that with the emergence of the ability to hold in mind more than one representation, the child's representational system begins to organize what were successive on-line self-representations ( S i ) as a separate concept, which I defme as the proper self:

(3)

sp

I speculate that this higher-order representation of the self serves as an organizing concept for all past states of the self (defined as Si - n ) and all future imagined states of the self (defined as Si + n ), and links them together into an organized temporal progression, where n denotes a unit of time. This allows us to expand (3) more formally:

(4)

Sp = ({Si- n, n > 0}, {Si }, {Si + n, n > 0})

168

DANIEL J. P O V I N E L L I

This formalism implies several things. First, each previous state of the self that is stored in memory (Si - n ) is a member of a set that in totality comprises a more general representation of the self (So). It also implies that, as n approaches 0, these states more closely approximate the organism's current state. Although there are obvious exceptions, an organism with a folk understanding of causality could (as a heuristic) assume that, for any previous state of themselves that they consider, as n approaches 0 that state has an increasing similarity to S i. 1 Similar reasoning applies to imagined future states. To summarize, I speculate that the ability to simultaneously hold in mind previous representations of Si allows for the child to establish a causal relation between these former states and Si 9In effect, this causal relation establishes the irreversible arrow of time as part of the child's folk psychology. Sp can thus be understood as the higher-order representation of the sell which holds the selfconcept together as an enduring entity through time with a past and a future. This view is consistent with some recent views concerning the timing of the onset of autobiographical memory (see Nelson, 1993). One important consequence of this is the construction of a temporal corridor along which the self progresses from past to present and (via imagination) into the future. It may be that this time line is viewed as deterministically moving in one direction, or it may be viewed as being cyclical in nature. At least, however, when the child considers or confronts a former or future representation of him- or herself, the child can group it as a particular instance of Sp. Thus, it becomes possible for the child to do more than simply recall and verbally report previous physical and mental states from the past, but to understand how one's current state its causally determined by one's previous states. Likewise, and more generally, the child can simultaneously consider multiple physical and/or psychological states of the self as referring to the same concept or entity, which I have defined as Sp. One probable consequence of this is

that coincident with the emergence of the capacity to embed representations of the self within propositions about the self the child discovers its own ontogeny. 2 Or, to borrow from William James (1890/1950) yet again, it creates the capacity to conceive of an "unbrokenness in the stream of selves" (p. 335). 3

Self-recognition in Comparative Perspective Having established the foundations for three different ways in which the developing infant or child processes and stores information about the self (one preconceptual, two conceptual), I now wish to move on to begin to consider the following question: Why is it that self-recognition in mirrors only occurs in organisms that possess conceptual knowledge of the self? After answering this, I will attempt to

THE UNDUPLICATED SELF

169

show how the formalisms established above can be used to generate some new (and counterintuitive) predictions about limitations of young preschoolers' ability for self-recognition. I begin by providing a quick sketch of what is currently known about self-recognition in mirrors in human infants and nonhuman primates.

Self-recognition in Nonhuman Primates In a widely known series of experiments, Gallup (1970) exposed chimpanzees and several different species of macaques to mirrors for a period of 10 days. Initially, most of the animals responded by engaging in a number of speciestypical social reactions, such as threatening, lip-smacking, play behavior, and sexual presentations. However, after 2 or 3 days, some striking species differences began to emerge~ Unlike the macaques, chimpanzees began to engage in a number of behaviors that Gallup labeled as "self-directed." These behaviors suggested that the subjects had discovered that the real source of the image was themselves. For example, the animals were reported to engage in repetitive movements of the limbs and exaggerated facial movements, and to use their hands to explore parts of themselves that they had never seen before (teeth, nose, ano-genital region) - - all while carefully monitoring the mirror image (see Figure 1). After 10 days of mirror exposure, the subjects were anesthetized and marked on the upper eyebrow ridge and ear with a bright red dye, which offered little or no olfactory or tactile cues. The significance of Gallup's procedure was that when the animals recovered they would have no way of knowing that they were so marked. After complete recovery from the anesthesia, the subjects were observed for a 30-minute control period in the absence of a mirror. Any attempts to touch the marked regions were noted. Next, the mirror was introduced, and again the number of mark-directed contacts was recorded. The chimpanzees made few if any contacts to the marked areas during the control period, but made a number of contacts in the mirror test. Indeed, the subjects often inspected their fingers i~mnediately after making contact with the marks, despite the fact that the marks left no olfactory or tactile cues. This pattern of results supported Gallup's initial impressions that the chimpanzees had correctly discovered the source of the mirror image. In contrast, the monkeys who were marked and tested in the same fashion made no attempts to touch the marks. These basic fmdings have been replicated many times, and have been extended to include orangutans (Gallup, McClure, Hill, & Bundy, 1971; Lethmate & Diicker, 1973; Suarez & Gallup, 1981; Calhoun & Thompson, 1988; Lin, Bard, & Anderson, 1992; Povinelli, Rulf, Landau, & Bierschwale, 1993). Likewise, the failure to find self-recognition in members of primate species outside the great ape-human clade has been widely replicated, despite some ingenious attempts to make the source of the image more obvious (Benhar, Carlton, & Samuel, 1975; Gallup, 1977a; Gallup, Wallnan, & Suarez, 1980; Anderson, 1984;

170

DANIEL J. POVINELLI

Itakura, 1987a, 1987b; Anderson & Roeder, 1989; Marchal & Anderson, 1993). Indeed, although many reasonable methodological criticisms have been raised concerning the repeated negative findings of self-recognition in monkeys, most have been empirically addressed and have been found wanting (see Gallup, 1977a; Gallup & Suarez, 1986; Gallup, Wallnau, & Suarez, 1980; Anderson & Roeder,

FIGURE 1. Chimpanzees are capable of using mirrors to engage in (a) contingent body movements, (b) contingent facial movements, and (c-f) self-exploratory behaviors. Some of these behaviors may indicate that the animals understand that the image in the mirror is equivalent to themselves (see text for details). Photographs by Donna T. Bierschwale.

THE UNDUPLICATED SELF

171

1989; Itakura, 1987a, b; Anderson, 1986; Povinelli, 1989). To be sure, there are debates about the distribution, ontogeny, patterns of emergence, and underlying cause of self-recognition in chimpanzees (Swartz & Evans, 1991; Lin, Bard, & Anderson, 1992; Povinelli et al., 1993; Mitchell, 1993). However, I do not believe we can seriously question the basic finding that many chimpanzees are capable of using a mirror to explore parts of themselves and are able to pass wellcontrolled mark tests (see Gallup et al., in press). Although Gallup (1970) originally suggested a conglomerate of self-directed behaviors that were indicative of self-recognition, researchers have recently distinguished between contingent body, contingent facial, and self-exploratory behaviors (Lin et al., 1992; Povinelli et al., 1993). In particular, Povinelli et al. have argued that contingent body and facial movements are not good indicators of self-recognition; in other words, they do not predict the presence of behaviors in which the animals seem to be using the mirror to explore parts of themselves (self-exploratory behaviors), nor are they good predictors of passing a mark test. Conversely, not all behaviors that might appear "self-exploratory" necessarily indicate that the subjects are using the mirror to explore themselves. Povinelli et al. (1993) suggest that some of this activity may be ambient-level self-grooming or scratching, or indeed some might be heightened levels of scratching caused by arousal (a phenomenon common in nonhuman primates, e.g., Maestripieri, Schino, Aureli, & Troisi, 1992). New techniques have been devised to control for this problem by recording the reactions of chimpanzees to mirrors or live video feedback, versus prerecorded videotape stimuli of other chimpanzees in a similar setting. These techniques reveal that young chimpanzees that display contingent body and facial movements (but not self-exploratory behavior) to the mirrors will initially display these "contingent" behaviors to the prerecorded videotape of other chimpanzees (Eddy, Gallup, & Povinelli, 1995). I shall return to the significance of this finding later, but for now I will simply note that we interpret this as evidence that long before chimpanzees are capable of recognizing themselves in mirrors, they learn through proprioceptive and kinesthetic feedback that they can control the movement of the mirror image. Thus, the young animals appear to learn procedural rules for manipulating the behavior of the "other" animal they see in a mirror, and when confronted with the same situation except that the stimuli is a video of others, they are duped into executing the same procedural rules.

Development of Self-recognition in Humans Amsterdam (1972) independently invented the mark test for use with human infants. Although similar to Gallup's, her methodology differs in several critical ways (see GaUup, 1994). Given that most subsequent investigators have followed a variant of her methodology, what follows is a description of a typical test of

172

DANIEL J. POVINELLI

infant self-recognition (e.g., Amsterdam, 1972; Schulman & Kaplowitz, 1977; Bertenthal & Fischer, 1978; Lewis & Brooks-Gunn, 1979; Zazzo, 1982; Johnson, 1983; Asendorpf & Baudonni~re, 1993). Infants are marked (usually by their mother) by having a location of their face (nose, forehead, chin, or cheek) wiped with a cloth containing some kind of wet marking substance (usually a cosmetic). Next, they are immediately presented with a mirror. The typical finding is that very few children younger than about 16-18 months display evidence of selfrecognition. However, by 18-24 months approximately 60-70% of infants will "pass" the test. What constitutes passing can range from reaching up to touch the mark, providing a correct verbal label for the image ("me" or the child's proper name), or even drawing attention to the mark in the mirror. Gallup (1994) has noted that not all of these are valid measures of self-recognition. The formal model I outline in this chapter will highlight some of the conceptual problems with accepting some of these behaviors as evidence of self-recognition. Distinguishing among the divergent kinds of measures of self-recognition in human infants is important because not all indicate the presence of Si. Although I shall return to this issue later, as prelude to my model it is important to note that only one of these measures (reaching up to touch the mark) can be construed as clear evidence that the infants understand that the image in the mirror refers to (or is about) themselves.

Self-recognition and Self-conception The studies of self-recognition in children and nonhuman primates described above have traditionally been used as a mean of exploring an organism's selfconcept. For instance, Gallup (1975) concluded that self-recognition in mirrors "would seem to necessitate an already established identity on the part of the organism making that inference" (p. 330). 4 Working with young children, Lewis and Brooks-Gunn (1979) argued that mirror self-recognition is indicative of the presence of a form of objective self-awareness. Mitchell (1993) has recently criticized the terminology used by Gallup and others and has proposed two models to account for the phenomenon. Both have attempted to explain mirror self-recognition in terms of mental operations in which the child's self-concept per se plays little or no role. It is important to note that Gallup has made two separate claims. His initial claim was that selfrecognition in mirrors was only possible with a well-integrated self-concept (Gallup, 1970, 1975, 1977b). The exact nature of this self-concept was not specified. More recently, Gallup (1982) argued that if self-recognition in chimpanzees implicates some kind of self-concept, and if that self-concept is elaborate enough, then chimpanzees may be capable of using their own experiences to model the experiences of others. He therefore developed a model predicting that

THE UNDUPLICATED SELF

173

chimpanzees should engage in mental state attribution, but that other species which fail to show evidence of self-recognition should not. Although a large portion of our own research agenda is designed to investigate whether chimpanzees attribute mental states to others, this issue is not considered in this chapter (see Povinelli & Eddy, in press). Instead, I focus on: a) the kind of self-knowledge that makes self-recognition in mirrors possible; and b) how that self-knowledge is distinct from that of older children.

Self-recognition in Mirrors Revisited I now use the formalisms adopted earlier to explain why self-recognition in mirrors is restricted to organisms that have developed the general capacity for representation (and in particular, the capacity to represent the self).

Self-recognition in Mirrors Derives from Conceptual Knowledge of the Self Let us begin with organisms that meet the following criteria: they a) are old enough to possess the capacity for representation (approximately 18-24 months); and b) have not yet learned to recognize themselves. First, like younger infants, the 18- to 24-month-old's perceptual system detects the contingency between its motion and the motion in the mirror. 5 However, assuming that the infants have already applied their representational capacities to the self (that is, they have constructed Si ), I propose that the infant will perform an agency mapping in the form of a causal proposition very rapidly: "That action by the mirror-image (object of perception) was caused by me (Si)." Thus, because they have the capacity to hold in mind a representation of Si while perceptually attending to other things, their mirror-image is tagged by the child as connected to (or being about.) S i. I speculate that very rapidly the infant takes the additional step that not only is the image about S i, but it is equivalent to Si. This equivalence is constructed precisely because every feature (physical and/or psychological) that the infant can represent about the self is also true of its mirror-image. For example, the 18month-old's limited attributional capacities (e.g., the ability to attribute desire), causes it to attribute to the mirror-image psychological states equivalent to its own. As the child reaches for a ball in front of a mirror while automatically monitoring (on-line) its state of desire for the ball, so too does it attribute to the mirror-image (which is reaching for the same object) the same desire-goal relation. The child therefore constructs the equivalence: (6)

mirror-image ~. Si

174

DANIELJ. POVINELLI

Thus, with the construction of a self-concept of the type outlined earlier, the answer to the question, "What is causing that?" becomes clear: "I (defined as Si) am." This is a conceptual form of knowledge of the sells agency. Finally, what about the infant prior to 18-24 months? Although younger infants are sensitive to contingency in the manner outlined by Bahrick and Watson (1985) and Meltzoff (1990), the absence of a capacity for mental representation leaves them with no ability to infer a referent for the mirror-image. If we assume that these infants only have access to schemata-, physiological-, and proprioceptive-based information about the self, then by definition they have no resources that allow them to perceptually attend to one object or event (for example, their image in a mirror) while simultaneously relating that thing to an object or event not present (see above). Thus, when they encounter an image of the self in a mirror, they may learn any number of procedural rules that result in specific payoffs. When infants (or animals) In:st look into mirrors and see objects or events that are behind them in real space, they may be duped into responding as if those objects or events were where they appear in the mirror. However, they can easily learn procedural rules whereby they respond to the appearance of reflected objects by turning around to their real location in space. For example, numerous researchers have demonstrated that with sufficient experience with mirrors, human infants and animals who do not pass a mark test can use a mirror to direct their visual or manual searches to the real location of the rewarding object or event (e.g., Anderson, 1986; Itakura, 1987a, 1987b; Povinelli, 1989; Robinson et al., 1990). Likewise, they can also learn that when they move, so does their mirrorimage. Indeed, because of this contingency, infants too young for self-recognition may even learn (through proprioceptive feedback) that they can control the movement of their mirror-image (see above; Povinelli et al., 1993; Eddy et al., 1995). Yet despite the construction and deployment of such sophisticated schemata, the image has no relation to anything else. More to the point, for the organism observing itself in a mirror, the image has no relation to the self precisely because (by definition) the self (S i ) cannot yet be represented. To summarize, I am hypothesizing that mirror self-recognition occurs because organisms with general representational abilities also form robust representations of themselves. This allows them to understand that the image in the mirror moving with them is about them, and ultimately refers to or is equivalent to them. Note, however, that I am not claiming they understand that their mirror image is a representation of themselves. Indeed, according to this model, organisms do not understand this until considerably later.

THE UNDUPLICATED SELF

175

Recognizing Parts of the Self? Some researchers will argue that I have sidestepped the question of how other phenomena such as facial or bodily recognition, imitation, or more importantly an understanding of mirror correspondence, contributes to self-recognition in mirrors. After all, some authors have identified some or all of these as key factors in providing a coherent account of self-recognition (Guillaume, 1926/1971; Parker, 1991; Mitchell, 1993). Yet according to the model outlined here, these phenomena are incidental to mirror self-recognition, not necessary features for a system to display behaviors indicative of self-recognition: self-exploratory behaviors in front of mirrors and/or reaching up to touch a previously unknown mark on the self. First, consider the question of how the child comes to know that the facial features seen in the mirror belong to him or her. Some researchers see this as the whole task of self-recognition in mirrors. But I believe that careful analysis reveals that this is a trivial component of the question of mirror self-recognition. On the one hand, infants too young to construct an Si still may be able to correctly label the featural cues they see in a mirror by using the words me or their proper name. As many authors have pointed out, verbal labeling of this kind may merely mean that the child or infant has learned from its parents that the face (that set of features) is correctly labeled using their proper name (i.e., "Mary") or the first person pronoun me (Gallup, 1975; Bigelow, 1981; Anderson, 1984b). In addition, by as early as 5 months of age infants may also show other evidence of discrimination of their own face from strangers, presumably because through previous exposure to mirrors their facial features are simply more familiar than a stranger's (e.g., Fadil, Moss, & Bahrick, 1993). On the other hand, the same infant who is now old enough to form the equivalence relation in (6) but who has not yet done so may have no idea what his or her facial features look like this, despite the fact that the infant has been able to discriminate his or her face from others for months. Some will insist that I have still not explained how children can know that the facial features seen in the mirror belong to them before they recognize themselves (Mitchell, 1993). Of course, it is true that prior to the very first time a child recognizes herself in a mirror certain aspects of her physical selfrepresentation may be incomplete. But I do not see a need to specify a unique inferential or deductive process by which the child incorporates details of his or her facial appearance (for example) into an updated si, physical. For the prototypical case, I will assume that the mirror-naive organism has little or no default inputs about its own facial features. The model that I am advancing argues that coincident with the advent of representational abilities, the child creates a cohesive self-identity that includes aspects of her physical and mental states. The child is constantly updating these self-representations, just as she does with other objects

176

DANIELJ. POVINELLI

and events in the world. Thus, prior to the child understanding that the image in the mirror is equivalent to herself, she has a concept of self (Si) and hence her encounter with a mirror (which is marked with perfect contingency between what the child sees in the mirror and Si ) leads the child to form the predicate relation (6) between Si and the mirror-image. Whereas earlier the child's perceptual system detected the contingency between her actions and external events (including those in a mirror, on a live video monitor, or those simply "caused" by its direct actions upon the world), by 18-24 months that perceptual system has been articulated to a conceptual system, which includes self-representations. Thus, children are able to hold in mind on-line internal representations of self (Si) as subject, with their perceptual systems free to attend to the mirror-image as the predicate. Once such a proposition is established, new information (in this case knowledge about the details of one's facial features) is simply incorporated into Si as the representation is updated. Thus, the process through which the child updates his si, physical from information provided via the mirror image would not seem to differ from a child updating his representations of other things in the world. For instance, imagine a 2-year-old child whose only experience with automobiles was from looking out the front window of his or her house as cars drove up into the driveway. With sufficient experience, the child could clearly form the concept of car and sufficiently generalize it to novel exemplars of the class. But now imagine that the child has the opportunity to go outside and actually explore a car for the first time. Consider all of the details that the child will now discover about each car that is visited (they have seats, license plates in the back, etc.). The child's new concept of automobiles includes new features that he or she did not know of before. In my view, this is no different than what occurs when children learn the details of their own facial features. Once the tagging occurs as described in (6), I suspect that children simply update their representation of their physical appearance. Up to this point I have provided an account of how an organism that is learning to recognize itself for the first time reaches the equivalence relation specified in (6). What about its subsequent interactions with mirrors? Here a slightly more complicated situation arises. Now that the organism has a more complete representation of its physical appearance, the exact triggering cause of the equivalence relation (6) is difficult to specify. Depending on the deployment of the attentional resources of the infant, the focus may be exclusively on physical similarity, or it may be on the psychological similarity of agency, which is embedded in both contingency and desire reflected in equivalent actions on the environment. However, as I shall show later, placing these two dimensions of the self in conflict with one another in the context of a visual re-presentation of the

THE UNDUPLICATED SELF

177

self has the potential to reveal the underlying differences among children of different ages.

Must Organisms that Display Self-recognition Understand Mirrors ? If the model I am offering is correct, then it means that one of the most widespread ideas about self-recognition m that it requires some practical l~owledge of how mirrors work m is incorrect. Mitchell (1993), for instance, argues in both of his models of mirror self-recognition that an organism needs to understand the property of "mirror correspondence," which is defined as the knowledge that "mirrors reflect accurate and contingent images of objects in front of them" (p. 298). He sees mirror correspondence as one of the necessary conditions for the organism to infer or deduct (depending on which of his theories one examines) that the image in the mirror is itself. However, according to the model offered here, the construction of the equivalence outlined in (6) does not require any such specific competence. In order for the child to construct an equivalence between her mirror image and her existing self-representations, she need not understand that mirrors accurately and contingently reflect what is in front of them. This would explain why there does not appear to be a reliable correlation between understanding mirror correspondence (an infant's ability to use mirrors to localize events or objects out of their perceptual field) and passing the mark test (Zazzo, 1982; Loveland, 1986; Robinson, Connell, McKenzie, & Day, 1990). Some infants who test positive for self-recognition using a mark test appear to understand this property of mirrors (or at least have formed procedural rules that make it appear as if they understand this property); others do not. Further, it would also explain why children reared in cultures that have few or no fabricated mirrors seem to be capable of recognizing themselves (passing a mark test) after only a few minutes of their first exposure to a mirror (Priel & de Schonen, 1986). My interpretation of these data is that children who pass the mark test may or may not understand the affordances of mirrors because understanding mirrors is not necessary in order to arrive at the equivalence relation given in (6). Another reason why the child's interactions with mirrors have seemed mysterious is because the mark test appears to involve finding something that is hidden from view (i.e., the mark on the face). For instance, Bertenthal and Fischer (1978) speculated that the correlation between the development of object permanence and self-recognition they obtained was probably because both involve skills related to "the ability to search for hidden objects" (p. 49). Mitchell's (1993) body-part objectification, object-permanence theory is perhaps the most formal statement of this view. But an implication of the formalism adopted in this essay is that the difference between a 14-month-old child who does not pass the mark test and an 18-month-old who does, has nothing to do with respective unsuccessful

178

DANIEL J. POVINELLI

versus successful "searches" for the mark. Further, as noted above, it is not that the 18-month-old has figured out that mirrors accurately reflect what is in front of them (an understanding of "mirror correspondence"), but rather that they have an existing on-line Si and can form the proposition described in (6). But none of this implies that they have actively searched for, and then found, the real location of what they are seeing in the mirror. Rather, they are simply using the mirror to gain access to either previously unknown aspects of the self (as in the case of the onset of self-recognition) or to explore some alteration of a previous selfrepresentation of the face (as in the case of the mark test after self-recognition has occurred). Thus, children (or chimpanzees) who recognize themselves in mirrors need not understand all (or even most) of the reflective properties of mirrors. They need not understand (although they may) that mirrors reflect things that are in front of them, nor need they understand mirrors as representational devices. They shnply need the conceptual capacity to form the predicate relation that the mirror image is equivalent to Si. This alone induces the child with a self-concept to explore his face, not the surface of the mirror. Another way of looking at this would be to say that the child could have a quite stable Si but could be quite confused about how or why it is framed in glass. On the other hand, children (or apes or monkeys) who have sufficient experience with mirrors early in life may come to form sets of procedural rules (schemata) for responding to mirrors. Conversely, this account also explains how both young infants and monkeys without representational capacities of the type discussed here could, with sufficient experience, learn procedural rules to react to mirrors appropriately but still not display behaviors indicative of self-recognition (e.g., Itakura, 1987a, 1987b; Robinson et al., 1990). Thus, self-recognition in mirrors requires neither an understanding of the reflective properties of mirrors in general nor the understanding that the mirror image is a representation of Si in particular. Finally, let me note that this model shares some features with Mitchell's (1993), in that proprioceptive matching contributes to the formation of the equivalence relation (6). But after that the models depart. Mitchell assumes that the limiting factor for the onset of self-recognition in young infants is either: a) the absence of an elaborate enough "kinesthetic-visual matching" capacity (which is present at birth, and quite elaborated by 9-14 months, e.g., Meltzoff & Moore, 1977; Meltzoff, 1990); or b) the absence of an understanding of mirror correspondence. In direct contrast, the model offered here specifies that: a) an understanding of mirror correspondence is not necessary to form the equivalence relation (6); and b) the kinds of sensitivity to contingency that are necessary for self-recognition are in place long before 18-24 months. What has not yet developed, and thus what can be described as the limiting factors for self-

THE UNDUPLICATED SELF

179

recognition, are the representational abilities necessary for the construction of the self-concept (Si).

Predictions of the Theory and Preliminary Tests If the theory outlined above is to be seriously entertained, it should generate some previously unarticulated predictions about the behavior of young children and chimpanzees at various stages in their development. Below I explore some of these predictions, one of which involves a previously unpredicted asynchrony in children's capacity to recognize themselves using live versus delayed visual feedback.

Interactions with Mirrors Before the Emergence of the Present Self First, the theory predicts that there ought to be little correlation between an organism's ability to use mirrors to locate objects or events using mirror cues and its ability to pass a mark test or display mirror-mediated self-exploratory behaviors. Thus, before they are able to pass a mark test of self-recognition, human infants and chimpanzees given sufficient experience with mirrors should be able to learn to use mirrors to locate objects that (for example) are really behind them. Likewise, other species in which self-recognition has never been demonstrated should also be able to use mirrors in this fashion. As I have indicated above, both of these predictions seem to be borne out by the existing data. As a rather large case in point, take our previous research with elephants: Although they displayed no evidence of recognizing themselves in mirrors, they showed a very sophisticated ability to use mirrors to locate hidden objects (Povinelli, 1989). A second prediction that we have explored empirically is the idea that organisms within a species whose members are capable of self-recognition in mirrors ought to be able to learn to manipulate a mirror-image using information about the self that is available though preconceptual channels (proprioceptive feedback). Thus, schemata-based self-knowledge should allow an organism to form procedural rules about its actions and contingent consequences in another location. Consistent with this view are the results from the studies described earlier, which indicate that before chimpanzees pass a mark test or display self-exploratory behaviors in front of mirrors, they learn to manipulate the image by engaging in contingent body and facial movements (see Figure 1; Povinelli et al., 1993). That their interpretation is not based upon a equivalence relation of the type outlined in (6) is apparent from the fact that they attempt the same manipulations when the stimuli are prerecorded images of other chimpanzees. Thus, the procedural nature of the self-knowledge is exposed.

180

DANIEL J. POVINELLI

Self-recognition in Mirrors and Beyond Next, a provisional acceptance of the model can generate some surprising predictions about what might happen if we manipulate either a) the contingency and/or b) the featural similarity of the child's self-image. Let me begin with contingency. Recall that 2- to 3-year-olds are assumed to possess a representational capacity such that they are able to represent their physical states, personal agency, desires, and perceptual experiences (and possibly their knowledge states) as Si. Second, their sensitivity to contingency leads them to construct the equivalence proposition that Si =_mirror image. But consider what would happen to such a system's response to a "re-presentation" of the self that was not contingent with its current actions (or desires, or knowledge states, for that matter). As a case in point, let us imagine that the image is of a series of events just previously performed by the child, described as the set: (7)

( { S i - 1 }, { S i - 2 } .... { S i - n } )

In this case, the detection of physical similarities of the image (i.e., bodily or facial recognition) should lead the system to provide a verbal description of the image using either me or the child's proper name. Indeed, if its attentional resources were maximally devoted to the featural cues, the organism might even momentarily form the equivalence relation in (6). However, as soon as the child's attentional resources focused upon the other (and I assume more salient) aspects of the re-presentation such as its agency, he or she should conclude that the image is not equivalent to Si. After all, neither agency nor desire-goal states appear to match; therefore, there is no reason to establish a straightforward proposition linking the two. An obvious alternative route to connecting the two is available to adults: relating the object of perception (the events on the monitor) to the representation S i through the representation Sp. Yet the model stipulates that this route is not available to the 2- to 3-year-old. Nonetheless, these younger children should still be capable of identifying the images using their proper nmne (or even me, if the child treats me and his/her proper name as denoting that set of featural cues).

How might we go about determining if young children actually reason in such a dissociated fashion about the re-presentations of a previous image of themselves? There are several ways, but let me first address a paradigm that we have explored (Povinelli, Landau, & Perilloux, in press). First, imagine that a child is playing a simple game where he or she looks under cups for stickers. One experimenter is directly playing with the child, and the other is sitting next to the child, praising her or him and patting him or her on the head after each sticker is found, hnagine also that the procedure is being videotaped so that the child's head is clearly

THE UNDUPLICATED SELF

181

visible. On the third triM, the second experimenter uses the act of praising the child as the vehicle to covertly place a large sticker on top of the child's head. Several additional trials are conducted to ensure that the child does not detect the marking. Next, the child is invited to watch what he or she did on television. Thus, two minutes after the child was marked, the child sees the events replayed on the monitor. 6 The critical question is, of course, what do the children do when the playback reaches the point at which the experimenter places the sticker on their heads? Contrary to what one might expect, the theory predicts that between 18-24 months and 3 years or so, children should not reach up to remove the sticker. They should be too young to construct Sp, and thus should be unable to infer that the sticker they see on their forehead in the video is currently on their head; thus, for them, Si is not causally related to the set ( { S i - 1 }, {Si - 2 } .... {Si- n }). Yet the theory predicts that if these same children are placed in front of a mirror, they should remove the sticker almost immediately precisely because they can form the proposition, Si =_..mirror image. Recall, however, that the theory also makes the explicit prediction that these younger children should have no trouble whatsoever identifying the child on the prerecorded videotape by using me or their proper name. Older 3-year-olds and most 4-year-olds ought to be in very different position. Their ability to consider multiple representations of the same thing simultaneously should allow them not merely to label the image using me or their proper name, but also to infer that what they are witnessing on the videotape is a particular instance of their proper self, a former S i temporally adjacent with their current S i. Thus, an inference of what is true of any given Si - n may also be true of Si is likely to be drawn, especially (as we have seen) as n approaches 0. The theory therefore predicts that unlike their younger counterparts, they ought to reach up to search for the sticker after the tape reveals the experimenter placing it on their heads. 7

Preliminary Tests To date, we have conducted three tests of the idea just described (Povinelli, Landau, & Perilloux, in press). In the first experiment, we executed the procedures exactly as described above using forty-two children ranging from 2 to 4 years of age as the subjects (ten 2-year-olds, sixteen 3-year-olds, sixteen 4-year-olds). As the model predicted, none of the 2-year-olds and only 25% of the 3-year-olds reached up to take the sticker off their heads after the tape showed the experimenter placing it on their heads. In contrast, a full 75% of the 4-year-olds reached up to remove the sticker within an average of 7 seconds after viewing that part of the tape. These results are especially dramatic when they are contrasted with studies of

182

DANIEL J. POVINELLI

mirror self-recognition, where up to 80% of the 2- and young 3-year-olds would be expected to pass the test. The second experiment was identical to the one described above, except that instead of filming the children using videotape, we took Polaroid snapshots of the children at two junctures in the procedure. The first photo was taken as the experimenter was praising the child and covertly placing the sticker on her head. The second photo was taken after the control trials. One of the experimenters introduced the child to a large stuffed gorilla that she had never seen before and explained that the other experimenter was going to take another picture, this time of the child, the gorilla, and the experimenter together. Thus, two snapshots were available to show the children: one that clearly depicted the experimenter placing the sticker on the child's head, the other depicting the child (with sticker on head) sitting with the gorilla and the experimenter. We hoped that these images would force in a more direct way the relation between what had just happened and the current state of affairs. Approximately 2-3 minutes after the second picture was taken, the main experimenter showed the child the photographs one at a time. During the presentation, the experimenter asked her a series of standardized questions designed to determine her ability to identify the images correctly. Finally, if the subject had not reached up to remove the sticker by the end of the presentation of the second photograph, she was given the stuffed gorilla again, presented with a mirror, and invited to look at themselves along with the main experimenter. A total of 60 children participated in this study, with 15 children in each of 4 age groups: young and old 3-year-olds, and young and old 4-year-olds. Consistent with the first experiment, only 13% of the young 3-year-olds reached up to their heads to remove the sticker while the photographs were presented and the questions were asked. In contrast (and in full accord with the theory), 85% of the young 3-year-olds who did not reach up while looking at the photographs did so when they were presented with the mirror. In contrast, by 4 years of age, 80-90% of the subjects reached up while looking at the first photograph. Other aspects of the results also support some of the predictions of the theory. For example, even the youngest children were able to provide a "correct" verbal label for their image: When the experimenter pointed to their hnage and asked, "Who is that?," 75% of the children in the youngest age group responded by stating their name or using the personal pronoun m e . However, there was an intriguing significant developmental difference in the use of the personal pronoun m e versus their proper names. The youngest children appeared not to discriminate between using their proper name or the personal pronoun m e . In contrast, by 4 years of age, the response profile had shifted completely. Virtually all of them responded by using the personal pronoun m e . Further linguistic evidence that the

THE UNDUPLICATED SELF

183

younger children did not interpret their photographic image as relating to Si comes from an analysis of the kinds of possessive verbal descriptions the children used when the experimenter pointed to the sticker in the photograph and asked, "Where is the sticker right now?" Most of the children in the youngest age group (after having just "correctly" identified the image by stating me or by using their proper name) described the sticker as being on his or her or the head, whereas only a few described its location using the first person possessive pronoun (my head). In striking contrast, not a single 4-year-old child used the third-person possessive pronouns. These patterns of answers are very consistent with the view that the younger children recognized and had a verbal label for their featural cues, but did not relate the image to their current present self. In addition, they show that even when the youngest children's attention was explicitly focused on the sticker in the image, they still failed to reach up to remove it. The third study produced results that suggest that the model as presented is incomplete. In this study, we directly compared two groups. One received delayed video feedback of the self after the child was marked with a sticker as in the first experiment, and the other received live video feedback of the self after the child was marked in an identical fashion. Thus, both groups observed themselves for two minutes with the stickers on their heads, with the only difference being the contingency of the image. We tested 48 subjects ranging in age from 2-3 years (mean age for both group = 35 months), with 24 subjects in each group. Although the group differences were in the direction predicted, the difference was not statistically significant (62% of the children in the Live Feedback condition reached up to remove the sticker, and 37% of the children in the Delayed Feedback condition did so, p< .07). The results surprised us for two reasons. First, a higher percentage of 2- and young 3-year-olds passed the Delayed Feedback test than had in our previous studies, and far fewer passed in the Live Feedback condition than would be expected based on previous studies of mirror selfrecognition. However, there are several intriguing possibilities that could explain these findings in the context of the theory. First, it is possible that the greater percentage of children passing the test in the Delayed Condition than did previously could be due to the fact that they did not see their image transformed from one state (not marked) to another state (marked) as they did in the first experiment involving delayed video. Ironically, we had originally speculated that seeing the whole marking event might make it easier for the younger children; in retrospect, the theory seems to predict the opposite. First, recall that the theory stipulates that the younger children can only consider one representation of themselves at a time. However, as they watch themselves on the monitor without a sticker, it is still true that the featural dimensions of the image are consistent with their current default inputs regarding si, physical. Thus, if all of their

184

DANIEL J. POVINELLI

attentional resources were focused on the featural cues at the expense of the contingency/agency cues, they could momentarily form the proposition that the image is equivalent to S i , but would have no reason to reach up to their heads because there are no marks yet. However, as soon as the child sees the experimenter reach up and act upon the image, he concludes that it cannot be about him because no one is putting a sticker on his head. Notice, however, that in the last experiment the child in the delayed group only saw himself in one state with a sticker on his head. However, in this case, when his attentional resources are initially devoted to the featural cues, this might cause him to form the equivalence relation, and thus reach up and remove the sticker. But as soon as his attention is drawn again to the agency discrepancy, he should be just as strong in his conviction that the image is not about S i 9 Although we are just beginning the experiments that will be necessary to tease these issues apart, it is worth reflecting upon the continuing sorts of spontaneous verbal comments made by these younger children when confronted with their delayed images. For example, one young girl in the Delayed Feedback test reached up immediately to her head when her image appeared, but then asked in confusion several seconds later why the girl on the screen did not take the sticker off her head, too. The general point is that the theory does not exclude the children from shifting back and forth between an interpretation of equivalence and nonequivalence, depending upon whether they focus (or the experimental paradigm forces them to focus) on the featural or contingency cues. Given the consistent negative pattern that has emerged with children 3 and younger confronting noncontingent stimuli, I conclude that the contingency factor is the most salient. Finally, why did so few children reach up in the Live Feedback condition? We speculate that it is because the best kind of stimuli to cause the equivalence relation is live symmetrical (or specular) feedback (i.e., mirrors, or live video in which the normal reversed image is made mirrorlike). We have been testing the idea that part of the motivation for the equivalence relation concerns not sensitivity to contingency, but a certain form of identical contingency ~ symmetrical contingency. This predicts that if the Live Feedback had been of a different kind a specular kind - - many more children would pass the test. Although there are somewhat uninteresting reasons why this might be the case (i.e., children develop scripts for dealing with mirrors, and antispecular images depart from those scripts), there may be more fundamental explanations. Although there is not room to expand upon this idea here, this sensitivity may be present in early infancy and may reflect natural selection for organisms' ability to detect when others are behaviorally connected or linked to them. A final issue I wish to raise concerns the temporal breadth of the present self as conceived by the 2-year-old child. To cast the question in empirical terms,

THE UNDUPLICATED SELF

185

imagine that we repeated our original delayed test of self-recognition, but instead of having the age of the child as our independent variable, we manipulated the interval between the time at which the marking event actually happened and the moment when the child saw the playback of the events. How close to perfect temporal contingency would be necessary before 2-year-olds consistently formed the proposition relating the image and Si ? Such experiments (if properly designed) might provide some insight into the duration of the troubling "specious present" at least for the 2-year-old child. Although we have yet to begin such studies, I suspect that this window is very narrow indeed.

Self-conception in Evolutionary Perspective The theory outlined in this chapter has implications far beyond the development of the self in human infancy. It also sets the stage for answering questions concerning the nature of the self-concept in other species that provide some initial reason to suspect that they form some kind of self-concept. Chimpanzees and orangutans represent obvious choices, given that members of both species have been demonstrated to show every bit as compelling evidence of self-recognition as 18to 24-month-old human infants. In this essay, I have committed to a domain-general view of the development of representational capacities in human infancy and childhood. However, ignoring the exact rate of development, it is not completely clear if these synchronies in development exist in other species, such as chimpanzees. Cast in slightly different terms, it is unclear which cognitive-developmental pathways are dissociable. Elsewhere I have tried to assess the current evidence concerning the homologous aspects of representational development between humans and chimpanzees as reflected in the domain of theory of mind (Povinelli, in press). A fair reading of the research to date leads to the conclusion that it is still too early to determine which aspects of cognitive development typical of the 18- to 24-month-old human infant occur in chimpanzees. Even for the ones that we can be reasonably certain exist in chimpanzees, we do not yet know if they develop in synchrony. This is an important point in the context of the theory outlined in this chapter because if chimpanzees develop representational capacities for some domains (such as objects) but not others (such as mental states), then it may have important implications for the scope of their self-concept. What if, for example, chimpanzees are only able to represent their physical or proprioceptive states? That is, what if for them Si is composed of only si, physical and si, psychological (agency) ? Such a representation would still allow them to represent the aspect of their psychology (their agency) that triggers the equivalence relation through the detection of perfect contingency.

186

DANIELJ. POVINELLI

In this case, we would still be warranted in claiming that their capacity to recognize themselves in mirrors reflects (is allowed by) the presence of a selfconcept, as Gallup (1970) originally speculated. However, in this case his later theory about the scope of that self-concept would be incorrect (Gallup, 1982). Finally, what about chimpanzees' conceptions of themselves as entities with a past and a future? Gallup (1982) speculated that the presence of a self-concept in chimpanzees (as indicated by their ability to recognize themselves in mirrors) left open the possibility that they might be able to "begin formulating questions about themselves in relation to historical as well as future events" (p. 242). From the view offered in this paper, the chimpanzee's capacity for such autobiographical memory and temporal projection depends directly upon its capacity to construct higher-order representations. As of yet, we have no definitive evidence concerning its abilities in this arena (see reviews by Cheney & Seyfarth, 1990; Whiten, 1993; Povinelli, 1993, in press; Tomasello & Call, in press). But this does not mean that we should conclude, as has Fraser (1987), that humans are the only species able to conceive of time far removed from the present. The extent of overlap in homologous cognitive developmental pathways among humans and their nearest relatives remains an open, empirical question. Thus, conducting explicit tests of self-recognition with chimpanzees using delayed feedback is a high priority for our own research program. Ultimately, such research will allow us to take a first step toward discovering if chimpanzees, like us, appreciate that they are unique, unduplicated selves trapped in an irreversible arrow of time.

NOTES 1. "Add to this character [of the present and distant selves belonging together] the farther [sic] one that the distant selves appear to our thought as having for hours of time been continuous with each other, and the most recent ones of them continuous with the Self of the present moment, melting into it by slow degrees; and we get a still stronger bond of union. And we think we see an identical bodily thing when, in spite of changes in structure, it exists continuously before our eyes, or when, however interrupted its presence, its quality'returns unchanged; so here we think we experience an identical Self when it appears to us in an analogous way. Continuity makes us unite what discontinuity might otherwise separate; similarity makes us unite what discontinuity might hold apart" (James, 1890/1950, p. 334; italics in original). 2. In a recent Master's thesis, Thomas Suddendorf (1994) makes a similar argument concerning the relation of metarepresentation and temporal projection into the past and future. Like this model, he sees this capacity as opening up the possibility for mental time travel. He also speculates on the question of whether chimpanzees develop this capacity. In contrast, I remain skeptical about the scope of the chimpanzee's theory of mind (Povinelli, 1993, in press). Although our ideas are quite similar in places, they

THE UNDUPLICATED SELF

187

have been derived independently. Rather than revise this manuscript to reflect his exposition, I refer the reader to his work directly. 3. "The various members of the collection [of the present and distant selves] are felt to belong with each other whenever they are thought at all. The animal warmth, etc., is their herd-mark, the brand from which they can never more escape. It runs through them all like a chaplet and makes them into a whole, which we treat as a unit, no matter how much in other ways the parts may differ inter se" (James, 1890/1950, p. 334). 4. In a later attempt to distinguish preconceptual forms of self-knowledge from the conceptual kind of knowledge needed for self-recognition in mirrors, Gallup (1977b) argued for a distinction between "self-sensation" and "self-perception." However, following Butterworth (1992), Gallup (1991) recently adopted a terminology more consistent with that used by developmental psychologists: "self-perception" versus "self-conception." Despite this, his distinction has always been between knowledge about the self coded in terms of proprioceptive and kinesthetic feedback versus a concept of self. 5. There have been numerous demonstrations of this sensitivity to contingency, ranging from demonstration that infants are sensitive to reciprocal behavioral patterns on the part of caregivers, to demonstration that infants detect the contingency between their own (visually obscured) movements and live feedback of that image on a video monitor (e.g., Bahrick & Watson, 1985). Meltzoff (1990) has shown that 14-montholds are sensitive to others who perform actions contingent with their own, and are especially sensitive to those who imitate the exact form of their behavior. From this kind of evidence it is possible to conclude that by very early in ontogeny, the infant is able to detect actions that are contingent with its own. The reason for this sensitivity is unknown, although detection of contingency in general plays a fundamental role in theories of the simplest forms of animal and human learning (Rescorla, 1967). 6. To my knowledge, two studies have exposed children to playbacks of their previously recorded visual images for the purpose of assessing self-recognition. Brooks-Gunn & Lewis (1984) recorded and coded the responses (affect, interest, imitation) of infants to prerecorded images of themselves or a same-sex age-mate. Zazzo (1982) also presented playbacks of images to young preschoolers. Apparently neither of these studies explicitly assessed self-recognition using a mark test. 7. Nelson (1991) has used Weist's (1986) four-stage model of the temporal systems that young children display from 18 months to 4 years to account for the linguistic referents to temporal events collected from the crib monologues of the child Emily. This model establishes increasingly sophisticated relations between speech time (ST, the here and now), event time (ET), and reference time (RT). The system begins at a point when ET and RT are fixed at ST. This implies that very young children of this age are able to talk about events not localized at the present. The final stage, a free RT system, is not achieved until about 3.5 to 4 years and is characterized by an ability to distinguish RT from ST, and ET from both RT and ST (Weist, 1986). Although it has primarily been used as a means of analyzing linguistic utterances, an analysis of the kind of temporalconceptual distinctions that become possible in this final free-RT system reveals important underlying parallels between Weist's (1986) four-stage model, Nelson's (1991) elaboration of it, and the model I have developed here.

188

DANIELJ. POVINELLI

ACKNOWLEDGMENTS

The theoretical ideas for this chapter were developed during my participation in meetings of the Infant Intentionality Group at Yale University during the spring of 1991. Gordon Gallup, Helen Perilloux, Timothy J. Eddy, Mark Povinelli, and Anthony Maida offered helpful discussions and advice on versions of the manuscript. This work was supported by National Institutes of Health Grant No. RR-03583-05 to the New Iberia Research Center and National Science Foundation Young Investigator Award SBR-85458111 to D.J.P. Address correspondence to Daniel J. Povinelli, Laboratory of Comparative Behavioral Biology, USL-New Iberia Research Center, 4401 W. Admiral Doyle Dr., New Iberia, LA 70560. Telephone: (318) 365 2411, FAX: (318) 373 0057.

REFERENCES

Amsterdam, B. (1972). Mirror self-image reactions before age two. Developmental Psychobiology, 5, 297-305. Anderson, J.R. (1984a). Monkeys with mirrors: Some questions for primate psychology. International Journal of Primatology, 5, 81-98. Anderson, J.R. (1984b). The development of self-recognition: A review. Developmental Psychobiology, 17, 35-49. Anderson, J.R. (1986) Mirror-mediated finding of hidden food by monkeys (Macaca tonkeana and M. fa,~cicularis). Journal of Comparative Psychology, 100, 237242. Anderson, J.R., & Roeder, J.J. (1989). Responses of capuchins monkeys (Cebus apella) to different conditions of mirror-image stimulation. Primates, 30, 581587. Asendorpf, J.B., & Baudonni~re, P-.M. (1993). Self-awareness and other-awareness: Mirror self-recognition and synchronic imitation among unfamiliar peers. Developmental Psychology, 29, 88-95. Bahrick, L.E., & Watson, J.S. (1985). Detection of intermodal proprioceptive-visual contingency as a potential basis for self-perception in infancy. Developmental Psychology, 21, 963-973. Baillargeon, R. (1987). Young infants' reasoning about the physical and spatial properties of a hidden object. Cognitive Development, 2, 655-664. Baldwin, D.A. (1993). Early referential understanding: Infants' ability to recognize referential acts for what they are. Developmental Psychology, 29, 832-843. Bartsch, K. & Wellman, H. (1995). Children talk about the mind. Oxford, Oxford University Press. Benhar, E.E., Carlton, P.L., & Samuel, D. (1975). A search for mirror-image reinforcement and self-recognition in the baboon. In S. Kondo, M. Kawai ,& S. Ehara (Eds.), Contemporary primatology: Proceedings of the 5th international congress of primatology, (pp. 202-208). New York: Karger. Bertenthal, B.I. & Fischer, K.W. (1978). Development of self-recognition in the infant. Developmental Psychology, 14, 44-50. Bigelow, A.E. (1981). The correspondence between self- and image-movement as a cue to self-recognition for young children. Journal of Genetic Psychology, 139, 11-26. Bischof-K6hler, D. (1988). Uber der Zusammenhang von Empathie und der Fahigkeit, sich im Spiegel zu erkennen [On the association between empathy and ability to recognize oneself in the mirror]. Schweizerische Zeitschrifl fur Psychotogie, 47, 147-159.

THE UNDUPLICATED SELF 189 Brownell, C.A., & Carriger, M.S. (1990). Changes in cooperation and self-other distinction during the second year. Child Development, 61, 1164-1174. Butterworth, G. (1992). Origins of self-perception in infancy. Psychological Inquiry, 3, 98-109. Butterworth, G., & Jarrett, N. (1991). What minds have in common is space: Spatial mechanisms serving joint visual attention in infancy. British Journal of Developmental Psychology, 9, 55-72. Calhoun, S., & Thompson, R.L. (1988). Long-term retention of self-recognition by chimpanzees. American Journal of Primatology, 15, 361-365. Cheney, D. L., & Seyfarth, R. M. (1990). How monkeys see the world. Chicago: University of Chicago Press. Eddy, T.J., Gallup, G.G., Jr. & Povinelli, D.J. (1995). Development of the ability of chimpanzees (Pan troglodytes) to distinguish mirror-images of self from video images of others. Manuscript submitted for publication. Fadil, C.A., Moss, L.E., & Bahrick, L.E. (1993, March). Infants' visual recognition of their own faces. Poster presented at the 1993 Meeting of the Society for research in Child Development, New Orleans, LA. Flavell, J.H. (1988). From cognitive connections to mental representations. In J.W. Astington, P.L. Harris, & D.R. Olson, (Eds.), Developing theories of mind (pp. 244-267). Cambridge: Cambridge University Press. Forguson, L., & Gopnik, A. (1988). The ontogeny of common sense. In J.W. Astington, P.L. Harris, and D.R. Olson (Eds.), Developing theories of mind. (pp. 226-243). Cambridge: Cambridge University Press. Fraser, J.T. (1987). Time the familiar stranger Amherst, MA: University of Massachusetts Press. Gallup, G.G., Jr. (1970). Chimpanzees: Self-recognition. Science, 167, 86-87. Gallup, G.G., Jr. (1975). Toward an operational definition of self-awareness. In R.H. Tuttle (Ed.), Socio-ecology and psychology of primates (pp. 309-341). The Hague, The Netherlands: Mouton. Gallup, G.G., Jr. (1977a). Absence of self-recognition in a monkey (Macaca fascicularis) following prolonged exposure to a mirror. Developmental Psychobiology, 10, 281-284. Gallup, G.G., Jr. (1977b). Self-recognition in primates: A comparative approach to the bidirectional properties of consciousness. American Psychologist, 32, 329-338. Gallup, G.G., Jr. (1982). Self-awareness and the emergence of mind in primates. American Journal of Primatology, 2, 237-248. Gallup, G.G., Jr. (1994). Self-recognition: Research strategies and experimental design. In S. Parker, R. Mitchell, & M. Boccia (Eds.), Self-awareness in animals and humans. (pp. 35-50). Cambridge: Cambridge University Press. Gallup, G.G., Jr., McClure, M.K., Hill, S.D., & Bundy, R.A. (1971). Capacity for selfrecognition in differentially reared chimpanzees. The Psychological Record, 21, 69-74. Gallup, G.G., Jr., Povinelli, D.J., Suarez, S.D., Anderson, J.R., Lethmate, J., & Menzel, E.W. (in press). Further reflections on self-recognition in primates.

Animal Behaviour.

Gallup, G.G., Jr., Wallnau, L.B., & Suarez, S.D. (1980). Failure to find selfrecognition in mother-infant and infant-infant rhesus monkey pairs. Folia Primatologica, 33, 210-219. Gopnik, A. (1982). Words and plans: Early language and the development of intelligent action. Journal of Child Language, 9, 303-318. Gopnik, A., & Slaughter, V. (1991). Young children's understanding of changes in their mental states. Child Development, 62, 98-110.

190

DANIELJ. POVINELLI

Guillaume, P. (1971). Imitation in children, 2nd Ed. Chicago: University of Chicago Press. (original work published 1926) Itakura, S. (1987a). Use of a mirror to direct their responses in Japanese monkeys (Macaca fuscata fuscata). Primates, 28, 343-352. Itakura, S. (1987b). Mirror guided behavior in Japanese monkeys (Macacafuscata fuscata). Primates, 28, 149-161. James, W. (1950). The principles of psychology, New York: Dover. (original work published 1890) Johnson, C.B. (1983). Self-recognition in infants. Infant Behavior and Development, 6, 211-222. Kagan, J. (1981). The second year: The emergence of self-awareness. Cambridge, MA.: Harvard University Press. Langer, J. (1986). The origins of logic: One to two years. New York: Academic Press. Leslie, A. (1987). Pretense and representation: Origins of "theory of mind." Psychological Review, 94, 412-426. Lethmate, J., & Ducker, G. (1973). Untersuchungen am ebsterkennen im spiegel bei orangutans einigen anderen affenarten. [Self-recognition by orangutans and some other primates.] Zeitschrift fur Tierpsychologie, 33, 248-269. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self New York: Plenum Press. Lin, A.C., Bard, K.A., & Anderson, J.R. (1992). Development of self-recognition in chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 106, 120127. Lillard, A.S. (1993). Pretend play skills and the child's theory of mind. Child Development, 64, 348-371. Loveland, K.A. (1986). Discovering the afffordances of a reflecting surface. Developmental Review, 6, 1-24. Macnamara, J. (1982). Names for things. Cambridge, MA: MIT Press. Maestripieri, D., Schino, G., Aureli, F., & Troisi, A. (1992). A modest proposal: Displacement activities as an indicator of emotions in primates. Animal Behaviour, 44, 967-979. Marchal, P., & Anderson, J.R. (1993). Mirror-image responses in capuchin monkeys (Cebus apeUa): Social responses and use of reflected environmental information. Folia Primatologica, 61, 165-173. Meltzoff, A.N. (1990). Foundations for developing a concept of self: The role of imitation in relating self to other and the value of social mirroring, social modeling, and self-practice in infancy. In D. Cicchetti & M. Beeghly (Eds.), The self in transition: Infancy to childhood, (pp. 139-164). Chicago: University of Chicago Press. Meltzoff, A.N., & Moore, M.K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 75-78. Mitchell, R.W. (1993). Mental models of mirror-self-recognition: Two theories. New Ideas in Psychology, 11, 295-325. Nelson, K. (1989). Monologue as the linguistic construction of the self in time. In K. Nelson (Ed.), Narratives from the crib (pp. 284-308). Cambridge, MA: Harvard University Press. Nelson, K. (1991). The matter of time: Interdependencies between language and thought in development. In S.A. Gelman & J.P. Bymes (Eds.), Perspectives on language and thought (pp. 278-318). Nelson, K. (1993). The psychological and social origins of autobiographical memory. Psychological Science, 4, 1-8.

THE UNDUPLICATED SELF 191 Olson, D.R. (1993). The development of representation: The origins of mental life. Canadian Psychology, 34, 1-14. Olson, D., & Campbell, R. (1993). Constructing representations. In C. Pratt & A.F. Garton (Eds.), Systems of representation in children: Development and use (pp. 11-26). New York: John Wiley & Sons. Parker, S.T. (1991). A developmental approach to the origins of self-recognition in great apes. Human Evolution, 6, 435-449. Perner, J. (1991). Understanding the representational mind. Cambridge, MA: MIT Press. Povinelli, D.J. (1989). Failure to find self-recognition in Asian elephants (Elephas maximus) in contrast to their use of mirror cues to discover hidden food. Journal of Comparative Psychology, 103, 122-131. Povinelli, D.J. (1993). Reconstructing the evolution of mind. American Psychologist, 48, 493-509. Povinelli, D.J. (in press). Chimpanzee theory of mind? The long road to strong inference. In P. Carruthers & P. Smith (Eds), Theories of theories of mind. Cambridge: Cambridge University Press. Povinelli, D.J., & Eddy, T.J. (in press). What young chimpanzees know about seeing. Monographs of the Society for Research in Child Development. Povinelli, D.J., Landau, K., & Perilloux, H.K. (in press). Self-recognition in young children using delayed versus live feedback: Evidence of a developmental asynchrony. Child Development. Povinelli, D.J., Rulf, A.R., Landau, K., & Bierschwale, D.T. (1993). Selfrecognition in chimpanzees (Pan troglodytes): Distribution, ontogeny, and patterns of emergence. Journal of Comparative Psychology, 107, 347-372. Priel, B., & de Schonen, S. (1986). Self-recognition: A study of a population without mirrors. Journal of Experimental Child Psychology, 41, 23 7-250. Rescorla, R.A. (1967). Pavlovian conditioning and its proper control procedures. Psychological Review, 71, 71-80. Robinson, J.A., Connell, S., McKenzie, B.E., & Day, R.H. (1990). Do children use their own images to locate objects reflected in a mirror? Child Development, 61, 1558-1568. Schulman, A.H., & Kaplowitz, C. (1977). Mirror image response during the first two years of life. Developmental Psychobiology, 10, 133-142. Suarez, S.D., & Gallup, G.G., Jr. (1981). Self-recognition in chimpanzees and orangutans, but not gorillas. Journal of Human Evolution, 10, 175-188. Suddendorf, T. (1994). Discovery of the fourth dimension: Mental time travel and human evolution. Unpublished Master's thesis. University of Waikato, Hamilton, New Zealand. Swartz, K.B., & Evans, S. (1991). Not all chimpanzees show self-recognition. Primates, 32, 483-496. Tomasello, M. & Call, J. (in press). Social cognition of monkeys and apes.

Yearbook of Physical Anthropology, 37. Weist, R.M. (1986). Tense and aspect. In P. Fletcher and M. Garman (Eds.), Language acquisition, 2nd Edition (pp. 356-374). Cambridge: Cambridge University Press. Whiten, A. (1993). Evolving theories of mind: The nature of nonverbal mentalism in other primates. In S. Baron-Cohen, H. Tager-Flusberg, D. Cohen, & F. Volkmar (Eds.), Understanding other minds, (pp. 367-396). Oxford: Oxford University Press.

192

DANIELJ. POVINELLI

Zahn-Waxler, C., & Radke-Yarrow, M. (1982). The development of altruism: Alternative research strategies. In N. Eisenberg (Ed.), The development of prosocial behavior (pp. 109-137). New York: Academic Press. Zazzo, R. (1982). The person: Objective approaches. In W.W. Hartup (Ed.), Review of child development research: Vol. 6(pp. 247-290). Chicago: University of Chicago Press.

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

193

CHAPTER 10

The Self as Reference Point: Can Animals D o Without It? EMANUELA CENAMI SPADA, FILIPPO AURELI, PETER VERBEEK and FRANS B.M. DE WAAL

Yerkes Regional Primate Research Center Emory University

The presence of a self in human infants and nonhuman animals has been investigated with similar methods. These methods have in common that they do not rely on verbal mediation (e.g., tests of mirror self-recognition; Amsterdam, 1972; Lewis & Brooks-Gunn, 1979; Gallup, 1970). Despite the shared methodology, research on preverbal members of our own species and members of other species (hereafter "animals") has been guided by quite different assumptions. In the case of preverbal children, developmental psychologists have been trying to answer questions such as: When does the process of forming the self begin? What are the most important aspects of this process? How does the self that we experience as adults develop? These questions derive from the assumption that every human being will eventually gain a level of awareness of the self that is characteristic of our species. For the purpose of this paper, the term self-awareness will refer to this mature conception of the self. Following the assumption of a species-typical level of self-awareness, one can investigate its antecedents as early as infancy. In the case of animals, the situation is not the same. Traditionally, such as in the case of radical behaviorism, the possibility of studying the animal self has simply been ruled out; an animal self was considered either nonexistent or at least unknowable. Nonetheless, a number of comparative psychologists and cognitive ethologists have recently begun to explore the possibility of animal selfawareness. Comparative psychologists have attempted to determine self-awareness in animals, particularly primates, by means of self-recognition tests with a mirror (e.g., Gallup, 1970). Cognitive ethologists (e.g., Griffin, 1976), on the other

194

E. CENAMI SPADA, F. AURELI, P. VERBEEK, ~ F.B.M. DE WAAL

hand, have sought animal self-awareness in natural behavior, particularly behavior that appears to rest on a high level of intentionality or cognition. The conclusions have been quite different. According to comparative psychologists, self-awareness is limited to humans and some of the great apes (reviewed in Parker, Mitchell, & Boccia, 1994), whereas cognitive ethologists believe self-awareness to be much more widespread (reviewed in Ristau, 1991). Whichever view one leans toward, it is evident that the assumptions underlying research on self-awareness in animals are quite different from those underlying research on preverbal children. Asking when the human self emerges and how it develops into adult self-awareness is not analogous to asking whether animals are self-aware. The former is a question of timing, the latter of occurrence. In the former case, the approach is justified because we know that normally children will reach the adult experience of self-awareness; in the latter case, however, we cannot reasonably expect a replication of all aspects of adult human self-awareness in other species. This, however, does not rule out the possibility of a sense of self in animals. Thus, before asking whether animals are self-aware, we think it is important to try to detect whether they show behavior that may allow us to consider a self at all. In this chapter, we propose a different approach. We will try to answer the question: Can animals do without a self as a reference point? Here, "self" is the ability of a living organism to be an active agent in its physical and social environment by means of a continuous monitoring of its position in relation to any environmental situation, i.e., danger, hunting, attack, etc. We propose that a self is a necessary condition for an active agent to adaptively solve the problems it is likely to encounter in the environment. Thus, we are not concerned with establishing antecedents for adult human self-awareness, but rather with determining if the reality of animal life includes a self as a necessary condition. In this attempt, we take for granted the material conditions for a self (i.e., the fact that basic structure and functioning of neurons and synapses are quite similar in all animals with organized central nervous systems), and we explore the necessity of postulating a self as a conditionl to give an adequate account for most of animal behavior. Memories, expectations, and plans, in this perspective, are elements of the sense of self. Whereas we do not wish to exclude the possibility of a sense of self underlying a single animal-object 2 relationship that rests on experience and memory, the more obvious area to explore is that of relationships with the environment that involve dynamic states of objects ~ or configurations of objects or interactions between objects. In these instances, a sense of self is the term from which the animal-object relationship proceeds. Although it could be argued that such more complex stimulus-response associations can also be acquired and remembered by a machine, we are not seeking to distinguish one more time

THE ANIMAL SELF AS REFERENCE POINT

195

between animals (or humans) and machines (see Gibson, this volume), but to reformulate the terms of the problem. If, for example, a stone located near the sea is moved in all directions by a sea current, the presence of a self as reference point seems quite unnecessary. 3 In contrast, from an evolutionary point of view, given the material conditions that animals share with us, the necessity of postulating a self as a reference point in animals seems worth exploring.

Self: A Necessary Condition for Infant and Animal Behavior In psychology, distinctions among several kinds of selves have been proposed for more than a century. Starting with James's (1980) distinction between the "I" and the "me," to more recent distinctions between the "implicit" or "explicit" self (Case, 1991), the "machinery of the self' and the "idea of me" (Lewis, 1994), and the "five kinds of selves" proposed by Neisser (1988), it has become evident that in order to study the nature and development of the self, it is important to consider different aspects separately. Although there exist important differences among aspects stressed by different investigators, these analyses generally share what may be called a "passage" from a preconceptual to a conceptual self. Traditionally, philosophers and psychologists have assumed that the conceptual self derives from verbally encoded concepts and cognitive representations. Within this cognitive framework, not surprisingly, infants were not considered active organisms able to participate in the formation of their own selves. They were only capable of reacting to external stimuli. Infants were viewed as passive organisms with limited cognitive and affective capacities, existing in a state of fusion with the environment (Preyer, 1887; Piaget, 1952). This contrasts with more recent research on the active role of infants in the construction of several aspects of the self based on direct perception of and interaction with the physical environment (Rochat, this volume). Furthermore, evidence for neonatal imitation (Meltzoff & Moore, 1977) as well as intersubjectivity (Trevarthen, 1980, 1993) suggests that sensitivity to ongoing social interaction starts very early. Infants appear to express a sense of self from birth through their ability to differentiate themselves at a perceptual level from the physical environment. In addition, this distinction hinges on the role played by "others" as well as "emotions" (Stem, 1985, 1993; Trevarthen, 1992). These recent empirical findings provide a useful starting point to explore similar phenomena in animals. If the core of human self-awareness, as suggested by Trevarthen (1993, p. 121), is "immediate, unverbalized, conceptless, totally atheoretical," we may be able to find shared aspects in animals. From this perspective, we need not be preoccupied with the idea that animals should match

196

E. CENAMI SPADA, F. AURELI, P. VERBEEK, & F.B.M. DE WAAL

adult human self-awareness, and we can therefore concentrate on examining the necessary conditions for animals to behave the way they do. Our hypothesis is that one necessary condition - - and not the least important one - - is a certain sense of self as a reference point for social, affective, ecological, and cognitive aspects of the interaction between animals and their environment. Furthermore, from the human literature, we can borrow the idea of different kinds of selves (e.g., Neisser, 1988), adapting it to investigate the animal self. We modify this concept of multiplicity by removing any connotation of hierarchical levels. In this way, we are not forced from the outset to determine which species has "higher" or "lower" selves, but can freely explore what aspects of the self are more or less evident in a given species. From a methodological point of view, the idea of finding a single paradigm (e.g., mirror self-recognition) to classify all species on a single dimension would seem ideal. However, we need to take into account the enormous diversity that exists at a perceptual level among animals. Homing behavior (i.e., movements to return to a specific or familiar site) provides interesting examples of species-typical cognitive capabilities and perceptual mechanisms. Birds migrating from one continent to another and back, displaced pets and pigeons finding their way home, salmon swimming back to their native stream, and insects exploring areas that are enormous relative to their body size all rely on quite different orientational information (reviewed in Papi, 1992). For a dog to find his way home, for example, he needs to be able to distinguish between familiar and unfamiliar landmarks. Similarly, when a dog persistently marks a site in his territory after another dog has urinated there, we need to assume that he olfactorily distinguishes his own urine markings from those of other dogs. Such self-recognition and preference for a place where one has spent time before are hard to imagine without a sense of self and its position in the "world." Although we can probably better understand and explain the behavior of species that rely upon sensory systems similar to ours (e.g., nonhuman primates), we should not overlook the diversity of perceptual capacities in other species. We need to leave open the possibility that we sometimes underestimate the complexity of a species' behavior because of our own limitations and biases in the perceptual domain. The importance of smells in forming the scene of a dog's world can, to a certain extent, be easily understood. However, not only do different species rely on different sensory channels (e.g., odors vs. visual cues), but a variety of organisms depend on stimuli that are completely foreign to us. For example, the hearing experience may range from species that hear through their feet (invertebrates such as roaches, spiders, and scorpions) to birds or mammals that, in order to

THE ANIMAL SELF AS REFERENCE POINT

197

communicate and move in their environment, use ultrasonic (bats and dolphins) or subsonic (elephants) frequencies. Bats and certain cetacea that use echolocation for orientation and hunting must possess the ability to pick out their own signals from among those emitted by (sometimes thousands of) conspecifics. Certainly, this form of self-recognition is much harder to comprehend for us than the one that occurs in a mirror. Jerison (1986) even speculates about collective perception by dolphins "listening in" on one another's return signals. Finally, in our view, the self does not correspond to a specific capacity or property that only a few species possess, but to a sense of the organism's own position vis-a-vis the physical and social environment that assists the effective functioning of most living species. Our analysis of the self in animals as a necessary condition for certain kinds of behavior thus needs to start from what animals do and how they do it. In the following section, we discuss a few examples of behavior related to the physical environment. Subsequently, we consider animals in their social environment. In our analysis, social environment refers to both conspecifics and other animals (e.g., predators and prey). Our examples of social behavior between conspecifics are mainly drawn from the primatological literature. This is not to say that similar examples could not be developed for other animals, but primate societies have been studied in greater detail than the social groups of most other species.

The Animal Self in the Physical Environment The uniqueness of each animal species' perceptual world is certainly not a new concept in biology; it is known as the Umwelt (von Uexktill, 1957). In this section, our aim is to account for the interaction between animals and the inanimate environment surrounding them. However, it must be remembered that even the inanimate environment is far from static. For a moving animal, it is constantly changing; therefore, an animal needs to continuously adjust its movements accordingly. Saharan desert ants of the genus Cataglyphis scavenge for other insects, and rely on vision to find food and to return to their nest (Wehner, Harkness, & Schmid-Hempel, 1983). These solitary ants can navigate a homeward course from a food source, even if the food source is in unknown terrain, and prior to any reinforcement provided by the food or a successful trip home (Gallistel, 1990). Dyer (1994) reviewed a series of observational and experimental studies by Wehner and his colleagues that offers insight into the remarkable homing behavior of these insects.

198

E. CENAMI SPADA, F. AURELI, P. VERBEEK, & F.B.M. DE WAAL

The researchers first established that reliance on familiar landmarks does not adequately explain the ants' homing behavior. Instead, the animals appear to employ the celestial compass together with cognitive mechanisms that allow them to integrate the distances and directions traveled over successive segments of their path relative to the celestial reference. Theoretically, this can be accomplished by either vector summation or distance-weighted arithmetic calculations of displacement over an outward path. Vector summation would require the ant to compute sine and cosine projections of direction and distance traveled, and to apply the Pythagorean theorem. In contrast, a distance-weighted calculation requires less complex computations and does not depend on the application of a fixed rule such as the Pythagorean theorem. As Dyer (1994) points out, in contrast with the long-standing assumption that vector summation underlies path integration in insects (e.g., Mittelstaedt, 1985), the experimental evidence strongly favors calculative mechanisms (Mueller & Wehner, 1988). The relevance of these data to the current discussion is that the suggested process requires the ant to continually update the position of the self in reference to the celestial compass. This compass, based on patterns of polarized light, is dynamic, which means that the ant needs to "interpret" it and derive its position from this interpretation. So, rather than storing traveling distances and angles in memory, without knowledge of where it is at each specific moment, the ant needs to be an active agent continuously keeping track of its position relative to its point of departure. Another example of the self as a referent in the physical environment is found in arboreal locomotion. For efficient locomotion to take place, animals must continuously evaluate their environment and the carrying capacity of the substratum. Accurate knowledge of their own characteristics, such as body gravity and the reach of arms and legs, is required to estimate where to move next. Arboreal mammals, for instance, must select branches that are strong enough to hold their weight and on which they can keep a finn grip to avoid the risk of a fall. Learning obviously plays an important role. In fact, when moving within a familiar area, arboreal mammals frequently follow the same paths, making it easier to evaluate the environment. The movements become almost automatic and the speed increases. However, a missing branch on a familiar path poses a challenge. The animal needs to find new support that allows continuation of its progression. Under these circumstances, it becomes evident that animals need to refer to their spatial position in relation to the limits of their own capacities before selecting the support for the next safe move. During quadrupedal locomotion, secure contact with the initial support is maintained while reaching for new support so as to avoid a fall if the support suddenly bends or breaks. All limbs are used in walking on horizontal branches and

THE ANIMAL SELF AS REFERENCE POINT

199

in climbing up or down vertical supports. The finn grip of the prehensile tail of some New World monkeys serves as an additional "safety anchor." This type of locomotion is often used while moving within the same tree or between trees in dense forests. However, different problems are encountered when animals need to negotiate passages between discontinuous supports such as between tree crowns. Animals capable of a long reach, such as orangutans (Pongo pygmaeus; MacKinnon, 1974) or spider and howler monkeys (genera Ateles and Alouatta; Cant 1986), often retain a grasp on the original support while crossing a small discontinuity in the canopy. This type of locomotion is obviously limited to discontinuities no wider than the length of the animal's reach. Animals need to carefully evaluate the distance between trees in relation to their own dimensions.

FIGURE 1. Arboreal locomotion depends on exquisite knowledge of one's own capacities (e.g., jumping distance) and of what the physical environment can afford (e.g., firmness of substrate). A capuchin monkey (Cebus apella) bridges a gap in the canopy through leaping. Photographed by Frans de Waal in Eastern Brazil. When the discontinuity between tree crowns is greater, different solutions are required. The solution that best maximizes speed is probably leaping (see Figure 1). Before the takeoff for a successful leap, it is critical to select a support for landing at a distance that falls within the animal's ability. Because a leap too short

200

E. CENAMI SPADA, F. AURELI, P. VERBEEK, & F.B.M. DE WAAL

could be fatal, learning by trial and error is unlikely to be the best process underlying such selection. A proper evaluation of the trajectory before the leap is essential for large gaps; once in the air, the individual possesses little control over the speed and direction of the leap. Better control is achieved by animals that glide between trees. The gliding of "flying" squirrels (genus Glaucomys) is a typical example (Wells-Gosling 1985). Although these animals cannot maneuver like free-flying birds or bats, they can make adjustments of their pathway and swerve to avoid obstacles. On a familiar gliding route, flying squirrels leap out and spread the patagium without other preparation. Before engaging on an unfamiliar route, however, they bob up and down or lean far left and fight in rapid succession. Sometimes they scurry sideways for a better view from a different angle. These brief movements probably provide a better judgment of relative position and distance, via parallax. Thus, flying squirrels seem to overcome the limitation of poor depth perception by the use of triangulation, judging potential gliding paths from two or more angles. This requires them to integrate information from different views, which implies a reference to the self in relation to the landing point, not only from the current position but also from positions occupied earlier. Such an ability is difficult to account for without reference to the self and its position in relation to objects in the environment.

The A n i m a l Self in the Social Environment

Coordination with Others Some animals change their appearance in response to both features of the physical environment and the presence of other animals. They thus adjust outward features of their own bodies according to the circumstances, hiding from some animals (e.g., predators) and displaying to others (e.g., potential mates). Caribbean reef squids (Sepioteuthis sepioidea), for example, use special postures and movements to convey information; they can use ink as a decoy and employ an almost endless range of color changes. A Caribbean reef squid may mimic vegetation (black sponges or corals) to pass unperceived by a predator. However, this performance often takes place in the vicinity of members of other species as well as of conspecifics. Although particular patterns may occur exclusively in either interspecific or intraspecific encounters, there are many patterns that occur in both situations. Both classes of onlookers perceive the same performances, yet they usually do not respond in the same way. For a potential predator, performances may be cryptic or baffling, whereas the same performance may carry important social information for a conspecific. For the performer, it is necessary to

THE ANIMAL SELF AS REFERENCE POINT

201

take into account its own position relative not only to conspecifics, but also to other nearby animals. Courting parties of these squids usually break up into trios of two males and one female before courtship in heterosexual pairs takes place (Moynihan & Rodaniche, 1982). The dominant or favored male assumes a distinctive color pattern. He remains dark on the side of his body facing the female while turning brilliantly silvery on the side turned away from her. As suggested by Moynihan (1985), the silver is repellent, probably a threat to discourage approaching rivals. Courting individuals may change relative positions rapidly and repeatedly. When and if so, the displaying male shifts his silver from side to side, always with appropriate orientation. The ability to suddenly change color patterns dependent on the animal's own position relative to mates and rivals would seem to require that the male makes use of the self as reference point. Reactions to predators are also interesting in relation to self-reference. When surprised and frightened in more variegated environments than white sand, reef squid, especially young individuals, may assume color patterns of either transverse bars or longitudinal streaks. When a group of Sepioteuthis is mildly disturbed by a potential predator, all the members of the group adopt the same patterns, either bars or streaks as the case may be. Their entire body, from front to end, conforms to the right image (Moynihan, 1985). These animals thus identify with conspecifics, adopting whatever outward appearance they see them adopt. If identification is the ability to be closer to one object in the environment than another, and to make the situation of the first object to some extent one's own, this is a basic ability indeed. It makes it possible to make others a continuation of the self, and to pay close attention to their situation so as to influence it or gain information from it. This ability may be present in a wide range of animals. Thus, Dugatkin (1992) has shown that female guppies (Poecilia reticulata) that have watched another female associate with a particular male tend to choose this male over others. This "l-want-what-she-has" principle has the power of reversing a female's independent mate preferences. Similarly, Fiorito and Scotto (1992) trained an octopus (Octopus vulgaris) to attack either a red or a white ball. After the training, another octopus was allowed to watch four demonstrations from an adjoining tank. When the balls were subsequently dropped in the spectator's tank, he attacked the ball of the same color as the first octopus. These experiments demonstrate that even some invertebrates and fish notice how members of their own species relate to the environment. The octopus identified with the other octopus, and the female guppy with the other female guppy, both letting their counterpart influence their attitude toward a stimulus. It is as if they "invested" the self into one object in the environment, closely

202

E. CENAMI SPADA, F. AURELI, P. VERBEEK, • F.B.M. DE WAAL

following the relation established by this object with other objects. The primate literature contains ample evidence for such so-called "triadic awareness," i.e., individual A's recognition of the relationship between B and C as well as of this relationship's impact upon A itself (de Waal, 1982). 4 The components of this triangle may be animate or inanimate. Thus, an ape imagining himself standing on top of two boxes reaching with a stick for a banana before he has actually brought about this situation by pulling over boxes and grabbing a sock (K0hler, 1925) is connecting the self to both an extension of the self (the tool) and an out-of-reach object (the food). Most often, however, triadic awareness is discussed in relation to situations in which all three components of the triangle are conspecifics. Due to their tendency to form alliances (i.e., two or more parties supporting each other against a third; Harcourt & de Waal, 1992; see Figure 2), the complexity of these social triangles is perhaps greater in primates than in any other animal group.

Triadic Social Relationships of Primates The need to postulate a self as referent during social interactions is most strongly felt when one considers how the relation between two individuals may be affected by the presence and actions of other individuals. In the case of dominance relationships, for instance, elegant experiments with Japanes e macaques (reviewed in Chapais, 1992) have confirmed the important role of third parties first documented by Kawai (1965). A young macaque who usually dominates another becomes subordinate in the absence of her allies and in the presence of at least one of the other juvenile's allies. The original dominance relationship is rapidly recovered when an ally of the formerly dominant individual is reintroduced. These changes in behavior seem to require a good grasp of the entire social environment. Kummer (1971) described the need of nonhuman primates to process large amounts of information for successful social functioning as follows: "In turning from one to another context at a rapid rate, the individual primate constantly adapts to the equally versatile activities of the group members around him. Such a society requires two qualities in its members: a highly developed capacity for releasing or suppressing their own motivations according to what the situation permits and forbids; and an ability to evaluate complex social situations, that is, to respond not to specific social stimuli but to a social field." (Kummer, 1971, p. 36) ~ Even though associative learning is certainly involved, a self would be a very helpful point of reference for the rapid evaluation of any new situation by relating the potential effects of the individual's own actions to the presence or absence of specific individuals. Further evidence for the importance of such self-reference is provided by the effects of previous experience. When a juvenile Japanese macaque (Macaca fuscata)

THE ANIMAL SELF AS REFERENCE POINT

203

is tested for the first time with a subordinate juvenile in the presence of a powerful ally of the latter, rank reversal between the juveniles occurs most of the time only after aggressive intervention of this ally against the first juvenile. By contrast, when the same juvenile is subsequently tested in a similar setting but with different individuals, rank reversal occurs mainly through spontaneous submission without the need of aggressive intervention (Chapais, 1992). Thus, the dominant juvenile is able to transfer the knowledge gained from previous experience and spontaneously displays submissive signals to minimize the risk of receiving aggression.

FIGURE 2. Primate society is characterized by coalition formation in which two or more parties support one another against a third. Three rhesus monkeys (Macaca mulatta), all members of the same matriline, confront a female (viewed from the back) from a dominant matriline. This single female is able to stand her ground because if she were to scream, all members of her own matriline would rush to her defense. Photographed by Frans de Waal at the Wisconsin Primate Center.

204

E. CENAMI SPADA, F. AURELI, P. VERBEEK, & F.B.M. DE WAAL

Another example of triadic awareness concerns hamadryas baboons (Papio hamadryas). Males of this species appropriate and herd females in an arrangement that is commonly referred to as a "harem." In earlier observations, Bachman and Kummer (1980) had noted that dominant males did not attempt to appropriate females after having seen a female interact with another male. This is intriguing because the dominance principle would predict that a dominant male would exert his power over subordinate males in order to gain access to the female. Bachman and Kummer (1980) decided to investigate this inhibition in an experimental situation. Six males and six females of a captive group of wild-caught baboons were tested. Although they were individually housed between tests, all baboons were familiar with one another from previous encounters in small groups. First, preferences between individuals were measured by means of a choice test in which each subject was given a choice to affiliate with one of two partners of the opposite sex. Next, individual males ("rivals") were allowed to watch a male and a female interact for 15 minutes. The rival males were kept in a small cage next to a larger enclosure that held the female and her male "owner." After 15 minutes, the rival male was allowed access to the pair so as to measure his tendency to "respect" their relationship, i.e., refrain from aggressive attempts to appropriate the female. It was found that the rival's respect did not correlate with either his own or the owner's preference for the female. Instead, it depended on the female' s preference for her current owner. The investigators speculated that it may be too costly for a rival to appropriate a female who strongly prefers her owner. Herding her, with a good chance of losing her again, may not be worth the effort to the rival male. Possibly, therefore, baboons are capable of assessing relationships among other members of their group. In terms of our discussion, the rivals not only appeared to assess the relationship between two individuals, but their subsequent action was a function of this assessment and a corresponding expectancy of the nature of subsequent interaction between themselves and one of the individuals. Studies of redirection of aggression provide further examples of predictions based on complex judgments of an entire "social field" (cf. Kummer, 1971, quoted above). An individual who receives aggression may subsequently attack a third individual, a response called redirection (Bastock, Morris, & Moynihan, 1953). Targets of redirection are usually individuals subordinate to the recipient of the original aggression. Redirection against such targets reduces reoccurrence of aggression against the redirecting individual herself (Aureli & van Schaik, 199 lb). Among nonhuman primates, however, redirection may target the former aggressor's kin or other close associates (Judge 1982; Smuts 1985; Cheney & Seyfarth 1986, 1989; Aureli & van Schaik 1991a; Aureli, Cozzolino, Cordischi,

THE ANIMAL SELF AS REFERENCE POINT

205

& Scucchi, 1992). Such redirection is a form of indirect retaliation: In the long run, its negative repercussions may make individuals refrain from aggression. In the short run, however, this type of redirection entails risks for the redirecting individual if the former aggressor defends the target of redirection. The fact that this type of redirection nevertheless occurs indicates that it is probably restricted to circumstances in which intervention by the former aggressor is strongly constrained.

@ @

@ @ FIGURE 3. Macaque society is characterized by a strict dominance hierarchy based on family relations. Kin-oriented redirection documented in Japanese macaques demonstrates that these monkeys possess intimate knowledge of their social system and of their own position within it. The drawing represenLs members of two families. A is the most dominant female in the group, and A1, A2, and A3 are her offspring. In the presence of family allies, all A-members dominate B and her offspring (B1, B2, and B3). The youngest member of the dominant family, A1, is not yet dominant over B without family support. If B is attacked by A (arrow 1), B will redirect against this vulnerable A-member (arrow 2). Among Japanese macaques, redirection against former opponents' kin takes place under special circumstances indeed (Aureli et al., 1992). When individual A attacks B (who belongs to a different family), the latter is usually subordinate to A (see Figure 3). It is risky, therefore, for B to attack A1, a relative of A, because

206

E. CENAMI SPADA, F. AURELI, P. VERBEEK, & F.B.M. DE WAAL

A1, too, is likely to be dominant over B, and A is likely to come to Al's aid. This is due to the strong family bonds and strict matrilineal hierarchy of these macaques, in which related females help each other and assume similar ranks. To avoid these problems, individual B must select a vulnerable target and favorable circumstances (e.g., absence of A). In fact, targets of kin-oriented redirection are usually juveniles subordinate to the redirecting individual, if unsupported by adult relatives. Of course, the vulnerability of these individuals changes with age, and so B must adjust her selection of targets over time. One possibility is for B to simply redirect to A1 when A is not around. For kin-oriented redirection to be advantageous, however, another condition should be met: A should be aware of the redirected aggression and change her aggressive attitude toward B over time to avoid retaliation against A1. Indeed, among Japanese macaques, A is within sight of the redirection against A1 in most of the cases, yet A rarely intervenes in favor of A1. This is possible because redirection occurs under circumstances in which it is risky for A to aggressively support her kin. For instance, redirection may consist of B joining other individuals already attacking a vulnerable kin of A. In this case, A is unlikely to intervene because doing so would entail facing several individuals at once, some of whom might be dominant. The study of kin-oriented redirection is relevant to our investigation of the animal self. In general, the conditions for kin-oriented redirection are such that a vulnerable target (A1) can be attacked within sight of the former aggressor (A) without the risk of aggressive intervention. There are many factors that animals must evaluate to select the appropriate conditions. Such conditions are rarely met, and this is the reason why kin-oriented redirection is a relatively rare behavior. Its rarity indicates that redirecting individuals need to be highly opportunistic and must rapidly evaluate situations that may be completely novel. The necessary judgment of the situation as a whole can probably only be achieved if the animal projects herself in the situation that might develop if she followed a certain course of action. This projection most likely rests on a comparison between the situation the animal is currently facing and previous expev'.ences. Such a comparison in turn would lead to a full assessment of the potentiality and risk associated with attack. Judgment of complex social situations is also reflected in so-called separating interventions (de Waal, 1978, 1982; see Figure 4). A separating intervention occurs when one individual breaks up an affiliative contact between two others through interposition, aggression, or threat display. In the chimpanzee (Pan troglodytes) colony of the Arnhem Zoo, the majority of such interventions is performed by adult males in response to an approach between other adults. The actor will either charge directly at the others or display in close vicinity until one of the two, or both, depart. The generally short response latency to a newly established contact and the abrupt ending of the male's display once the targeted

THE ANIMAL SELF AS REFERENCE POINT 207

FIGURE 4. A sequence of a separating intervention among adult male chimpanzees at the Arnhem Zoo. Top: The alpha male (left) notices his ally (right) sitting together with his chief rival (center). He faces the two others with hair erect. Middle: When the alpha begins to display, staring menacingly at the rival male, his ally gets up and walks away. Bottom: The alpha charges closely past the rival male, lifting an arm over him (a dominant gesture), while his ally looks on. His ally then permanently leaves the scene, and the alpha has thus successfully separated the two. Such intervention suggests an understanding in chimpanzees of the possible implications of alliances among others: This alpha male would never have come to power without his ally's backing. Photographs by Frans de Waal, from de Waal (1982).

208

E. CENAMI SPADA, F. AURELI, P. VERBEEK, & F.B.M. DE WAAL

contact has been terminated strongly suggest goal-directedness. Similar interventions have been observed in wild chimpanzees (Nishida, in press). Separating interventions represent the other side of the coin of alliances. Rather than by individual fighting abilities, dominance relations mnong male chimpanzees are largely decided by alliances. The male with the most effective and committed supporters tends to achieve the upper hand. Separating interventions may be interpreted as an attempt to protect the actor's own alliances while isolating rivals, i.e., as a "divide-and-rule" strategy that prevents the formation of alliances hostile to the actor. This hypothesis is based on the observation that interveners almost exclusively target contenders for their own rank at moments when these contenders are in proximity to suitable alliance partners, such as other adult males or high-ranking females. Thus far, this hypothesis has offered the best prediction of the distribution of separating interventions in the Arnhem colony, including changes in the targeted contacts following changes in the configuration of alliances. If true, we see again a situation in which an individual judges the relationship between two objects in his environment and how this relationship might affect himself based on knowledge of how the self relates to either one of them.

The Animal Self in Relation t~::,,Other Selves To make a beeline return after a tortuous journey away from the nest as described earlier for desert ants m suggests that the self serves as reference point in path calculations. Things become considerably more complex, however, when one individual possesses spatial knowledge that another tries to obtain or take advantage of. This may require that the latter individual connects two reference points: that of the self and the other. Pioneering research in this area was conducted with a small group of juvenile chimpanzees by Menzel (1974). The investigator would take one juvenile out into a large enclosure to show it hidden food. Subsequently, he would take the "knower" back to the waiting group, and release all of them together. Would the others appreciate the knower's information, and if so, would they develop strategies to exploit it? Menzel describes the attempt of a dominant chimpanzee, Rock, to outwit an equally canny subordinate, Belle. Because Rock tended to take Belle's food away, Belle quickly developed a counter-strategy: Belle accordingly stopped uncovering the food if Rock was close. She sat on it until Rock left. Rock, however, soon learned this, and when she sat on one place for more than a few seconds, he came over, shoved her aside, searched her sitting place, and got the food. Belle next stopped going all the way. Rock, however, countered by steadily expanding the area of his search through the grass near where Belle sat. Eventually, Belle sat farther and farther away, waiting until Rock looked in the

THE ANIMAL SELF AS REFERENCE POINT

209

opposite direction before she moved toward the food at a l l - and Rock in turn seemed to look away until Belle started to move somewhere. On some occasions Rock started to wander off, only to wheel around suddenly precisely as Belle was about to uncover the food. In other trials when we hid an extra piece of food about 10 feet away from the large pile, Belle led Rock to the single piece, and while he took it she raced for the pile. When Rock started to ignore the single piece of food to keep his watch on Belle, Belle had temper tantrums (Menzel, 1974, p. 134-135). To assume that another individual possesses valuable information - - as Rock seemed to do m and to try to mislead another as to the whereabouts of food m as Belle seemed to do ~ may require conceiving others as knowing agents separate from oneself. This certainly seems a tempting explanation, although simpler accounts, based on quick learning and anticipation of the other's actions, cannot be ruled out. Menzel (1974) was the first to formulate questions relating to the attribution of knowledge, feelings, and intentions of others. This ability, now often phrased as the possession of a "Theory of Mind" about others (Premack & Woodruff, 1978), has in recent years become the focus of research covering both child and nonhuman primate behavior (Buttersworth, Harris, Leslie, & Wellman, 1991; Whiten, 1991; Cheney & Seyfarth, 1990; Povinelli, this volume). Differentiation between the knowledge states of self and other would seem to rest on differentiation between self and other, which is another way of saying that the strategies of Rock and Belle required that they perceived the self as separate from their social environment. Intentional deception ~ which rests on an evaluation of what others know and expect ~ has been documented mainly in apes. Deception can be defined as the deliberate projection, to one's own advantage, of a false image of past behavior, knowledge, or intentions. In its most complete form, this requires understanding of how one's actions come across, mad what the outside world is likely to read into them. Chimpanzees will, for instance, quickly collect a mouthful of water from a faucet in their cage when they see a stranger approaching, then wait with a perfect poker face until they can let the intruder have it. Some are such experts that they trick even people who are thoroughly mindful of the possibility. The ape will stroll around in his cage as if occupied with something else only to swing around at the right moment, when he hears his victim behind him (Hediger, 1985). The first body of evidence demonstrating that chimpanzees do the same to one another was put together by de Waal (1982), who saw them wipe unwanted expressions off their faces, hide compromising body parts behind their hands, and act blind and deaf when others tested their nerves with noisy intimidation displays. It is not hard to see how concern about signals emitted by one's own body relates to an understanding of the self. However, deception remains a controversial area largely because of the qualitative nature of the current evidence. On the other hand,

210

E. CENAMI SPADA, F. AURELI, P. VERBEEK, & F.B.M. DE WAAL

it should not surprise anyone if this behavior is sporadic; it is hard to see how deception could work if it were common and predictable. For reviews of deception in primates see Whiten and Byrne (1988), and de Waal (1992); a wider range of species is treated in Mitchell and Thompson (1986). Other possible examples of mental state attribution in monkeys and apes concern role-taking (Povinelli, Nelson, & Boysen, 1992), expressions of sympathy toward victims of aggression (de Waal & Aureli, in press), and observational learning (Visalberghi & Fragaszy, 1990; Tomasello, Kruger, & Ratner, 1993). Needless to say, this is a hotly debated area of research with few firm conclusions (e.g., Heyes, 1993). Further observations and experiments are needed to clarify how animal selves relate to other selves.

Conclusion The variety of the animal kingdom is so vast that the list of examples to explore a sense of self is limitless. However, even from the small number of situations here described, the presence of the self as a reference point appears to be a necessary condition to explain and understand most animal behavior. Thus, we feel comfortable answering the question "Can animals do without a self?" in the negative. However, the question "Is the self as a necessary condition also sufficient to maintain self-awareness in animals?" still needs to be formulated in a way that takes species differences into account, as well as the great variety of Umwelten. In this sense, our exploration of the animal sense of self contrasts with approaches based on a single measure of self-awareness (e.g., mirror selfrecognition). For, from an evolutionary point of view, it is hard to accept that only humans and a couple of their closest relatives would be self-aware, leaving all other animals without a self (see also Gibson, this volume). This way, we would merely draw the demarcation line between humans and animals a bit l o w e r - by admitting a few species into our domain m while keeping a Cartesian dualism for the remaining species. Sharp dividing lines m regardless of whether they place our species in a class by itself or create a slightly more inclusive elite are to be treated with reservation. Theoretical justification, of course, is not the same as empirical demonstration. At this point, a reasonable question may arise: "Should all animals have a sense of self within this framework?" Certainly, our examples cannot resolve this question one way or another. However, by recalling the contrast proposed at the beginning of this chapter between the movements of inanimate and animate objects, our attempt might become clearer. That is, we first need to agree

THE ANIMAL SELF AS REFERENCE POINT

21 1

that not only are animals able to learn by associating specific stimuli in their physical and social environment, but also, in order to cope with different environmental situations, of remembering them. They not only carry out behavior, but also plan it in advance and exert control over it. A stone moved by sea currents only "performs." In contrast, an animal that as we have described acts as an agent in a cease.lessly varying (physical and social) environment seems to require a reference point (the self) as a necessary condition for its actions. The difference between "performance" and "experience" (e.g., Crook, 1983) is what Heidegger (1983) analyzed as the difference between "a stone without a world" or "an animal with a poor world" ("poor," in his sense, is of course in contrast with the human world). But, how "poor" might their world be compared with ours? This is probably the hardest question to answer, as it implies several interlinked considerations. The first consideration regards the difficulty of projecting ourselves into a completely different sensory perception system, e.g., our (in)ability to understand "what it is like to be a bat" (Nagel, 1974). The second m connected to the first M is our unavoidable anthropocentric bias to best understand animal behavior closest to ours (Mitchell, Thompson, & Miles, in press) and therefore to consider it as somehow "richer" (i.e., possessing "more self'). The third is our misleading tendency to see the self as a property that is either present or absent, rather than a relation between the acting agent and the environment. Instead, we emphasize how the animal self is formed on the basis of reciprocal influences and dynamic interactions with surrounding objects: it is ecological, social, and cognitive. Any attempt to measure the complexity of the selves of different species requires full awareness of these variables. In this sense, the investigation of the self in animals is infinitely more complex than the study of the self in preverbal children. However, a promising approach to further explore issues related to the self m not only in animals, but also in young children - - is the use of naturalistic observation in conjunction with experimental work. If, as suggested by Trevarthen (1993), the core of human selfawareness can be considered nonverbal and unconceptualized, there are good prospects for its study in animals.

NOTES

1. Following Kant's Critique of Pure Reason (1934), we could term this condition "condition of possibility" (Bedingung der M6glichkeit), i.e., a general (non ad hoc) condition that although it cannot be empirically observed as such must be postulated if one is to make sense of certain interrelated empirical phenomena (in this case, animal behavior).

212

E. CENAMI SPADA, F. AURELI, P. VERBEEK, & F.B.M. DE WAAL

2. With object, we mean both inanimate objects and other animals. 3. On this radical, if unusual way of approaching our problem, see the seminar offered by Martin Heidegger, Die Grundbegriffe der Metaphysik. Welt - Endlichkeit Einsamkeit, during the academic year 1929-1930. 4. A concept closely related to "triadic awareness" is Cheney and Seyfarth's (1990, pp. 72-86) "nonegocentric social knowledge." This phrase refers to the fact that monkeys and apes learn aspects of social relationships in which they themselves are not directly involved, such as the hierarchy among others, or the matriline to which each group member belongs. It emphasizes the ability of A to observe interactions between B and C and to evaluate the B-C relationship, whereas the phrase triadic awareness considers how A needs to understand the B-C relationship because of its implications for the A-B and A-C relationships. For the present discussion, the phrase triadic awareness is preferable as it emphasizes this connection to the self.

ACKNOWLEDGMENTS Writing of this chapter was made possible by support from the National Institutes of Health grant RR-00165 to the Yerkes Regional Primate Research Center. We are grateful to Stefano Velotti, caro Leopardi, and the editor for comments on previous versions of the manuscript. REFERENCES Amsterdam, B. (1972). Mirror self-image reactions before age two. Developmental Psychobiology, 5, 297-305. Aureli, F., & van Schaik, C. P. (1991a). Post-conflict behaviour in long-tailed macaques (Macaca fascicularis): I The social events. Ethology, 89, 89-100. Aureli, F., & van Schaik, C. P. (1991b). Post-conflict behaviour in long-tailed macaques (Macaca fascicularis): II Coping with the uncertainty. Ethology, 89, 101-114. Aureli, F., Cozzolino, R., Cordischi, C., & Scucchi, S. (1992). Kin-oriented redirection among Japanese macaques: an expression of a revenge system? Animal Behaviour, 44, 283-291. Bachman, C. & Kummer, H. (1980). Male assessment of female choice in hamadryas baboons. Behavioral Ecology and Sociobiology, 6, 315-321. Bastock, M., Morris, D., & Moynihan, M. (1953). Some comments on conflict and thwarting in animals. Behaviour, 6, 66-84. Butterworth, G. E., Hams, P. L., Leslie, A. M., & Wellman, H. M. (1991). Perspectives on the child's theory of mind. Oxford: Oxford University Press. Cant, J. G. H. (1986). Locomotion and feeding postures of spider and howling monkeys: Field study and evolutionary interpretation. Folia Primatologica, 46, 114. Case, R. (1991). Stages in the development of the young child's first sense of self. Developmental Review, 11, 210-230. Chapais, B. (1992). The role of alliances in social inheritance of rank among female primates. In A. H. Harcourt, & F. B. M. de Waal (Ed.), Coalitions and alliances in human and other animals (pp. 29-59). Oxford: Oxford University Press.

THE ANIMALSELF AS REFERENCE POINT 213 Cheney, D. L., & Seyfarth, R. M. (1986). The recognition of social alliances by vervet monkeys. Animal Behaviour, 34, 1722-1731. Cheney, D. L., & Seyfarth, R. M. (1989). Redirected aggression and reconciliation among vervet monkeys, Cercopithecus aethiops. Behaviour, 110, 258-275. Cheney, D. L., and Seyfarth, R. M. (1990). How monkeys see the world: Inside the mind of another species. Chicago: University of Chicago Press. Crook, J. H. (1983). On attributing consciousness to animals. Nature, 303, 11-14. de Waal, F. B. M. (1978). Exploitative and familiarity-dependent support strategies in a colony of semi-free-living chimpanzees. Behaviour, 66, 268-312. de Waal, F. B. M. (1982). Chimpanzee politics. Baltimore: Johns Hopkins University Press. de Waal, F. B. M. (1992). Intentional deception in primates. Evolutionary Anthropology, 1, 86-92. de Waal, F. B. M., and Aureli, F. (in press). Reconciliation, consolation, and a possible cognitive difference between macaque and chimpanzee. In Russon, A. E., Bard, K. A., and Parker, S. T. (Eds.) Reaching into thought: The Minds of the great apes,. Cambridge: Cambridge University Press. Dugatkin, L. A. (1992). Sexual selection and imitation: Females copy the mate choice of others. American Naturalist, 139,1384-1389. Dyer, F. C. (1994). Spatial cognition and navigation in insects. In L. A. Real (Ed.). Behavioral mechanisms in evolutionary ecology (pp. 66-98). Chicago: The University of Chicago Press. Fiorito, G., and Scotto, P. (1992). Observational learning in Octopus vulgaris. Science, 256, 545-547. Gallistel, C. R. (1990). The organization of learning. Cambridge, Mass: MIT Press. Gallup, G.G., Jr. (1970). Chimpanzees: Self-recognition. Science, 167, 86-87. Griffin, D.R. (1976). The question of animal awareness: Evolutionary continuity of mental experience. New York: Rockfeller University Press. Harcourt, A. H. & F. B. M. de Waal (Eds.). (1992). Coalitions and alliances in humans and other animals. Oxford: Oxford University Press. Hediger, H. (1955). Studies in the psychology and behaviour of animals in zoos and circuses. London: Buttersworth. Heidegger, M. (1983). Die Grundbegriffe der Metaphysik. Welt- Endtichkeit Einsamkeit. Frankfurt am Main: Klostermann Verlag. Heyes, C. M. (1993). Anecdotes, training, trapping and triangulating: Do animals attribute mental states? Animal Behaviour, 46, 177-188. James, W. (1890). The principles of psychology. Henry Holt & Co. Jerison, H. J. (1986). The perceptual worlds of dolphins. In R. J., Schusterman, J. A., Thomas, & Wood, F. G. (Eds.). Dolphin cognition and behavior: A comparative approach, (pp.141-166). London: Erlbaum. Judge, P. G. (1982). Redirection of aggression based on kinship in a captive group of pigtail macaques. International Journal of Primatology, 3, 301. Kant, I. (1934). Critique of pure reason (2nd ed.). (J. M. D. Meiklejohn.). London: J.M. Dent & Sons Ltd., Everyman's Library. (Original work published 1787) Kawai, M. (1965). On the system of social ranks in a natural troop of Japanese monkeys I, II. In K. Imanishi, & S. A. Altmann (Eds. and Trans.), Japanese monkeys (pp. 66-86). Atlanta: Emory University Press. (Reprinted from Primates, 1958, 1, 111-148. K6hler, W. (1925). The mentality of apes. New York: Vintage Books. Kummer, H. (1971). Primate societies: Group techniques of ecological adaptation. Arlington Heights, IL: AHM Publishing Corporation.

214

E. CENAMI SPADA, F. AURELI, P. VERBEEK, ~; F.B.M. DE WAAL

Lewis, M. (1994). Myself and me. In S. T. Parker, R. W. Mitchell, & M. L. Boccia (Ed.), Self-awareness in animals and humans: Developmental perspectives (pp. 20-34). Cambridge: Cambridge University Press. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. New York: Plenum Press. MacKinnon, J. (1974). The behaviour and ecology of wild orang-utans (Pongo pygmaeus). Animal Behaviour, 22, 3-74. Meltzoff, A.N., & Moore, M.K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 75-78. Menzel, E. W. (1974). A group of young chimpanzees in a one-acre field. In: A. M., Schrier, & F. Stollnitz (Eds.), Behavior of nonhuman primates, (Vol. 5, pp. 83153). New York: Academic Press. Mitchell, R. & Thompson, N. (1986). Deception: Perspectives on human and nonhuman deceit. SUNY Press, New York. Mitchell, R. W., Thompson, N.S. & Miles, L. H. (Eds.). (in press). Anthropomorphism, anecdotes and animals. New York: SUNY Press. Mittelstaedt, H. (1985). Analytical cybernetics of spider navigation. In F. G. Barth (Ed.). Neurobiology of arachnids (pp. 298-316). New York: Springer Verlag. Moynihan, M. & Rodaniche, A. F., (1982). The behavior and natural history of the Caribbean reef squid Sepioteuthis sepioidea. Advances in Ethology, 25, 1-150. Moynihan, M. (1985). Communication and non-communication by cephalopods. Bloomington: Indiana University Press. Mueller, M. & Wehner, R. (1988). Path integration in desert ants, Cataglyphis fortis. Proceedings of the National Academy of Sciences U.S.A., 85, 5287-5290. Nagel, T. (1974). What is like to be a bat? Philosophical Review, 83, 435-50. (Reprinted in T. Nagel, 1979. Mortal questions. London: Cambridge Univerisity Press). Neisser, U. (1988). Five kinds of self-knowledge. Philosophical Psychology, 1(1), 35-59. Nishida, T. (in press). In W. C. McGrew, L. Marchant, & T. Nishida (Eds.). Great apes societies. Cambridge: Cambridge University Press. Papi, F. (Ed.). (1992). Animal homing. London: Chapman & Hall. Parker, S. T., Mitchell, R. W., & Boccia, M. L. (Eds.). (1994). Self-awareness in animals and humans: Developmental perspectives. Cambridge: Cambridge University Press. Piaget, J. (1952). The origin of intelligence in children. New York: International Universities Press. Povinelli, D. J., Nelson, K. E., and Boysen, S. T. (1992). Comprehension of social role reversal by chimpanzees: Evidence of empathy? Animal Behaviour, 43, 633640. Premack, D., & Woodruff, G. (1978). Does the chimpanzee have a theory of mind? Behavioral and Brain Sciences, 1,515-526. Preyer, W. (1887). L'~xne de l'enfant. Paris: Alcan. Ristau, C.C. (Ed.). (1991). Cognitive ethology: The minds of other animals. Hillsdale, NJ: Lawrence Erlbaum Associates. Smuts, B. (1985). Sex and friendship in baboons. New York, New York: Aldine. Stem, D. N. (1985). The interpersonal world of the infant. New York: Basic Books. Stern, D. N. (1993). The role of feelings for an interpersonal self. In U. Neisser (Ed.). The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 204-215). Cambridge, MA: Cambridge University Press. Tomasello, M., Kruger, A., and Ratner, H. (1993). Cultural learning. Behavioral and Brain Sciences, 16, 495-552.

THE ANIMALSELFAS REFERENCEPOINT 215 Trevarthen, C. (1980). The foundations of intersubjectivity: Developmental of interpersonal and cooperative understanding in infants. In D. R. Olson (Ed.), The social foundations of language and thought: Essay in honor of Jerome Bruner. New York: Norton. Trevarthen, C. (1992). The functions of emotions in early infants communication and development. In J Nadel & L. Camaioni (Eds.), New perspectives in early communicative development. London: Routledge. Trevarthen, C. (1993). The self born in intersubjectivity: The psychology of an infant communicating. In U. Neisser (Ed.). The perceived self" Ecological and interpersonal sources of self-knowledge (pp.121-173). Cambridge: Cambridge University Press. Visalberghi, E., and Fragaszy, D. M. (1990). Do monkeys ape? In Parker, S., & Gibson, K. (Eds.). "Language" and intelligence in monkeys and apes: Comparative developmental perspectives, (pp. 247-273). Cambridge: Cambridge University Press. von Uexktill, J. (1957). A stroll through the worlds of animals and men. In C. H. Shiller (Ed. and Trans.), Instinctive behavior: The development of a modern concept (pp. 5-80). New York: International Universities Press. (Original work titled "Streifztige durch die Umwelten von Teiren und Menscen" published 1934) Wehner, R., Harkness, R. D. & Schmid-Hempel, P. (1983). Foraging strategies in individually searching ants. Stuttgart: G. Fischer Verlag. Wells-Gosling, N. (1985). Flying squirrels. Smithsonian Institution Press. Whiten, A. (1991). Natural theories of mind; Evolution, development and simulation of everyday mindreading. Oxford: Blackwell. Whiten, A., & Byrne, R. W. (1988). Tactical deception in primates. Behavioral and Brain Sciences, 11, 233-273.

This Page Intentionally Left Blank

PART II

Research

This second part brings together recent experimental research on self-knowledge at the origins of development. It is divided into three sections, according to particular research themes: The Self ReveaIed in Posture and Action (Schmuckler; Jorgensen, Suomi, & Hopkins; Van der Meet & Van der Weel; Jouen & Gapenne; Bertenthal & Rose); Perceptual Origins of the Self (Bigelow; Bahrick; Watson; Rochat & Morgan); and Social Origins of the Self (Stem; Reed; Tomasello). Although the common feature of this collection of chapters is their particular emphasis on empirical research, they are not limited to mere reviews of experimental facts. Most of the chapters provide additional theoretical views on the problem of the early sense of self. These theoretical views are grounded in recent empirical findings, within new experimental perspectives, and using novel research paradigms.

SECTION 1

The Self Revealed in Posture and Action

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Roehat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

221

CHAPTER 11

Self-knowledge of Body Position: Integration of Perceptual and Action System Information MARK A. SCHMUCKLER

University of Toronto, Scarborough College

There has been a recurrent fascination in psychology with describing and understanding the growth of self-knowledge in children. In general, this interest in the self has been widespread, covering a diverse array of ideas. For example, in her review of developmental perspectives on the self, Hatter (1983) provides a partial list of the aspects of the self that have been studied by psychologists. In her words: [W]e encounter self-recognition, self-concept, self-image, self-theory, selfesteem, self-control, self-regulation, self-monitoring, self-evaluation, selfcriticism, self-reward, self-perception, self-schematas, self-referent thought, selfconsciousness, self-awareness, and self-actualization, to name the most prevalent exemplars (Harter, 1983, p. 276). Related to such widespread interest, numerous theoretical proposals have been offered for understanding both the functioning of self as well as its development (see Harter, 1983, for a review). One recent theoretical formulation offered by Neisser (1988, 1991, 1993) undertakes a cognitive analysis of the self and its development by focusing on the different forms of information available to people. In Neisser's view, "... people have access to five basically different kinds of information about themselves. Each kind specifies a different aspect of the individual and thus implicitly defines a different sort of self' (Neisser, 1993, p. 3). The first of these selves, the "ecological self," concerns the relationship between the physical self and the physical environment, focusing on the individual as an active, causal agent in the world. According to Neisser, the information for this aspect of self is perceptual, with one directly perceiving one's self as acting within

222

MARK A. SCHMUCKLER

the world. Because the ecological self is based on perceptual information, this form of self-knowledge is evident early in infancy, potentially from birth. A second form of self, and one also based on perceptual information, is the "interpersonal self," which involves the individual considered from the point of view of engaging in social interaction with others in the world. Again, because the interpersonal self is based on perception, such self-knowledge appears during infancy. Neisser's other forms of the self, the "conceptual," "temporally extended," and "private self," all involve the understanding of different components of the self, covering aspects such as one's beliefs and assumptions about one's self; one's life narratives and memories of the self; and one's thoughts, images, and feelings. These forms of self are based on a variety of types of information and arise in development at varying ages coincident with the child's ability to think, remember, and reason about the self. Neisser's characterization of the self is coherent and wide ranging, and also provides a framework for examining and understanding the self. Although there exists developmental data pertinent to each of these five selves, this chapter will focus on the first: the ecological self. The most intriguing aspect of the ecological self is its focus on an individual's knowledge of his or her physical body in relation to the external world. As such, a prime example of the ecological self might involve self-knowledge of one's body position in space, along with how one's body subsequently moves through space. Viewed in this way, one pertinent research question is whether young infants show evidence of self-knowledge of body position and movement. A second point, and one related to the first, is the recognition that such selfknowledge is given through directly available perceptual information. Building on the theoretical ideas of James Gibson (1966, 1979), Neisser assumes that knowledge of the world begins with perception, with our perceptual systems sensitive to information invariantly specifying the objects and events within the world. Moreover, not only are young infants sensitive to such information, but they play an active, exploratory behavior role in uncovering new information (E. J. Gibson, 1987). This focus raises obvious questions about the general nature of the perceptual input specifying our perception of the world, and more specific to the current context, the information underlying one's self-knowledge of one's body position. For Neisser (1991), this information is primarily visual, given that proprioceptive and/or kinesthetic sources (while providing information about movement) vary tremendously with changes in body position and must+ accordingly be continuously scaled or calibrated to the visual input. It is important to realize, however, that such knowledge is potentially available via all of these sources of information.

SELF-KNOWLEDGE OF BODY POSITION

223

If knowledge of body position and movement can arise through multiple input systems, then it becomes critical to explain how these systems not only provide such information, but are also combined and/or integrated by perceivers. Correspondingly, the question now shifts from the ways in which self-knowledge of body position and movement is specified visually to how such self-knowledge is specified through visual and action systems. Understanding the coordination of visual and action systems in providing knowledge about body position is the focus of the work described in this chapter. The question of whether infants have self-knowledge of their body position, and if so, what the nature of the information is for such knowledge, can be examined across a range of scales or levels. At one level, one can ask about knowledge of limb position: Do infants know where their limbs are in space and how they are moving? This question can be explored within the context of intermodal perception in infancy. At a different level of analysis, one can ask whether infants have knowledge of their general body posture. Essentially, how is one's body oriented with respect to the ground plane? These issues are studied within the context of research on postural control. Another level of analysis involves knowledge of one's dynamic body position: In this case the question is, what do infants know about moving and guiding themselves through the world? These issues can be explored within the context of work on visually guided locomotion. Finally, one can ask about infants' knowledge of their global body position relative to the external environment. Where are we in space, in relation to important objects and/or landmarks? Such issues fall within the province of research on spatial orientation. A common theme of these questions is whether infants exhibit these various forms of self-knowledge of body position and movement, and how such knowledge might be due to integrating visual information with proprioceptive and kinesthetic system information.

Self-knowledge of Limb Position and Movement The first context in which to examine self-knowledge of body position involves recognition of limb position and movement: This question can be interestingly explored within the realm of intermodal perception in infancy. Generally, there exists strong evidence that young infants coordinate information arising from different perceptual systems. Numerous studies suggest that by the age of 5 months, infants recognize object properties such as shape, substance, and texture on the basis of visual and haptic system information (see Bushnell & Boudreau, 1991, 1993; Rose & Ruff, 1987; Spelke, 1987, for reviews). Is there any corresponding evidence that infants use intermodal information for recognizing

224

MARK A. SCHMUCKLER

their own limb position and movements? This form of intermodal perception requires integrating visual information with kinesthetic and proprioceptive inputs, which results in knowledge of limb movement and position. Two experimental results have demonstrated that infants, in fact, do evidence such intermodal perception (Bahrick & Watson, 1985; Rochat & Morgan, 1995). In a series of studies, Bahrick and Watson (1985) examined visual-proprioceptive intermodal perception of the leg movements of 3- and 5-month-old infants. In this work, infants kicked their legs while sitting in an infant seat such kicking provided proprioceptive information for leg movement. While kicking their legs, infants participated in a preferential looking task with two video monitors containing different visual images. On the first monitor, infants saw a live, online version of their own leg movements. Because the visual movement on this display was correlated with their own movement, it was referred to as the "contingent" display. On the second monitor, infants saw a videotape of a different child in the same situation (or, in one of the experiments, a previously recorded videotape of their own legs). Because the movement in this display was unrelated to their own movement, it was called the "noncontingent" displa(. If infants discriminate between these displays, they should show preferential fixation to one of the monitors. According to Bahrick and Watson (1985), the most likely basis for this discrimination would be the detection of the contingency between the proprioceptive information for movement and the visual movement occurring on one of the monitors. The results of a series of studies convincingly demonstrated intermodal perception, with the 5-month-old (but not 3-month-old) infants preferentially fixating one of the displays. Somewhat counterintuitively, infants in these studies did not prefer an intermodal match, but instead preferred to fixate the noncontingent display. Although various explanations for such preferential fixation can be offered, this work does demonstrate that 5-month-old infants detect their own leg movements on the basis of visual and proprioceptive information. Rochat and Morgan (1995) have extended these findings by examining the nature of the visual information necessary for performing such visualproprioceptive intermodal recognition. Similar to Bahrick and Watson (1985), the preferential fixation of 3- and 5-month-old infants to two on-line video images of their moving legs was observed. In one study, left-right spatial directionality was reversed on one of the video monitors, thereby producing a mismatch between the perceived proprioceptive direction of movement and the seen visual direction of movement; on the other, left-fight spatial directionality was retained. Infants in this situation looked longer and kicked more while attending to the left-right reversed monitor, which suggests that this display was perceived as spatially noncongruent.

SELF-KNOWLEDGE OF BODY POSITION

225

A recent series of experiments conducted in my laboratory (Schmuckler, 1994) provides strong convergent evidence for the findings of Bahrick and Watson (1985) and Rochat and Morgan (1995). In this work, visual-proprioceptive perception of ann and hand movements (rather than leg movements) was examined, with the focus again on exploring the visual information necessary for intermodal perception. In this work, 5-month-old infants performed hidden arm and hand movements (as opposed to the leg movements of Rochat & Morgan, 1995) while simultaneously viewing a contingent display (an on-line image of their own hand) and a noncontingent display (a previously recorded videotape of a different child in the same situation). The logic of these experiments was identical to that of Bahrick and Watson (1985) and Rochat and Morgan (1995): If infants perceive their own limb movementS, they should preferentially fixate one of the two displays. Three experiments tested this hypothesis. An initial study was designed as a simple replication of the previous experiments, to ensure that infants have visualproprioceptive intermodal detection of arm and hand movements. The second experiment extended these results by again exploring the importance of spatial directionality. Similar to Rochat and Morgan (1995), the left-right dimension of the video image was reversed for the contingent display, thereby producing a situation in which a physical arm and hand movement in one direction resulted in an image in which the hand seemed to move in an opposite direction. A third experiment investigated the importance of the point of observation of the hidden limb by providing a relatively novel view of this limb. In this study, the camera focused on the child's hand was positioned on the floor, facing upwards, producing an image displaying the palm of the hand with the fingers pointed downwards and the wrist and ann of the hand at the top of the screen. Such an image is novel in that it does not correspond to the image of the hand naturally seen from one's own view (an "egocentric" view) or from that of another individual looking at someone (an "observer" view). Figure 1 presents the results of these three experiments, shown in terms of infants' percent looking times toward the contingent and noncontingent displays. Replicating and extending Bahrick and Watson (1985) and Rochat and Morgan (1995), Experiment 1 demonstrated 5-month-olds' significant preferential fixation toward the noncontingent display, thereby suggesting visual-proprioceptive intermodal discrimination. In contrast, infants in Experiment 2 did not preferentially fixate either the contingent or noncontingent display, which again demonstrates that left-right spatial reversal disrupts intermodal perception. Experiment 3 once again produced significant preferential fixation of the noncontingent display, implying that the point of observation for the hidden limb is relatively unimportant, with infants able to detect their own limb movements despite seeing this limb from a novel orientation. Interestingly, the interpretation

226

MARK A. SCHMUCKLER

of these results converges with those of Rochat and Morgan (1995), despite differences in the experimental setup of these studies, as well as the specific pauem of results of this work. As such, these studies provide a nice example of the principle of converging operations using discrimination measures (Garner, Hake, & Erikson, 1956; Proffitt & Bertenthal, 1990).

FIGURE 1. The mean proportion of looking time, and standard error, toward the contingent and noncontingent visual displays. Equivalent looking (50%) toward the displays is notated with a dotted line. Together, these results provide compelling evidence that 5-month-old infants coordinate visual and proprioceptive inputs for detecting their own limb movements. Such a finding strongly implies perception of own limb movements, a finding in keeping with Neisser's (1988, 1991, 1993) characterization of the ecological self. The next series of experiments explores evidence for infants' selfknowledge of body posture and orientation.

SELF-KNOWLEDGE OF BODY POSITION

227

Self-knowledge of Body Posture and Upright Orientation A second context for examining self-knowledge of body position involves recognition of body posture and/or upright orientation on the basis of visual and action system information. What is the information used for maintaining body posture? Obviously, one knows about body posture through vestibular, kinesthetic, and proprioceptive inputs (e.g., Nashner & McCoUum, 1985). Less obvious, however, is that visual information is critical for the control of stance and posture. One of the most compelling demonstrations of this effect was provided by David Lee and colleagues (Lee & Aronson, 1974; Lee & Lishman, 1975; Lishman & Lee, 1973), in their experiments using the "moving room." A moving room, which is an enclosure consisting of three walls and a ceiling that moves back and forth atop a stationary floor, simulates the optical information arising from a loss of posture. For an observer standing within a moving room, movement of the walls produces a perceived loss of stability in the opposite direction, resulting in compensatory postural sway in the same direction as room movement. Intriguingly, these compensatory responses occur despite the fact that both vestibular and kinesthetic inputs indicate postural stability. The moving room has also been used to examine visual control of posture in developmental contexts. Lee and Aronson (1974), for example, found that newly standing infants were dramatically influenced by such visual information, with wall movements producing postural instabilities such as sways, staggers, and falls; similar results have been found in other studies examining both static (Bertenthal & Bai, 1989; Stoffregen, Schmuckler, & Gibson, 1987) and dynamic (Schmuckler & Gibson, 1989; Stoffregen, Schmuckler, & Gibson, 1987) postures. Other researchers have provided evidence that these postural reactions are present early in infancy (Bertenthal & Bai, 1989; Butterworth & Hicks, 1977; Gapenne & Jouen, 1994; Jouen, 1984), appearing coincident with the onset of crawling (Bertenthal & Bai, 1989; Higgins, 1992; Higgins, Campos, & Kermoian, 1993; but see Jouen, 1984, for conflicting results). These moving room results support the idea that infants and toddlers have knowledge of body posture and uptight orientation on the basis of perceptual and action system information. Interestingly, although it is clear that even young infants use visual information for postural control, there are also dramatic developmental changes in this visual control of posture. One area that has been examined involves postural responses to moving visual information as a function of the frequency and/or amplitude of that movement (Brandt, Dichgans, & Koenig, 1973; Lestienne, Soechting, & Berthoz, 1977; Stoffregen, 1986; van Asten, Gielen, & van der Gon, 1988). For example, van Asten et al. (1988) observed that adults' postural compensation to visual rotations around the line of sight fell off

228

MARK A. SCHMUCKLER

for rotations above the frequency of about 0.3 Hz; accordingly, it appears that adults do not use relatively high-frequency visual information for balance control. In contrast, there exists compelling evidence that young infants and toddlers do use high-frequency visual information for maintaining equilibrium (Bai, 1991; Delorme, Frigon, & Lagace, 1989). Bai (1991), for example, examined the postural responses of seated 5-, 9-, and 13-month-old infants to visual oscillations occurring at 0.3 and 0.6 Hz. In this work, the 9- and 13-month-old infants responded to both speeds of movement, although there was no systematic postural response for the 5-month-old, prelocomotor infants. In the same vein, Delonne et al. (1989) found appropriate postural compensations to visual oscillations of 0.52 Hz for standing children. Similar developmental differences between infants/young children and adults have been found for the amplitude (i.e., gain) and timing (e.g., latency to respond) components of postural sway. I am currently exploring (Schmuckler, 1995a) such developmental differences in the visual control of posture by focusing on children between the ages of 3 and 6 years. This age range is of interest because it conceivably represents a transitional period in the adoption of adultlike postural control (Ashmead & McCarty, 1991; Shumway-Cooke & Woollacott, 1984). Although the details of this work are outside of the purview of this chapter (because of the somewhat older subjects involved), its findings are intriguing. A series of studies have replicated the results of Bai (1991) and Delorme et al. (1989) in finding that children use high-frequency visual information for postural control; accordingly, children react in a nonadultlike fashion in terms of the frequency of their postural response. In contrast to these findings, children responded in an adultlike manner in the timing of their postural reactions, relative to the visual movement. Finally, the amplitude of children's postural responses demonstrated a mixture of adult- and nonadultlike responding. Together, these findings suggest that the postural sway of 3- to 6year-old children is characterized by both adultlike and nonadultlike responding; speculatively, this mixed developmental profile might represent a transitional state in postural control. More generally, these findings are revealing in light of the question of infant's and children's self-knowledge of body posture, supporting the idea that children know about their body's uptight orientation on the basis of perceptual and action system information.

Self-knowledge of Body Movement and Action Capabilities The previous section suggested that infants and children have self-knowledge of their body's uptight orientation in static postures. It is also possible to examine self-knowledge of body movements in dynamic situations by looking at toddlers'

SELF-KNOWLEDGE OF BODY POSITION

229

knowledge of their own large-scale body movements through the environment. One domain in which such questions can be explored is that of visually guided locomotion. Visually guided locomotion, which refers to the ability to guide ones' self through the world while avoiding obstacles and moving through openings, has been suggested as a fundamental skill involved in the growth of independent mobility, and requires sophisticated integration of visual information with motor behavior (Gibson & Schmuckler, 1989). Visually guided locomotion has been examined both developmentally and with adult subjects. Work with adults has focused on delineating the optical information important for the control of locomotion (J. J. Gibson, 1958; Lee, 1974; Lee & Lishman, 1977) or with examining the effects of visual guidance on kinematic parameters of gait (Patla, 1989; Patla, Prentice, Robinson, & Neufield, 1991; Patla, Robinson, Samways, & Armstrong, 1989; Warren, Young, & Lee, 1986). In contrast, developmental work has explored the growth of skills necessary for visually guided locomotion, such as spatial orientation (e.g., Cornell, Heth, & Broda, 1989; Herman, 1980; Rieser, Doxsey, McCarrell, & Brooks, 1982), the development of detour behavior (Heth & Cornell, 1980; Lockman, 1984; Mckenzie & Bigelow, 1986), the perception of surface properties relevant for independent locomotion (Gibson et al., 1987), and so on. Some research with young toddlers has focused specifically on the impact of visually guided locomotion on kinematic parameters of gait (Palmer, 1987, 1989; Schmuckler, 1990, 1993a; Schmuckler & Gibson, 1987). Schmuckler and Gibson (1989), for example, examined the impact of guidance around obstacles on responses to optical flow information imposed by a moving room. Examining children between 1 and 3 years of age, divided into three groups based on their locomotor experience (novice, intermediate, expert), they found increased postural perturbation in response to movement of the room when children performed a route-finding situation (walking around obstacles), relative to when no obstacles were present in the child's path. In contrast, there were no differences in postural perturbations as a function of the presence versus absence of obstacles when children stood still within the room, which indicates that the increased response in the presence of obstacles was specific to route finding. Subsequent analyses of the videotapes explored gait perturbations (noticeable accelerations and decelerations in walking speed) as a function of visual guidance conditions. Interestingly, and in contrast to analyses of postural instabilities, gait perturbations produced in response to room movements decreased in the presence of obstacles, relative to when no obstacles were present, with this difference disappearing for the oldest, most experienced group of walkers. Thus, visual guidance produced a more rigidly controlled locomotor style at younger ages. This finding is in keeping with the idea that novice performers of a motor act try to

230

MARK A. SCHMUCKLER

reduce the degrees of freedom relating to that behavior (Bernstein, 1967; Newell & Corcos, 1993; Woollacott & Sveistrup, 1994). With increasing skill and experience, however, these actors release this stiffness, which results in less constrained movements. In a similar vein, Schmuckler (1990, 1993a) explored the impact on kinematics of gait of locomoting through environments requiring varying degrees of visual guidance. In this work, 14-month-old toddlers locomoted along a path free of obstacles (the "free locomotion" condition), a narrow path requiring regulation of gait but no route-finding (the "controlled locomotion" condition), and a path requiring guidance around obstacles (the "guided locomotion" condition). Analyses of gait kinematics revealed that visual guidance conditions significantly impacted on the child's step size (but not on walking speed), with step size systematically decreasing as the environment required higher degrees of visual guidance. Again, this finding is in keeping with the idea that visually guided locomotion produces a more rigid gait, with this rigidity expressed as a more conservative step size. A final series of experiments (Schmuckler, 1993b, 1995b) examined visually guided locomotion using a somewhat different route-finding situation m children's navigation over barriers. Although the details of this work are, again, beyond the scope of this chapter, it will be briefly described. The primary goal of this research was to examine whether visually guided locomotion was related in some way to children's intrinsic knowledge of their own locomotor abilities, or to action capabilities, or was scaled to the child's body size. For example, recent research with older children and adults suggests that action in the world is often "bodyscaled," that is, related to physical body dimensions (Burton, 1992; Carello, Grosofsky, Reichel, Solomon, & Turvey, 1989; Konczak, Meeuwsen, & Cross, 1992; Mark, 1987; Mark, Baillet, Craver, Douglas, & Fox, 1990; Mark & Vogele, 1987; Pufall & Dunbar, 1992; Warren, 1984; Warren & Whang, 1987). One project examining such questions (Schmuckler, 1993b) explored 12-, 18-, and 24-month-old toddler's abilities to successfully cross over a barrier to reach their parent, seated on the other side. Crossing behavior to barriers of various heights was coded using a scheme by Adolph (in press), in which crossing was categorized as a "success" (crossing the barrier without disturbing it), a "failure" (attempting to cross the barrier but ultimately knocking it over), or a "refusal" (not attempting to cross the barrier). The results of this study revealed that at all ages, children successfully crossed the barriers at low heights, with the mean number of failures increasing as the barrier height increased. Interestingly, as the barrier continued to rise in height, the number of failures in barrier crossing actually decreased, accompanied by a corresponding increase in the number of refusals shown by children. This finding that children will refuse to even attempt to cross

SELF-KNOWLEDGE OF BODY POSITION

231

barriers at certain heights is very much in keeping with the idea that toddlers have knowledge of their body capabilities, with their subsequent action reflecting this self-knowledge. Subsequent analyses examined whether the ability to successfully cross over barriers was related in some fashion to the child's actual capabilities for action in the world. Toward this end, thresholds for barrier crossing were calculated as the barrier height at which successfully crossing occurred 50% of the time. Not surprisingly, mean crossing thresholds increased with age. These thresholds were then investigated with reference to the factor(s) that might account for variation in these thresholds between the different age groups. Toward this end, crossing thresholds were normalized by dividing each child's threshold by either body dimensions or locomotor skill measures. If thresholds are related to these dimensions, then this normalization should remove the differences between the ages. The different normalizing factors investigated included the child's standing height, sitting height, leg length, body weight, walking experience (as given by parental report), and age. Intriguingly, the only factor that eliminated the differences in crossing thresholds between the age groups was each child's walking experience. Generally, the studies on visually guided locomotion support the idea that toddlers have self-knowledge of both their body movement and their action capabilities. The pattern of success, failure, and refusal rates suggest that children had knowledge of their locomotor capabilities, in that they did not attempt to cross barriers at heights that could not yet be successfully negotiated. In addition, crossing thresholds were related to children's overall locomotor skill, as indexed by walking experience. This result is especially interesting in light of the current evidence suggesting that for adults and older children, the successful accomplishment of a number of motor activities (e.g., stair climbing, sitting, reaching, and gap crossing) are related to body dimensions (Burton, 1992; Carello et al., 1989; Konczak et al., 1992; Mark, 1987; Mark et al., 1990; Mark & Vogele, 1987; Pufall & Dunbar, 1992; Warren, 1984; Warren & Whang, 1987). In contrast, research with infants and toddlers has found such relationships with skill measures, not body dimensions. Adolph (in press), for example, found that toddler's abilities to successfully climb up and down slopes was related primarily to walking skill.

Self-knowledge of Body Orientation in Space The final area to be described concerns knowledge of body position on a more global, environmental scale. In this case, the primary issue involves examining

232

MARK A. SCHMUCKLER

infants' and toddlers' spatial orientation abilities, or self-knowledge of body position in space, relative to environmental landmarks. As in the previous sections, one of the focal concerns here involves understanding how perceptual and action systems contribute to knowledge of where one is in the world, as well as how such knowledge is modified or updated subsequent to some form of movement through the ~nvironment. For example, when an observer moves through the world, there exist multiple sources of information for how the world changes that accompany this movement. Assuming the environment is lit, visual input provides information for the changing spatial relationships in the world. Along with this visual information is information arising from the movement of one's body w kinesthetic, proprioceptive, and vestibular inputs that specifies movement through space. Rieser and colleagues (Rider & Rieser, 1988; Rieser, 1979, 1983; Rieser, Guth, & Hill, 1986; Rieser & Heiman, 1982; Rieser & Rider, 1991) have elegantly demonstrated that both older children and adults use such "body movement" information for spatial orientation. In contrast to the information accompanying self-movement, when an object moves relative to one's self, the only information specifying such movement is visual. Accordingly, movement of objects in the world produces only a single source of information for a change in spatial relationships. How do these differences in the available information for spatial orientation affect infants' and toddlers' knowledge of the spatial layout? Bremner (Bremner, 1978a, 1978b; Bremner & Bryant, 1977) explored this question by examining 9month-old infants' abilities to find a hidden object (a modified stage IV search task), following either infant self-movement through the world or following object movement through the world. In keeping with the above analysis on the informational sources available for spatial updating, infants successfully retrieved the toy more often subsequent to self-movement, relative to object movement, through the world. This result has been replicated by other researchers examining 6- to 7-month-old infants (Bai & Bertenthal, 1992) and suggests that the availability of multiple cues for spatial change leads to better spatial updating than the presence of only a single source of information for such change. One way in which these findings are limited is that they do not provide a thorough exploration of the relative importance of visual and body movement information for spatial orientation. A pair of experiments recently completed in my laboratory attempted to evaluate the impact of these two input sources on spatial updating. In these studies, infants between the ages of 9.5 and 18 months were seated in front of a table and saw a toy hidden in one of two locations (different colored cups) on the table. Prior to being allowed to search and retrieve this toy, infants experienced one of two displacements: Either they were moved 180 ~ around the table (the "infant displacement" condition), or the table was

SELF-KNOWLEDGE OF BODY POSITION

233

moved 180 ~ around them (the "table displacement" condition). Accordingly, infant versus table displacements manipulated the presence versus absence of body movement information. Visual information was manipulated by performing infant and table displacements with the lights on or off. Using these manipulations enabled us to systematically examine the impact of visual and body movement information on infant spatial orientation and to assess infants' abilities to integrate these inputs during spatial updating. These two experiments were distinctive in that they provided different cues for successful performance. In the first experiment, the hidden toy always ended up on the same side of the table (relative to the infant) throughout the experiment, with the visual characteristics of this hiding position (i.e., the color of the cup) varying randomly from trial to trial. In the second experiment, the location of the hidden toy was signaled visually, with the color of the correct location constant throughout the experiment. However, the actual side on which the hidden toy was located varied throughout the study. Thus, in the first study, infants needed only learn a particular motor sequence to retrieve the hidden toy (e.g., search in the cup to their fight), whereas in the second study, infants needed only learn a color association (e.g., search in the red cup). Both experiments also explored search performance developmentally, with the first study testing 9.5-, 14-, and 18-monthold infants, and the second study employing 9.5- and 16-month-old infants. Examining search performance with these different age groups was of interest, given previous results (e.g., Acredolo, 1978, 1979; Acredolo & Evans, 1980) that suggested significant developmental change in spatial orientation abilities across this age range. The results of these experiments are presented in Figure 2, which graphs the mean number of times infants successfully retrieved the toy (out of four possible trials) as a function of body movement information (infant versus table displacement) and visual information (lit versus dark environment) conditions. Because there were no significant age effects in either study, these data are shown averaged across age. The most interesting result was the significant interaction between visual and body movement information conditions. In both studies, search performance was best when both visual and body movement sources provided information for spatial updating. In contrast, search in the remaining three conditions led to equivalent, or worse, performance. Intriguingly, cuing the correct location visually (Experiment 2) led to somewhat better search performance than cuing the location with a motor response (Experiment 1), although both studies produced the same pattern of results. Together, these findings suggest that visual and body movement information interact in spatial updating, with visual information possibly playing a more important role than body movement information.

234

MARK A. SCHMUCKLER

FIGURE 2. The mean number of successful searches (out of 4), as a function of body movement and visual conditions. The top panel shows search performance when the location of the hidden toy was signaled using a motor cue, and the bottom panel shows search performance when the location of the hidden toy was signaled using a visual cue. Chance performance is notated with a horizontal line.

More generally, the results of these studies are in keeping with the idea that infants have self-knowledge of their global body position in space. This ability to update their spatial orientation, while limited, is present at a fairly young age (9.5 months), although it appears to undergo little developmental change (at least as measured by the current paradigm) across the age range examined.

SELF-KNOWLEDGE OF BODY POSITION

235

Summary and Condusions A variety of different experimental paradigms and results have supported the idea that infants perceive their own body position in space. This knowledge has been seen to exist at a variety of "levels," from knowledge of individual limb movements to knowledge of one's spatial orientation within the larger environment. Accordingly, this work provides evidence in support of Neisser's (1988, 1991, 1993) notion of the ecological self, or knowledge of the physical self in relation to the physical environment. Because this information for self arises through directly available perceptual information, the ecological self is presumably present early in life. The studies described in this chapter support this hypothesis, suggesting that the ecological self might be a very early component or form of the growing child's self-concept. In keeping with Neisser's emphasis on the importance of perceptual information, the work described here explored the information giving rise to this self-knowledge. One inescapable conclusion drawn from this work is that there are multiple sources of information that can be used for knowing about one's body position and movement in space. This chapter has concentrated on the use and integration of visual and action system information, and has found that even young infants integrate such information to inform themselves about their body's orientation. Thus, although Neisser has focused on the importance of visual information for the ecological self, it seems clear that the ecological self is a joint product of a multitude of information sources. A number of issues and questions have been studiously avoided in this chapter. One conspicuously absent concern is the issue of whether the types of self-knowledge exhibited by infants in these studies represent implicit or explicit awareness. Unfortunately, given the obvious limitations of dealing with infant subjects, it is difficult to imagine how one would gain direct experimental insight into this question. Hence, the question of whether this self-knowledge is implicitly or explicitly represented remains unresolved. Speculatively, one possibility is that the explicitness of the different types of self-knowledge discussed in this chapter change as this knowledge becomes more global. For example, visualproprioceptive intermodal perception of limb movements seems, intuitively, to be an unlikely candidate for explicit self-knowledge, particularly with 5-month-old subjects. In contrast, the research on visually guided locomotion and spatial orientation intuitively seems to convey more explicit aspects of self-knowledge of body position and capabilities. Although one obvious confounding factor is the general age difference of the participants in these situations, the idea that explicit self-knowledge might increase with increasing scales of body movement and position is an intriguing possibility.

236

MARK A. SCHMUCKLER

Finally, although it seems clear that infants make use of different forms of information for self-knowledge of body position, the use and integration of such information must undergo developmental change. For example, the research briefly described on the visual control of posture outlined a fascinating developmental progression in the use of certain types of visual information for balance control, and highliglited the fact that infants and young children make use of visual information to which adults appear insensitive. Similarly, the work on spatial orientation implies a significant developmental change in spatial updating, although this suggestion arises through the notable failure to find developmental differences across a wide age range. One of the more general goals of future research is to identify the ways in which the use and integration of information sources change as a function of age, as well as identifying the factors underlying such changes. In sum, the work described in this chapter has provided evidence that infants, at a very young age, express some self-knowledge pertaining to their body position and movement, and their capabilities for action. More globally, this work has attempted to understand how children perceive and act, and how perceiving and acting within the environment fits into the larger framework of the emergence of the child as an active, independent agent in the world. NOTES The research described in this report and the writing of this manuscript were supported by a grant from the Natural Sciences and Engineering Research Council of Canada. Correspondence concerning this article can be addressed to: Mark A. Schmuckler, Division of Life Sciences, University of Toronto, Scarborough College, Scarborough, Ontario, Canada, M1C 1A4, or by electronic mail at: [email protected]. REFERENCES

Acredolo, L. P. (1978). Development of spatial orientation in infancy. Developmental Psychology, 14, 224-234. Acredolo, L. P. (1979). Laboratory versus home: The effect of environment on the 9month-old infant's choice of spatial reference system. Developmental Psychology, 15, 666-667. Acredolo, L. P., & Evans, D. (1980). Developmental changes in the effects of landmarks on infant spatial behavior. Developmental Psychology, 16, 312-318. Adolph, K. E. (in press). A psychophysical assessment of toddlers' ability to cope with slopes. Journal of Experimental Psychology: Human Perception and Performance. Ashmead, D. H., & McCarty, M. E. (1991). Postural sway of human infants while standing in light and dark. Child Development, 62, 1276-1287. Asten, W. N. J. C. van, Gielen, C. C. A. M, and van der Gon, J J. D. (1988). Postural movements induced by rotations of visual scenes. Journal of the Optical Society of America, 5, 1781-1789.

SELF-KNOWLEDGEOF BODYPOSITION 237 Bahrick, L. E., & Watson, J. S. (1985). Detection of intermodal proprioceptive-visual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21, 963-973. Bai, D. L. (1991). Visual control of posture during infancy. Unpublished doctoral dissertation., University of Virginia. Bai, D. L., & Bertenthal, B. I. (1992). Locomotor status and the development of spatial search skills. Child Development, 63, 215-226. Bernstein, N. (1967). Coordination and regulation of movement. New York: Pergamon Press. Bertenthal, B. I., & Bai, D. L. (1989). Infants' sensitivity to optical flow for controlling posture. Developmental Psychology, 25, 936-945. Brandt, T., Dichgans, J., & Koenig, E. (1973). Differential effects of central versus peripheral vision on egocentric and exocentric motion perception. Experimental Brain Research, 23, 471-489. Bremner, J. G. (1978a). Spatial errors made by infants: Inadequate spatial cues or evidence of egocentrism? British Journal of Psychology, 69, 77-84. Bremner, J. G. (1978b). Egocentric versus allocentric spatial coding in 9-month-old infants: Factors influencing the choice of code. Developmental Psychology, 14, 346-355. Bremner, J. G., & Bryant, P. E. (1977). Place versus response as the basis of spatial errors made by young infants. Journal of Experimental Child Psychology, 23, 162-171. Burton, G. (1992). Nonvisual judgment of the crossability of path gaps. Journal of Experimental Psychology: Human Perception and Performance, 18, 698-713. Bushnell, E. W., & Boudreau, J. P. (1991). The development of haptic perception during infancy. In M. A. Heller & W. Schiff (Eds.), The psychology of touch (pp. 139-161). Hillsdale, NJ: Erlbaum. Bushnell, E. W., & Boudreau, J. P. (1993). Motor development and the mind: The potential role of motor abilities as a determinant of aspects of perceptual development. Child Development, 64, 1005-1021. Butterworth, G., & Hicks, L. (1977). Visual proprioception and postural stability in infancy: A developmental study. Perception, 6, 255-262. Carello, C., Grosofsky, A., Reichel, F. D., Solomon, Y., & Turvey, M T. (1989). Visually perceiving what is reachable. Ecological Psychology, 1, 27-54. Comell, E. C., Heth, D., & Broda, L. S. (1989). Children's wayfinding: Responses to instructions to use environmental landmarks. Developmental Psychology, 25, 755-764. Delorme, A., Frigon, J., & Lagace, C. (1989). Infants' reactions to visual movement of the environment. Perception, 18, 667-673. Gapenne, O., & Jouen, F. (1994, June). Temporalproperties of the visuo-postural coupling in newborns. Poster presented at the 9th International Conference on Infant Studies, Paris, France. Garner, W. R., Hake, H. W., & Erikson, C. W. (1956). Operationalism and the concept of perception. Psychological Review, 63, 149-159. Gibson, E. J. (1987). Introduction essay: What does infant perception tell us about theories of perception? Journal of Experimental Psychology: Human Perception and Performance, 13, 515-523. Gibson, E. J., Riccio, G., Schmuckler, M. A., Stoffregen, T. A., Rosenberg, D., & Taormina, J. (1987). Detection of the traversability of surfaces by crawling and walking infants. Journal of Experimental Psychology: Human Perception and Performance, 13, 533-544.

238

MARKA. SCHMUCKLER

Gibson, E. J., & Schmuckler, M. A. (1989). Going somewhere: An ecological and experimental approach to development of mobility. Ecological Psychology, 1, 3-25. Gibson, J. J. (1958). Visually controlled locomotion and visual orientation in animals. British Journal of Psychology, 49, 182-194. Gibson, J. J. (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin. Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. Harter, S. (1983). Developmental perspectives on the self-system. In P. H. Mussen (Ed.), Handbook of Child Psychology, Vol. IV: Socialization, personality, and social development (E. M. Hetherington, Vol. Ed.) (pp. 275-385). New York: John Wiley and Sons. Herman, J. F. (1980). Children's cognitive maps of large-scale spaces: Effects of Psychology, 29, 126-143. Heth, C. D., & Cornell, E. H. (1980). Three experiences affecting spatial discrimination learning by ambulatory children. Journal of Experimental Child Psychology, 30, 246-264. Higgins, C. (1992, May). The relation between self-produced locomotion and postural compensation to optic flow. Paper presented at the International Conference on Infant Studies, Miami Beach, FL. Higgins, C., Campos, J., & Kermoian, R. (1993, March). The influence of creeping on infant postural compensation to optic flow. Paper presented at the 1993 meetings of the Society for Research in Child Development, New Orleans, LA. Jouen, F. (1984). Visual-vestibular interactions. Infant Behavior and Development, 7, 135-145. Konczak, J., Meeuwsen, J. J., & Cross, M E. (1992). Changing affordances in stair climbing: The perception of maximum climbability. Journal of Experimental Psychology: Human Perception and Performance, 18, 691-697. Lee, D. N. (1974). Visual information during locomotion. In R. B. MacLeod and H. L. Pick (Eds.), Perception: Essays in honor of James J. Gibson (pp. 250-267). Ithaca, NY: Cornell University Press. Lee, D. N., & Aronson, E. (1974). Visual proprioceptive control of standing in human infants. Perception & Psychophysics, 15, 529-532. Lee, D. N., & Lishman, J. R. (1975). Visual proprioceptive control of stance. Journal of Human Movement Studies, 1, 87-95. Lestienne, F., Soechting, J., & Berthoz, A. (1977). Postural readjustments induced by linear motion of visual scenes. Experimental Brain Research, 28, 363-384. Lishman, J. R., & Lee, D. N. (1973). The autonomy of visual kinaesthesis. Perception, 2, 287-294. Lockman, J. J. (1984). The development of detour ability during infancy. Child Development, 55, 482-491. Mark. L. S. (1987). Eyeheight-scaled information about affordances: A study of sitting and stair climbing. Journal of Experimental Psychology: Human Perception and Performance, 10, 361-370. Mark, L S., Baillet, J. A., Craver, K. D., Douglas, S. D., & Fox, T. (1990). What an actor must do in order to perceive the affordance for sitting. Ecological Psychology, 2, 325-366. Mark, L. S., & Vogele, D. (1987). A biodynamic basis for perceived categories of action: A study of sitting and stair climbing. Journal of Motor Behavior, 19, 367394.

SELF-KNOWLEDGEOF BODY POSITION 239 McKenzie, B. E., & Bigelow, E. (1986). Detour behavior in young human infants. British Journal of Developmental Psychology, 4, 139-148. Nashner, L. M., & McCollum, G. (1985). The organization of human postural movements: A formal basis and experimental synthesis. The Behavioral and Brain Sciences, 8, 135-172. Neisser, U. (1988). Five kinds of self-knowledge. Philosophical Psychology, 1, 3559. Neisser, U. (1991). Two perceptually given aspects of the self and their development. Developmental Review, 11, 197-209. Neisser, U. (1993). The self perceived. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge. (p. 3-24). Cambridge, MA: Cambridge University Press. Newell, K., & Corcos, D. M. (1993). Issues in variability and motor control. In K. M. Newell and D. M. Corcos (Eds.), Variability and motor control (pp. 1-12). Champaign, IL: Human Kinetics. Palmer, C. F. (1987, April). Between a rock and a hard place: Babies in tight spaces. Paper presented at the Biennial Meetings of the Society for Research in Child Development, Baltimore, MD. Palmer, C. F. (1989, April). Max headroom: Toddlers tocomoting through doorways. Paper presented at the Biennial Meetings of the Society for Research in Child Development, Kansas City, MO. Patla, A. E. (1989). In search of laws for the visual control of locomotion: Some observations. Journal of Experimental Psychology: Human Perception and Performance, 15, 624-628. Patla, A. E., Prentice, S. D., Robinson, C., & Neufield, J. (1991). Visual control of locomotion: Strategies for changing direction and for going over obstacles.

Journal of Experimental Psychology: Human Perception and Performance, 17,

603-634. Patla, A. E., Robinson, C., Samways, M., & Armstrong, C. J. (1989). Visual control of step length during overground locomotion: Task-specific modulation of the locomotor synergy. Journal of Experimental Psychology: Human Perception and Performance, 15, 603-617. Proffitt, D. R., & Bertenthal, B. I. (1992). Converging operations revisited: Assessing what infants perceive using discrimination measures. Perception & Psychophysics, 47, 1-11. Pufall, P. B., & Dunbar, C. (1992). Perceiving whether or not the world affords stepping onto and over: A developmental study. Ecological Psychology, 4, 1738. Rider, E. A., & Rieser, J. J. (1988). Pointing at objects in other rooms: Young children's sensitivity to perspective after walking with and without vision. Child Development, 59, 480-494. Rieser, J. J. (1979). Spatial orientation of six-month-old infants. Child Development, 50, 1078-1087. Rieser, J. J. (1983). The generation and early development of spatial inferences. In H. L. Pick & L. C. Acredolo (Eds.), Spatial orientation in natural and experimental settings (pp. 39-71). New York: Plenum. Rieser, J. J., Doxsey, P. A., McCarrell, N. S., & Brooks, P. H. (1982). Wayfinding and toddlers' use of information from an aerial view of a maze. Developmental Psychology, 18, 714-720. Rieser, J. J., Guth, D. A., & Hill, E. (1986). Sensitivity to perspective structure while walking without vision. Perception, 15, 173-188.

240

MARKA. SCHMUCKLER

Rieser, J. J., & Heiman, M. L. (1982). Spatial self-reference systems and shortest-route behavior in toddlers. Child Development, 53, 524-533. Rieser, J. J., & Rider, E. A. (1991). Young children's spatial orientation with respect to multiple targets when walking without vision. Developmental Psychology, 27, 97-107. Rochat, P., & Morgan, R. (1995). Spatial determinants in the perception of selfproduced leg movements in 3- to 5-month-old infants. Developmental Psychology, 31 (4). Rose, S. A., & Ruff, H. A. (1987). Cross-modal abilities in human infants. In J. D. Osofsky (Ed.), Handbook of infant development, 2nd Ed., (pp. 318-362). New York: Wiley. Schmuckler, M. A. (1990). Issues in the development of postural control. In H. Bloch and B. I. Bertenthal (Eds.), Sensory-motor organizations and development in infancy and early childhood (pp. 231-236). Dordrecht: Kluwer Academic Publishers. Schmuckler, M. A. (1993a). Perception-action coupling in infancy. In G. J. P. Savelsbergh (Ed.), The development of coordination in infancy (pp. 137-173). Advances in Psychology Series. North-Holland: Elsevier. Schmuckler, M. A. (1993b, March). Knee's up Mother Brown: Toddlers' stepping over barriers. Poster presented at the 60th meetings of the Society for Research in Child Development, New Orleans, LA. Schmuckler, M. A. (1994, June). Infant's visual proprioceptive intermodal recognition. Poster presented at the 9th International Conference on Infant Studies, Paris, France. Schmuckler, M. A. (1995a, March). Children's postural sway in response to high and low frequency visual oscillation. Poster presented at the Biennial Meetings of the Society for Research in Child Development, Indianapolis, Indiana. Schmuckler, M. A. (1995b, March). The influence of spatial extent and transparency on toddIer's crossing of barriers. Poster presented at the Biennial Meetings of the Society for Research in Child Development, Indianapolis, Indiana. Schmuckler, M. A., & Gibson, E. J. (1989). The effect of imposed optical flow on guided locomotion in young walkers. British Journal of Developmental Psychology, 7, 193-206. Shumway-Cooke, A., & Woollacott, M. J. (1985). The growth of stability: Postural control from a developmental perspective. Journal of Motor Behavior, 17, 131147. Spelke, E. (1987). The development of intermodal perception. In P. Salapatek & L. Cohen (Eds.), Handbook of Infant Perception, Vol. 2: From perception to cognition (pp. 233-273), Orlando, FL: Academic Press. Stoffregen, T. A. (1986). The role of optical velocity in the control of stance. Perception & Psychophysics, 39, 355-360. Stoffregen, T. A., Schmuckler, M. A., & Gibson, E. J. (1987). Use of central and peripheral optical flow in stance and locomotion in young walkers. Perception, 16, 113-119. Warren, W. H. (1984). Perceiving affordances: Visual guidance in stair climbing.

Journal of Experimental Psychology: Human Perception and Performance, 10. 683-703. Warren, W. H., & Whang, S. (1987). Visual guidance of walking through apertures: Body-scaled information for affordances. Journal of Experimental Psychology: Human Perception and Performance, 13, 371-383.

SELF-KNOWLEDGEOF BODYPOSITION 241 Warren, W. H., Young, D. S., & Lee, D. N. (1986). Visual control of step length during running over irregular terrain. Journal of Experimental Psychology: Human Perception and Performance, 12, 259-266. Woollacott, M. H., & Sveistrup, H. (1994). The development of sensorimotor integration underlying posture control in infants during the transition to independent stance. In S. P. Swinnen, J. Massion, & H. Heuer (Eds.), Intertimb coordination: Neural, dynamic, and cognitive constraints. San Diego: Academic Press.

This Page Intentionally Left Blank

Tile Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

243

CHAPTER 12

Using a Computerized Testing System to Investigate the Preconceptual Self in Nonhuman Primates and Humans MATTHEW J. JORGENSEN and STEPHEN J. SUOMI

NIH Animal Center WILLIAM D. HOPKINS

Yerkes Regional Primate Research Center Emory University and Berry College

Since Gallup (1970) first developed his mark test of self-recognition, numerous studies have attempted to determine whether different species of nonhuman primates can recognize their reflection in a mirror. Yet despite 25 years of research, most studies of self-recognition in nonhuman primates have ignored questions concerning the antecedents of mirror self-recognition. In other words, comparative psychologists have rarely asked what cognitive components are necessary and sufficient for the evolution and development of self-recognition (Mitchell, 1993). This chapter attempts to address such questions by investigating a simpler form of self-perception that exists prior to the emergence of self-recognition. We will attempt to show that alternative methods are available that more accurately tap into this preconceptual self. After describing our investigations with nonhuman primates, we will also present preliminary investigations with human children and adults using similar methods.

Problems with the Mark Test Although the mark test clearly has been one of the most important tools used in the study of self-recognition, there are still problems with the method. For example, although chimpanzees as a species are considered to be self-aware, not all

244

M.J. JORGENSEN, S.J. SUOMI, & W.D. HOPKINS

chimpanzees are known to pass the mark test (Swartz & Evans, 1991; Povinelli, Rulf, Landau, & Bierschwale, 1993). For instance, Povinelli et al. (1993) reported that only 21 of 105 chimpanzee subjects (or 20% of the sample) passed the mark test as defined by these investigators. Such a low proportion of subjects passing the mark test clearly calls into question the sensitivity of this measure for selfrecognition. There are also inter-species inconsistencies. Most gorillas do not pass the mark test (Ledbetter & Basen, 1982; c.f. Patterson, 1990), yet they are more closely related to chimpanzees than are orangutans, which do pass the test (Lethmate & Ducker, 1973). Although there have been attempts to explain these inconsistencies (Povinelli, 1993), the questions are still unresolved. This leads to the second problem, that of interpretation. Do negative results indicate that subjects do not possess the ability to recognize themselves in a mirror? Or is the mirror test not an appropriate method for investigating self-recognition in certain nonhuman primates (or even certain individuals)? Although these problems have been addressed before (see Mitchell, 1993; Parker, Mitchell, & Boccia, 1994 for reviews), this chapter deals with a more fundamental problem. The mark test does not provide any insights into the cognitive components necessary for selfrecognition, nor does it test what precursors to self-recognition may exist prior to successful mark test performance. In short, we believe that experimental questions of the origins of self can be addressed by means other than the mark test.

Precursors to Self-recognition Parker (1991), for example, has suggested that human developmental models should be applied to studies with primates in order to more thoroughly understand the mechanisms underlying self-recognition. Specifically, she recommended using Lewis and Brooks-Gunn's (1979) ontogenetic model as a framework for understanding phylogenetic trends in nonhuman primates. As Lewis and BrooksGunn (1979) state, "visual self-recognition is only one aspect of self-recognition, one which in fact may be the last to develop" (p. 25). Prior to self-recognition, some sort of proprioceptive recognition may develop that is not detectable by the use of mirror tests. Other researchers have voiced similar concerns regarding the usefulness of the mark test at elucidating the foundations of self-recognition in both humans and nonhuman primates (e.g., Butterworth, 1992; Neisser, this volume; Rochat, this volume). As the existence of this book attests, there is a new wave of research that is more concerned with an aspect of the self that is fundamentally different from what traditional mirror studies have attempted to understand. As Neisser (1991) puts it: "In my view, it is unwise to think of the self as a single entity with a single

VIDEO-TASKS AND THE PRECONCEPTUAL SELF

245

course of development. There are many sources of self-knowledge, each giving rise to a different aspect of the self' (p. 197). Following Neisser's suggestions, we attempted to focus on an aspect of the self that was more grounded in perception than representation. In other words, we were interested in what visual and kinesthetic perceptual abilities may precede the development of self-recognition. As Butterworth (1992) noted, "self-conception (as revealed by [the mark test], for instance) is ontogenetically derived from processes not themselves revealed by mirror tasks" (p. 105). It was with this perspective on the self that we started our experiments with nonhuman primates.

Video-tasks The initial focus of our research was to develop procedures for investigating the antecedents of self-recognition in nonhuman primates. In particular, we were interested in applying recent developments in computerized testing procedures to the question of self-recognition. Rumbaugh, Richardson, Washburn, SavageRumbaugh, and Hopkins (1989) had developed a computerized testing system for the purpose of conducting controlled studies of primate cognition in rhesus monkeys (see Richardson, Washburn, Hopkins, Savage-Rumbaugh, & Rumbaugh, 1990 for technical details). Their animals readily learned to perform these tasks, and their success opened up opportunities for a variety of other research applications. The basic task involved manipulating a joystick to move a cursor (a two-dimensional image on a computer monitor) into various targets on the computer screen. For the initial training program (called the SIDE task), subjects learned to move the cursor into stationary targets located on the edge of the screen. The size of the targets became smaller as the subject's proficiency increased. Finally, once the SIDE task was mastered, subjects were required to move the cursor into a moving target (called the CHASE task). Rumbaugh et al. (1989) suggested that the ability of monkeys to solve this task was evidence of some form of self-recognition. Their argument was that the subject's ability to distinguish between the cursor they controlled and the target they did not control required capabilities similar to those required for most tests of self-recognition. They concluded that successful performance on this video-task required "all of the cognitive components needed for mirror recognition, unless indeed there is a metaphysical basis of a higher level of self-awareness for the phenomenon of mirror recognition" (p. 38). However, this conclusion seems untenable, given the fact that both monkeys and human infants can engage in such discriminations but fail to recognize themselves in standard mirror tests (Lewis & Brooks-Gunn, 1979; Anderson, 1984). Although Rumbaugh et al. (1989) claimed

246

M.J. JORGENSEN, S.J. SUOMI, • W.D. HOPKINS

that "it is only by discriminating that the relation between its hand movements and those of the cursor is different from the relationship between its hand movements and those of the target that the monkey can learn which of the moving images on the screen is under the control of its own hand" (p. 37), other explanations are perhaps more plausible. For example, animals could solve the task without any knowledge of what image they controlled. A strategy could be followed in which the animal moves the joystick in any direction that causes the two images to move closer together. No knowledge of the relationship between hand movements and cursor movements is needed to solve the problem because the cursor/target distance would be the only salient cue. Such results do, however, seem to suggest that the recognition of such a visual-kinesthetic contingency may be the basis of a simpler form of selfrecognition. For example, Bahrick and Watson (1985) claim that the ability to discriminate between contingent and noncontingent movement may be fundamental to the development of self-perception in human infants. Heyes (1994) has further suggested that video-tasks "required the animal to keep its eyes on the screen and to use the cursor position as a source of novel, displaced visual feedback on the position of its hand" (p. 916). She also argues that this novel, displaced visual feedback is an essential component to mirror self-recognition. In short, the relationship between self-awareness and video-tasks is not clear. Further investigations are needed to determine what knowledge of the self is necessary for the successful completion of video-tasks. Given the flexibility of the computerized testing system, we were interested in developing alternative videotasks that more directly assessed the strategy that animals use to solve such visualkinesthetic discrimination problems. Rather than attempt to develop tests that were directly analogous to the mark test, we were more interested in asking questions that were related to more basic self-perception abilities. Therefore, our main questions was: Is it necessary for subjects to recognize that they control the physically displaced movements of images on the computer screen in order to successfully complete video-tasks?

Initial Studies with Capuchins and Chimpanzees We first trained 3 capuchin monkeys (Cebus apella) and 3 chimpanzees (Pan troglodytes) on the basic computer tasks SIDE and CHASE. The capuchins (2 females and 1 male) were tested in group cages and were videotaped in order to determine individual performance. The chimpanzees (2 males and 1 female) were also tested in their group cages, but they learned to take turns at the apparatus and thus videotaping was not necessary. Although the testing environments were

VIDEO-TASKS AND THE PRECONCEPTUAL SELF

247

slightly different for the two species, the videotapes indicated no obvious social influences on performance due to group testing for either species. The apparatus was a modification of the NASA/LRC Computerized Test System (LRC-CTS) developed by Rumbaugh et al. (1989). It consisted of a microcomputer (IBM compatible XT, AT, or 386) with a joystick (KC3; Kraft Systems, Inc.) and a 33 cm (13") color monitor. For the capuchins, a 6cm-long joystick was mounted approximately 3 cm below the center of the monitor and was attached to a cart positioned outside of the cage. The capuchins were rewarded with 97-mg banana-flavored pellets (Noyes, Inc), which were delivered via a Gerbrands pellet dispenser (G5120). For the chimpanzees, the joystick was mounted onto a piece of opaque Plexiglas that was attached directly to the outside mesh of the animals' cage. They were rewarded with small pieces of fruit, peanuts, or a squirt of fruit-flavored juice whenever a trial was successfully completed. The entire testing apparatus is more thoroughly described by Richardson et al. (1990). SIDE Task. For the SIDE task, the cursor was a white circle (2 cm in diameter) positioned in the center of the computer screen at the beginning of each trial. The target was a blue, 3cm-thick border located on the edge of screen. At first, this border existed on all four sides of the screen. The subject's task was to manipulate the joystick so that the cursor would move into any of these four target walls. Whenever the subject moved the joystick, the cursor would move across the screen in the direction identical to that of the joystick. If the subject ceased to manipulate the joystick, the cursor would remain stationary. As the subject's performance improved, the number of possible target walls decreased from 4 down to 3, 2, and ultimately 1. Then, as performance continued to improve, the length of the one wall would decrease from 26 cm (a full wall) to 17 cm in length (Wall l-A), then down to 9 cm (Wall l-B), and ultimately to 2 cm (Wall l-C). Specifically, contacting the target in less' than 5 seconds, averaged over 5 consecutive trials, resulted in a decrease in the number/size of the target walls. Average response times greater than 20 seconds, on the other hand, produced an increase in the number/size of the target walls. In all conditions, the positions of the target walls on the edge of the screen (top, bottom, left, or right) were randomly assigned for each trial. Results of the SIDE task training indicated that both species were able to learn to move the cursor into the stationary targets. Figure 1 contains the average proportion of correct responses for each species during the first 100 trials of each target condition of the SIDE task. Performance was at the level of chance for each species during the 4 full-wall conditions. Repeated measures analysis of variance (ANOVA) were performed (with eta-square h2 ) a s a measure of effect size. Results indicated significant main effects for species and target size as well as a significant species-by-target size interaction. Further analysis of the interaction

248

M.J. JORGENSEN, S.J. SUOMI, •

W.D. HOPKINS

showed that it was only after the task became more difficult that species differences emerged. Tests of simple main effects revealed that the chimpanzees performed at higher proportions during the one full wall, Wall l-A, and Wall 1-B conditions. Interestingly, there was no significant difference at the hardest level (Wall I-C).

0.8

o 0.6

.o o 0.4 gg, t~

0.2

0

4

I

I

I

I

I

I

3

2

1

1-A

1-]3

1-C

Target Wall

FIGURE 1. Proportion of correct first responses on the SIDE task for both species. Values represent species averages across the first 100 trials for each target condition. CHASE Task. The CHASE task was similar to the SIDE task except that the target did not remain stationary. The subject was required to move the cursor (now a 1 x 1 cm white plus sign) into a moving target (a 2.5 x 2.0 cm blue rectangle). At the start of each trial, the cursor was positioned in the center of the screen while the target was randomly positioned near the periphery of the screen. The cursor moved at a rate of approximately 5 cm/s while the target moved in a random billiard ball pattern (see below) across the screen at a rate of approximately 4 cm/s that was contingent upon joystick manipulation. In other words, both the cursor and target moved only when the joystick was manipulated. If the joystick was not manipulated, then both the cursor and target remained stationary. Specifically, the billiard ball target-movement routine consisted of an initial movement in a diagonal direction followed by rebounds off the edges of the screen. Whenever the target reached the edge of the computer screen, it would reflect off the edge of the screen and move in the opposite direction, shifted 90 degrees from

VIDEO-TASKS AND THE PRECONCEPTUAL SELF

249

its original direction. For example, if the target was moving up and to the left and hit the top edge of the screen, it would then begin to move down and to the left from that location. Results of the CHASE task indicated that both species showed rapid generalization from the SIDE task to the CHASE task. During the first 100 trials of the CHASE task, the chimps completed 78% of their trials, whereas the capuchins completed 53%. The mean response time during the first 100 trials was 20.65 seconds for the capuchins and 14.15 seconds for the chimps. Overall, both the capuchins and the chimpanzees were able to master the basic video-tasks without much difficulty. Although the performance of the chimpanzees was generally better than that of the capuchins, large species differences only occurred when the task became increasingly more difficult.

Competitive CHASE Task Once we knew that both species were able to successfully complete the SIDE and CHASE video-tasks, we set about developing alternative programs that would more specifically test whether the subjects understood which image they controlled on the computer screen. As Heyes (1993) states, self-directed behaviors shown in response to mirror exposure require subjects to "distinguish sensory inputs generated by the state and operations of its own body from other sensory inputs, and to detect contingencies, with the former category, between direct feedback and novel, displaced visual feedback" (p. 186). In order to test this idea, we added a competitive component to the basic CHASE task. Now instead of merely moving the cursor into the target image, subjects had to move the cursor into the target before a competing, computercontrolled image (CCI) contacted the target. The idea was that the subjects would not be able to correctly complete a majority of trials unless they understood which image they controlled. We tested the same 3 capuchin and 3 chimpanzee subjects using this new task. At the beginning of each trial, three images appeared on the computer screen: 1) the cursor image, which was always a white plus symbol; 2) the target image, which was always a blue square; and 3) the computer-controlled image (CCI), which varied across conditions (details described below). Movement of the joystick caused all three images to move. When no joystick movement was made, none of the three images moved. In every condition, the target moved in the same billiard ball pattern used in previous tasks. A correctly completed trial occurred if the cursor contacted the target before the CCI. For correct trials, appropriate auditory feedback was given and a food reward was delivered. An incorrect trial occurred if

250

M.J. JORGENSEN, S.J. SUOMI, t~ W.D. HOPKINS

the CCI contacted the target before the cursor; then a low frequency noise and no reward was given. The parameters of the computer-controlled image varied along two dimensions: its image and its method of movement. The CCI was either "cued" or "not-cued." The cued CCI was a red circle. This allowed for visual discrimination between the cursor and the CCI (i.e., one was a white plus sign and the other was a red circle). In the not-cued condition, the CCI was a white plus; in other words, it was identical to the cursor image and thus overt visual discrimination between the two images was not possible. Therefore, when the CCI was not cued, the subject had to discriminate between the cursor and CCI based solely on the correspondence between the movements of the joystick and the movements of the images on the screen. The computer-controlled image either did not move ("none") or else it moved according to either a "dumb" or a "smart" method. When following the dumb method, the CCI moved in a manner identical to the target movement: It moved in a different, random billiard ball pattern across the screen. When following the smart method, the CCI moved according to the following algorithm: 1) it moved in a random direction; 2) if the distance between the computer and target images got smaller, then it continued to move in that direction; 3) otherwise it moved in another random direction and repeated the algorithm until the trial ended. This method was much more efficient than the dumb method (as shown in simulations in which the dumb and smart strategies competed against each other). The different dimensions of the CCI were systematically altered during three phases of testing, with each phase consisting of four testing sessions. In each phase, the CCI was cued for the first session, then not cued for the next two sessions, and then cued again for the final session (following an ABBA design). In the first (training) phase, the CCI always remained stationary. In the second phase, the CCI always followed the dumb method. In the final phase, the CCI always followed the smart method. Because the first session was used primarily as a training period in order to familiarize the subjects with the procedure, no results will be discussed. For the other two phases of testing, performance measures were obtained by combining the first 15 trials of the first and last sessions (the A's of the ABBA design) and combining the first 15 trials of the second and third sessions (the B's of the ABB A design). These data provided the percentage of correct trials for each combination of CCI parameters (see Table 1). Overall, only the chimpanzees were able to successfully complete a majority of their trials. The chimpanzees correctly completed 74% (p<.01, binomial probability), and the capuchins correctly completed 61% of their trials (ns, binomial probability). Repeated measures analyses of variance were performed with

VIDEO-TASKS AND THE PRECONCEPTUAL SELF

251

one between-subjects factor (species) and two within-subjects factors (method and image). Results indicated that the chimpanzees correctly completed significantly more trials than the capuchins. There was an overall effect of computer method; however, there was also a significant interaction between species and method. According to tests of simple main effects, the capuchins correctly completed significantly fewer trials when the CCI followed the smart method (30.3%) compared to when it followed the dumb method (75.7%). On the other hand, the chimpanzees' performance was not affected by the CCI method (64.7% and 82.0% for the smart and dumb method, respectively). Neither species showed significant decrements in performance when the CCI was identical to the cursor image.

TABLE 1. Percentage of Correct Trials for Capuchins and Chimpanzees.

Species

CCI Imase

CCI Method Dumb Smart

Capuchins

Cued Not Cued Cued Not Cued

78.0 73.3 80.7 83.3

Chimpanzees

36.7 24.0 68.0 61.3

These data suggest that the chimpanzees were capable of recognizing that they controlled the cursor image and were able to use that knowledge to successfully solve the task. The capuchins, on the other hand, had difficulty with the task, and these results suggest that they were not readily able to recognize which image they controlled on the computer screen. These data further indicate that it was when the demands of the task increased that the species differences became most pronounced. For example, when the CCI was cued and followed the dumb method (the easiest condition), the chimpanzees correctly completed 73% of their trials while the capuchins correctly completed 78%. However, when the CCI was not cued and followed the smart method (the most difficult condition), the chimpanzees correctly completed nearly twice the number of trials than the capuchins (69% compared to 36%). This suggests that the chimpanzees did not need either the visual cue or the slower CCI method in order to accurately complete a majority of their trials. The capuchins, on the other hand, were incapable of successfully completing a majority of the trials when the task was most difficult.

252

M.J. JORGENSEN, S.J. SUOMI, & W.D. HOPKINS

Studies w i t h H u m a n s In light of the fact that there is no basic information on video-task performance in humans as they apply to our experimental question of interest, we decided to test children and adults on our competitive version of the CHASE task. Our rationale for testing the children was to examine the extent and limits of video-task performance in subjects of comparable ages to our chimpanzee subjects. Assuming that similar mechanisms underlie the performance of each species, we sought to determine whether humans would resemble other primate species in performance. The human subjects we tested were 4 children, 2 males and 2 females, between 5 and 6 years of age (mean = 5.7) and 11 college students, 4 males and 7 females (mean age = 19.1 years). The children were enrolled at a local Montessori school. The humans were tested using a computer system and software identical to those used with the capuchin and chimpanzee subjects. The subjects were tested in a gymnasium or small conference room away from their classroom. Each subject was tested on the same four conditions of the competitive task, in which the CCI was either cued or not cued and followed either the dumb or smart movement method. However, due to the limited attention span of the children, we modified the testing procedure in order to facilitate data collection. Unlike the nonhuman primates, the human subjects were given one test session per condition, with each condition counterbalanced across subjects. Each subject received 15 trials per session. Prior to testing, the subjects were given as many practice trials as they wanted in order to feel comfortable with the task demands. For practice trials, we had the subjects complete trials during which the CCI remained stationary on the computer monitor (phase 1 for the nonhuman primates). We reasoned that these types of trials would familiarize the subjects to the task demands. Both the children and the adults were verbally instructed that the goal of the task was for them to determine which of the two cursors on the monitor was under their control. We further emphasized that the objective of the task was for them to direct a collision between "their" cursor and the target as quickly as possible. The children were rewarded at the end of the test session with a piece of candy. The results of the children and adult subjects are depicted in Table 2. Overall, the children completed 77% of their trials and the adults completed 86.2%, both significantly greater than chance (p<.01, binomial probability). The differences between the children and the adults were significant. Similar to the nonhuman primates, there was also a significant effect of method, with subjects correctly completing more trials when the CCI followed the dumb method (90.0%) than when it followed the smart method (77.6%). However, unlike the nonhuman primates, the humans correctly completed significantly more trials when the CCI

VIDEO-TASKS AND THE PRECONCEPTUAL SELF

253

Was cued (88.3%) than when it was not cued (79.3%). None of the interactions was significant. TABLE 2. Percentage of Correct Trials for Children and Adults. CCI Method Groups

CCI Image

Dumb

Smart

Children

Cued Not Cued Cued

93.3 78.0 93.5

73.5 63.3 86.7

Not Cued

89.6

75.3

Adults

Due to the fact that the testing procedures for the humans and the nonhuman primates were slightly different, no statistics will be presented. However, both groups were given the same conditions of the same video-tasks and therefore, it is worthwhile to at least make a guarded comparison of their performances. |

lOO 1

,,

121Capuchins ! C~anz~s

gO

n CNdmn m Adults

c.)

o

60

q.)

40

20

Most D~cult FIGURE 2. Species by CCI condition during competitive CHASE task. "Easiest" refers to the condition when the CCI followed the dumb method and was cued. "Most Difficult" refers to the condition when the CCI followed the smart method and was not cued.

254

M.J. JORGENSEN, S.J. SUOMI, • W.D. HOPKINS

Figure 2 shows the results for each of the four groups on the easiest and most difficult conditions of the competitive CHASE task. The easiest condition occurred when the CCI was cued and followed the dumb algorithm. The most difficult condition occurred when the CCI was not cued and followed the smart algorithm. As the graph shows, all four groups performed at approximately the same level when the CCI was not very efficient and there was a visual cue available to distinguish between the cursor and CCI. However, when the CCI was identical to the cursor image and moved according to a more efficient algorithm, the performance of the capuchin monkeys dropped dramatically. The performance for the chimpanzees and the two human groups dropped as well, but was not nearly as pronounced.

Conclusions The results of these preliminary experiments suggest that chimpanzees, young children, and adults are able to recognize which image they control on the computer screen. The capuchin monkeys, on the other hand, appear to be either unable to recognize which image they control or else are unable to properly use that knowledge to complete the task. The performance of the capuchins was most strongly affected by the efficiency of the computer-controlled image, which suggests that they may need more time in order to assess which image they control. When comparing the performance of the different groups, the chimpanzees performed better than the capuchins, and the adults performed better than the children. All groups had a more difficult time completing the task when the computer-controlled image followed the more efficient strategy. In contrast to the nonhuman primates, the children and adults performed better when the computercontrolled image was visually distinct from the cursor. Why the per/ormance of the capuchins and chimpanzees was not influenced by the visual cue is not clear. Although these studies must be considered preliminary, we believe that the methods described will provide a useful alternative means for assessing the preconceptual self in both humans and nonhumans. The competitive CHASE task requires the subjects to: 1) recognize which image they control; and 2) rapidly act upon that knowledge before the CCI contacts the target first. We argue that this is one of the main advantages of the video-task paradigm over other contingency recognition methods. Other techniques, such as video or mirror recognition, only require the subjects to recognize a contingency between their movements and the movements of the images presented. Video-tasks, on the other hand, require subjects to actively use that knowledge in order to bring about the desired result.

VIDEO-TASKS AND THE PRECONCEPTUALSELF 255 It is the competitive part of the task we have presented that adds an extra dimension to the behavior under study. Although many questions remain unanswered, we believe that the video-task techniques described here will provide many insights into the preconceptual self in both human and nonhuman primates. We hope to develop future studies that will take full advantage of video-task technology, and will ask more specific questions about the nature of the preconceptual self.

ACKNOWLEDGMENTS

Financial support was generously provided by grant NS-29574 to WDH, grant RR00165 to the YRPRC, and a Predoctoral IRTA Fellowship from NICHD to MJJ. The authors would like to thank Mindy Babitz, Stacey Bales, Laura Empey, Charles Hyatt, Ellen Johnson, Lynn Rousseau, and Brandi Woods for help in collecting the data. A portion of these findings was presented at the 17th Annual Meeting of the American Society of Primatologists, Seattle, Washington, July 27-31, 1994. Correspondence concerning this chapter should be sent to Matthew J. Jorgensen, Division of Behavioral Biology, Harvard Medical School, New England Regional Primate Research Center, One Pine Hill Drive, P. O. Box 9102, Southborough, MA, 01772-9102.

REFERENCES

Anderson, J. R. (1984). Monkeys with mirrors: Some questions for primate psychology. International Journal of Primatology, 5, 81-98. Bahrick, L. E., & Watson, J. S. (1985). Detection of intermodal proprioceptive-visual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21, 963-973. Butterworth, G. (1992). Origins of self-perception in infancy. Psychological Inquiry, 3, 103-111. Gallup, G. G. (1970). Chimpanzees: Self-recognition. Science, 167, 86-87. Heyes, C. M. (1993). Anecdotes, training, trapping, and triangulation: Do animals attribute mental states? Animal Behaviour, 46, 177-188. Heyes, C. M. (1994). Reflections on self-awareness in primates. Animal Behaviour, 47, 909-919. Ledbetter, D. H., & Basen, J. A. (1982). Failure to demonstrate self-recognition in gorillas. American Journal of Primatology, 2, 307-310. Lethmate, J., & Ducker, G. (1973). Untersuchungen zum Selbsterkennen im Spiegel bei Orang-utans und einigen anderen Affenarten. Zeitschrift fbir TierpsychoIogie, 33, 248-269. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self New York: Plenum Press. Mitchell, R.W. (1993). Mental models of mirror self-recognition: Two theories. New Ideas in Psychology, 11, 295-325.

256

M.J. JORGENSEN, S.J. SUOMI, • W.D. HOPKINS

Neisser, U. (1991). Two perceptually given aspects of the self and their development. Developmental Review, 11, 197-209. Parker, S. T. (1991). A developmental approach to the origins of self-recognition in great apes. Human Evolution, 6, 435-449. Parker, S. T., Mitchell, R. W., & Boccia, M. L. (1994). Expanding dimensions of the self: Through the looking glass and beyond. In S. T. Taylor, R. W. Mitchell, & M. L. Boccia (Eds.), Self-awareness in animals and humans: Developmental perspectives (pp. 3-19). Cambridge: Cambridge University Press. Patterson, F. (1990). Mirror behavior and self-concept in the lowland gorilla. American Journal of Primatology, 20, 219-220. Povinelli, D. J. (1993). Reconstructing the evolution of mind. American Psychologist, 48, 493-509. Povinelli, D. J., Rulf, A. B., Landau, K., & Bierschwale, D. (1993). Self-recognition in chimpanzees (Pan troglodytes): Distribution, ontogeny, and patterns of emergence. Journal of Comparative Psychology, 107, 347-372. Richardson, W. K., Washburn, D. A., Hopkins, W. D., Savage-Rumbaugh, E. S., & Rumbaugh, D. M. (1990). The NASA/LRC computerized test system. Behavior Research Methods, Instruments, & Computers, 22, 127-131. Rumbaugh, D. M., Richardson, W. K., Washburn, D. A., Savage-Rumbaugh, E. S., & Hopkins, W. D. (1989). Rhesus monkeys (Macaca mulatta), video-tasks, and implications for stimulus-response spatial contiguity. Journal of Comparative Psychology, 103, 32-38. Swartz, K. B., & Evans, S. (1991). Not all chimpanzees (Pan troglodytes) show selfrecognition. Primates, 32, 483-496.

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

257

CHAPTER 13

Move YourSelf, Baby! Perceptuo-motor Development from a Continuous Perspective AUDREY L.H. VAN DER MEER and F. RUUD VAN DER WEEL

University of Edinburgh

Explaining Development What do we mean when we say that people, especially infants and children, have developed? In general, we mean that an individual's behavior changes with age. Consider the following example: A 9-month-old baby stands while holding onto something; nine months later she can walk the stairs alone. Here we have an intraindividual change in perceptuo-motor behavior that is related to the age of the tested child. Theoretically, one can distinguish between continuity and discontinuity in development. A continuous view on development holds that all changes in behavior are predictable from earlier behavior. In contrast, there is the view that development is essentially discontinuous in nature. From such a perspective, "climbing the stairs alone" at the age of 18 months would then not develop gradually from "standing while holding onto something." Instead, from a discontinuity perspective "walking the stairs" should be considered a completely novel behavior. For years, developmentalists adopting a discontinuous view of development have been tempted to speculate about the changes in internal processes and structures that cause qualitatively different behaviors to emerge in a child's behavioral repertoire as he or she gets older.

Development as a Discontinuous Process Traditionally, the behavioral repertoire of a newborn baby was thought to consist of a set of reflexes. Thus, newborns were believed to respond to external stimuli in a very stereotyped manner, without showing any voluntary control.

258

A.L.H. VAN DER MEER d~; F.R. VAN DER WEEL

Reflexes can be very simple, such as the knee-jerk reflex, or they can involve more complex movements, as seen in the stepping reflex. Because of the tendency to consider newborns as reflexive creatures, their movements have been dismissed as unintentional. However, developmentalists agree that, for instance, the reaching behavior of a 4-month-old baby is clearly intentional and purposeful. So what are the underlying processes that turn the involuntary reflexes of the newborn into voluntary actions later on in the baby's life? Many explanations of development still implicitly rely on the nativist concept of neural maturation (e.g., Forssberg, 1985, 1989). Development occurs through innate, biologically determined predispositions and without influence from the environment. McGraw (1945), for example, explained the development of motor skills in terms of maturation of the brain (See Figure 1). She believed that initially, movements were under midbrain control and therefore involuntary. As the cortex matured, however, the reflexes were suppressed and then disappeared completely only to be replaced by more sophisticated behaviors. This cortical inhibition theory is an example of a discontinuous view on development. Development here is seen as a process of qualitative differentiation, where unobservable changes in the brain are assumed to cause development.

Ji. ,

~

i

i:

1

i

2

'

FIGURE 1. Maturation of the brain in progress. When Zelazo (1976; Zelazo, Zelazo, & Kolb, 1972) showed that the stepping reflex in infants persists for longer if extensively practiced, it seemed he was arguing for a more continuous view on development. If, with practice, neonatal stepping movements can be maintained and even increased in frequency, then the cortical inhibition theory must be incorrect. However, Zelazo (1983) argues that, with use, newborn reflexive stepping (requiring no cortical control) converts into an instumental behavior, through the "intrinsically" rewarding effects of the

MOVE YOURSELF, BABY!

259

upright position, and this instrumental response forms the basis for the stepping pattern of independent walking. Inspired by Piaget (1952), Zelazo believes the trigger for the onset of voluntary walking is a major cortical or cognitive change toward the end of the first year. Thus, when it comes to explaining development, Zelazo still advocates a discontinuous view on development when he holds cognitive transitions in the brain responsible for qualitative changes in development. Development as a Continuous Process

Alternatively, perceptuo-motor development can be regarded as a continuous process in which (changes in) coordinated movements are not orchestrated by executive centers in the brain, but emerge spontaneously from the interplay between subsystems within an overall system that includes both infant a n d environment (Kugler, Kelso, & Turvey 1982). Development, then, is not seen as due to discrete changes in the central nervous system. Instead, the opposite is argued for, with neurophysiological changes in the brain resulting from the system as a whole adapting to new levels of organization at more peripheral levels. Thus, from a continuous perspective on infant development, developmental change is explained as due to changes inherent to the dynamics of the system, without invoking cortical explanations when more simple ones are sufficient. Brain development is no longer thought to be the single most important developmental driving force. Instead, there are many contributors to movement outcome (c.f. Bemstein, 1967), all of which change dramatically in the first year of life. Examples of infant studies departing from a discontinuous perspective on perceptuo-motor development are widely available in the literature. Examples of studies using a more continuous view of general infant development, on the other hand, are few and far between. However, recent advancements within Gibson's (1979) ecological approach to perception and action in general, and the dynamical systems approach (Kugler, Kelso, & Turvey, 1980) and direct perception approach (Lee et al., 1991) in particular, have made it possible to adapt continuous theories about perceptuo-motor coordination to continuous theories about perceptuo-motor development, while emphasizing the mutual relationship between the organism and the environment. In what follows, we will describe, in short, the fundamental principles underlying the dynamical systems approach and the direct perception approach and show how these principles can be applied to the field of perceptuomotor development.

260

A.L.H. VAN DER MEER & F.R. VAN DER WEEL

Dynamical Systems Approach In an attempt to solve Bernstein's degrees of freedom problem, the dynamical systems approach to movement coordination and development explains (the development of) coordinated behavior as an emergent property resulting from the interplay between many subsystems, as opposed to resulting from central, prescriptive instructions to the skeleto-muscular system. The fundamental tenet of the dynamical systems approach is that organized patterns in complex systems are the result of a self-organizing process and that this process is consistent with, although not reducible to, the laws of physics (Vereijeken, 1993).

Dynamical Systems Approach to Movement Central to the dynamical systems approach to movement is the notion of selforganization (e.g., see Prigogine & Stengers, 1984; Haken, 1983). A nice example of the process of self-organization is the coherence in a vortex in a river. The emerging order in a vortex is not forced upon the system from the outside; it thus does not result from higher, or central, or commanding, or controlling instructions (see Figure 2). Instead, it emerges spontaneously from the whole system of the inflowing water, the stones, the plants, the sand, etc. (Meijer, 1988). The concept of self-organization goes very much against the informationprocessing notion of a motor program (e.g., Schmidt, 1988). The motor program, per definition, is an internal representation of a movement pattern that is prestructured in advance of the movement itself. Functionally similar to a computer program, it consists of a prescribed set of highly detailed commands from higher neural centers to the skeleto-muscular apparatus, whose role is to carry out these instructions. It is important to note that from a motor-program perspective, both motor learning and motor development are explained in terms of an individual's information-processing abilities. In the dynamical systems approach, however, the emphasis is on the lawful basis for coordination to occur (Turvey, 1990). The explanation is derived from first principles: It never takes the form of introducing a special mechanism (like a motor program) that contains or represents the coordination before it appears. Von Hoist's remarkable experiments on the centipede provide a good opportunity to contrast traditional motor program theory with the dynamical systems approach to motor control. By amputating leg pairs until only three such pairs were left, Von Hoist (1937) transformed the centipede's gait (a pattern in which adjacent legs are about one-seventh out of phase) into that of a six-legged insect. Further, when only two pairs of legs were left, the asymmetrical gait of the quadruped was observed. It is hard to imagine that the nervous system of the

MOVE YOURSELF, BABY!

261

centipede possessed stored programs for these gaits in anticipation of legs being amputated by a keen experimenter. Rather, given a novel configuration, the system appears to adopt spontaneously those modes of locomotion that are dynamically stable (Kelso & Scholz, 1986).

Ck GP-

FIGURE 2. The "self"-organizing vortex.

The Compatibility of Movement Coordination and Development The issues of movement coordination and its development are commonly treated as being separate with respect to different time scales: the immediate time scale of here and now (i.e., movement coordination) and the longer time scale of movement development. However, when adopting a continuous view of perceptuomotor development, this separation is not fruitful. Namely, in the dynamical systems approach, movement coordination and movement development are part and parcel of the same process (Kugler et al., 1982). Perceptuo-motor development consists of successive solutions to a coordination problem on an immediate time scale and the adaptations of those solutions over a longer time scale to ultimately achieve more coordinated behavior. Thus, there is a unity and continuity between movement coordination and development. Consequently, fundamental principles used in the dynamical systems approach to investigate movement control also apply when it comes to investigating movement development.

262

A.L.H. VAN DER MEER c~; F.R. VAN DER WEEL

Dynamical Systems Approach to Development In an attempt to explain developmental change without reverting to unobservable changes in the brain that cause development to occur, the dynamical systems approach has adopted a much more continuous view on development. Development is no longer thought to be due to one causal factor (usually thought to be a discrete change in brain development). Instead, through the process of selforganization, the dynamical systems approach views development as a product of interplay between many factors in a system that encompasses both infant and environment. Thelen and her coworkers have adopted a dynamical systems approach to development (Thelen, 1983; Kamm, Thelen, & Jensen, 1990; Thelen & Smith, 1994). Specifically, two developmental transitions are put in a different light: the disappearance of newborn stepping and the onset of independent locomotion. In the case of newborn stepping, they show that the response disappears because of increasingly heavy legs (Thelen & Fisher, 1982; Thelen, Fisher, & RidleyJohnson, 1984; Thelen, 1985). This offers a much more parsimonious explanation than introducing obscure causal explanations such as cortical maturation (McGraw, 1945), cognitive structures (Zelazo, 1983), or a succession of hierarchical, patterngenerating circuits (Forssberg, 1985) to account for the disappearing response. With respect to the onset of independent walking, Thelen (1986; Thelen, Ulrich, & Niles, 1987) elegantly demonstrates that long before infants can walk alone, they show frequent, coordinated, alternating steps with many characteristics of more mature walking when supported with their feet on a small treadmill. So it seems that the ability to generate maturelike steps is in place long before actual independent locomotion, but that the performance of these movements is constrained by the infant's inability to provide postural support and stability to stretch the legs backwards. If you provide the infant with postural support and mechanically stretch his leg backwards, then mature stepping emerges spontaneously. More recently, the dynamical systems approach has also been applied to the development of reaching and grasping (Wimmers, Savelsbergh, Beek,& Hopkins, 1993). Wimmers and colleagues showed that the transition from nonreaching to reaching behavior, and from reaching without grasping to reaching with grasping in young infants can be considered a nonequilibrium phase transition. Thom (1975) defined an elementary transition (or catastrophe) as a sudden change in a behavioral variable induced by a small and continuous change in an independent variable or control parameter. Viewed this way, developmental stages that may seem to appear suddenly in an infant's behavioral repertoire, caused by so-called discontinuous developments in the brain, are in fact the result of an entirely unspecific continuous change in the control parameter. Thelen has proposed that

MOVE YOURSELF, BABY!

263

for independent walking, the control parameter is a critical combination of strength and balance. In the development of reaching, the muscle/fat ratio of the ann can be considered a control parameter (Wimmers et al., 1993). When this ratio increases gradually over time, a qualitative jump from nonreaching to reaching behavior can be observed.

Direct Perception Approach The second pillar on which the ecological approach to perception and action rests is direct perception. The remainder of this chapter will concentrate on the direct perception approach. One core idea of this approach is the theory of affordances (Gibson, 1979). In this theory, the environment is said to constitute a collection of possiblities for action, where invariant energy patterns specify those affordances, and in which perception is the pickup of such information. A second core idea of the direct perception approach is that perception and action are tightly coupled, and that perception involves active exploration: As perceptual information that is relevant for action is extended in space and time, activity is essential for the pickup of the information (Michaels & Beek, 1994). The aim of the direct perception approach is to quantify how prospective perceptual information for movement control changes (or remains constant) over time, as a function both of the actions of the observer and structures and changes in the environment. Instead of presupposing an unstructured, meaningless, fiat, two-dimensional retinal image as a starting point for visual perception (indirect perception), the direct perception approach takes a structured, meaningful, and infinitely rich ambient optic array as its point of departure. The concept of optic array (or optic flow) can be defined as the changing pattern of light reflected to a moving eye (Gibson, 1961). The optic array is of fundamental importance in understanding direct perception because it stands outside particular perceptual systems. It is the input available to each and every one of them. The job of the perceptual system is to pick up the information in the flow field. The concept of optic array has been refined over the last decades, and mathematical descriptions of it have been developed. For instance, the "tau margin" (Lee & Young, 1985) specifies the time it would take a surface element to reach the nearest position to the point of observation under constant velocity. The more generally defined "tau function" (Lee, 1991) is a feature describing the timing and spatial information available in the optic array more continuously (see Figure 3). The tau function of x is generally defined as x divided by the rate of change of x, in symbols:

264

A.L.H. VAN DER MEER & F.R. VAN DER WEEL

t(x)=x/~. For example, when hitting an approaching ball, the tau function would not only provide information about when to initiate the hitting action, as the tau margin would, but it would provide information about how to control the hitting action as well (Van der Weel, Van der Meer, & Lee, 1995). However, before such continuous perceptual information for movement control can be picked up adequately in developing infants, some lessons in practical optics have to be learned.

/

.~.~

\~.J

:,'~.

FIGURE 3. Continuous prospective perceptual information for movement control.

Direct Perception in Development The direct perception approach to development argues that there is unity and continuity between the development of perceptual and motor skills. Rather than assuming that the baby, under the influence of cortical maturation, develops voluntary motor skills to which perception then needs to be mapped, the direct perception approach holds that perceptual and motor skills develop hand in hand. This suggests that development in general is a gradual, continuous process that begins at, or even before, birth. Next, an illustrative example will be given of how spontaneous arm movements develop into later successful reaching. Traditionally, the emergence of successful reaching and grasping is described as a discrete step in development that suddenly emerges at around 4 months of age. Newborn babies have long been thought to be reflexive creatures, responding to physical stimuli in a compulsory and stereotyped manner. However, more and

MOVE YOURSELF, BABY!

265

more research to date indicates that newborns can move their arms in a purposeful way (Bower, Broughton, & Moore, 1970; Von Hofsten, 1982; Butterworth & Hopkins, 1988). Would newborn babies counteract the effects of weights pulling on their wrists? If it could be demonstrated that newborns take gravity into account when making arln movements, then this would show that these spontaneous movements cannot be characterized as stereotyped or reflexive. If it could also be shown that spontaneous arm-waving movements were under visual control, then it is very likely that these movements have an exploratory function. We recorded spontaneous arm-waving movements while newborns lay supine facing to one side (Van der Meer, Van der Weel, & Lee, 1995). The babies were allowed to see only the ann they were facing, or only the opposite ann via a video monitor, or neither arm because of occluders. Small weights pulled on their wrists in the direction of the toes. The babies opposed the perturbing force so as to keep an arm up and moving normally, but only when they could see the arm, either directly or via video. These results suggest that while looking at their waving arms, newborn babies are developing visual control of movement. In a recent experiment (Van der Meer, 1995), we investigated the nature and possible functional significance of neonatal arm movements in more detail. Would manipulating where the baby sees her arm have an influence on where she holds her arm? This experiment was carried out under dim lighting conditions, with the baby lying on her back facing one arm. A narrow beam of light (7 cm in diameter) was shone over the baby's nose or chest in such a way that the arm the baby was facing was barely visible, unless it was moved into the otherwise invisible beam of light. The babies spent most of the recording time with their ann in the light. As a result, they held their arms significantly higher when the light was level with the nose than when it was level with the chest. Preliminary results suggest the babies also slowed down the arm when in the light, indicating sophisticated control rather than excited thrashing of the arms.

Developing the Self What functional significance might neonatal arm waving have? In order to be able to successfully direct behavior in the environment, the infant needs to establish a bodily frame of reference for action. Because actions are guided by perceptual information, setting up a frame of reference for action requires establishing informational flow between perceptual input and motor output. It also requires learning about body dimensions and movement limitations. Acting successfully entails perceiving environmental properties in relation to oneself. It is not objects per se that organisms perceive, but what these objects afford for action (Gibson, 1979). What any given object affords depends on the size

266

A.L.H. VAN DER MEER • F.R. VAN DER WEEL

and action possibilities of the perceiver. Young infants' exploratory actions inform them about the consequences of their actions in the environment (Gibson, 1988). At the same time, a self-produced action provides the baby with important information about itself. It is this dual process of perceiving oneself and perceiving the consequences of one's actions that provides very young infants with invaluable information about themselves, in terms of their action capabilities and bodily characteristics. Obviously, during infancy new skills are constantly appearing, and bodily dimensions are changing rapidly. As new action possiblities emerge, infants have to update their perceptions, and vice versa, in a never-ending circular cycle (see Figure 4).

/

F I G U R E 4. The perception-action information.

cycle:

Gathering

action-relevant perceptual

It is widely known that young infants spend many hours looking at their hands. And so they should, for a vast amount of lessons in practical optics have to be learned in those early weeks before reaching for objects can emerge. First of all, infants have to learn that the hands belong to the self, that they are not simply objects, but that they can be used to touch all sorts of interesting objects in the environment. In order to successfully reach out and grasp objects in the environment, infants also have to familiarize themselves with their own body dimensions in units of some body-scaled or, more generally, action-scaled metric (Warren, 1984; Warren & Whang, 1987). In other words, infants have to learn to perceive the shapes and sizes of objects in relation to the arms and hands, as

MOVE YOURSELF, BABY!

267

within reach or out of reach, as graspable or not graspable, in terms of their affordances for manipulation. Interestingly, in a study where hemiparetic cerebral palsied children had to knock an approaching ball off a track, we found that the children started the hitting action earlier with their affected arm (Van der Weel et al., 1995), thus compensating for the fact that this arm moved slower than the unaffected arm. Because cerebral palsy is a congenital disorder, the children were used, so to speak, to it being slower and more difficult to control and took this into account when initiating their actions. Research with clumsy children intercepting a moving toy reports similar findings (Ftirsstrom &Von Hofsten, 1982). All this information about the self has to be incorporated into a bodily frame of reference for action in those early weeks before reaching for objects "emerges." We have all experienced this process of incorporation, namely when learning new perceptuo-motor skills. For instance, tennis rackets, skis, golf clubs, and other extensions of the human body, such as false teeth and cars, first have to be incorporated into our habitual frame of reference before we can use them to their full potential (Tamboer, 1988). At first we experience those instruments as unmanageable barriers between ourselves and the environment. However, once incorporated into our "bodily" frame of reference, they increase our action possibilities considerably and are almost regarded as our own body part~s. In this context, we would like to speculate about the role of early arm movements for distance perception in general. Professor Henk S tassen (1994, personal communication) is a mechanical engineer from Delft University of Technology, The Netherlands. He designs artificial arms for babies who are born with two stumps because of genetic disorders or because their mothers had taken the drug thalidomide in the 1960s during pregnancy to prevent miscarriage. He observed that if you fit babies with artificial arms early (around 2 to 3 months), they do not seem to have any problems avoiding obstacles as soon as they learn to walk. However, if the arms are fitted too late, the babies will have tremendous problems perceiving distance, and they will initially bump into walls and obstacles when they start walking. Gibson (1979) suggested that we perceive distance in relation to our nose length. Stassen's observations would suggest that we scale distance according to our arm length, as within reach or out of reach. During infancy, new skills are constantly appearing, and bodily dimensions are changing rapidly. In general, the bodily frame of reference has to be updated during life to accommodate changes in action capabilities and body characteristics. Sudden changes in action capabilities, as in after a stroke, show this very clearly, as do rapid changes in body size in pregnancy and adolescence. Teenagers, for example, can be notoriously clumsy; they undergo such sudden growth spurts that their bodily frames of reference need to be updated nearly daily.

268

A.L.H. VAN DER MEER 8~; F.R. VAN DER WEEL

Successfully reaching out and grasping objects in the environment requires infants to be familiar with their own body dimensions. As infants wave their arms while supine, they learn about their own body and its dimensions through vision. It seems likely that a fast-growing organism will constantly need to recalibrate the system controlling movement, and visual proprioceptive information (Lee & Aronson, 1974) is least susceptible to "growth errors." This being so, our findings could have practical implications for babies with visual deficits and for the early diagnosis of premature babies at risk of brain damage. If early ann movements have an important function for later reaching, then infants with signs of hypoactivity and/or spasticity of the arms should be monitored closely with respect to retardation in developing reaching, and possibly other perceptuo-motor skills, too. In such cases, early intervention should concentrate on helping the baby explore his arms and hands, both visually (Gibson, 1988) and nonvisually (Fraiberg, 1977). A simple intervention technique that could be used on babies with a visual deficit is the use of brightly colored, high-contrast mittens, or a string of bells around the baby's wrists. It is a well-known phenomenon that reaching out is the first developmental milestone that blind babies fail to reach on time. The sound of bells always accompanying that particular proprioceptive feeling when the arms move might enable the baby to establish a stable bodily frame of reference for reaching based on nonvisual exploration of the self.

Prospective Control in Development In order to act successfully, a baby needs not only to be able to perceive environmental properties in relation to the self, as explained above, but the infant also needs to anticipate future events in the environment, which requires prospective control (Lee, 1993). For instance, if we see a 2-year-old going around bumping into things, we think something clearly must be wrong so much do we take prospective control for granted. To avoid colliding with objects, the consequences of continuing the present course of action (such as heading in a particular direction [Warren et al., 1991] or braking with a particular force [Lee, 1976]) must be perceived and evasive action taken in time. However, collisions are not necessarily to be avoided. Getting around the environment in fact depends on bringing about collisions of the feet with the ground, the hands with a (moving) toy, and so on. These collisions have to be carefully controlled if they are to secure the required propulsive force while avoiding injury to the body. Thus, precise prospective control is needed. The where, when, and how of the collision must be perceived ahead in time, and the body must be prepared for it. Therefore, a crucial aspect of animal-environment coupling is predictive perceptual information.

MOVE YOURSELF, BABY!

269

According to the direct perception approach, this type of continuous prospective control information is, in principle, available in the ambient optic array via the tau function (Lee, 1991). But are infants capable of picking up such higher-order control information straight from birth? One possibility is that at first the infant's raw perceptual systems are only sensitive to lower-order variables such as distance and velocity and that, with learning, the infant becomes more and more sensitive to the higher-order variables such as tau (Michaels & Beek, 1994). It does not seem likely that infants perform at random until they stumble onto the proper higher-order variable, and thereafter "perceive directly." Rather, merely correlated variables might be used early, and may perhaps guide the search for and/or may come to be selected as integral parts of a higher-order informational complex. Take for example reaching for a moving toy. From an information processing account, catching a moving toy is considered more difficult than reaching for a stationary toy (simply because there is more information to process) and should therefore develop later. Interestingly, however, as soon as infants start reaching for stationary tOys, they can also catch moving toys quite well (Von Hofsten, 1979, 1983). Catching a moving toy requires quite advanced timing and anticipation skills. It makes no sense to move the hand to the place where the toy was last seen because by the time the hand gets there, the toy will have moved further along its trajectory. Thus, reaching for a moving object requires prediction of the future location of the object, which in turn requires prospective control of head, eye, and arm movements. To test how prospective control of reaching develops, we investigated reaching abilities in 11-month-old infants for a toy moving at different speeds (Van der Meer, Van der Weel, & Lee, 1994). The toy was occluded from view by a screen during the last part of its approach to force the infants to make use of predictive information. The infants clearly anticipated the moving toy because they shifted their gaze across to the exit side of the screen and started to move their arms forward before the toy had disappeared behind the occluder. Furthermore, the records of the trajectory of the hand showed that as soon as the hand started to move, it was aimed at the catching place rather than at the current position of the toy. In addition, the actions of shifting gaze and moving the hand forward were both prospectively geared to certain times - - rather than distances m before the toy would reappear, consistent with Lee's tau function. In a longitudinal experiment reported in the same study, we then investigated the development of predictive reaching in infants between the ages of 16 and 48 weeks. As soon as the infants started reaching (around 20 weeks of age), they anticipated the reappearance of the moving toy with their gaze, suggesting that this ability is a prerequisite for the onset of reaching for moving toys. However,

270

A.L.H. VAN DER MEER & F.R. VAN DER WEEL

anticipation with hand movement of the disappearance of the toy and the ability to gear actions prospectively to the time (instead of the distance) the toy was away from the catching place developed relatively late (around 40 weeks of age) and marked the transition to successfully catching faster moving toys. Thus, the information used by infants for prospectively controlling the timing of shift of ga~e and of movement of hand appears to change with age. Infants up to 32 weeks of age use a strategy of shifting gaze and starting to move the hand when the toy reaches certain positions (Distance Strategy). This strategy has one important drawback because it means that as the speed of the toy increases, the available time to arrive with the gaze at the catching place and to start the hand moving is reduced. Thus, the Distance Strategy is not efficient because it entails moving the hand increasingly faster the faster the toy moves, until the hand can move no faster. Older infants show more skill. From about 40 weeks, infants shift their gaze and start reaching when the toy is certain times, rather than distances, away from the catching place. They thus keep the time to gaze arriving and hand starting about constant (Time Strategy), so that as the speed of the toy increases, the distance of the toy from the catching place increases. In this way, the infants make available the same average time to catch successfully whether the toy is moving slowly or quickly. Would preterm, low-birthweight babies who are neurologically at risk of brain damage show a similar development of anticipation and timing skills, and, if not, is their lowered ability indicative of brain damage? We investigated these questions and found that the onset of reaching, anticipation of the moving toy with gaze and hand, and the switch from the Distance Strategy to the more efficient Time Strategy was delayed in almost all premature infants (Van der Meet, Van der Weel, Lee, Laing, & Lin, 1995). However, most preterm infants had caught up with their full-term peers by the last session (48 weeks of age), corrected for prematurity. However, two infants from the at-risk group still seemed to use the Distance Strategy when shifting their gaze and starting to move their hand. These infants also showed the least anticipation of the toy' s reappearance with their hand. The same two children were unique in the at-risk group in having neurologically abnormal scores, and were officially diagnosed at 18 and 21 months as suffering from mild to moderate cerebral palsy. Thus, poor development of prospective skill on the catching task might serve as an indicator of brain damage. Greater understanding of both normal and abnormal development of use of perceptual information in prospectively guiding action might therefore have important diagnostic and therapeutic consequences. The mastering of reaching and grasping normally develops very early, and it provides a foundation for more specific perceptuo-motor skills that rely on these abilities. Catching is such a case. It requires the pickup of predictive information and quite advanced timing

MOVE YOURSELF, BABY!

271

skills. Thus, if there is a problem with basic catching skills, then more complex skills such aswalking and speaking m skills that are highly dependent on correct timing m are also likely to be affected later on in life. In sum, the direct perception approach describes the apparent discontinuous step from nonreaching to successful reaching as a gradual, continuous process spanning many weeks. From very early on in life, babies show an interest in their moving arms. As newborns wave their arms and hands while supine, they are learning about their own body and its dimensions through vision, as well as about the consequences of their movements on the environment. At the same time, an action provides information about oneself. By looking at their waving arms, newborn babies discover and learn about all the relationships essential for successful reaching and grasping: They establish a stable bodily frame of reference for reaching. Once established, the frame of reference then allows the baby to negotiate actions that require taking into account properties of the environment, such as toys moving at different speeds. As the baby approaches 1 year of age, she has quite advanced timing and anticipation skills, which shows clear evidence of prospective control of movement.

Summary and Conclusions In this chapter, a distinction was made between a discontinuous and a continuous perspective of perceptuo-motor development. The emergence of reaching and grasping was taken as an example to show how perceptuo-motor development can be regarded as a continuous process that starts at, or even before (e.g., Prechtl, 1986), birth. Instead of introducing discontinuous, obscure causal explanations, such as cortical maturation or cognitive structures, to explain the apparent "jump" from nonreaching to reaching, the ecological approach to development provides more continuous explanations in terms of a gradually increasing fat/muscle ratio of the arms (Wimmers et al., 1993) and changes in the pickup of action-relevant perceptual information for reaching (Van der Meer et al., 1994). Clearly, the explanations provided by the dynamical systems approach and the direct perception approach are formulated at a functional level of description, which includes both infant and environment. "But what are the mechanisms?" many students of developmental psychology will undoubtedly ask. Traditionally, developmental psychologists have searched for contributing factors at the neurological level that cause discrete developmental changes to occur at the functional (behavioral) level. However, according to the ecological approach, meaningful explanations should be at their own level of functioning. Causal

272

A.L.H. VAN DER MEER & F.R. VAN DER WEEL

relationships do not exist between, say, the level of neurological functioning and the functional level, only at their own (Gibson, 1994). A discontinuous perspective on infant development culminates into the nature-nurture debate: Either through endogenous factors such as cortical maturation or through exogenous factors such as learning (or a combination of both endogenous and exogenous factors), discrete developmental changes are presumed to occur at the neurological level (Van der Meer & Van der Weel, 1993). Although traditional developmentalists agree with the proposition that nature and nurture are not separable, that we cannot attribute causality to one or the other alone, the dichotomy stubbornly persists. People still often refer to certain achievements in infancy as "innate" or "learned" and then drop the problem as if it were solved (Gibson, 1994). The search for factors either from within the environment or from within the individual supposedly causing development, has led, we think, to a preoccupation with identifying a single developmental driving force, usually located within the organism. But, according to the ecological approach to development, there is always an interaction of many factors in continuous development. Developmental change is not thought to be caused by factors either within the environment or within the individual. Instead, development is seen as an emergent property of the relevant systems under study, rather than a property of the infant alone, such as a cognitive structure. After a quarter of a century's worth of research with newborn babies, it can now safely be stated that motor actions are not simply reflexive, even in very young infants. Babies can act spontaneously, and they come equipped with perceptual systems that can be used to observe the environmental consequences of their actions. At the same time, an action provides information about oneself. It is this dual process of perceiving oneself and perceiving the consequences of selfproduced actions that provides very young infants with invaluable knowledge about themselves, in terms of their action capabilities and bodily characteristics. From this viewpoint, it follows that the emergence of successful reaching and grasping is not a discrete step in development occurring sometime around 4 months of age. Instead, the development of reaching should be seen as a gradual, continuous process, starting immediately after birth with spontaneous, selfinitiated ann movements that are brought more and more under visual control, finally culminating in the successful catching of fast-moving toys.

ACKNOWLEDGMENTS

The research reported in this chapter was supported by a grant to the first author from the Medical Research Council (U.K.) and by a grant to the second author from the

MOVE YOURSELF, BABY!

273

Scottish Office Home & Health Department. We thank Jane Lloyd for making the illustrations. REFERENCES

Bernstein, N.A. (1967). The coordination and regulation of movements. Oxford: Pergamon Press. Bower, T.G.R., Broughton, J.M., & Moore, M.K. (1970). Demonstration of intention in the reaching behavior of neonate humans. Nature, 228, 679-681. Butterworth, G., & Hopkins, B. (1988). Hand-mouth coordination in the newborn baby. British Journal of Developmental Psychology, 6, 303-314. Forssberg, H. (1985). Ontogeny of human locomotor control. I. Infant stepping, supported locomotion, and transition to independent locomotion. Experimental Brain Research, 57, 480-493. Forssberg, H. (1989). Infant stepping and development of plantigrade gait. In C. von Euler, H. Forssberg, & H. Lagercrantz (Eds.), Neurobiology of early infant behavior (pp. 119-128). Stockholm: Stockton. Forsstr0m, A., & Hofsten, C. von (1982). Visually directed reaching of children with motor impairments. Developmental Medicine and Child Neurology, 24, 653-661. Fraiberg, S. (1977). Insights from the blind. New York: Basic Books. Gibson, E.J. (1988). Exploratory behavior in the development of perceiving, acting, and the acquiring of knowledge. Annual Review of Psychology, 39, 1-41. Gibson, E.J. (1994). Has psychology a future? Psychological Science, 5, 69-76. Gibson, J.J. (1961). Ecological optics. Vision Research, 1, 253-262. Gibson, J.J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. Haken, H. (1983). Synergetics: An introduction (3rd ed.). Berlin: Springer Verlag. Kamm, K., Thelen, E., & Jensen, J.L. (1990). A dynamical systems approach to motor development. Physical Therapy, 70, 763-775. Kelso, J.A.S., & Scholz, J.P. (1986). Cooperative phenomena in biological motion. In H. Haken (Ed.), Synergetics of Complex Systems in Physics, Chemist~. and Biology (pp. 506-534). Berlin: Springer Verlag. Kugler, P.N., Kelso, J.A.S., & Turvey, M.T. (1980). On the concept of coordinative structures as dissipative structures. I. Theoretical lines of convergence. In G.E. Stelmach & J. Requin (Eds.), Tutorials in motor behavior (pp. 3-47). Amsterdam: North-Holland. Kugler, P.N., Kelso, J.A.S., & Turvey, M.T. (1982). On the control and coordination of naturally developing systems. In J.A.S. Kelso & J.E. Clark (Eds.), The development of movement control and coordination (pp. 5-78). New York: Wiley. Lee, D.N. (1976). A theory of visual control of braking based on information about time-to-collision. Perception, 5, 437-459. Lee, D.N. (1993). Body-environment coupling. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 43-67). Caambridge, MA: Cambridge University Press. Lee, D.N., & Aronson, E. (1974). Visual proprioceptive control of standing in human infants. Perception and Psychophysics, 15, 529-532. Lee, D.N., Reddish, P.E., & Rand, D.T. (1991). Aerial ducking by hummingbirds. Naturwissenschaften, 78, 526-527.

274

A.L.H.VAN DER MEER 8~ F.R. VAN DER WEEL

Lee, D.N., & Young, D.S. (1985). Visual timing of interceptive action. In D.J. Ingle, M. Jeannerod, & D.N. Lee (Eds.), Brain mechanisms and spatial vision (pp. 1-30). Dordrecht: Nijhoff. McGraw, M.B. (1945). The neuromuscular maturation of the human infant. New York: Columbia University Press. Meijer, O.G. (1988). The hierarchy debate: Perspectives for a theory and history of movement science. Amsterdam: Free University Press. Michaels, C., & Beek, P.J. (1994). The state of ecological psychology. In R. Guski & W.D. Heine (Eds.), Proceedings of the third European workshop on ecological psychology (pp. 1-21). Bochum: Ruhr-University Press. Piaget, J. (1952). The origins of intelligence in the child. New York: Basic books. Prechtl, H.F.R. (1986). Prenatal motor development. In M.G. Wade & H.T.A. Whiting (Eds.), Motor development in children: Aspects of coordination and control (pp. 53-64). Dordrecht: Martinus Nijhoff. Prigogine, I., & Stengers, I. (1984). Order out of chaos. London: Fontana. Schmidt, R.A. (1982). Motor control and learning: A behavioral emphasis. Champaign, IL: Human Kinetics. Tamboer, J.W.I. (1985). Mensbeelden achter Bewegingsbeelden. Haarlem: De Vrieseborch. Thelen, E. (1986). Treadmill-elicited stepping in seven-month-old infants. Child Development, 57, 1498-1506. Thelen, E. (1983). Learning to walk is still an "old" problem: A reply to Zelazo. Journal of Motor Behavior, 15, 139-161. Thelen, E. (1985). Developmental origins of motor coordination: Leg movements in human infants. Developmental Psychobiology, 18, 1-22. Thelen, E., & Fisher, D.M. (1982). Newborn stepping: An explanation for a "disappearing reflex." Developmental Psychology, 18, 760-775. Thelen, E., Fisher, D.M., & Ridley-Johnson, R. (1984). The relationship between physical growth and a newborn reflex. Infant Behavior and Development, 7, 479493. Thelen, E., & Smith, L. (1994). Dynamic systems in development. Cambridge, MA: MIT Press. Thelen, E., Ulrich, B.D., & Niles, D. (1987). Bilateral coordination in human infants: Stepping on a split-belt treadmill. Journal of Experimental Psychology." Human Perception and Performance, 13, 405-410. Thom, R. (1975). Structural stability and morphogenesis. Reading, MA: Benjamin. Turvey, M.T. (1990). Coordination. American Psychologist, 45, 938-953. Van der Meer, A.L.H. (1995). Keeping the arm, in the limelight: The functional significance of neonatal arm movements. Manuscript in preparation. Van der Meer, A.L.H, & Van der Weel, F.R. (1993). Les conduites d' atteinte de 1'objet chez le b6b6. In V. Pouthas & F. Jouen (Eds.), Les comportements du b~Jb~." Expression de son savoir? (pp. 233-244). Liege: Mardaga. Van der Meer, A.L.H., Van der Weel, F.R., & Lee, D.N. (1994). Prospective control in catching by infants. Perception, 23, 287-302. Van der Meer, A.L.H., Van der Weel, F.R., & Lee, D.N. (1995). The functional significance of arm movements in neonates. Science, 267 693-695. Van der Meer, A.L.H., Van der Weel, F.R., Lee, D.N., Laing, I.A., & Lin, J.P. (1995). Development of prospective control of catching a moving toy in premature at-risk infants. Developmental Medicine and Child Neurology, 37, 145-158. Van der Weel, F.R., Van der Meer, A.L.H., & Lee, D.N. (in press). Measuring dysfunction in basic movement control in cerebral palsy.

MOVE YOURSELF, BABY!

275

Vereijken, B. (1993). Learning new movement skills: Dynamics in action. In C.A.M. Doorenbosch et al. (Eds.), Learning motor skills: Proceedings of the third symposium of the Graduate Institute of Human Movement (pp. 47-60). Enschede: CopyPrint 2000. Von Hofsten, C. (1979). Development of visually directed reaching: The approach phase. Journal of Human Movement Studies, 5, 160-178. Von Hofsten, C. (1982). Eye-hand coordination in newborns. Developmental Psychology, 18, 450-461. Von Hofsten, C. (1983). Catching skills in infancy. Journal of Experimental Psychology: Human Perception and Performance, 9, 75-85. Von Hoist, E. (1937/1973). The behavioral physiology of animals and man." The collected papers of Erich von Holst. Coral Gables, FL: The University of Miami Press. Warren, W.H. (1984). Perceiving affordances: Visual guidance of stair climbing. Journal of Experimental Psychology: Human Perception and Performance, 10, 683-703. Warren, W.H., Mestre, D.R., Blackwell, A.W., & Morris, M.W. (1991). Perception of circular heading from optical flow. Journal of Experimental Psychology." Human Perception and Performance, 17, 23--48. Warren, W.H., & Whang, S. (1987). Visual guidance of walking through apertures: Body-scaled information for affordances. Journal of Experimental Psychology: Human Perception and Performance, 13, 371-383. Wimmers, R.H., Savelbergh, G.J.P., Beek, P.J., & Hopkins, B. (1993). Are there phase transitions in the development of eye-hand coordination? Research and clinical center for child development: Annual report (1992-1993) (pp.123130). Sapporo, Japan: Hokkaido University. Zelazo, P.R. (1976). From reflexive to instrumental behavior. In L.P. Lipsitt (Ed.), Developmental psychobiology: The significance of infancy. Hillsdale, NJ: Erlbaum. Zelazo, P.R. (1983). The development of walking: New findings and old assumptions. Journal of Motor Behavior, 15, 99-137 Zelazo, P.R., Zelazo, N.A., & Kolb, S. (1972). "Walking" in the newborn. Science, 176, 314-315.

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

277

CHAPTER 14

Interactions Between the Vestibular and Visual Systems in the Neonate FRANCOIS JOUEN and OLIVIER GAPENNE

Centre Nationale de la Recherche Scientifique-U.R.A 654

To perceive the world is to coperceive oneself. - J.J. Gibson (1979)

Neurobiology and cognitive psychology bear evidence of fundamental paradigm shifts and modifications from a theoretical point of view (Engel & K0nig, 1993). From a classical cognitivist point of view, cognitive processes were assumed to be based on algorithmic computations dependent on predictable and formalizable rules. However, in very recent years with the application of nonlinear dynamical approaches to development, a new idea has emerged that some functions such as action, perception, or memory arise from complex interactions in highly distributed neuronal networks (e.g., Edelman, 1992; Thelen & Smith, 1994). Unlike traditional information-processing systems, these networks are shaped by learning and experience-dependent plasticity. As a consequence, information processing is based on the self-organization of patterns of activity rather than on explicit application rules. These new conceptions stress the idea of massive parallel and distributed processing in the brain (Engel, KOnig, Kreiter, Schillen, & Singer, 1992). Rather than being parts of a highly specialized hierarchy, various brain areas are thought to form complex networks that are activated simultaneously and that contain numerous input and output pathways. Both the ontogenetic development of this massive cortical network and its topological changes during childhood or adulthood can be considered as self-organized processes with local rules that give rise to globally ordered patterns (von der Malsburg & Singer, 1988; Engel & KOnig, 1993). Apart from these changes in neurophysiological theories, cognitive and perceptual theories have also emerged (Neisser, 1993). Drawing on

278

F. JOUEN & O. GAPENNE

phenomenological tradition and the Gibsonian approach, Varela, Thompson, and Rosch (1991) have emphasized the role of action, which they consider as inseparable from perception. According to this view, cognition can be understood as an active constructive capacity of "enacting" a world. Engel and KOnig (1993) also propose that what neuroscience has to explain is not how the brain acts as a "world-mirroring" device, but how the brain is responsible for "world-making." As a consequence, neural states can be considered as basically representing the "capacities of actively structuring situations" that the organism experiences. These new theoretical views seem very useful for a study of the origins of self-perception. As depicted by Butterworth (1992) and Rochat (1993), Gibson's theory of perception offers some interesting ways of studying the development of self in infancy. Gibson's approach is based on the ecological premise that perception presents two distinct poles, subjective and objective, specified in terms of variant and invariant properties of sensory stimulation. Invariant properties correspond to unchanging aspects of the environment (e.g., the force of gravity), whereas variant aspects of stimulation have subjective reference because they are transformations (e.g., angular or linear accelerations) that occur as a consequence of body movements. According to Butterworth's formulation (1992), "the inherent 'duality' of sensory stimulation implies that there is information for the distinction between self and nonself inherent in perception." This chapter, concerned with the interaction between vestibular'and visual systems in the neonate, focuses on the study of how the newborn and the young infant are able to code information specifying a distinction between self and nonself: Self-perception requires having a body that is clearly differentiated from the environment. The onset of such self-perception involves sophisticated sensory motor systems that are able to analyze information about the environment as well as body position and that produce accurate movements (e.g., Neisser, this volume). The chapter is divided into two main sections. The first one is related to the vestibular sensory system. Rather than summarizing the anatomy and physiology of the labyrinth, which can be found elsewhere (e.g., Guedry, 1974; Benson, 1982), we shall focus on some neural characteristics of the vestibular system that are involved in self-perception. In the second section, experimental data about early vestibular-visual interactions will be presented. This section will emphasize the role of vision in providing information for movement of the self in conjunction with interoceptive information provided by the mechanical and vestibular systems.

The Vestibular Sensory System Unlike the other sensory systems, the vestibular system is not considered one of the classical senses. The sensations generated when the receptors of the inner ear

VESTIBULAR AND VISUAL SYSTEM INTERACTIONS

279

are stimulated have been discussed for a long time. Is the sensation of motion, such as is experienced after body motion, the direct consequence of a change in the afferent discharge from sensory receptors in the vestibular apparatus? Or is it the result of a complex stimulation of muscles, joints, and cutaneous receptors from the postural adjustment that is the primary response to a vestibular stimulus (e.g., seen the discussion about the vestibular nature of the Mort reflex in the neonate Andr6-Thomas & Ajuriaguerra, 1948; Eviatar & Eviatar, 1978)? The recent demonstration that vestibular afferents project to the cerebral parietal cortex (Guldin, Akbarian, & Grtisser, 1992) lends neurophysiological support to the idea that there are characteristic sensations and percepts that depend primarily on information provided by the labyrinthic system (Brandt, Dieterich, & Danek, 1994). The main function of the vestibular sensory system is to provide information about the movement and orientation of the head and body, especially the regulation of postural and motor activity at a subcortical level. Studies of human and experimental animals with vestibular lesions have clearly established that the primary function of the vestibular apparatus is to transduce both angular and linear accelerations of the head and to signal its relative position to the specific force of gravity, i.e., the gravitational vertical. This information is basically used to maintain postural equilibrium in conjunction with information from other mechanoreceptors that signal forces acting on or within the body, as well as the relative orientation of body segments (Jouen, Lepecq, & Gapenne, 1993). When studying the vestibular sensory system, one finds that some characteristics are very striking, if not paradoxical. On one hand, this system's structural organization is very simple: a peripheral vestibular apparatus that lies only one synapse away from the central vestibular midbrain neurons. Unlike the complexity of visual or auditory stimuli, the physical characteristics of vestibular stimuli are only defined in terms of direction and intensity of acceleration of the head. On the other hand, the vestibular sensory system is highly complex. This complexity is first related to its function. Apart from the control of postural activity, the vestibular sensory system is involved in more sophisticated cognitive processes relative to self-motion perception and spatial orientation. The second point about the complexity of the vestibular system concerns its multisensory nature. The natural stimulation of the vestibular system, as it occurs with head motion and body displacement, is always multisensory (vestibular, somatosensory, and exteroceptive). The different sensory cues thus provide redundant information for spatial orientation and postural control, among which vestibular information may play a overriding role. This redundant multisensory functioning is necessary. Indeed, the simple nature of both the peripheral end organs and the vestibular stimulation makes, in many cases, the coding of the acceleration direction

280

F. JOUEN & O. GAPENNE

impossible without redundancy. For example, a positive acceleration in one direction cannot be distinguished from a negative acceleration in the opposite direction. When moved at a constant velocity (i.e., with a null acceleration), one cannot perceive the direction of motion without vision. In a similar manner, headbody tilting and linear acceleration produce a similar vestibular stimulation.

The Peripheral VestibularApparatus The labyrinth is basically made up of three curved tubes, the semicircular canals, which are approximately orthogonal to one another. These structures open into the utriculus, which communicates at a sublevel with another sacklike structure, the sacculus. The three orthogonal canals located in each inner ear are identified as anterior, posterior, and lateral. In humans, the plane of the lateral canal lies in the horizontal plane when the head is tilted forward about 30 degrees. With the head in this position, the planes of the anterior and posterior canals are vertical and lie about 45 degrees to the sagittal and coronal planes. The sensory cells of a particular canal are optimally stimulated by all angular acceleration acting in the plane of the associated canal. Moreover, any angular acceleration of the head will modify the activity of sensory cells of at least one pair (left and right) of canals. This first vestibular subsystem, which has a very low response threshold (0.1 to 0.2 deg/sec2), can be considered as an accurate transducer system for detecting angular accelerations of the head and body in any plane of space. The otolith organs (utriculus and sacculus), also orthogonally oriented receptors, are involved in the detection of the gravito-inertial forces acting on the body and are sensitive to linear accelerations (thresholds: 10 -3 to 5.10 -3 g). The utricular macula is roughly parallel to the lateral canal, (i.e., horizontal) and is involved in the coding of horizontal linear acceleration. The saccular macula is located on the medial wall of the sacculus. Its major plane is parallel to the sagittal plane of the head, (i.e., vertical) and is thus perpendicular to the utricular macula. The saccular macula is involved in the detection of vertical (up and down) linear acceleration. Already at the peripheral level, the vestibular system represents an elegant transducer of head motion, providing the organism with cues that allow it to determine its orientation and motion in any plane within a gravito-inertial reference system. In a similar way, the coding of the direction and the magnitude of self-motion is found inside the peripheral end-organs themselves where polarization maps of the sensory hair cells are found. First-order afferent signals originating from the peripheral vestibular apparatus have been extensively investigated in various species, from fish to primates (see Lacour & Borel, 1993, for a review). The resting discharge rate of peripheral vestibular neurons varies but is typically about 90 spikes/sec in monkeys and 35 spikes/sec in cats. The morphological

VESTIBULAR AND VISUAL SYSTEM INTERACTIONS

281

polarization of the cells corresponds to a functional polarization that allows the perception of the direction and the magnitude of accelerations. For labyrinthine organs, the resting discharge increases when the cilia of the sensory cells are directed toward a specific cell (the kinocilium) and decreases when the hairs are deflected away from the kinocilium. Modulation of primary afferent activity is reflected by the membrane potential of the sensory cells: Depolarizing currents increase the resting discharge, whereas hyperpolarization has an opposite effect. Inside otolithic receptors, hair cells are spatially organized in reference to the striola, i.e., a physical boundary between regions of opposite polarizations. With this specific organization, otolithic afferents are excitatory for one direction and inhibitory for the opposite direction for both symmetrical end-organ receptors.

Central Vestibular System First-order afferents from the semicircular canals and the otolithic receptors terminate in the ipsilateral vestibular complex, which is one of the largest nerve structures in the midbrain. First-order afferents also send collaterals that cross the contralateral vestibular complex and also reach the cerebellum. Each vestibular nuclear complex is composed of four main nuclei (lateral, medial, superior, and descending) and several cell groups that have direct or indirect relations with the vestibular nuclei (see Gerrits, 1990; Lacour & Borel, 1993, for an extensive review). The efferent connections from the vestibular nuclear complex to the spinal levels constitute the anatomical and physiological foundations of the vestibulospinal reflexes and of postural control of the head and limbs. The vestibular nuclear complex is not only involved in compensatory postural reflexes, but also in headeye coordination and orienting behavior via the reticulo-spinal and the tecto-bulbospinal systems. The vestibular control of the head-neck system is mainly exerted by the medial vestibulo-spinal tract arising from the medial and descending vestibular nuclei. This tract is bilaterally distributed to the cervical and thoracic levels. The axons of the vestibulo-spinal neurons exhibit a fast conduction velocity (around 60 m/sec), which is compatible with the short compensatory postural responses. These vestibulo-spinal pathways, basically involved in the control of neck muscles, exert facilitatory and inhibitory influences on spinal motoneurons according to a very complex pattern necessary for the accurate tuning of body posture. Control over the basal postural activity of the head-neck system also involves the lateral vestibulo-spinal tract. This tract, originating from the lateral vestibular nucleus, contains rapidly conducting fibers (90 m/sec). Its chief function is to convey information from the sacculus and the utriculus. Connections with neck motoneurons are disynaptic (Ikegami, Sasaki, & Uchino, 1994). Output from the

282

F. JOUEN t~ O. GAPENNE

lateral tract facilitates the a and g motoneurons of the extensor muscles and has an opposite effect on the flexor muscles. Latencies around 2.5 msec have been recorded in neck muscle motoneurons after a direct stimulation of the utricular branch of the vestibular nerve (Ikegami et al., 1994). Control of axial tonus and postural adjustments of body segments are achieved via mediation of the vestibulospinal pathways, described above. From various experiments conducted on animals (see Lacour & Borel, 1993, for a review), it appears that the coordination of the head, neck, back, and limb musculature can be performed by single vestibulo-spinal neurons acting on different motoneuronal pools located at different spinal levels. Another very interesting result is the recent discovery, in animals, of simultaneous branching patterns from secondary-order vestibular neurons to oculomotor and spinal levels. Using intracellular injections of HRP (horseradish peroxidase), Berthoz, Yoshida, Vidal, and McCrea (1981) demonstrated that 70% of the neurons located in the medial vestibular nucleus of the cat had branching axons reaching the contralateral abducens oculomotor nucleus and the spinal cord. Such double connections have been identified in secondary vestibular neurons related to the posterior and anterior semicircular canals. This means that via these extensive branching patterns, a given single sensory signal is distributed to multiple target structures involved in the orienting behavior or in eye-head coordination. Neuronal activity within the vestibular nuclear complex is closely related to spatial orientation mechanisms. For example, neurons located in the pontine reticular formation, which fire in relation to eye movements, directly project onto the vestibular neurons. Around 40 to 50% of secondary vestibular neurons exhibit neuronal modulation during occular saccades, and most of them show an inhibition of their firing rate. Twenty percent of secondary vestibular neurons are directly affected by eye position signals. For example, turning the eyes to the right induces an increase in the activity of neurons located on the left side. This in turn produces an increase in the EMG activity of dorsal neck muscles on the right part of the body. Such a correlation between eye position in space and vestibular cell discharge has been found in horizontal and vertical eye deviations (Lacour & Borel, 1993). Lastly, numerous studies concerning visual-vestibular connections have clearly demonstrated that optokinetic and vestibular stimulation in opposite directions induces similar neuronal modulation at the level of second-order vestibular neurons (Dichgans & Brandt, 1978). As discussed in the introduction of this section, the vestibular nucleus complex is a rather complicated integrative structure in which signals from the vestibular apparatus, conveying information about head motion and orientation, are combined with signals from joint receptors related to the position and movement of the whole body. Information about movement from retinal receptors also

VESTIBULAR AND VISUAL SYSTEM INTERACTIONS

283

reaches the vestibular nuclear complex. This clearly suggests that the vestibular nuclear complex represents a major brainstem integration center for the perception of self-motion and for the regulation of postural and occulomotor activities involved in body spatial orientation. Neuro-anatomical evidence for ascending fiber systems connecting to the cortex has long been equivocal. However, the demonstration of short-latency (2.5 ms) evoked potentials in the central posterior-inferior nucleus of the thalamus upon stimulation of the contralateral vestibular nerve implies that the second-order vestibular neurons are directly connected to the thalamus (Benson, 1982). This is important because GlOsser and coworkers have clearly demonstrated the existence of thalamic connections to cortical areas that play an important role in the control of head movement in space (see Akbarian, Grtisser, & Guldin, 1992). Recent speculations on vestibular cortex function in humans are based on the existence, in monkeys, of multiple projections from thalamus to a kind of "inner circle" (Brandt et al., 1994) of the vestibular cortical representation. All microelectrode recordings from areas PIVC (parietal insular vestibular cortex), 7 and 3aV, which are located in the parietal lobe, demonstrate that the neurons are multisensory, responding not only to vestibular stimulation, but also to somatosensory and optokinetic stimuli. These structures appear to be a major multisensory integration center for spatial orientation and visuo-motor functions. Cortico-cortical connections (Guldin, Akbarian, & Glosser, 1992), which involve frontal and parietal areas, have been also described. Parietal temporal area T3, which receives optokinetic information via the medial, lateral, and inferior pulvinar, is also involved in this huge cortico-cortical circuit responsible for cognitive processes related to the perception of self-motion and spatial orientation. Whatever the existence of a vestibular cortex in humans, it is noteworthy that the majority of vestibular activity is controlled by brainstem structures. The question now becomes one of whether the vestibular structures involved in self-perception are functionally mature at birth.

Development of the Vestibular System The fetal development of the vestibular system has been extensively studied using the sophisticated techniques of electronic microscopy (Anniko, 1983; Lavigne-Rebillard, Deschesne, Pujol, Sans, & Escudero, 1985). The human vestibular apparatus develops simultaneously with the auditory system. The vestibular structures can be identified by the 6th week of gestation. By the 7th week, the three semicircular canals, including the cristae ampullares, the maculae of the sacculus, and the utriculus, start developing. Deschesne & Sans (1985) and Lavigne-Rebillard et al.(1985) found that the development of the vestibular system consists of two rapid phases of maturation separated by a pause between 9 and 12

284

F. JOUEN •

O. GAPENNE

weeks. By 14 weeks, vestibular receptors are developed and appear similar to those of adults. However, the semicircular canals are about 55% of their final size. According to Bast and Anson (1949), the vestibular apparatus reaches maximum size around midterm. At birth, the vestibular system is fully mature: The semicircular canals and the otolith organs are entirely formed (Dayal, Farkashidy, & Kokshanian, 1973), hair cells of peripheral receptors are functional (Rosenhall, 1972), and the fibers of the eighth vestibular nerve show a degree of myelination comparable to that observed in adults (BergstrOm, 1973). At present, there is some indirect evidence that vestibular organs do function in utero. A vestibular response similar to the Moro reflex has been described by Hooker (1952) and Wyke (1975). Elliot and Elliot (1964) suggest that during the 5th month, the fetus starts to orient itself by exhibiting a kickinglike behavior. Fetus self-orientation has been discussed in relation to the birth position (Previc, 1991). Lastly, Lecanuet and his coworkers have shown that near term, fetuses respond to passive maternal displacements by specific modulations of their heart rate (Porton, Jacquet, & Lecanuet, 1994). There is a surprising lack of data on the functional development of primary vestibular neurons in humans. Existing studies come from neurophysiological researches with animals. Curthoys (1979, 1983) has demonstrated that the horizontal semicircular canals in rats can respond to angular accelerations immediately after birth. The sensitivity is comparable to adult values at around four days after birth. A steady decrease in time constant to long-duration angular acceleration, and a decrease in the phase lag for acceleration at frequencies between 0.02 and 1.0 Hz have been observed during the first 20 days. These changes in time constant, phase lag, and high-frequency gain could be attributed to receptorafferent-efferent system development (Curthoys, 1983). So far, there is no evidence of functional development in the primary saccular and utricular neurons. Nevertheless, otolith organs are known to develop earlier than semicircular canals during fetal development (Sans, Pujol, & Marty, 1968).

Vestibular Responses in Neonates At present, direct evidence that vestibular organs do function in neonates comes from research about vestibular-ocular responses and primitive reflexes such as the Moro reflex. For example, Eviatar and Eviatar (1978) have recorded pre- and post-rotatory nystagnus, induced by body rotation, in 75% of neonates and in only 25% of premature infants. Ocular doll's eye reflex in response to vestibular stimulation during body tilting or head rotation also induces compensatory horizontal smooth eye movements as soon as birth. Different studies have focused on the effect of vestibular-postural stimulation on the visual system (see Jouen & Bloch, 1981, for a review). Placing newborns in an upright position increases the

VESTIBULAR AND VISUAL SYSTEM INTERACTIONS

285

duration of fixation on visual targets and induces an improved horizontal smooth pursuit (Frederickson & Brown, 1975; Gregg, Haffner, & Komer, 1976). Another example of the maturation of otolithic organs was found in a series of experiments about head-fighting responses (Jouen, 1984). In one experiment, we found that babies under 3 months, who do not yet have a muscular control of head and neck antigravity muscles, are, however, able to detect otolithic stimulation provided by a 25 degree lateral body tilting by compensating for body tilting with a continuous horizontal head rotation. This head-turning brings the head to exactly the same final horizontal position as in older infants reacting to body tilting via a vertical head-righting. Tonic neck and labyrinthine reflexes belong to the category of primitive reflexes involved in early equilibrium reactions (Capute et al., 1978, 1982). Tonic neck reflexes originate from proprioceptive receptors in the neck extensors and are found in neonate (Peiper, 1962). Tonic labyrinthine reflexes are closely connected to the tonic neck reflexes. They are thought to be mediated by the medial and lateral vestibulo-spinal tracts and the reticulo-spinal pathway with primary afferents in the otoliths and perhaps the neck extensors. Little is known about their appearance, strength, and disappearance in normal children. In cerebral palsied children, these reflexes have strong effects on the regulation of muscle tonus (Illingworth, 1978). They are involved in basic postural activities such as the change of body position (by body rolling) or the extension of the head when prone. The tonic labyrinthine reflex in prone position (TLP) has been systematically studied in 149 infants followed from birth up to 2 years of age by Capute et al. (1982). For a child held in prone suspension, the position of the limbs changes with respect to the position of the head in space and the orientation of the labyrinths. When the neck is extended by 45 degrees, the limbs extend, whereas they flex when the neck is flexed by 45 degrees. The TLP is present in 80% of infants at 2 weeks of age and persists throughout the first 18 months, with a maximum occurrence between 4 and 6 months. Stimulating vestibular and neck receptors by turning the infant's head affects the position of his/her arms and legs (Gesell, 1938; Peiper, 1962, 1963; Turkewitz, Gordon, & Birch, 1965): The limbs on the side toward which the face is turned will extend, and the limbs on the opposite side will be in flexion. This reflex is accompanied by a modification of the muscle tonus regulation: The side of the body where the face is turned is more tonic, whereas the opposite flexed limbs are less tonic. This pattern seems relatively rare in premature infants (Mellier & Jouen, 1985; Allen & Capute, 1986) but reaches a peak frequency around 6 to 8 weeks after birth (Coryell & Michel, 1978; Coryell & Cardinalli, 1979). This reflex usually disappears around the third month.

286

F. JOUEN & O. GAPENNE

The Moro reflex has been widely studied in preterm infants (Allen & Capute, 1986). As early as 25 weeks post-conception, a Moro reflex is observed. This vestibular reaction is observed in 80% of 30-week-old fetuses. Eviatar and Eviatar (1978) describe a reaction in neonates in response to a vertical acceleration that is similar to the Moro reflex. To summarize, from birth, the vestibular sensory system is able to tune postural and oculomotor activities involved in head-eye coordination and body spatial orientation relative to gravity. Moreover, different data suggest that vestibular sensitivity is not only peripheral but also central: Vestibular responses can be habituated in neonates and young infants (Omitz et al., 1979). This is important because experiments with animals have demonstrated that vestibular habituation requires a central coding of information relative to self-motion. In other words, the brainstem structures involved in the control of the vestibular system are mature and functional at birth (Stein & Meredith, 1993).

Early Vestibular-Visual Interactions Gibson (1966) coined the term visual proprioception to emphasize the role of vision in providing information for movement of the body in reference to interoceptive information provide by the mechanical and vestibular systems. According to Gibson, visual proprioception specifies self-motion via dynamic transformations of the optic array that occur when an observer moves through stable space (Butterworth, 1992). Many experiments have demonstrated during the last two decades that very young infants are sensitive to the optical information specifying self-motion. Lee and Aronson (1974) were the first to show that infants may use visual information to control their posture. Babies who had recently learned to stand on their own were tested on a motionless floor, within a moveable room. The infants faced the interior end wall, and the whole structure, except the floor, was moved so that the end wall approached or receded. Subjects compensated for a nonexistent loss of balance signaled by the optic flow pattern and fell in the direction of the optic flow. If the end wall moved away from the baby, the infant fell forward, and if the wall moved toward the baby, the infant fell over backward. Information from peripheral vision is particularly important for maintaining postural stability. Pope (1984) showed that movement in the center of the visual field did not result in significant postural adjustment, whereas a slight movement in the periphery was sufficient to induce a complete loss of stability. These results have been replicated by Bertenthal and coworkers (see Bertenthal, 1990, for a review), with infants aged between 12 and 15 months. Subsequent developmental studies demonstrated that vision does not acquire proprioceptive functioning as a

VESTIBULAR AND VISUAL SYSTEM INTERACTIONS

287

result of motor development. Butterworth and Hicks (1977) found that infants too young to walk would nevertheless compensate for visually specified motion when seated in a moving room. This research has been extended by Pope (1984) and Jouen (1986) to even younger infants. They investigated the role of proprioception in the control of stable head posture in 2-month-old infants. Infants too young to sit without external body support and incapable of independent locomotion were nevertheless able to make directionally appropriate compensatory postural movements in response to optic flow. Generating an optic flow by a sequentially displaced pattern of 15 electroluminescent diodes, Jouen (1988) and Jouen & Lepecq (1989) provided evidence for directionally appropriate head movements by newborn infants that suggests that visual proprioceptive control of posture is an early coordination available from birth. In this section, we shall focus on the origins of such a sensitivity in the neonate. Basic control of head movements in neonates will be examined in order to demonstrate the functional properties of the coupling between the visual and vestibular systems at birth. Visually induced modulation of spontaneous head oscillations will first be analyzed before we focus on the calibration of head motion to peripheral optical flow. Before presenting illustrative experimental findings obtained with static and dynamical optical information, we shall briefly summarize the rationale and techniques we used in the different experiments devoted to the study of the vestibular-collic system in newborn infants.

An Important Postnatal Control Parameter: Gravity The head can be considered as a more or less ovoid sphere that is, in the human mature organism, slightly tilted by 25 degrees compared to the gravitational vertical. This orientation has geometrical implications because it optimizes the position of the horizontal canal in the transverse plane and places the utriculus orthogonally to gravity. Movements of this spherical mass are clearly constrained by the mechanical properties of the head-neck joint. Because this joint is in a posterior position and thus is not aligned with the head mass center, the torque required to avoid head-forward falls is a function of the distance between these two points. The spatial lag between these two points constitutes a critical factor in head postural instability. Moreover, two other mechanical factors usually contribute to this instability in newboms. The first is the absence of a curvature at the cervical level, which reduces the support surface of the head. The second is the important inertia of the head, which results from the high mass ratio between head and body. It is also important to consider that beyond the active forces involved in head postural control, the organism must also manage passive forces such as gravity and inertia (e.g., Thelen et al. [1991] on kicking responses). Consequently, the

288

F. JOUEN & O. GAPENNE

functionally mature dorsal cervical extensors assume a critical role in the stabilization of the head, even if tonic activity must be conceived as a more global synergy between extensors and flexors. Beyond the structural and functional limitations of these muscles, adultlike forms of head postural control can be induced as early as the first days of life. Transient postural stabilization of the head over the trunk can be observed in newborns in a sitting position, even if the head is submitted to slow and large oscillatory movements. An infant pulled up from a supine to a sitting position can produce an alignment of the head-trunk set (AmielTison & Grenier, 1980). In a similar way, Grenier's clinical work (1981) has shown that in seated newborns, passive cervical support induces the progressive emergence of head stabilization, hypertony of the body axis, hypotony of the limbs, and reaching movements of the upper limbs toward objects placed in front of them.

Basic Neural Networks and Visual-Vestibular-Postural Coupling No anatomical study has demonstrated direct connections between the axons of the retinal ganglion cells and the motoneurons involved in the tonic activity of the dorsal cervical extensors. Consequently, modulation of muscular activity by vision is assumed to involve indirect neural tracts. However, the different converging inputs arising from sensory systems must have a similar dimension (Jouen, 1990). For example, inputs representing head velocity that originate from ~ e vestibular system must have the same coding dimension as the visual signal representing visual motion. This means that to be compared with vestibular signals, visual inputs do not require sophisticated pattern-analyzing properties. Visu~ pathways should only convey information about the direction and the velocity of motion in order to interact with vestibular input. Various neurons have been described in the central nervous system; they are related to peripheral vision and have appropriate direction and velocity coding to be used by the oculomotor system or to modulate the activity of the peripheral vestibular system (Stein & Meredith, 1993). Apart from the visual projection to the vestibular nuclei, brainstem structures that relay visual information are indirectly connected to the vestibular neurons. The accessory optical system (AOS) receives input from the retina and relays visual signals to the vestibulo-cerebellum (Simpson, Soodak, & Hess, 1979). AOS neurons have large receptive fields and respond optimally to textured patterns of 20 degrees. Most of the cells are sensitive to field movement between 0.05 and 1 degree/sec. Neurons responding over a similar velocity range have also been described in the rabbit retina. Single-cell recordings in the nucleus of the optic tract (NOT) of the pretectum have demonstrated a particular sensitivity of these cells to large and slow horizontally moving patterns. Precht and Cazin (1979) and Precht & Strata (1980) have shown that in rats and cats, lesions in the NOT eliminate or

VESTIBULAR AND VISUAL SYSTEM INTERACTIONS

289

modify the responses of central vestibular neurons to optokinetic stimuli. The distribution of preferred direction of AOS and pretectal neurons reveals an organization along spatial planes that coincides exactly with the orthogonal organization of the vestibular semicircular canals. As information in the vestibular system is also coded in vector components in these three planes, AOS and pretectal neurons would be expected to be directly involved in yisual-vestibular interactions. Neurons of the superior colliculus are also known to respond greater to moving visual stimuli than to stationary ones. Vestibular influences also have been demonstrated in the lateral geniculate body (Papaioannou, 1973). Another possible pathway consists of the cortical projections from primary visual areas to the pontine nuclei, as demonstrated by Brodal (1978). Single neurons in the pontine nuclei are movement- and direction-sensitive and have large receptive fields. Moreover, the fibers in the pontine nuclei project to the cerebellum and subsequently to the vestibular nuclei. All of these brainstem visual structures involved in the coding of the direction of and optical information specifying self-motion are known to be mature at birth (Banks & Salapatek, 1984). In other words, the structures involved in the control of visual vestibular interactions are mature and functional at birth (Stein & Meredith, 1993). This clearly suggests that visual proprioceptive control of head posture is a condition present at birth. In a recent study (Jouen et al., 1993), we underlined three functional aspects that stress the key role of head control in the coupling of perception-action behavior, space-environment coding and self-motion perception. First, the head contains most of the receptors, including visual as well as vestibular receptors. The perception of accurate and stable sensory inputs requires the static and dynamical control of the head-neck set in reference to space (Paillard, 1971). Of course, the receptors and their support are reciprocally linked because they contribute to head stability: Gain stability increases the coherence of sensory flow, and gain coherence increases head stability. Second, the cephalic segment plays a critical role in the postural chain and in the development of postural control. Head stability is a prerequisite for body stability. The head may be an essential reference in a top-down postural organization (Assaiante & Amblard, 1993). Third, the head contributes to the set of oriented behaviors and possesses an important teleonomic function (Paillard, 1971). This point is crucial: Orientation and stabilization are complementary (Berthoz, 1989).

Effect of Static Visual Stimulation on Spontaneous Head Activity As we have suggested before, head postural stability is fundamentally assumed by the tonic activity of the cervical dorsal extensors. This reflexlike activity is permanently modulated by direct sensory reafferences (somatic) and by less-direct tracts coming from central nervous structures such as the vestibular nuclei, which

290

F. JOUEN & O. GAPENNE

convey multisensory signals. We first hypothesized that the neonatal head-neck system's postural oscillations were modulated by the simple presence of a peripheral static and unstructured light source (Jouen & Gapenne, 1994). To test this hypothesis, we placed newborns (mean age: 4.5 days) in a specially designed infant seat that allowed them to remain in a seated position (reclined 25 degrees). They had a 14-inch monochrome monitor on both sides of their head. The monitors were parallel to each other and 31 cm apart. The whole apparatus was placed inside a dark experimental chamber. The response device was composed of two air bags orthogonally positioned along the occipital-temporal part of the skull. The pressure in the air bags was sensitive to any shift in the center of head mass. Head pressure was continuously monitored by two pressure transducers and was digitally sampled by a 12-bit A/D converter at a 60 Hz rate. The integration of left and right signals reflected the head movements along the anterior-posterior sagittal X-axis. An infrared video camera was also used for recording the infant's behavioral state. Based on this experimental setup, we compared neonate head oscillations in two experimental conditions (see Figure 1). In the first condition (upper part of Figure 1), the two monitors were switched off, and the newborns were in total darkness for the whole duration of the trial. In the second condition (lower part of Figure 1), monitors were switched on (white blank screens). As can be seen in Figure 1, low-frequency and high magnitude head oscillations increased when the light was switched off. The oscillatory postural behavior was then analyzed by the method of signal processing and, more specifically, by Fourier transformation. Classical spectral analysis revealed that head oscillations occurred in a very low-frequency band (see Figure 2). The power or the magnitude of each frequency band decreased as an inverse function of the frequency. This means that about 80% of the spectral energy belonged to the 0-0.5 Hz range. In other words, spontaneous head oscillations belonged to a very low-frequency band. This result indicates a remarkable comparability between spontaneous head oscillations in newborns and body oscillations in older infants and adults. Many experiments that have used a force plate for measuring postural sway in adults or infants (e.g., Powell & Dzendolet, 1984) have demonstrated that spontaneous body oscillations belong to the same low-frequency band (0-0.5 Hz). Our results clearly demonstrate that a functional multimodal control of head oscillation is present at birth: Head oscillations belong to a very low-frequency band, and the power of these low frequency oscillations increases when the light is off. These findings are similar to those found in older infants and adult subjects in erect postures. They also indicate that from birth, the vestibulo-collic tract behaves as a multimodal efferent system controlling head stability.

VESTIBULAR AND VISUAL SYSTEM INTERACTIONS

291

..................... :i ..................

, =,

.•

d

v fry or'll rr I~T'

11 time

22 (seconds)

.>4oo 2200 20oo (/)

1800 1600 1400 1200 1000 0

t-,,-

r~w,

11 time

22

(seconds)

FIGURE 1. Spontaneous head oscillations in neonates recorded in darkness (upper part of the graph) and in light (lower part of the graph) (from Jouen & Gapenne, 1994).

Peripheral Visual Movement and Head Control in Neonates Since the first observations of Wood (1895), it has become clear that the human postural system is sensitive to the movement of visual surroundings. When a static subject is submitted to a linear antero-posterior movement of the peripheral visual field, his or her body tilts backwards. Similarly, forward tilts are observed for a postero-anterior visual movement. The postural system is thus tied to the direction of visual flow. Psychophysical studies have described the

292

F. JOUEN & O. GAPENNE

Mean Relative Power Spectra from 0 to 12 Hz.

100

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

80

o~ 6o o o

Q. o _> .,...

r(0 O

E

40

20

0-0,5

1,5-2

3-3,5

4,5-5

6-6,5

7,5-8

9-9,5

10,5-11

frequency band (Hz) B B Darkness ~

Light

FIGURE 2. Mean relative power spectra of head oscillations in two experimental conditions (Light and Darkness). These spectra indicate that head oscillations are mainly of very low-frequency bands in both conditions (from Jouen & Gapenne, 1994).

VESTIBULAR AND VISUAL SYSTEM INTERACTIONS

293

characteristics of this visuo-postural coupling in adults according to geometry (angular or linear), profile (square or sinusoidal), direction, surface, or velocity of the visual input (see Andersen, 1986, for a review). As we have suggested above, the modulation of synergic muscular activity by visual inputs could essentially be performed by subcortical neural networks that are clearly mature at birth. Variations of the direction and velocity of visual flow that are encoded in these networks (e.g., the neural loop retina-AOS-vestibular nuclei-spinal motoneurons) differentially modulate the postural system. Of course, this does not exclude the participation of cortical structures as early as the first days of life. Although we have already shown that the sensitivity to optical flow is present at birth in human infants (Jouen, 1988; Jouen & Lepecq, 1989), a recent study demonstrates that this coupling does not function in an "all or none" way, but is truly modulated by both the direction and velocity of the visual flow (Jouen, Lepecq, & Gapenne, 1995). Mean amplitude of head response increases according to the increase in flow velocity. In other words, the gain of the neonatal visuo-postural coupling is linear

and is approximately equal to 1 (see Figure 3).

14

~ ,,.., E

12

:s (D

10

~

8

~

4

D_

j

2

4

6

8

10

12

i

i

l

l

f

w

I

I

I

I

4

6

8

10

12

.S 2

L

Incremental

Field

Flow

Velocity

14

/

FIGURE 3. Head movement amplitude as a function of the increment of the visual flow velocity: observed (dots) and theoretical (line) gain in neonates (from Jouen, kepecq, & Gapenne, 1995).

294

F. JOUEN & O. ~3APENNE

These results indicate that this sensitivity is clearly modulated by the directional and kinematic properties of optical flow. Mean head pressure is direction-specific. When exposed to backward flow, newborns react by a backward leaning of the head. Impressively, mean head pressure is also velocity-sensitive. The magnitude of the head pressure significantly and linearly increases with the velocity of optic flow. The neonates' head postural responses are remarkably consistent with global postural reactions exhibited later on in development. Indeed, the backward leaning of the head in response to backward flow is virtually identical to the backward leaning of the whole body observed under similar conditions in older infants, children, and adults (see Bertenthal, 1990).

Adults Gain / I1) "o c

0,2

0,0 7

0,6

0,4

i

0,8

i

i

,,,

1,0

7

6

6

2~ t-

s9

5

8- -

4

t'~

.,..,

~_

o

a

E

2

J

J

-,j

f

t

J

4

3

2

o

1 0,O

1 0,2

I 0,4 Incremental

I 0,6 Field Flow Velocity

ii

i

i

0,8

1,0

1

i

FIGURE 4. Ankle rotation amplitude as a function of the visual flow velocity' observed (dots) and theoretical (line) gain in adults (from Lestienne et al., 1977). As can be seen in Figure 4, the variation of neonates' head response magnitude with flow velocity is virtually identical to global postural reactions observed in adults, in whom the magnitude of body leaning has been shown to linearly increase with optic flow velocity up to 1.1 m/sec (Lestienne, Berthoz, Mascot, & Koitcheva, 1976; Lestienne, Soechting, & Berthoz, 1977). Some of the major characteristics of global postural reactions in response to optic flow observed

VESTIBULAR AND VISUAL SYSTEM INTERACTIONS

295

throughout infancy, childhood, and adulthood are thus present as soon as birth at a segmentary postural level. The present findings thus suggest that learning is not necessary for the emergence of optic flow sensitivity, although experience, and particularly self-produced segmental and global mobility, may play a role in the subsequent development of visual postural coupling (Gibson, 1966; Bertenthal, 1990).

Conclusion As Stein & OguztOreli (1978) have proposed, the integration of signals from different sensory receptors contributes to the synchronization of muscular activity and hence to postural regulation. From this point of view, the redundancies between vestibular and visual inputs have a critical role in postural control and self-motion perception. In the first part of this chapter, we proposed that from birth, the peripheral vestibular organs deal with the resultant of gravity forces and physical accelerations. The presence of a simple peripheral functional organization possibly induces early adapted postural behaviors involved in the perceived self. The complexity of the central vestibular projections also suggests that the vestibular system does not work independently of other sensory systems. It has long been shown that the intensity, spatial and temporal resolution, distance, position in visual space, etc., of a light source are many factors that modulate postural stability. The role of visual input in postural regulation can be further demonstrated by its elimination. For instance, frequency analysis has shown that visual suppression induces a significant increase of the spectral power in the very low-frequency band (F< .05 Hz) in neonates as well as in older infants and adults. How can this coupling between visual and postural systems be considered as a basic dynamical mechanism? We must remember that since the origins of life, light has played a critical role in the relationships between plants or animals and their environment. For example, in the unicellular Euglena viridis, the trajectory of its displacement is already modulated by the presence of a light source, which activates the eye spot or stigma. Since early in evolution, light has oriented organisms and led them to organize their motor and postural systems as a function of light properties. As the complexity of optical systems and the differentiation of nervous systems developed, more complex properties of visual inputs (direction and velocity) were taken into account. As shown by the works of GOtz (1968) and Reichardt and Poggio (1976), the wild type Drosophila (which possesses an optic ganglion) produces an optomotor response during flying, when exposed to a rotational visual field. In other words, an insect's postural adjustments can be visually driven. More

296

F. JOUEN & O. GAPENNE

recently, the contribution of genetic mechanisms to this visuo-postural behavior have begun to be studied (reviewed in Ferrus & Canal, 1994). This type of response, as suggested by Owen and Lee (1986), has been preserved during phylogenesis because it can be measured in pigeons (Friedman, 1975), ducks (Butterworth & Henty, 1991), human infants, and adults (Lishman & Lee, 1973; Lee & Aronson, 1974). Butterworth and Henty's results on postural responses in newborn domestic ducks also suggested that the functioning of this visuo-postural coupling does not require any great behavioral experience. To use the terms of Bullinger (1991), the visual postural interaction may correspond to biologically determined adaptations (see also Berthoz, 1993; Yardley, 1992). The results of our studies suggest that some functional properties of the visuo-postural coupling are maintained or preserved during ontogenesis as they have been through phylogenesis. Thus, learning may not be necessary for the emergence of optic flow sensitivity. However, different important points must be made to avoid misunderstanding. First, this suggestion does not mean that the neural networks that generate these functions have stayed similar during natural (phylogenesis) or individual (ontogenesis) history. Second, it must not be considered as a nondevelopmental position because during development these basic functionings coalesces. What about the developmental function of this visual-vestibular-postural coupling? It has been proposed that the visual movement surrounding organisms affects their postural systems mainly during the transition phase. For instance, Lee and Aronson (1974) and Pope (1984) showed that postural deviations under peripheral visual flow decreased according to the postural maturation of new walkers. In a more global study based on electromyographic and postural cues, Sveistrup and coworkers (1992) confirmed this developmental fact. Vision should thus serve as a calibrating function of perception-action coupling when new postural abilities emerge and when body-environment relationships are drastically modified. This suggests that if learning is not necessary for the emergence of optic flow sensitivity, experience m and particularly self-produced segmental and global mobility m may however play a critical role in the subsequent development of visual postural coupling. Given all these facts, we consider that visual-vestibular-postural coupling has become one of the most fascinating problems in the perception-action domain and gives rise to a number of interesting models: dynamical (Sch0ner, 1991), ecological (Koenderinck, 1986; Lee, 1993; Turvey & Carello, 1986), and computational (Perrone, 1992). Regardless of the heuristical importance of these models, the demonstration of optic flow sensitivity in neonates is fundamental for the understanding of the perception of self. Visual proprioception specifying selfmovements and hence optic flow sensitivity is sufficient to specify the distinction

VESTIBULARAND VISUALSYSTEM INTERACTIONS 297 between "self' and "the world." Our results clearly show that some of the major characteristics of global postural reactions that allow discrimination of the self from the environment are present from birth.

ACKNOWLEDGMENT

The authors thank Sylvie Margules for her helpful reading of the manuscript. REFERENCES

Akbarian, S., Grtisser, O.J., & Guldin, W.O.(1992). Thalamic connections of the vestibular cortical fields in the squirrel monkey (saimiri sciureus). The Journal of Comparative Neurology, 326, 423--441. Allen M.C., & Capute A.J.(1986). The evolution of primitive reflexes in extremely premature infants. Pediatric Research, 20, 1284-1289. Amiel-Tison, C., & Grenier, A. (1980). Evaluation neurologique du nouveau-n( et du nourrisson. Paris: Masson. Andersen, G.J. (1986). Perception of self-motion: Psychophysical and computational approaches. Psychological Bulletin, 99(1), 52-65. Andr6-Thomas, A.S., & Ajuriaguerra, J. (1948). L'axe corporel, musculature, et innervation. Paris: Masson. Anniko, M. (1983). Embryonic development of vestibular sense organs and their innervation. In R. Romand (Ed.), Development of auditory and vestibular systems (pp. 375-423). New York: Academic Press. Assaiante, C., & Amblard, B. (1993). Ontogenesis of head stabilization in space during locomotion in children: Influences of visual cues. Experimental Brain Research, 93, 499-515. Banks, M. S., & Salapatek, P. (1984). Infant visual perception. In P.H. Mussen (Ed.),

Handbook of Child Psychology. Vol. H: Infancy and Developmental Psychobiology (pp. 435-471). New York: John Wiley & Sons. Bast T.H., & Anson, B.J. (1949). Temporal Bone and the Ear. IL: Thomas Springfield. Benson, A.J. (1982). The vestibular sensory system. In H.B. Barlow & J.D. Molton (Eds.), The Senses: Cambridge texts in physiological sciences (pp. 333-368). Cambridge: Cambridge University Press. Bergstr6m, B. (1973). Morphology of the vestibular nerve. I. Anatomical studies of the vestibular nerve in man. Acta of Otolaryngology, 76, 162-172. Bertenthal, B.I (1990). Application of biomechanical principles to the study of perception and action. In: H. Bloch & B.I. Bertenthal (Eds.), Sensorimotor organizations and development in infancy and early childhood (pp. 243-264). Dordrecht: Kluwer Academic Publishers. Bertenthal, B.I, & Bai, D.L. (1989). Infants' sensitivity to optical flow for controlling posture. Developmental Psychology, 25, 936-945. Berthoz, A. (1989). Coop6ration et substitution entre le syst~me saccadique et les "r6flexes" d'origine vestibulaire" faut-il reviser la notion de "r~flexe." Revue Neurologique, 145, 513-526. Berthoz, A. (1993). Leqon Inaugurale au Coll~ge de France. Chaire de Psychophysiologie de la Perception et de l'Action.

298

F. JOLYEN& O. GAPENNE

Berthoz, A., Yoshida, K., Vidal, P.P, & Mc Crea, R. (1981). Horizontal eye movement sensitivity in second order vestibular neurons in the cat. In B. Cohen (Ed.), Vestibular and Oculomotor Physiology (pp. 144-157). New York: Annals of the New York Academy of Sciences (Vol. 175). Brandt, T., Dieterich, M., & Danek, A. (1994). Vestibular cortex lesions affect the perception of verticality. Annals of Neurology, 35(4),403-412. Brodal, P.(1978). The cortico pontine projections in the rhesus monkey. Origin and principles of organization. Brain, 101,251-283. Bullinger, A. (1991). Vision, posture et mouvement chez le b6b6: Approche developpementale et clinique. In F. Jouen & A. Henocq (Eds.), Du nouveau-nd au nourrisson (pp. 48-61). Paris: PUF. Butterworth, G. (1992). Origins of self-perception in infancy. Psychological Inqui~., 3(2), 103-111. Butterworth, G., & Hicks, L. (1977). Visual proprioception and postural stability in infancy: A developmental study. Perception, 6, 255-262. Butterworth, G., & Henty C. (1991). Origins of the proprioceptive function of vision: Visual control of posture in one day old domestic chicks. Perception, 20, 381386. Capute, A.J., Accardo, P.J., Vining, E.P., Rubenstein, J.E., Walcher, J.R., Harryman, S. & Rose, A. (1978). Primitive reflexes profile: A pilot study. Physiological Therapy, 50, 1061-1065. Capute, A.J., Watchel, R.C., Palmer, F.B., Shapiro, B.K. & Accardo, P.J. (1982). A prospective study of three postural reactions. Developmental Medicine and Child Neurology, 24, 314-320. Coryell J.F, & Michel G. (1978). How supine preferences of infants can contribute toward the development of handedness. Infant Behavior and Development, 1, 245257. Coryell J.F, & Cardinalli, N. (1979). The asymmetrical tonic neck reflex in normal fullterm infants. American Journal of Occupational Therapy, 12, 225-260. Curthoys, I.S. (1979). The development of function of horizontal semicircular canal primary neurons in the rat. Brain Research, 167, 41-52. Curthoys, I.S. (1983). The development of function of vestibular primary neurons. In R. Romand (Ed.), Development of auditory and vestibular systems (pp. 425-457). New York: Academic Press. Dayal, V.S., Farkashidy, J., & Kokshanian, A. (1973). Embryology of the ear. Canadian Journal of Otolaryngology, 2, 136-142. Deschesne, C., & Sans, A. (1985). Development of vestibular receptor surfaces in human fetuses. American Journal of Otolaryngology, 6, 378-387. Dichgans, J., & Brandt T. (1978). Visual-vestibular interaction: Effects on self-motion perception and postural control. In R. Held, H. Leibowitz & H.L. Teuber (Eds.), Handbook of Sensory Physiology, Vol. VIII, Perception. New York: Springer Verlag. Edelman, G.M. (1992). Bright air, brilliant fire: On the matter of mind. New York: Basic Books. Elliot, G. B., & Elliot, K. A. (1964). Some pathological, radiological, and clinical implications of the precocious development of the human ear. Laryngoscope, 74,1 160-1171. Engel, A.K., K6nig, P., Kreiter, A.K., Schillen, T.B., & Singer, W. (1992). Temporal coding in the visual cortex: New vistas on integration in the nervous system. Trends in Neurosciences, 15, 218-226.

VESTIBULARAND VISUALSYSTEMINTERACTIONS 299 Engel, A.K., & K6nig, P. (1993). Paradigm shifts in neurobiology: Toward a new theory of perception. In R. Casati and G. White (Eds.), Philosophy and Cognitive Sciences (pp. 131-138). Wien: Austrian Ludwig Wittgenstein Society. Eviatar, L., & Eviatar, A. (1978). Neurovestibular examination of infants and children. Advances in Otorhinolaryngology, 23, 169-191. Ferrus, A., & Canal, I. (1994). The behaving brain of a fly. Neurosciences, 17, 479486. Frederickson W.T., & Brown, J. (1975). Posture as determinant of visual behavior in newborns. Child Development, 46, 579-582. Friedman, M.B. (1975). Visual control of head movements during avian locomotion. Nature, 255, 67-69. Gesell, A. (1938). The tonic neck reflex in human infants. Journal of Pediatrics, 13, 455-464. Gerrits, N.M. (1990). Vestibular nuclear complex. In The human nervous system (pp. 863-888). New York: Academic Press. Gibson, J.J (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin. Gibson, J.J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. G6tz, K.G. (1968). Flight control in Drosophila by visual perception of motion. Kybernetic, 4, 199-208. Gregg, C., Haffner, M.E., & Korner, A.L. (1976). The relative efficacy of vestibular proprioceptive stimulation and uptight position in enhancing visual pursuit in neonates. Child Development, 47, 309-314. Grenier, A. (1981). La "motricit6 lib6r6e" par fixation manuelle de la nuque au cours des premieres semaines de la vie. Archives Frangaises de P~diatrie, 38, 557-561. Guedry, E.E. (1974). Psychophysics of vestibular sensation. In H.H. Kornhuber (Ed.), Handbook of Sensory Physiology (pp. 5-154). New York: Springer. Guldin, W.O., Akbarian, S., & Grtisser, O.J. (1992). Cortico-cortical connections and cytoarchitectonics of the primate vestibular cortex: A study in squirrel monkeys (Saimiri sciureus). The Journal of Comparative Neurology, 326, 375-401. Hooker, D. (1952). The prenatal origin of behavior. Lawrence, KS: Kansas University Press. Illingworth, R.S. (1978). Le D~veloppement psycho-moteur de l'enfant. Paris: Masson. Ikegami, H., Sasaki, M., & Uchino, Y. (1994). Connections between utricular nerve and neck flexor motoneurons of decerebrate cats. Experimental Brain Research, 98, 373-378. Jouen, F. (1984). Visual-vestibular interactions in infancy. Infant Behavior and Development, 7, 135-145. Jouen, F. (1986). La contribution des r6cepteurs visuels et labyrinthiques h la detection des mouvements du corps propre chez le nourrisson. Ann~.e Psychotogique, 86, 169-192. Jouen, F. (1988). Visual-proprioceptive control of posture in newborn infants. In B. Amblard, A. Berthoz, & F. Clarac (Eds.), Posture and gait: Development, adaptation and modulation (pp. 59-65). Amsterdam: Elsevier. Jouen, F. (1990). Early visual vestibular interactions and postural development, In H. Bloch & B.I. Bertenthal (Eds.), Sensory motor organization and development in infancy and early childhood (pp. 199-215). Dordrecht: Kluwer. Jouen, F., & Bloch, H. (1981). Le r61e des informations visuelles dans les premiers contr61es posturaux, Annde Psychologique, 81, 197-221. Jouen, F., & Lepecq, J.C. (1989). La sensibilit6 au flux optique chez le nouveau-n6. Psychologie Frangaise, 34, 13-18.

300

F. JOUEN & O. GAPENNE

Jouen, F., & Gapenne, O. (1994). Rtgulation de la posture ctphalique chez le nouveaunt. Revue d'Oto-Neuro-Ophtalmologie, 28(1), 28-34. Jouen, F., Lepecq, J.C., & Gapenne, O. (1993). Frames of reference underlying movement coordination in infants. In G.J.P. Savelsbergh (Ed.), The Development of Coordination in Infancy (pp. 237-263). Amsterdam: Elsevier Science Publisher. Jouen, F., Lepecq, J.C., & Gapenne, O. (in press). Optical sensitivity in neonates.

Child Development.

Koenderinck, J.J. (1986). Optical flow. Vision Research, 26, 161-180. Lacour, M., & Borel, L. (1993). Vestibular control of posture and gait. Archives Italiennes de Biologie, 131, 81-104. Lavigne-Rebillard, M., Deschesne, C., Pujol, R., Sans, A., & Escudero, P (1985). Dtveloppement de l'oreille interne pendant le premier trimestre de la grossesse. Difftrenciation des cellules sensorielles et formation des premibres synapses. Annales d'Oto-Laryngologie (Paris), 102, 493--498. Lee, D.N. (1993). Body-environment coupling . In U. Neisser (Ed.), The perceived self: Ecological and interpersonal sources of self-knowledge (pp. 43-67), Cambridge, MA: Cambridge University Press. Lee, D.N., & Aronson, E. (1974). Visual postural control of standing human infants. Perception and Psychophysics, 15, 529-532. Lestienne. F., Soechting, J., & Berthoz A. (1977). Postural readjustments induced by linear motion of visual scenes. Experimental Brain Research, 28, 363-384. Lishman, J.R., & Lee, D.N. (1973). The autonomy of visual kinaesthesis. Perception. 2, 287-294. Malsburg, C. von der, & Singer, W. (1988). Principles of cortical networks organization. In P. Rakic & W. Singer (Eds.), Neurobiology of the Neocortex (pp. 69-99). New York: Wiley. Mellier, D., & Jouen, F. (1985). Postural activities as behavioral organization cues in preterm infants. Paper presented at the 8th Congress of ISSBD, Tours, France. Neisser, U. (1993). The perceived self" Ecological and interpersonal sources of selfknowledge. Cambridge MA: Cambridge University Press. Ornitz, E. M., Atwell, C. W., Walter, D. O., Hartmann, E. E., & Kaplan, A. R.(1979). The maturation of vestibular nystagmus in infancy and childhood. Acta Otolaryngology, 88, 244-256. Owen, B.M., & Lee, D.N. (1986). Establishing a frame of reference for action. In M.G. Wade and H.T.A. Whiting (Eds.), Motor development in children: Aspects of coordination and control (pp. 287-308). Dochdrecht: Martinus Nijoff Publishers. Paillard, J. (1971). Les dtterminants moteurs de l'organisation spatiale. Cahiers de Psychologie, 14, 261-316. Papaioannou, J. (1973). Electrical stimulation of vestibular nuclei: Effects on spontaneous activity of lateral geniculate nucleus neurons. Archives Italiennes de Biologie, 11, 217-233. Peiper A. (1962). Rtflexes de posture et de mouvements chez le nourrisson. Revue de Neuropsychiatrie Infantile, 10, 411-530. Peiper A. (1963). Cerebral function in infancy and childhood. New York: Consultants' Bureau. Perrone, J.A. (1992). Model for the computation of self-motion in biological systems. Journal of the Optical Society of America, 9, 177-194. Pope, J.M. (1984). Visual proprioception in infant postural development. Doctoral Dissertation, University of Southampton.

VESTIBULARAND VISUALSYSTEM INTERAC'I'IONS 301 Porton, I., Jacquet, A.Y., & Lecanuet, J.P. (1994, June). Fetal response to maternal rocking and tilting. Paper presented at the International Conference for Studies in Infancy Meeting, Paris. Powell, G.M., & Dzendolet, E. (1984). Power spectral density analysis of lateral human standing sway. Journal of Motor Behavior, 16, 424-441. Precht, W., & Cazin, L. (1979). Functional deficits in the opto-kinetic system of the albino rat. Experimental Brain Research, 37, 183-186. Precht, W., & Strata, P. (1980). On the pathway mediating optokinetic responses in vestibular nuclear neurons, Neuroscience. 5, 777-787. Previc, F.H. (1991). A general theory concerning the prenatal origins of cerebral lateralization in humans. Psychological Review, 98, 299-334. Reichardt, W., & Poggio, T. (1976). Visual control of orientation behavior in the fly. Part I. A quantitative analysis. Quarterly Reviews of Biophysics, 9, 311-375. Rochat, P. (1993). Connaissance de soi chez le b~b~. Psychologie Fran~aise, 338 (1), 41-51. Rosenhall, U. (1972). Vestibular macular mapping in man. Annals of Otorhinolaryngology, 81, 339-351. Sans, A., Pujol, R., & Marty, R. (1968). Etude du r6flexe de redressement dans la p~riode post-natale chez divers mammif~res. Psychologie Franfaise, 13, 351353. Sch6ner, G. (1991). Dynamic theory of action-perception patterns: The "moving room" paradigm. Biological Cybernetic, 64, 455-462. Simpson, J. I., Soodak, R. E., & Hess, R. (1979). The accessory optic system and its relation to the vestibulo-cerebellum. Prog Brain Research, 5, 715-724. Stein, B.E., & Meredith, M.A. (1993). The merging of senses. Cambridge, MA: MIT Press. Stein, R.B., & Oguz6reli, N.M. (1978). Reflex involvement in the generation and control of tremor and clonus. In J.E. Desmeth (Ed.), Progress in Clinical Neurophysiology: Physiological tremor, pathological tremor and clonus, Vol. V (pp. 28-50). New York: Karger. Sveistrup, H., Foster, E.C., & Woollacott, M.H. (1992). Changes in the effect of visual flow on postural control across the lifespan. In M. Woollacott & F. Horak (Eds.), Posture and gait: Control mechanisms, Vol. H (pp. 224-227). Eugene, OR: University of Oregon Books. Thelen, E., & Smith L. (1994). A Dynamic system approach to the development of cognition and action. Cambridge MA: MIT Press. Thelen, E., Jensen, J.L., Kamm, K., Corbetta, D., Schneider, K, & Zernicke, R.F. (1991). Infant motor development: Implications for motor neuroscience. In G. Stelmach & J. Requin (Eds.), Tutorials in Motor Neuroscience. Dordrecht: Kluver Academic Press. Turkewitz G., Gordon E., & Birch H. (1965). Head turning in the human neonate: Spontaneous patterns. Journal of Genetic Psychology, 107, 143-158. Turvey, M.T., & Carello, C. (1986). The ecological approach to perceiving-acting: a pictorial essay. Acta Psychologica. 63, 133-155. Varela, F.J., Thompson, E., & Rosch, E. (1991). The embodied mind: Cognitive science and human experience. Cambridge, MA: MIT Press. Wood, R.W. (1895). The "haunted swing" illusion. Psychological Review, 2, 277278. Wyke, B.(1975). The neurological basis of movement. A developmental review. In K. Holt (Ed.), Movement and Child Development. London: SIMP. Yardley, L. (1992). Motion sickness and perception: A reappraisal of the sensory conflict approach. British Journal of Psychology, 83, 449-471.

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

303

CHAPTER 15

Two Modes of Perceiving the Self BENNETF I. BERTENTHAL and JAMES L. ROSE

University of Virginia

The focus of this chapter is on the visual perception of self by infants. For reasons that will become apparent, it behooves us to begin with a clarification of what is meant by visual perception. The most common response by students, colleagues, and the lay public is that visual perception is concerned with recognition and identification behaviors that are certainly relevant to the perception of self. As adults, we are quite adept at recognizing pictures and other representations of our faces, body images, gait patterns, etc. The empirical evidence suggests that visual recognition of self emerges at a relatively young age, and thus this behavior has played a central role in the study of the early development of self. Visual perception of self is not, however, limited to recognition. One of the principal contributions of Gibson's theory (1966, 1979) is its claim that vision also functions proprioceptively as an integral component in the perception of selfmotion and the control of actions. A number of recent studies on visual control of postural stability suggest that infants perceive self-motions in the perspective transformations of the optical array long before they recognize themselves (Bertenthal & Bai, 1989; Butterworth & Hicks, 1977; Lee & Aronson, 1974). These findings are especially provocative because they challenge the traditional view of propriocepfion as restricted to somatosensory, kinesthetic, and vestibular information. Indeed, some of the most recent evidence suggests that infants are sensitive to the proprioceptive information in vision even before they are sensitive to analogous information from muscles, joints, and skin receptors (Woolacott & Sveistrup, 1994). In this chapter, we review evidence from a number of innovative paradigms showing that infants visually perceive proprioceptive and exteroceptive information about self. A question that is central to our exposition is whether these two perceptual functions are developmentally dissociable. Following Gibson (1979), a number of researchers consider proprioceptive and exteroceptive sources

304

B.I. BERTENTHAL •

J.L. ROSE

of information as two sides of the same coin, and implicitly assume that perception of self and object are mediated by the same perceptual process. We take exception to this point of view for reasons that are outlined in the next section describing evidence for two dissociable visual pathways. This position leads us to caution researchers from sugggesting that different forms of self-perception are developmentally continuous. As an alternative, we propose that recognition and representation of the self follow a distinct developmental trajectory that is different than the trajectory followed by visual proprioception.

Dissociable Pathways for Processing Visual Information It is commonly assumed that the traditional function of vision is to provide a three-dimensional representation of the world from the two-dimensional retinal mosaic of changing luminances and wavelengths (e.g., Man, 1982). Although many researchers (e.g., Livingstone & Hubel, 1987; Schiller, Logothetis, & Charles, 1990) now propose modular processing of different dimensions of sensory information, such as luminance, color, and motion, it is still expected that these different pathways will eventually converge in a unified representation that makes contact with both thought and action. The implication of this thesis is that both thought and action are mediated by the same perceptual information. It is no longer clear, however, that this position is consistent with current findings on the structure and function of the visual system. In this section, we review evidence suggesting that the visual system is parsed into two functionally dissociable pathways. One pathway is concerned with the recognition and representation of the visual world, whereas the other pathway is concerned with the visual control of different motor responses, such as ocular tracking, reaching, and grasping. From an evolutionary perspective, the principal function of vision is to ensure distal control of actions (Goodale, 1983). Natural selection operates at the level of survival and is concerned primarily with how well the animal forages for food, avoids predators, finds mates, etc. (Milner & Goodale, 1993). Research on nonmammalian vertebrates reveals that many of the visual control systems mediating behavior are subserved by functionally independent neural substrates. For example, Ingle (1983) showed that for frogs, visually elicited feeding and guidance around barriers are mediated by separate visual modules, each with separate input and output pathways. More generally, the evidence suggests that visuomotor mechanisms in nonmammalian vertebrates are modularly organized and function independently (Goodale, 1988). Of course, the mammalian visual system is more complex and organized differently. Nevertheless, recent evidence (discussed below) suggests that the modular organization of visuomotor functions seems to

TWO MODES OF PERCEIVING THE SELF

305

have been preserved and dissociated from those functions concerned with the recognition and representation of the visual world. In a seminal paper on the structural organization of the visual system, Ungerleider and Mishkin (1982) distinguished two general pathways following different projections from the primary receptive area of the visual cortex in primates. The dorsal stream projects primarily to areas within the posterior parietal cortex, and the ventral stream projects primarily to areas within the inferior temporal cortex. A series of behavioral studies on monkeys with lesions in either the posterior parietal or inferotemporal cortex suggested that the ventral stream was dedicated toward processing information concerned with the object's qualities ("what" is the objec0, whereas the dorsal stream was concerned with processing information about the spatial location of the object ("where" is the object). From the perspective of hindsight, it is significant that so little attention was directed toward the striking differences in tasks used to measure performance of the dorsal and ventral pathways. The implicit assumption was that the locus of the differences was on the input side of processing, and thus the output side would depend primarily on whether or not the necessary information was processed. This functional dissociation in visual processing continues to be widely cited as a valid division of labor by the visual system, but recent evidence suggests that the anatomical division proposed by Ungerleider and Mishkin (1982) might be more profitably conceptualized as a "what" and "how" system. Milner and Goodale (1993) suggest that the dorsal stream should be extended to include areas in the premotor and prefrontal cortex in order to account for a set of semi-independent modules for the on-line control of actions. By contrast, the ventral stream processes similar visual information but uses this information for the off-line control of visual recognition and representation. An extensive review by Milner and Goodale (1993) of the neurophysiological correlates of the dorsal and ventral streams in monkeys suggests that the ventral pathway plays a central role in perceptual identification of objects, while the dorsal pathway mediates the sensorimotor transformations required for visually guided actions directed at those objects. These two pathways process much of the same information. For example, both pathways process information about size, orientation, and shape, and probably information about spatial relations, including depth. The difference is in how the information is used. In general, the ventral stream is concerned with the enduring characteristics of objects that are stored in memory so that they can be recognized when seen again from the same or different vantage points. To accomplish this task, it is essential that objects be perceived within an object-centered coordinate system so that changes in size, shape, color, lighmess, and location do not disrupt recognition. The situation is quite different for visuomotor mechanisms that are mediated by the

306

B.I. BERTENTHAL •

J.L. ROSE

dorsal stream. The function of this system is to support a nesting of skilled behaviors, such as visually guided reaching and grasping, which demand considerable coordination between movements of the fingers, hands, arms, head, and eyes. Thus, it is necessary that objects are perceived within a viewer-centered frame of reference because the location, size, and orientation of the objects must be encoded relative to the observer. These relations between viewer and object tend to change continuously, which implies that so does the visual coding within this pathway. In sum, the tasks mediated by these two systems will differ with regard to the role of past experience and visuomotor coordination. Thus, for example, the ventral system informs you that the object that you are now holding in your hand matches your representation of a book; the dorsal system ensures that you are able to grasp the book correctly and position it so that it can be read. Although the preceding evidence is based on studies with monkeys, it appears that a similar functional dissociation in visual processing applies to humans as well. Some of the most compelling evidence derives from patients who have suffered damage to one of the two perceptual systems but not to the other. For example, patients with optic ataxia have suffered damage to the posterior parietal region. These patients show deficits in positioning their fingers or orienting their hand when reaching for an object (Perenin & Vighetto, 1988). Similarly, these patients may show difficulties in adjusting their grasp to the size of an object that they intend to pick up (Damasio & Benton, 1979). Recently, Jakobson, Archibald, Carey, and Goodale (1991) conducted a quantitative study of grasping movements in a patient who showed profound deficits in spatial attention, gaze, and visually guided reaching. Although this patient revealed no difficulty in recognizing line drawings of common objects, her grasp was poorly scaled to the size of the object: She often opened her hand as wide for small objects as for large objects. Taken together, these studies suggest that damage to the dorsal pathway (i.e., parietal lobe) impairs the sensorimotor transformations necessary for adjusting reaching and grasping to the size, shape, and orientation of objects, even though this same information is used successfully for recognizing and describing objects. The phenomenon of "blindsight" provides complementary evidence for a proposed functional dissociation in the human visual system. Over two decades ago, Weiskrantz, Warrenton, Sanders, and Marshall (1974) reported that some patients with striate lesions were capable of reaching out and touching objects even though they reported no perception of these objects. More recently, Goodale, Milner, Jakobson, and Carey (1991) conducted a systematic series of studies on a patient, DF, with visual agnosia. In one experiment, DF showed that she was capable of adjusting the orientation of her hand and wrist in anticipation of placing a hand-held card through a slot, even though she could not judge the orientation of

TWO MODES OF PERCEIVING THE SELF

307

the slot either verbally or manually. DF also showed that she was capable of scaling the distance between her finger and thumb in preparation to grasp solid objects of different dimensions, even though she was unable to judge the width of these same objects. These findings converge with those from patients with visual ataxia to suggest a functional dissociation in the processing of visual information. Contrary to the earlier proposal by Ungerleider and Mishkin (1982), object properties are processed by both pathways. The difference resides in the ventral pathway using this information for object recognition and categorization tasks, and the dorsal pathway using this information for controlling the responses of motor behaviors so that they are tuned to the specific dimensions of the object. Currently, the evidence for this dissociation in the human visual system is still piecemeal, and much additional empirical research is necessary to validate specific predictions. Nevertheless, this dissociation is consistent with other psychophysical evidence (Goodale, 1988), as well as developmental evidence (Bertenthal, 1993), and for this reason the functional implications of this dissociation for interpreting perceptual research seem noteworthy and important. In a recent paper, one of us (Bertenthal, 1993) proposed that this dissociation in visual processing is not restricted to object perception and manipulation, but extends to other categories of knowledge besides objects, such as people, events, and even written words, as well as to other skilled actions, such as posture and gait. A similar proposal was formulated by Neisser (1989) in an effort to show that the ecological aproach to visual perception was concerned with very different issues than cognitive science approaches. Both proposals share some similar claims regarding the processing differences between the two visual systems. The first concerns the way in which past experience is used. The object recognition system (which is mediated by the ventral pathway) includes processes that make contact with information perceived at some prior time and stored in some representational form. Successful recognition will depend on both how the optic array is parsed and on the representational format of the stored information. As this information becomes stored in more abstract forms, recognition will generalize to a broader class of objects. (For example, lions, tigers, goats, and sheep will all be recognized as animals.) By contrast, the perception and control of actions (which is mediated by the dorsal pathway) is directed toward present information, and if anything, includes a prospective view toward new information. This future-oriented perspective is necessary because the inertia of body segments demands that individuals anticipate and plan for changes in coordinated movements in order to ensure smooth and continuous actions (Hofsten, 1993; Lee, 1993). The second difference is related to the first and involves the representational format. Visual information is represented by the recognition system in a form that

308

B.I. BERTENTHAL •

J.L. ROSE

preserves modality-specific information. This provision is necessary to allow for recall of specific featural information, such as the image of a face or the sound of a person's voice. Intermodal information is represented in the form of spatial and temporal correspondences between different sensory channels, but this representation is not meant to replace the modality-specific representation. By contrast, visual information is represented by the perceptual control system in an amodal format. This format is necessary to ensure that all sensory information is transformed into the same body-scaled information necessary for modulating the forces of the body to perform coordinated movements. For example, posture is specified by proprioceptive, vestibular, and visual information (Lishman & Lee, 1973). When an individual detects that support is perturbed, it is not important to determine which sensory input channel specified this loss of balance. The goal is simply to restore equilibrium, and this involves scaling the compensatory forces to the perceived displacement. A final difference concerns the observer's awareness of the perceptual information. Recent discussions on the difficulty of defining conscious processing of thoughts and actions (e.g., Milner & Rugg, 1992) suggest that this difference will appear the most controversial. Nevertheless, this processing distinction is central to our interpretation of the developmental data on self-perception, and therefore we wish to avoid unnecessary confusion by emphasizing that our comments are restricted to the processing of the perceptual information per se. It is our impression that the "waters become much more muddied" when considering the relation between consciousness and higher-level thought processes (cf. Milner, 1992). By definition, perceptual recognition requires explicit and conscious processing of visual information so that it is recognized or stored for future recall. Moreover, it is necessary that observers know what they've seen if they are to perform successfully on a recognition task, which involves measures such as verbal report or discrimination. Although subliminal perception might be viewed as a possible challenge to this claim, the evidence for such a process is strongly contested and will not be considered further at this time. Let us now consider the role of consciousness in the perceptual control of actions. The evidence is fairly consistent in demonstrating that perceptual control does not require explicit processing of the visual information. This claim is supported by the previously reviewed neuropsychological studies, especially those involving patients who report not seeing objects but successfully reaching for them nonetheless (Weiskrantz et al., 1974). In addition, numerous psychophysical studies reveal that it is possible to verbally report a misperception even when the visuomotor response reflects veridical perception of the visual information (e.g., Goodale & Milner, 1992). For example, Lisberger and Movshon (cited in Sejnowski, 1991) report that observers

TWO MODES OF PERCEIVING THE SELF

309

are quite successful in tracking an equiluminant target defined exclusively by color, even though they see the target move much slower than one defined by luminance differences. Apparently, the perceptual recognition system is fooled by the equiluminant target information, but not the perceptual control system, which perceives the information veridically. Taken together, these results confirm that it is possible for visuomotor modules to use veridical perceptual information that is not accessible to consciousness and verbal report. In sum, the logical and empirical evidence reviewed above suggests that the perception and recognition of objects is functionally dissociable from the perception and control of actions. This latter process always requires proprioceptive perception of the self, whereas the self is but one of many objects recognized and represented by the former process. The perception of self is available to both systems, but the function and significance of this information differs considerably between the two systems. Visual proprioception is foundational to the control of actions and thus will be available from the first evidence of spatially coordinated behaviors. By contrast, recognition of self emerges gradually with the development of a representation of the self. In a recent paper on self-perception by infants, Butterworth (1992) suggested that "self-knowledge may originate in processes of self-perception" (p. 104). The implication was that knowledge of self is rooted in proprioception of self, especially visual proprioception. This claim appears to rely on an incomplete analysis of visual proprioception and its relevance for learning about the self. A review of the research will show that some measures of self-perception, such as mirror recognition, rely on a process of visual perception and representation. This form of representation establishes the infant as an object in the world of objects. Other measures of self-perception, such as visual proprioception, provide additional evidence that the self is perceived, but this information is restricted to its visuomotor function and not accessible to the infant in the form of self knowledge.

Visual Recognition of Self Until very recently, studies of self-perception by infants have focused exclusively on their recognition of themselves. The most common measure of this behavior is mirror self-recognition, which has a long and venerable history in the developmental literature. Darwin (1877) reported that his 9-month-old son was able to recognize his own reflected image in a mirror. Other baby biographers offered anecdotal reports of mirror self-recognition during the second year of life. These reports were provocative, but the evidence was somewhat subjective and could not eliminate alternative interpretations, such as conditioning a child to

310

B.I. BERTENTHAL •

J.L. ROSE

produce a specific vocalization in front of the mirror. It was not until the development of the rouge technique (Amsterdam, 1972; Gallup, 1970) that an objective method for studying mirror self-recognition became available. This technique involves surreptitiously placing some rouge on the face of the child, and then observing his or her response to the "blemish" that appears on the face. Most studies (e.g., Bertenthal & Fischer, 1978; Lewis & Brooks-Gunn, 1979) report that infants begin to detect the rouge by 15 to 21 months of age. It is also relevant that cross-species comparisons reveal that only chimpanzees and orangutans recognize themselves in mirrors when tested with an analogue of the rouge task. Monkeys, for example, fail to detect the "rouge" on their faces even after 2,400 hours of experience with mirrors (Gallup, 1977). It is generally agreed by researchers that detecting the rouge on the face requires some representation of the appearance of the face (Butterworth, 1992; Harter, 1983). Presumably, the rouge is perceived as a discrepancy that leads the infant to try to touch it or remove it by using the mirror to guide his or her hand movements. The conclusion that this task requires representation seems reasonable and straightforward, but hypotheses concerning how this representation develops are still somewhat speculative. For example, Bertenthal and Fischer (1978) suggested over 15 years ago that the representation of the self emerges gradually in a developmental sequence of theoretically related stages, but the evidence was essentially restricted to a logical analysis of the tasks. A somewhat more precise hypothesis was suggested by Gibson (1993), who speculated that representation emerges from the detection of visual-proprioceptive contingencies. In other words, infants perceive that their actions are contingently related to the changing image in the mirror and gradually recognize the self-similarity between body parts and the reflected image. By extrapolation, infants will eventually recognize that the reflected image of their face matches what they look like. In general, this hypothesis seems quite reasonable, but it is nevertheless somewhat perplexing as to why representation is not reported until 18 months of age. Recent evidence from the developmental literature suggests that other objects and their properties are represented by 4 to 5 months of age (Baillargeon, 1993; S pelke, 1994). One reason for this conundrum is that alternative interpretations for the criterion responses in mirror recognition are not ruled out by current data. In the rouge task, for example, the criterion behavior for concluding representation is when infants touch the red mark on their faces. It is conceivable, however, that this task taps much more than a simple representation of the face. For example, trying to touch or remove the rouge may reflect a level of self-consciousness or concern about physical appearance that wasn't present previously. This conjecture is certainly consistent with reports by other theorists (Kagan, 1984; Mahler, Pine, & Bergman, 1975) that infants become much more self-conscious of their

TWO MODES OF PERCEIVING THE SELF

311

behaviors during the second year of life. If this hypothesis is true, it might be possible to find evidence of representation at younger ages than reported in the mirror recognition studies. Interestingly, some preliminary evidence consistent with this speculation was reported recently by Fadil, Moss, and Bahrick (1993). These investigators tested 5- and 8-month-old infants for visual preference of their own face versus the face of a peer. The results revealed preference of the unfamiliar face at both ages, suggesting some representation of the face by 5 months of age. Although further confirmation of this finding is required, it seems much more in line with some of the other more recent developmental evidence on representation. In theory, studies investigating the detection of visual proprioceptive contingencies might reveal additional evidence for recognition of self. Currently, however, the evidence supporting this interpretation is less compelling than implied by some commentators. The problem is that the detection of a contingency between visual and proprioceptive information does not necessarily require any appreciation of self-similarity between actions and visual feedback. Let's consider the paradigm pioneered by Papousek and Papousek (1974), and later refined by Bahrick and Watson (1985) for investigating this issue. In this paradigm, infants are presented with a contingent video image of their own face or legs and a noncontingent image of the face or legs of a peer. Five-month-old infants show preferential looking to the noncontingent image, presumably because they detect the contingent image as themselves, and thus the noncontingent image is less familiar and hence more interesting. Although this interpretation is plausible, it remains somewhat speculative because infants might show a similar preference for any noncontingent event. The resolution of this interpretive impasse requires a comparable test with other events, such as a contingent or noncontingent activated mobile, to determine whether infants would show the same preference when the noncontingency does not involve perception of themselves. Until this issue is resolved, the age at which the detection of visual-proprioceptive contingencies first specify the self will remain uncertain. Although some questions relating to age of onset of self-recognition remain unresolved, it is apparent that the preceding measures of self-recognition require explicit information about the self. This information is available from the behaviors that are reflected in the mirror or video monitor. Of course, the coordinated behaviors performed in front of the mirror also require perceptual information about the self, but this information should not be confused with the visual feedback presented by the mirror or video monitor. This first form of perceptual information is propriospecific and is necessary for the control of actions. By contrast, the perceptual information, i.e., the mirror image, that results from the actions of the infant is processed by the perceptual recognition system. It is thus the feedback from spatially coordinated behaviors, such as

312

B.I. BERTENTHAL 8z J.L. ROSE

touching or tracking the mirror image, and not the visual proprioceptive information controlling the action, that contributes to the development of a representation of the self. In the next section, we will examine in greater detail why the proprioceptive perception of self necessary for controlling actions does not contribute directly to the development of self-recognition.

Perception of Self-motion A number of recent commentaries on the perception of self by infants claim that this information is directly perceived through actions directed at the environment (Butterworth, 1992; Gibson, 1993; Neisser, 1993). All of these commentaries follow Gibson (1979) in asserting that perception of self and perception of spatial layout are two poles of the same process. It is for this reason that vision, like all other modalities, provides a proprioceptive as well as an exteroceptive source of information about the world. Yet the relation between this proprioceptive function of vision and explicit perception of self is not as clear-cut as implied by some theorists. It is our contention that spatially coordinated behaviors, such as visual tracking, reaching, and postural control, do not allow direct access to the perception of self and thus do not represent direct evidence of self-perception. In order to explicate our reasons for this reservation, we will focus on the perception of self-motion, which has received considerable attention by developmental and psychophysical researchers. Let's begin by discussing the optical information available for specifying selfmotions. When objects move in the visual field, they produce a local deformation in the geometric structure of the optical flow field that specifies the speed and direction of the moving object. By contrast, movements of the eyes and head produce a global change in the structure of the optical flow field (Gibson, 1979). This change is lawfully related to the speed and direction in which the observer is moving. Adult observers report compelling sensations of self-motion when stimulated with visual fields of moving texture simulating either translation or rotation (Andersen, 1986). This information not only specifies the direction and speed of movement, but also induces postural compensations in standing adults (Howard, 1986). One of the first reports of this phenomenon dates back to an old fairground device known as the "haunted swing." People entered a boatshaped chamber, and artificial scenery was slowly swung back and forth outside the windows. The perception of the moving scenery induced a compelling illusion that the chamber was rocking, and participants experienced all the sensations of real movement, including loss of postural stability and vertigo (Wood, 1895). A similar procedure

TWO MODES OF PERCEIVING THE SELF

313

for inducing postural sway was introduced many years later by Lishman and Lee (1973). Participants stand inside a large illuminated box that swings back and forth along a path that is parallel to the direction of gaze. They typicany report that the walls are not moving, but that they themselves are swaying. The perceived direction of self-motion is opposite to the direction of optical flow, which induces participants to sway in synchrony with the optical flow in an attempt to restore postural equilibrium. This coupling between perception and action suggests that visual information is perceived as specifying body movement in much the same way that mechanical information from the skin, joints, muscles, and vestibular receptors specifies self-motion. Indeed, Lee and Lishman (1975) present evidence suggesting that the coupling between proprioception and muscular responses is initially tuned by visual information, which explains why new postures are especially dependent on visual information for maintaining postural equilibrimn. In spite of the precise control exercised by vision in modulating postural responses, it is generally agreed that postural equilirium is regulated automatically. This implicit control of the configuration of body segments is necessary so that the observer can devote attentional resources to other activities, such a~s steering or visually exploring the spatial layout (Dichgans & Brandt, 1978; Liebowitz & Post, 1984). Recent investigations reveal that visual control of posture ensures that the adult postural control system responds automatically and accurately to perceived perturbations while standing. For example, Anderson and Dyre (1989) instructed observers to remain stationary while observing a projection of random dots that were sinusoidally oscillating at four different frequencies. Observers could not identify the four frequencies in the stimulus display, yet spectral analyses of their postural responses revealed that they were oscillating at the four driving frequencies presented in the display. In another study, Asten, Gielen, and Denier van der Gon (1988) measured postural responses of observers instructed to stand upright and stationary while viewing a rotating windmill pattern. The results revealed a clear correlation between the frequency of the visual stimulus and the frequency of postural sway for conditions involving either a constant or a variable driving frequency. The above examples reveal that the postural control system involves complex and rapid sensorimotor transformations that necessitate perception of the covariation between the optical flow and the inertial forces of the body. Although these transformations demand precise adjustments of the body, they are performed without explicit guidance. Reports from observers in our own lab viewing flowfields while standing on a force plate reveal that they are completely unaware of specific postural corrections that are produced in response to the changing stimulus display.

314

B.I. BERTENTHAL ~; J.L. ROSE

Much of the evidence for infants' perception of self-motion is based on a similar paradigm. A brief review of the evidence will show that infants, like adults, show rapid and precise postural corrections to optical flow. It is precisely because this system is functionally so successful at such a young age that we suggest it represents an example of a visuomotor module akin to that mediated by the dorsal or perceptual control pathway. From a somewhat different vantage point, it could be said that visual control of posture represents an encapsulated system that is cognitively impenetrable during early development (Rozin, 1976).

FIGURE 1. Schematic drawing of moving room. Depicted inside the room is a child falling backwards in response to the room moving toward the child. This compensatory response would occur if the child perceived the optical flow produced by the room movement as specifying a forward sway rather than a movement of the walls. The first study to test infants' postural responses to optical flow was conducted by Lee and Aronson (1974), who tested infants just beginning to stand on their own in the previously described swinging room. (A similar version of this apparatus used in our own studies is depicted in Figure 1. Note that it is referred to as a "moving room" because the walls are suspended on wheels that glide along a track.) Results from this investigation revealed that infants would compensate to the visually specified sway by responding with a sway, stagger, or fall in the appropriate direction. Because these initial results involved infants between 13 and 16 months, it remained unclear whether perception of self-motion necessitated experience with locomotion (cf. Butterworth and Hicks, 1977). More recent research confirms that self-produced locomotor experiences are unnecessary for perception of self-motion. Butterworth and Hicks (1977) reported that 11month-old infants who sat in the moving room would lean in the appropriate

TWO MODES OF PERCEIVING THE SELF

315

direction when the walls were moved. Bertenthal and Bai (1989) replicated that finding with a more quantitative measure of postural sway. In their study, 5-, 7-, and 9-month-old infants were passively supported in a seat, and displacements of posture were monitored by pressure transducers located under the seat. Interestingly, the findings revealed that 9-month-old infants showed postural responses when the whole room or just the side walls moved. Seven-month-old infants responded systematically when the whole room moved, and 5-month-old infants showed no evidence of systematic responding (see Figure 2).

0.70-

I 15-monfh-olds E;~ 7-monfh-olds k--N-~9 - m o n f h - o l d s

0 . 6 0 -taJ ft.. O _J tO Z

0 . 5 0 --

Ld

0.20-

L9 Z 3:: 0

0.40 0.30 -

0.10 0.00

-

-0,10-0.20-

!

WHOLE ROOM

!

!

SIDE WALLS

FRONT WALL

FIGURE 2. Mean magnitude of postural sway (change in slope) as a function of age and wall m o v e m e n t condition. Positive values r e p r e s e n t directionally appropriate responses; negative values represent directionally inappropriate responses (from Bertenthal & Bai, 1989).

One interpretation for the preceding responses is that infants are not sensitive to optical flow for specifying self-motion until 7 to 9 months of age. Yet another interpretation is that 5-month-old infants lack the necessary muscle strength and coordination to sit without support; thus, they are unsuccessful in adjusting their posture even though they perceive the optical flow as specifying self-motion. This latter interpretation is apparently correct because younger infants control the position of their heads in response to changes in the moving room. Pope (1984) reports that 2-month-old infants show changes in backward head pressure when stimulated in a moving room. More recent evidence by Jouen (1990; Jouen &

316

B.I. BERTENTHAL •

J.L. ROSE

Gapenne, this volume) suggests that perception of self-motion is present at birth because neonates show systematic adjustments of their head posture in response to an oscillating flowfield. In spite of the importance of the preceding findings, they present a somewhat incomplete understanding of how infants adjust their postural responses to optical flow. The only question addressed by these studies is whether infants show directionally appropriate compensatory responses to the perception of self-motion. More specific questions concerning how infants use visual information to modulate their postural responses and when they begin to scale their compensatory responses to the perceived displacement are just beginning to receive attention. In a recently completed investigation (Bertenthal, Rose, & Bai, 1995), we tested developmental changes in infants' scaling of the muscular torques necessary to maintain postural equilibrium while sitting in a moving room. Infants between 5 and 13 months of age sat in a specially designed seat that was attached to a force plate. During 10-second trials, the walls and ceiling of the room were sinusoidally oscillated at one of two frequencies (.3 and .6 Hz) and amplitudes (9 and 18 cm) to determine whether sway responses by infants were coupled to the specific dimensions of the optical flow information. In this study, entrainment, or the covariation between sway and room movement, was defined as the magnitude of the sway frequency at the driving frequency of the room. For comparison purposes, the magnitude of postural sway at the same frequency was calculated on trials when the room moved at a different frequency. The results revealed significant entrainment by infants at all four ages (see Figure 3). Two points about these results are especially noteworthy. First, the magnitude of entrainment increased steadily during the period of time when sitting develops (i.e., 5 to 9 months of age). Additional results from this study as well as an ongoing longitudinal study (Rose & Bertenthal, in press) suggest that the improvement in entrainment is systematically related to the development of sitting. Of course, this finding is not surprising because it is quite natural that the coupling between perception and action would improve as experience with this posture increases. Second, and somewhat more surprising, is the finding that 5month-old infants show postural compensations to the optical flow even before they control their sitting sufficiently to maintain postural equilibrium. Apparently, visual control of sitting does not merely follow the development of this new posture, but actively contributes to the tuning of the necessary muscle synergies, as suggested previously by Lee and Lishman (1975). The moral from this most recent research is that postural responses are controlled at a level of specificity that demands perception of self-motion as well as perception of the inertial forces that accompany this motion. It is also necessary for infants to appreciate the configuration of the relevant body segments to produce

TWO MODES OF PERCEIVING THE SELF 317

the necessary postural compensation. This analysis thus suggests that infants show very specific evidence of visual proprioception at very young ages. It is clear, however, that this perception is implicit in the organization of the behavior, and it is not monitored any more directly than would be true for the control of other actions, such as the control of the vocal apparatus to produce speech sounds. If theorists were inclined to argue otherwise, it would be tantamount to claiming that nonhumans also experience explicit information about the self in that they, too, show the same perceptual modulation of their actions (Mitchell, 1992). Thus, a more parsimonious interpretation for postural compensations to optical flow is that the self-specifying information is automatically and implicitly mapped to the motor response synergies.

20

I

Same Freq. Different Freq.

I.H

o15 Z < rr <

>10 Z ILl O tr

tu 5 12.

5

7 9 AGE (months)

13

FIGURE 3. Mean magnitude of postural sway (percent variance) calculated at the wall movement frequency (Same freq.) and at the opposing frequency (Different freq.) as a

function of age. (from Bertenthal et al., 1995). This scenario for visual proprioception is quite different than the one suggested by the results of visual self-recognition studies. In essence, the self is defined exclusively as an ensemble of muscle synergies that are responsive to perceptual information about spatial orientation. Perceptual control of each new posture, such as sitting, standing, etc., develops gradually as the child learns to scale the muscle torques to the perceived information. There is no evidence of transfer of perceptual control from one posture to another because each represents an autonomous visuomotor module (Bertenthal et al., 1995). Accordingly,

318

B.I. BERTENTHAL•

J.L. ROSE

perception of self defined in this fashion is piecemeal, implicit, and organized by the goal of maintaining postural equilibrium. It is unlikely that this source of information about the self contributes later to an integrated notion of self because it is not represented in a form accessible to other information about the self. Before concluding this section, we are obliged to clarify a subtle but important distinction between the perceptual control of actions and the perception of the consequences of those actions. It is well documented that even young infants perceive some consequences of their own actions (e.g., Alessandri, Sullivan, & Lewis, 1990; Rovee-Collier & Gekoski, 1979; Watson & Ramey, 1972). The recognition of these consequences informs infants of their own causal agency (Bullinger & Chatillon, 1983). As self-produced actions become better coordinated and more complex, the infant's appreciation of his or her own agency will become better defined and more differentiated. Thus, the visual proprioception involved in regulating actions does not contribute directly to self-knowledge, but does so indirectly by eliciting coordinated behaviors that produce observable consequences. This point is best illustrated by an example. It is well documented that locomotor infants avoid the deep side of the visual cliff (Bertenthal & Campos, 1990). Although it was originally presumed that depth perception was the rate-limiting factor in their performance, more recent results challenge this interpretation. First and foremost, the cumulative evidence on the development of depth perception reveals that infants perceive depth long before the development of independent locomotion (Yonas & Owsley, 1987). Moreover, the findings on infants' avoidance of the deep side reveal that this behavior is not evident immediately following the onset of independent locomotion (Bertenthal & Campos, 1984). Most infants do not show avoidance until sometime between one and two months following the onset of crawling. It is this latter finding that is pivotal to our interpretation of infants' performance. As first suggested by Bertenthal, Campos, and Barrett (1984), infants begin to avoid the deep side not when they perceive the depth, but when they perceive the consequences to themselves of the apparent drop-off. The perception of these consequences is not immediately apparent to infants with the onset of independent locomotion. As infants start to crawl and explore the consequences of this action, they begin to perceive and anticipate the implications of their actions on others as well as on themselves. One implication of a perceived drop-off is danger to the self, which explains why infants will gradually learn to avoid the deep side of the cliff. It is thus through the development of a new motor response (i.e., crawling) that infants gain new experiences with their surroundings and learn more about the consequences of their own actions. This position is similar to the view espoused by Gibson and Adolph (1992) that infants learn new affordances through perceptual

TWO MODES OF PERCEIVING THE SELF

319

exploration; but unlike that position, learning about self demands that selfproduced consequences are represented and accessible for future interactions with the surround. It is also not sufficient to posit some visuomotor program or module to explain the emergence of these new behaviors that guide actions (e.g., Rader, Bausano, & Richards, 1980). As discussed previously, visuomotor modules represent a direct and autonomous coupling between perception and action. This coupling is present even before motor responses, such as sitting or head control, are capable of independent functioning. Yet, in the case of avoidance of the deep side of the cliff, there is no evidence of modulation in precrawling infants, as assessed with heart rate (Campos, 1976), nor any evidence of modulation immediately following the onset of crawling, i.e., infants do not initially show any avoidance of the deep side. For these reasons, we contend that visual cliff performance is not a consequence of visuomotor control, but rather a consequence of visuomotor exploration and learning about the self that ensues from this exploration.

Concluding Remarks It is commonplace for theorists to propose multiple forms of the self. James (1890) distinguished between an existential self defined as the agent of activity and a categorical self embodying those properties that define the self as a unique object in the world. More recently, Neisser (1988) distinguished between five forms of self defined by different sources of information, ranging between self-produced actions and sociocultural experiences. In this chapter, we propose another distinction concerning the difference between implicit and explicit perception of self. The former construct is rooted in spatially coordinated behaviors that begin to appear early in development and become more systematic with practice and experience. By contrast, the latter construct is rooted in the development of representation. Both forms of self-perception are important to the early development of the infant's adaptation to the world. Infants learn about people, objects, and events through actions that enable them to explore the important variant and invariant properties in their world. The development of visual pursuit, head and trunk control, reaching, grasping, sitting, and crawling are all examples of actions that contribute to the exploration of the world. Visual proprioception optimizes the coordination of these exploratory actions to ensure that they are modulated to local conditions. This form of selfperception is modular, automatic, and defined by the pattern of muscular torques necessary to ensure spatially coordinated behaviors. Although this form of

320

B.I. BERTENTHAL•

J.L. ROSE

perception is implicit, it is foundational to all that is learned through selfexploration. Explicit perception of self focuses on what is learned from actions. By definition, learning involves some form of storage or representation of results so that these same results are recognized and anticipated in the future. This representation includes objects as well as events, and thus encompasses visual selfrecognition as well as recognition of self-produced consequences. Unlike hnplicit perception of self, which is organized in modular fashion, explicit perception is not context-specific and allows for generalization across different experiences. It is this generalization that allows infants to appreciate that the self is a unitary and coherent object among an ever-expanding assemblage of other objects in the world. By way of conclusion, we wish to emphasize that this proposed distinction between implicit and explicit perception of self is still somewhat speculative, and its significance rests on whether it provides any additional conceptual clarity for understanding the early development of self. It is certainly premature to judge the impact of this processing distinction, but we look forward to these ideas undergoing refinement and reformulation as more is learned about the early development of self-perception. ACKNOWLEDGMENTS

The writing of this chapter was supported by NIH grant HD16196 and a grant from the John D. and Catherine T. MacArthur Foundation. JLR was supported in part by a predoctoral fellowship from NIH grant HD0723. REFERENCES

Alessandri, S. M., Sullivan, M. W., & Lewis, M. (1990). Violation of expectancy and frustration in early infancy. Developmental Psychology, 26, 738-744. Amsterdam, B. (1972). Mirror self-image reactions before age two. Developmental Psychobiology, 5, 297-305. Andersen, G. J. (1986). Perception of self-motion: Psychophysical and computational approaches. Psychological Bulletin, 99, 52-65. Andersen, G. J., & Dyre, B. P. (1989). Spatial orientation from optic flow in the central visual field. Perception & Psychophysics, 45, 453-458. Asten, W. N. J. C. van, Gielen, C. C. A. M., & Denier van der Gon, J.J. (1988). Postural movements induced by rotations of visual scenes. Journal of the Optical Society of America, 5, 1781-1789. Bahrick, L. E., & Watson, J. S. (1985). Detection of intermodal proprioceptivevisual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21, 963-973. Baillargeon, R. (1993). The object concept revisited: New directions in the investigation of infants' physical knowledge. In C. E. Granrud (Ed.), Visual perception and cognition in infancy (pp. 265-316). Hillsdale, NJ: Erlbaum.

TWO MODES OF PERCEIVING THE SELF 321 Bertenthal, B. I. (1993, March). Emerging trends in perceptual development. Master Lecture presented at the Meetings of the Society for Research in Child Development, New Orleans. Bertenthal, B. I., & Bai, D. L. (1989). Infants' sensitivity to optical flow for controlling posture. Developmental Psychology, 25, 936-945. Bertenthal, B. I., & Campos, J. J. (1984). A reexamination of fear and its determinants on the visual cliff. Psychophysiotogy, 21, 413-417. Bertenthal, B. I., & Campos, J. J. (1990). A systems approach to the organizing effects of early locomotor experience. In C. Rovee-Collier & L. P. Lipsitt (Eds.), Advances in infancy research (Vol. 6, pp. 51-98). Norwood, NJ: Ablex. Bertenthal, B. I., Campos, J. J., & Barrett, K. C. (1984). Self-produced locomotion: An organizer of emotional, cognitive, and social development in infancy. In R. N Emde & R. J. Harmon (Eds.), Continuities and discontinuities in development (pp. 175-210). New York: Plenum Press. Bertenthal, B. I., & Fischer, K. W. (1978). Development of self-recognition in the infant. Developmental Psychology, 14, 44-50. Bertenthal, B. I., Rose, J. L., & Bai, D. L. (1995). Perception-action coupling in the development of visual control of posture. Manuscript submitted for publication. Bullinger, A., & Chatillon, J. (1983). Recent theory and research of the Genevan school. In J. H. Flavell & E. M. Markman (Eds.), Handbook of child psychology: Cognitive development (Vol. 3, pp. 231-262). New York: Wiley. Butterworth, G. (1992). Origins of self-perception in infancy. Psychological Inquiry, 3, 103-111. Butterworth, G., & Hicks, L. (1977). Visual proprioception and postural stability in infancy: A developmental study. Perception, 5, 255-263. Campos, J. J. (1976). Heart rate: A sensitive tool for the study of emotional development in the infant. In L. P. Lipsitt (Ed.), Developmental Psychobiology: The significance of infancy (pp. 1-34). Hillsdale, NJ: Erlbaum. Damasio, A. R., & Benton, A. L. (1979). Impairment of hand movements under visual guidance. Neurology, 29, 170-178. Darwin, C. R. (1877). A biographical sketch of an infant. Mind, 2, 285-294. Dichgans, J., & Brandt, T. (1978). Visual-vestibular interaction: Effects on selfmotion and postural control. In R. Held, H. W. Liebowitz, & H. L. Teuber (Eds.), Handbook of sensory phsyiology (Vol. 8, pp. 755-804). Heidelberg: Springer. Fadil, C. A., Moss, L. E., & Bahrick, L. E. (1993, March). Infants' visual recognition of their own faces. Paper presented at the Meetings of the Society for Research in Child Development, New Orleans, LA. Gallup, G. G. Jr. (1970). Chimpanzees: Self-recognition. Science, 167, 86-87. Gallup, G. G. Jr. (1977). Absence of self-recogntiion in a monkey (Macaca fascicutaris) following prolonged exposure to a mirror. Developmental Psychobiology, 10, 281-284. Gibson, E. J. (1993). Ontogenesis of the percieved self. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 4367). Cambridge, MA: Cambridge University Press. Gibson, E. J., & Adolph, K. A. (1992). The perceived self in infancy. Psychological Inquiry, 3, 119-120. Gibson, J. J. (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin. Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin.

322

B.I. BERTENTHAL & J.L. ROSE

Goodale, M. A. (1983). Vision as a sensorimotor system. In T. E. Robinson (Ed.), Behavioral approaches to brain research (pp. 41-61). New York: Oxford University Press. Goodale, M. A. (1988). Modularity in visuomotor control: From input to output. In Z. Pylyshyn (Ed.), Computational processes in human vision: An interdisciplinary perspective. Norwood, NJ: Ablex. Goodale, M. A. (1993). Visual pathways supporting perception and action in the primate cerebral cortex. Current Opinion in Neurobiology, 3, 578-585. Goodale, M. A., & Milner, A. D. (1992). Separate visual pathways for perception and action. Trends in Neuroscience, 15, 20-25. Goodale, M. A., Milner, A. D., Jakobson, L. S., & Carey, D. P. (1991). A neurological dissociation between perceiving objects and grasping them. Nature, 349, 154156. Harter, S. (1983). Developmental perspectives on the self system. In M. Hetherington (Ed.), Handbook of child psychology: Socialization, personality, and social development (Vol. 4., pp. 275-385). New York: Wiley. Hofsten, C. von (1993). Prospective control: A basic aspect of action development. Human Development, 36, 253-270. Howard, I. P. (1986). The perception of posture, self-motion, and the visual vertical. In K. R. Boff, L. Kaufman, & J. P. Thomas (Eds.), Handbook of perception and human performance. Vol. 1. Sensory processes and perception (pp. 181-186). New York: Wiley. Ingle, D. J. (1983). Brain mechanisms of localization in frogs and toads. In J. P. Ewert, R. R. Capranica, & D. J. Ingle (Eds.), Advances in vertebrate neuroethology (pp. 177-226). New York: Plenum Press. Jakobson, L. S., Archibald, Y. M., Carey, D. P., & Goodale, M. A. (1991). A kinematic analysis of reaching and grasping movements in a patient recovering from optic ataxia. Neuropsychologia, 29, 803-809. James, W. (1890). The principles of psychology (Vol. 1). New York: Holt. Jouen, F. (1990). Early visual-vestibular interactions and postural development. In H. Bloch & B. I. Bertenthal (Eds.), Sensory-motor organization and development in infancy and early childhood (pp. 199-215). Dordrecht: Kluwer. Kagan, J. (1984). Continuity and change in the opening years of life. In R. N. Emde & R. J. Harmon (Eds.), Continuities and discontinuities in development (pp. 15-40). New York: Plenum Press. Lee, D. N. (1993). Body-environment coupling. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 68-88). Cambridge, MA: Cambridge University Press. Lee, D. N., & Aronson, E. (1974). Visual proprioceptive control of standing in human infants. Perception & Psychophysics, 15, 529-532. Lee, D. N., & Lishman, J. R. (1975). Visual proprioceptive control of stance. Journal of Human Movement Studies, 1, 87-95. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. New York: Plenum. Liebowitz, H. W., & Post, R. B. (1984). The two modes of precessing concept and some implications. In J. Beck (Ed.), Organization and representation in perception (pp. 343-363). Hillsdale, NJ: Erlbaum. Lishman, J. R., & Lee, D. N. (1973). The autonomy of visual kinaesthesis. Perception, 2, 287-294. Livingstone, M. S., & Hubel, D. H. (1987). Psychophysical evidence for separate channels for the perception of form, color, movement, and depth. Journal of Neuroscience, 7, 3416-3468.

TWO MODES OF PERCEIVING THE SELF 323 Mahler, M., Pine, F., & Bergman, A. (1975). The psychological birth of the human infant. New York: Basic Books. Marr, D. (1982). Vision. San Francisco: Freeman. Milner, A. D. (1992). Disorders of perceptual awareness m commentary. In A. D. Milner & M. D. Rugg (Eds.), The neuropsychology of consciousness (pp. 139158). San Diego, CA: Academic Press. Milner, A. D., & Goodale, M. A. (1993). Visual pathways to perception and action. In T. P. Hicks, S. Molotchnikoff, & T. Ono (Eds.), Progress in brain research (pp. 317-337). Amsterdam: Elsevier. Milner, A. D., & Rugg, M. D. (1992). The neuropsychology of consciousness. San Diego, CA: Academic Press. Mitchell, R. W. (1992). Developing concepts in infancy: Animals, self-perception, and two theories of mirror self-recognition. Psychological Inquiry, 3, 127-130. Neisser, U. (1988). Five kinds of self knowledge. Philosophical Psychology, 1, 3559. Neisser, U. (1989, August). Direct perception and recognition as distinct perceptual systems. Paper presented at the Annual Conference of the Cognitive Science Society, Ann Arbor, MI. Neisser, U. (1993). The self perceived. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 3-21). Cambridge, MA: Cambridge University Press. Papousek, H., & Papousek, M. (1974). Mirror image and self-recognition in young human infants: I. A new method of experimental analysis. Developmental Psychobiology, 7, 149-157. Perenin, M. T., & Vighetto, A. (1988). Optic ataxia: A specific disruption in visuomotor mechanisms. I. Different aspects of the deficit in reaching for objects. Brain, 111, 643-674. Pope, M. J. (1984). Visual proprioception in infant posturat development. Unpublished doctoral dissertation. University of Southampton, England. Rader, N., Bausano, M., & Richards, J. E. (1980). On the nature of the visual-cliff avoidance response in human infants. Child Development, 51, 61-68. Rose, J. L., & Bertenthal, B. I. (in press). A longitudinal study of the visual control of posture in infancy. Proceedings of the eighth international conference on

perception and action.

Rovee-Collier, C. K., & Gekoski, M. J. (1979). The economics of infancy: A review of conjugate reinforcement. In H. W. Reese & L. P. Lipsitt (Eds.), Advances in child development and behavior (Vol. 13, pp. 195-255). New York: Academic Press. Rozin, P. (1976). The evolution of intelligence and access to the cognitive unconscious. In J. M. Sprague & A. A. Epstein (Eds.), Progress in psychobiology and physiological psychology (Vol. 6, pp. 245-280). New York: Academic Press. Schiller, P. H., Logothetis, N. K., & Charles, E. R. (1990). Functions of the colouropponent and broad-band channels of the visual system. Nature, 343, 68-70. Senjowski, T.J. (1991). Back together again.Nature, 352, 669-670. Spelke, E. (1994). Initial knowledge: Six suggestions. Cognition, 50, 431-445. Ungerleider, L. G., & Mishkin, M. (1982). Two cortical visual systems. In D. J. Ingle, M. A. Goodale, & R. J. W. Mansfield (Eds.), Analysis of visual behavior (pp. 549586). Cambridge: MIT Press. Watson, J. S., & Ramey, C. T. (1972). Reactions to response-contingent stimulation in early infancy. Merrill-Palmer Quarterly, 18, 219-227. Weiskrantz, L., Warrington, E., Sander, M. D., & Marshall, J. (1974). Visual capacity in the hemianopic field following restricted occipital ablation. Brain, 97, 709729.

324

B.I. BERTENTHAL •

J.L. ROSE

Wood, R. W. (1895). The "haunted-swing" illusion. Psychological Review, 2, 27.7278. Woollacott, M. H., & Sveistrup, H. (1994). The development of sensorimotor integration underlying posture control in infants during the transition to independent stance. In S. P. Swinnen, J. Massion, & H. Heuer (Eds.), Interlimb coordination: Neural, dynamical and cognitive constraints (pp. 371-389). San Diego, CA: Academic Press. Yonas, A., & Owsley, C. (1987). Development of visual space perception. In P. Salapatek & L. Cohen (Eds.), Handbook of infant perception: From sensation to perception (Vol. 1, pp. 80-122). Orlando, FL: Academic Press.

SECTION 2

Perceptual Origins of the Self

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

327

CHAPTER 16

The Effect of Blindness on the Early Development of the Self ANN E. BIGELOW

St. Francis Xavier University

Development involves an interrelationship among perceptual, motor, cognitive, social, and emotional factors. Consequently, a deficit in one area of development will impact on the development of other areas. At no time in life is this more apparent than in infancy, when developmental processes are so intimately integrated. Perceptual abilities affect children's understanding of the world and of their own experience in it. Thus, although blindness is a perceptual deficit, it has important implications for the child's early development of the self.

The Importance of Vision for the Development of the Early Self The formation of the early self hinges on infants developing an awareness that they can affect their physical and social environments through their own actions. Self-action allows infants to detect their own efficacy in producing external events in ways that become predictable. This awareness, in turn, influences infants' abilities to perceive the affordances or possibilities for self-actions in given situations. Gibson (1993) calls this awareness of self-agency the epitome of perceiving oneself. Such an awareness is most easily acquired through the use of vision because children can see that their behavior has causal consequences in the external world. Vision is the primary sense modality in infancy because so much of the information babies use to learn about themselves and the environment comes to them through their visual systems. The organization of early visual perception virtually assures that infants attend to the elements in the world that are most important to their survival and development. When alert, young infants visually

328

ANN E. BIGELOW

search for things to inspect and then visually explore them in an increasingly systematic fashion (Haith, 1980). They coordinate their visual experience with information from other modalities such that they are likely to look at things that sound and to visually inspect things they are tactually manipulating (Spelke & Cortelyou, 1981; White, 1969). Such visual inspection enhances knowledge of the world and of the effects of their own actions on the world. Vision readily allows for intermodal transfer of information because many attributes of the physical environment detected in other modalities are shared by vision. These include such attributes as shape, movement, and rhythm. Thus, infants visually recognize things experienced only tactually or auditorily and vice versa (see Sued, 1993, for a review). Infants are particularly likely to be visually attentive to people, partly because faces have characteristics that are attractive to them. Infants are visually attracted to the high contrast of the face, the animation and internal movement of the face, and its visual complexity (Maurer, 1985). Therefore, infants are likely to find people to be especially interesting, which, of course is adaptive, and the visual system plays a major role in this attraction. As babies learn to discriminate specific special people in their world, they turn to them in greeting, brighten, smile, and reach out to them in anticipation of a familiar interaction. They have learned what Lamb (1981a, 1981b) calls "perceived personal effectance": that their actions have predictable consequences in the world. Such knowledge develops most easily in early social interactions. People respond contingently to infants' behaviors more consistently and reliably than do objects, and it is through repeated contingent behaviors that babies most readily learn their effect on the world (Watson, 1979, 1985). By 2 to 3 months, infants have learned the association between their own action and its effect on the social environment. They become distressed if mothers become still-faced during face-to-face interaction (Tronick, Als, Adamson, Wise, & Brazelton, 1978) or if the immediate video playback of their mothers is switched to a delayed playback (Murray & Trevarthen, 1985). Thus, they develop expectation and anticipate specific consequences to their actions. Vision is central to these developments. The nature of visual stimulation itself also allows infants to exercise more control over that stimulation than does stimulation from the other modalities. Infants can visually follow things that are interesting and can look away (or close their eyes) when the visual stimulation becomes too overwhelming. Indeed, there is evidence that infants do regulate the level of their visual stimulation in just such a manner (Brazelton, Koslowski, & Main, 1974; Stern, 1974, 1985). Such regulation is more difficult in the other modalities. Nonlocomotor infants cannot readily stop hearing external auditory stimulation that is overwhelming, nor can they do much to pursue interesting external sounds if the sounds change. With respect to tactile stimulation, they can drop objects they no longer wish to touch,

BLINDNESS AND THE SELF

329

but grasping them again or reaching for alternative objects is not a simple matter in the beginning. Likewise, olfactory and taste sensations are less easily regulated. In a very real sense, infants can more actively control their level of visual stimulation than they can stimulation from the other modalities.

How Blindness Affects the Development of the Early Self So what happens when infants are born without vision? Fraiberg (1977), who pioneered the study of the development of blind children by extensively studying ten totally blind infants, noted that young blind infants are typically passive children who do not demand much and are content to lie in their cribs seemingly impassive to environmental or social stimuli. She notes rather dramatically, but probably accurately, that for blind infants, persons and things manifest themselves in an unpredictable, random fashion, emerging from the void only to fade back into the void. The passive behavior of these infants is exactly what would be expected from beings who have no way, or no known way, of controlling or predicting what is happening to them or around them (Hiroto & Seligman, 1975; Seligman & Maier, 1967). Blind infants have difficulty discovering that their actions and behaviors have an effect on the social and physical environment. Even their discovery of their own bodies is slowed. There is no fascination with watching the hands, midline finger play is delayed, and discovery of the feet is late (Fraiberg, 1977). Without visual feedback, the proprioceptive sense of selfmovement does not readily lead to the discovery of the body. Thus, the challenge to the blind infant to formulate an early sense of self is greater than it is for sighted children. Play and language are the developmental areas that have been specifically studied in populations considered to be at risk for the development of self: populations such as mentally handicapped children, severely abused children, and autistic children (Cicchetti, Beeghly, Carlson, & Toth, 1990; Cicchetti, 1991; Hobson, 1990). Children's development of symbolic play and their linguistic reference to self and others are thought to reflect their emerging understanding of the distinctiveness of self and nonself. Blind children show abnormalities and delays in these areas. Their early use of language is restricted to co~mnents on self-actions or desires (Andersen, Dunlea, & Kekelis, 1984; Bigelow, 1987; Dunlea, 1984), they have difficulties with the use of first- and second-person pronouns (Andersen et al., 1984; Fraiberg, 1977), and they do not readily engage in pretend play (Dunlea, 1984; Fraiberg & Adelson, 1973). These findings suggest that blind children are, indeed, experiencing difficulties in the development of the self. The fact that most blind children overcome such early

330

ANN E. BIGELOW

difficulties in the understanding of the self is testimony to the flexibility of human intelligence. The investigation of how they do so both affirms and challenges some of the emerging ideas of early development of the self. In discussing the early development of the self, Neisser (1988, 1991, 1993) proposes that, prior to representation, the child acquires aspects of the self that are specified through perception, particularly perception of the self in activity with the social or physical environment. There are two forms of the self that arise from perception: the "interpersonal self," which connects and separates the self and other people; and the "ecological self," which connects and separates the self and the physical environment. These aspects of self are proposed to begin at birth and continue to exist throughout life, even after the emergence of representation and of the conceptual self. Unlike forms of the self developed later, these early senses of the self are acquired directly through experience and are not constructed by the child. The interpersonal self is concerned with the perception of the self in relation to others and is facilitated and evidenced through mutual gaze, contingent gestures, reciprocal vocalization, and later by pointing and joint attention. What specifies perceptions relevant to the interpersonal self is the contingency of each person's action on those of the partner, either temporally, as in turn-taking, or spatially, as in eye contact (Neisser, 1991). The ecological self concerns the self perceived in terms of the ongoing relationship with the local physical environment. Knowledge of the layout of the environment seen from the perspective of the self and knowledge of how that perspective changes with movement through space positions the ecological self in the environment. The ecological self includes knowledge of the environmental effects of self-action, the effects of one's current actions, and the affordances available to future actions, that is, what one is doing and what one might do. For example, it allows for knowledge of what is within reachable space and what is not. Vision is important to both aspects of the early sense of self. Approaches and actions of others and the way objects present different angles and perspectives as the child moves through space are most easily noted through vision. It follows, therefore, that children born without vision are necessarily at risk for the development of these early forms of self-knowledge.

Blindness and the Development of the Interpersonal Self Blindness strains the parent-child relationship and puts the development of the infant's interpersonal self in jeopardy. There is no turning to face the parent in greeting nor reaching out to be picked up. It is difficult to determine where the child is attending or what the child has noticed. Fraiberg (1977) advises that

BLINDNESS AND THE SELF

331

parents of blind children must learn to look in different places for their babies' responses. Blind infants' hand movements, for example, may be more indicative of their emotional responses than are their facial movements. The blind baby who remains turned away from the parent may move the hands excitedly upon hearing the mother's voice and become immobile upon hearing a stranger's voice. Parents must learn to read such recognition cues in order to establish interactive social patterns with their children. These cues are often minute and difficult to detect initially, particularly if parents are emotionally overwhelmed with the prospect of raising a handicapped child and are questioning their adequacy in meeting the task. This was illustrated for me on one particular visit to the mother of a blind child about 6 weeks of age. The mother was distressed at her baby's unresponsiveness and passivity, and she questioned whether he had other disabilities in addition to blindness. The baby was indeed inactive and appeared listless. I asked the mother to interact with her child in a way she especially enjoyed. After some hesitation, she took her baby in her arms and rolled gently back and forth on the bed talking softly to him as she repeatedly kissed him on one cheek and then the other. After a few minutes, I asked her to stop in mid-roll. She did, and the baby, who had been inactive during this procedure, slowly turned the other cheek. To the mother's delight, he showed anticipation of the coming kiss and knowledge of their intimate ritual. Most parent-child interactions, so important to the establishment of the babies' sense of shaping their own experience, begin with the parent doing most of the work, i.e., being more responsive to the infant than vice versa, elaborating on the interaction within the infant's zone of proximal development. Yet, for most parents of normal infants, this is done without awareness. Babies' visual responsiveness allows them to increase their role in the interactions at rates that are experienced by parents as smooth and natural. Children without vision, however, have more difficulty detecting patterns in social interactions, and their responses to such interactions are often less obvious than those of sighted infants. Blindness makes it difficult for infants to perceive what others are attending to and therefore to understand the emotional reactions of others (Hobson, 1993). Many important avenues for the formation of the interpersonal sell such as mutual gaze and pointing, are absent in blind infants; joint attention and social referencing are impaired. Thus, blindness presents a challenge to the development of the interpersonal self. Blind children's emerging ability to sense the effect of their behavior on the actions and behaviors of others is greatly influenced by the patience, sensitivity, and effort of their parents. If parents meet the challenge presented by their children's blindness, creative interactions are possible. The children can learn that others are responsive to their behavior and that they can influence the actions and, eventually, the feelings and thoughts of others.

332

ANN E. BIGELOW

Blindness and the Development of the Ecological Self There are parallels between the ecological and interpersonal selves (Neisser, 1991). In both, the critical information is produced by the child's own actions, either by moving or acting in the environment or by mutually responding. Perceptions'relevant to the two emerging selves often co-occur. Yet the ecological and interpersonal selves are distinct. They are based on different information, can be salient on different occasions, and are independently at risk for pathology. Autistic children have been shown to have disturbances in interpersonal self-development, but not in ecological self-development (Hobson, 1990, 1993; Loveland, 1993). They have difficulty acquiring concepts of others as people with their own subjective lives and have difficulty understanding themselves to be objects of others' thoughts and feelings. Thus, their relationships with others are impaired. Yet they have little trouble relating to objects and the physical environment. The examination of what is known about blind children's development suggests that their primary pattern of disturbance in the development of the early sense of self is the reverse of that of autistic children. Although blindness strains the development of the interpersonal self in ways discussed above, it is the development of the ecological self that is most at risk. It is blind children's difficulty in understanding themselves in relation to the physical environment that presents the major obstacle to early self-knowledge. A child's perceptual abilities allow for the detection of the nature of the relationship between the child and the environment, and they influence the child's cognitive interpretation of the experience and thus the child's reaction to the experience (Cicchetti, 1991). Blindness is a perceptual deficit, but it has important implications for the development of infants' motor and cognitive abilities, as well as social and communicative skills. Many of these effects are due to the difficulty blind children have in acquiring knowledge of the self within the physical environment. Children born blind are developmentally challenged in a number of ways. One of the most obvious areas of developmental delay is in certain motor skills, specifically reaching, crawling, and walking (Adelson & Fraiberg, 1974; Fraiberg, 1968). Other motor skills, such as sitting and standing, are developed on a par with sighted peers. Yet milestone behaviors that involve extending the self into the environment are substantially delayed. Sighted infants first reach for objects when they are about 5 months of age, crawl at about 7 months, and walk at about 12 months (Bayley, 1969). Fraiberg found that the blind children she followed first reached for an object at an average age of 10 months (Fraiberg, 1968), crawled at an average age of 13 months, and walked at an average age of 19 months (Adelson & Fraiberg, 1974). However, Fraiberg's criteria for these motor

BLINDNESS AND THE SELF

333

skills were quite liberal. For instance, reaching was credited the first time an infant grasped an object in a session that had 20 to 30 trials, and Fraiberg noted that such a demonstration by the blind infant "would actually correspond to the very early stages of visually directed prehension, under 5 months, when a sighted baby swiping at an object within range makes a first lucky contact with the object, grasps it, and brings it to his mouth" (Fraiberg, 1968, p. 284; italics in original). Other researchers report delays in the development of these motor skills that are longer than those cited by Fraiberg. Norris, Spaulding, and Brodie (1957) studied the development of 66 blind children and found that only 50% of them were walking independently at 24 months. Blind babies are physically ready to reach and move into the environment at the same age as sighted infants, yet they do not do so. Fraiberg (1977) speculated that the reason this occurs is that blind infants have difficulty understanding that objects are located in space and that they could gain access to these objects through their own actions; that is, they have difficulty understanding themselves in relation to the physical environment. Reaching, crawling, and walking are on a continuum of behavior having to do with infants' increasing abilities to interact with and explore the environment. Fraiberg (1968) thought that reaching was the critical skill leading to locomotion. Once her blind infants began to reach to external sound cues, crawling followed shortly thereafter. Reaching is one of the first self-initiated contacts blind children make with the external world. Reaching for sighted children is initially stimulated by vision. For blind children, reaching is more closely tied to their cognitive development. Blind children have difficulty understanding that objects to reach for exist, that these objects have a permanence of their own that is independent of sell and that they can use theft bodies to obtain and explore these objects. The development of their reaching skill gives clues to their emerging understanding of space and objects and to themselves in relation to the physical world. Bigelow (1986) investigated blind children's use of sound and touch in eliciting a reach or search by presenting five totally blind infants with ten sound and touch tasks every two months until mastery. Table 1 shows the tasks and the ages of the infants at mastery. (The mastery criterion was two consecutive sessions with direct reaches to the object on the majority of the trials.) The children mastered the tasks at different ages, yet the sequence with which they showed their mastery is revealing with respect to their growth of object and spatial awareness without vision. The initial reaches were to objects that were touching their bodies (Task 1). The addition of sound produced by such objects (Task l b) did not facilitate their reaching, probably because tactile sensation is such a powerful cue in eliciting and directing their reach. Later they reached

TABLE 1. Age in Months for Mastery of the Reaching Tasks for Five Blind Children. 1

Tasks

(1) (2) (3) (4) (5)

Age in months when first seen:

17

2 13

Subjects 3 32

4~

4 11

5 15

Age at Mastery Toy is touching the body: (a) quiet toy (b) sound toy Sound toy is pulled away at chest level Sound toy is sounded (a) at chest level front (b) to the right or left of chest level front Sound toy is arced around the head Toy is dropped (a) quiet toy

>

Z Z o

20 23 23

13 13 13

32 32 32

13 12 14

23 23 23

13 13 13

32 32 32

15

24

13

a

14

15 20

c~ t'-'

9

BLINDNESS AND THE SELF

335

reached out in the direction of last contact for objects they had just previously held. If such an object was making a continuous sound (Task 2), they reached for that object developmentally before they reached for an object that they had just previously held but that made no sound (Task 8). Thus, sound began to facilitate their reaching. However, even after sound alone began to elicit and direct their reaches (Task 3), if that sound was in directional conflict with the location of previous tactile contact with the object, the infants would reach in the direction of previous tactile contact. For example, in Task 9, sounding objects were pulled away from the children at the midline front and moved while continuing to sound in a horizontal arc to the children's fight or left. The precriterion reaches were to search for the object at the midline front, despite the fact that if the children were simply presented with a sounding object at midline fight or left (Task 3b), they reached directly to it. Likewise, if the children were playing with a continuously sounding object that dropped to the floor (Task 5b), they initially searched for the object in midair where they lost contact with it and only later were able to direct their search to the floor. Thus, it is location of last physical contact with the object rather than external cues from the object that initially directs blind children's search for the object. Their understanding of space is primarily bodycentered, and the locations of objects are perceived to be in relation to body contact rather than independent of such contact. It was Fraiberg's (1977) conjecture that, for blind children, the area directly in front of their bodies at midline was the first space to have subjective reality. B igelow (1986) documented that reaching for sounding objects in positions higher or lower than the midline (Task 7) was more difficult than reaching for sounding objects at chest level (Task 3), which suggests that Fraiberg's speculation was correct. The general precriterion reaches to Bigelow's tasks were also illustrative of the early subjective reality of the midline space. Typically, early reaches to the tasks consisted of horizontal sweeps of the air in the front of the body or bringing the hands together at midline. These are adaptive responses if it is difficult to locate a sounding object precisely in the area directly in front of the body. However, these precriterion responses occurred to silent objects and to sounding objects clearly not directly in front of the body. Thus, early in the development of searching strategies, a random horizontal search in front of the body was activated regardless of whether the object was sounding in that area or out of it, or was silent. Such a strategy was reactivated when the children were confronted with more difficult tasks, even though they no longer used it on easier tasks. It is probable that such a strategy originated because the children had prior experience discovering objects at the midline. Indeed, parents of blind children often hand their children objects, and the midline is the most frequent area from which they

336

ANN E. BIGELOW

do so. From such experience, the children may generalize to search the midline for desired objects. However, such a search is subjective and body-centered. Blind children's responses to Bigelow's sound and touch tasks suggest that they become aware that objects continue to exist even though their physical contact with the objects has ceased; they do search for objects. But their search must overcome a subjectively circumscribed space that is limited to the space directly in front of them or to the space in which they last contacted the object. They must move from random searches in front of their bodies or searches based on previous physical contact with objects to the use of cues that come from the objects for which they are searching; their search for objects must become autonomous from their previous actions or current body positions. Their understanding of the physical environment and of themselves in the physical environment must become less body-focused. 1 Lilli Nielsen (1992) has developed a miniature environment to teach blind children, with and without mental handicaps, to reach for and explore objects, which utilizes their initial subjective, body-centered space construction. Children lie on their backs in three-sided cubicles (30 x 60 x 60 cm for infants, larger cubicles for older children) called "Little Rooms." Tactually interesting objects are attached to the ceilings and inside walls of the Little Rooms. Through random hand and arm movements, the children contact the objects. With frequent sessions in the Little Rooms in which the specific objects and their positions remain constant, the children learn to reach for specific objects and to explore and compare them with other specific objects. The experience with contacting particular objects in nonvarying positions relative to their own bodies results in expectation of such contact and anticipation of particular tactile and auditory experience as a consequence of specific self-actions. The children begin to perceive their own actions as having predictable consequences in the physical environment. This experience stimulates interest in, and expectation of, using their bodies to locate and explore objects in the larger world outside of the Little Rooms. 2 Fraiberg (1977) took blind children's reaching to objects on sound cues alone as an indication that the children were in Stage 4 of object permanence; the children behaviorally indicated their belief in the existence of external objects even though the objects could not be seen. Sighted children also begin to search for occluded objects on sound cues alone in Stage 4 (Bigelow, 1983; Freedman, Fox-Kolenda, Margileth, & Miller, 1969; Uzgiris & Benson, 1980). 3 Objects that are surreptitiously hidden and then sounded elicit a search in sighted children only after entry into Stage 4. Prior to this, sound coming from the hidden objects does not direct a search. However, there is evidence that reaching for sounding objects in the dark is an earlier acquired skill than reaching for sounding objects that are

BLINDNESS AND THE SELF

337

hidden with occluding objects. Clifton and colleagues (Clifton, Perris, & Bullinger, 1991; Clifton, Rochat, Litovsky, & Perris, 1991; Perris & Clifton, 1988) found that infants 6 1/2 to 7 months of age (which is typically younger than infants in Stage 4) use sound to direct their search in the dark. Wishart, Bower, and Dunkeld (1978) tested infants between 4 and 12 months of age and found a Ushaped function with respect to reaching to sounding objects in the dark. They found early reaches at 4 and 5 months followed by a sharp decline in reaching until about 10 months, which is within the normal range for Stage 4 (8-12 months [Flavell, 1963]). Stack, Muff, Sherriff, and Roman (1989) tested infants between 2 and 7 months in a partial replication of the Wishart et al. (1978) study. They found no evidence for a U-shaped function in that they did not find an early propensity to reach to sounds in the dark. Although it is not clear w h y , for sighted children, reaching to sounding objects in the dark is easier than searching for sounding objects that are occluded (Clifton, Perris, & Bullinger, 1991), it is evident that auditory cues from unseen objects elicit reaches neither as frequent nor as accurate as do visual cues. However, coincident with Stage 4, there appears to be an increase in the use of sound to direct search. Thus, a Stage 4 understanding of objects and space may facilitate both blind and sighted children's ability to use auditory cues to find objects. Bigelow's (1986) ten sound and touch tasks have been judged to tap the development of object permanence in blind children from Stage 3 (Task 1) through Stage 4 (Tasks 2-7) and into Stage 5 (Tasks 8 [transitional task] - 10). These tasks have been used to investigate the relation between cognition and locomotion in blind children. Bigelow (1992) examined the development of selfproduced locomotion for the three blind children who mastered the complete sequence of tasks. Table 2 shows their development of locomotion as it relates to their development of object permanence. Fraiberg (1968) noted that blind children begin to crawl shortly after they begin to search on sound cues alone R the behavior she took to indicate the emergence of Stage 4 of object permanence. The results of Bigelow's (1992) study support Fraiberg's finding. There is a growing body of evidence from sighted children that crawling is related to the emergence of Stage 4 behavior. Crawling infants demonstrate advanced object permanence compared to noncrawling infants and are more likely to use allocentric search strategies (Bertenthal, Campos, & Barrett, 1984; Campos, Hiatt, Ramsey, Henderson, & Svejda, 1978; Horobin & Acredolo, 1986; Kermoian & Campos, 1988). Thus, for both blind and sighted children, their initial self-produced locomotion may be related to cognitive advances. For blind children, walking also is associated with cognitive change. Bigelow (1992) found that walking coincided with, or was subsequent to, the onset of Stage 5, despite considerable age differences in the

ge in Months for Three Blind Children's Mastery of L o c o m o t i v e Skills and Search Behaviors Indicative of ect Permanence. 1

Children 2

3

Age of Sighted Children's Acquisition

s when first seen

17

13

32

; 3 (reach for toy touching body)

20

*

*

4a

4 (reach for sound toy pulled away aest level)

23

*

*

8a

23

*

*

7b

28

17

35

30

21

34

',ition between Stages 4 & 5 (reach for t toy pulled away at ches t level) 5

(reach for sound toy pulled away

12 a

BLINDNESS AND THE SELF

339

children. Blind children appear to have a protracted Stage 4 and a more extended period between learning to crawl and learning to walk than sighted children. Although Fraiberg (1977) did not speculate on the development of object permanence in blind children beyond Stage 4, she noted that the blind children she followed also had a longer interval between the learning of these two locomotor skills. Blind children are as physically mature as their sighted peers, yet they know less about objects and space than their sighted agemates. To release themselves from the four-point tactile contact with the environment that crawling provides, the children have to be confident in their spatial/object awareness. They may need a more cognitively advanced knowledge of space and of how to locate objects within space in order to walk independently. In Stage 5, blind children locate objects by attending to the external cues coming from the objects even when the locations of such cues are in conflict with locations of last physical contact with the objects. This ability suggests an increased understanding of the location of objects independent of the child's own body space and position. Even after beginning to walk autonomously, however, blind children at times resort to a body-centered understanding of space, as evidenced in a 2-year-old blind child I observed who had been walking for a few months. When traveling through the kitchen to explore his favorite drawer, he accidentally bumped into a metal utensil that was lying on the floor. He stepped back and carefully felt down his leg to explore the space at the end of his foot in search of the bumped object, unaware that in stepping back he was now approximately a foot and a half away from the utensil. Fraiberg (1977) proposed that blind children's reaching is a product of their cognitive understanding of objects and space and that reaching and the object knowledge it implies leads to locomotion. The results of Bigelow's (1992) study support and extend her proposal. Cognitive advances in object/spatial knowledge preceded, or coincided with, advances in walking as well as crawling. Yet for sighted children, reaching and self-produced locomotion have been thought to precede advances in object/spatial awareness; indeed, locomotion has been proposed as an epigenetic event facilitating infants' understanding of space and objects (Bertenthal et al., 1984). The discrepancy may be due to differences in the way blind and sighted children acquire spatial knowledge. For both blind and sighted children, there is a cyclical relation between extension into the physical environment and advancement in object knowledge. For blind children, knowledge of the existence of objects, of how to locate objects, and that one can operate effectively in space and influence one's environment leads to the extension of self into the world. This is also true for sighted children, but sighted children acquire such knowledge directly and easily from visual information. They initially reach for objects they see. They locate objects visually and are visually motivated

340

ANN E. BIGELOW

to locomote. As sighted children move through the environment, the changing perspective produced by the flow patterns in the optic array clearly specifies where they are, where they are going, what objects are available to act upon, and where these objects are located. At a glance, sighted children have access to an immediate updating of the spatial relations between objects, and between objects and themselves. This increased knowledge and understanding further motivates them to explore and locomote. For them, one of the major incentives for walking over crawling is that they can manipulate objects while moving and exploring the environment from various perspectives (Jones et al., 1990). For blind children, the cyclical relationship may be initially generated by advancement in object/spatial awareness rather than by reaching or locomotion. They are faced with the difficult task of having to learn about objects and space without visual information. For them, reaching and subsequent forms of locomotion may be delayed until certain cognitive understandings of objects and space are acquired. As blind children learn more about objects and how to locate them in space, they become more motivated to explore and move into that space.

The Contributions of the Study of Blind Children to the Understanding of the Early Development of the Self The investigation of the development of object and spatial awareness in blind children is important to the understanding of the development of these children per se and to the development of intervention strategies that could be of use to them. Yet the study of atypical children also can be used to inform the thinking about developmental processes formed by the study of normal children. In what ways do blind children's difficulty in understanding themselves in relation to the physical environment, which underlies their delays in reaching and locomotion, inform our thinking about the development of the ecological self'? Blind children's difficulty in acquiring knowledge of objects and space affirms the importance of vision for the development of this early sense of self in relation to the physical environment. Vision is the spatial modality par excellence. The spatial information that vision allows cannot be perceived as richly or as easily in any other modality. Many studies with older blind children and blind adults demonstrate that persons blind from birth have continued difficulty with spatial understanding (Bigelow, 1991; Casey, 1978; Dodds, Howarth, & Carter, 1982; Fletcher, 1980, 1981a, 1981b; Hollyfield & Foulke, 1983; Lockman, Rieser, & Pick, 1981; Rieser, Guth, & Hill, 1982; Rieser, Hill, Talor, Bradfield, & Rosen, 1992; Rieser, Lockman, & Pick, 1980; Rosencranz & Suslick, 1976). Although blind children can readily acquire the ability to navigate to new locations once

BLINDNESS AND THE SELF

341

they are led to them, they are at a disadvantage when asked to synthesize information from different routes to produce novel routes or detours, or to synthesize spatial knowledge among objects that were not encountered sequentially. Knowledge of the overall layouts of environments is difficult because blind children must acquire knowledge of the spatial relations among objects, and between self and objects, primarily through sequential discovery. Spatial and event information that can be acquired by sighted children in a single glance must for blind children be sequentially explored, synthesized, and reconstructed. Stable landmarks used to locate the self in space are necessarily fewer for blind children, and they have less reliable information about them. Blind children's spatial orientation is usually based on sound changes, air currents, and echolocation, as well as on tactile information (Millar, 1988). The information acquired from the study of blind children also challenges some of the ideas about the development of the ecological self gleaned from the study of normal infants. In particular, it brings into question the notion that the ecological self is generated exclusively from perceptual information, and it points to the possibility that, at least for blind children, cognition also plays a role in the formation of this early sense of self. In the absence of vision and the rich spatial information it provides, blind children come to perceive their surroundings as independent from themselves. They develop the understanding that their movement changes their position in space and thus their position to objects located in that space. Their route to this understanding, however, is delayed and may be necessarily different from that of sighted children. For blind children, the ecological self is not perceived directly as it is for sighted children; rather, it is, at least in part, cognitively constructed. Cognition depends on information from perception and self-action. Thus, the interplay between cognition and perception is present in both blind and sighted children, but, because of blind childre~fs less reliable and less precise information about the physical world, the early sense of the ecological self may depend more on cognitive advances than on perception alone as in sighted children. For sighted infants, reaching is initiated by vision n seeing attractive objects and seeing their own hands move and sensing the control of that movement as a means of acquiring the objects. Sight allows children to observe the changing relation between themselves and their environments as they move through space. The constant transformation of their visual array during movement provides information about the spatial relations among objects and how the relative position of locations to self changes. For blind infants, reaching may be more closely tied to their conceptual development N to the understanding that objects exist in the external world, objects to which they can have access through their own actions. This

342

ANN E. BIGELOW

understanding is manifested initially in Stage 4 of the sensorimotor period. For blind children, reaching for objects on external cues indicates an understanding that self exists in the physical world with objects in space. The task becomes one of sorting out where self is in the physical environment relative to key objects and landmarks, and therefore, what affordances are available and what the possible effects of self-action on the specific environment are. The initial sense of self in the physical environment is body-centered. The external physical world exists and self is positioned in it, but self defines the physical environment. Space depends on body orientation and action. By Stage 5, objects' locations have become independent of self-actions. The search for objects becomes independent of previous physical contact and is guided directly from external cues from the objects themselves. Walking begins; self-produced locomotion is less tied to physical contact with the environment. For blind children, knowing where one is in the physical environment and mapping the spatial layouts of physical spaces are lifelong challenges. Yet from Stage 4 on, blind children appear to perceive self as existing in the physical world with other objects. Deciphering where to locate self and objects within the physical world becomes their focus. Sighted children also search actively for objects they cannot see in Stage 4, and difficulties focus on where to search. It is during this stage that children become proficient in the use of landmarks to direct their search and in allocentric search strategies (see Harris, 1983, for a review). It is also during this stage that sighted children show the first coordinations between the interpersonal and the ecological selves (Neisser, 1991). They become capable of joint attention to a person and an object at the same time, understanding that the specific person is attending simultaneously to them and to the object. It may be that it is in Stage 4 that the interpersonal self begins to inform the ecological self (and vice versa), which is of tremendous advantage to blind children. Indeed, without sensitive, caring others with whom to form relations, blind children are at risk, not only for autisticlike behavior, which suggests difficulties with the development of the interpersonal self, but also for delays in reaching and locomotion that are substantially beyond those of blind infants in more caring family circumstances (Fraiberg, 1977). Such extended delays in these motor skills suggest increased difficulties with the development of the ecological self. Blind children's ability to notice the effect of their actions on the environment is greatly reduced compared to sighted infants. Their ability to perceive the affordances or possibilities for self-action in given situations is limited. Therefore, they must construct their physical world in a way that is not necessary for sighted infants. Their understanding of self in relation to the environment is initially formed from a body-centered understanding of objects and space. Separating the physical world from self relies more heavily on cognitive

BLINDNESS AND THE SELF a d v a n c e m e n t s than on perceptual i n f o r m a t i o n alone.

343

A l t h o u g h the u n d e r l y i n g

p r o c e s s in the f o r m a t i o n of the e c o l o g i c a l self is the same for both blind and sighted children m

that o f n o t i n g the effect o f self-action on the p h y s i c a l

e n v i r o n m e n t m the route to the ecological self is more cognitively taxing for blind infants than it is for sighted infants. For them, the d e v e l o p m e n t of the ecological self m a y d e p e n d on the m u t u a l l y facilitative e f f e c t . o f cognitive and p e r c e p t u a l processes.

NOTES 1. Interestingly, this body-centered perspective has been proposed to underlie some of the language and play difficulties detected in older blind infants and preschoolers (Andersen et al., 1984; Dunlea, 1984). Difficulties in the early use of language and in pretend play have been used to identify disturbances in the early formation of the self in other atypical populations (Cicchetti, 1991; Cicchetti et al., 1990). Yet the differences blind children show in their use of language and in play with objects are understandable, given their different knowledge of themselves in relation to objects and space. Fraiberg and Adelson (1973) described blind children's lack of interest in pretend play. When a 3-year-old blind child was provided with a doll and a washtub filled with water, she ignored invitations to give the doll a bath and instead stepped into the washtub herself. Blind children's difficulty with locating objects in space may make it easier to put self in relation to an object than to put two objects in relation to each other. In addition, many of the toys used in pretend play are miniatures of much larger objects (e.g., dolls, toy cars, and trucks). The toy object represents the larger object in overall shape and features, but often the overall shape may be unknown to blind children. The toys do not feel, sound, or smell anything like the blind children's tactile, auditory, or olfactory experience with the larger objects. Thus, the children do not recognize them as representations of the larger objects and do not play with them accordingly. Blind children use their early language primarily to refer to their own actions, wishes, and desires. Like sighted children, their language initially is aligned closely with their perceptual experience, but this experience is reduced to only what they are interacting with or currently doing. It is the close perceptual experiences of touch, smell, and taste that are commented upon. Referents to the experience of others or events they have not participated in, both of which might be perceived with interest by sighted children, are rare in blind children. Their understanding of themselves in the physical world is restricted in a way it is not for sighted children. 2. In the original test of the effectiveness of the Little Room, 9 blind-only (mean age 12.8 months) and 6 blind-mentally handicapped (mean age 12.5 months) infants were placed in the Little Room for two to eight 20-minute sessions (Nielsen, 1992). After only two sessions, spatial activities while in the Little Room increased similarly for both blind-only and blind-mentally handicapped infants, which indicates that the passive behavior characteristic of blind children is not due to mental ability. Because mentally handicapped children take longer to learn associations, it is important that they be exposed to stimulation that teaches associations between self-action and resultant environment change from an early age. The Little Room is used with blind-only and blind-mentally handicapped children ranging in age from infancy to middle childhood. 3. Fraiberg's (1968) blind children reached for objects on sound cues alone at an average age of 10 months. Sighted children begin to use sound cues to locate occluded objects

344

ANN E. BIGELOW

between the ages of 8 and 12 months (Bigelow, 1983; Freedman, Fox-Kolenda, Margileth, & Miller, 1969; Uzgiris & Benson, 1989). Thus, Fraiberg's blind children used sound to locate objects at approximately the same age as sighted children. However, as indicated earlier in this chapter, Fraiberg's criteria for reaching to sounding objects was liberal. Bigelow (1986) found that blind children reached for objects on sound cues alone at later ages than Fraiberg's subjects, and there was variability in the ages at which Bigelow's subjects demonstrated this ability (see ages of mastery for Task 3 on Table 1 of this chapter). Blind children's rate of cognitive advancement varies considerably and, undoubtedly, is influenced by a number of factors, e.g., age of diagnosis, time spent in hospitals, reaction and adaptability of parents to their children's blindness, accessibility to services for the blind, socioeconomic status of the family. Thus, age is less meaningful in determining the function of blind children than is their cognitive level.

ACKNOWLEDGMENTS

The research conducted by the author described in this chapter was aided by Social and Behavioral Sciences Research Grant No. 12-20, from the March of Dimes Birth Defects Foundation, by the St. Francis Xavier Research Council, and by an equipment loan from the Nova Scotia Department of Social Services. Gratitude is expressed to the blind children who participated in the research and their families, to Bernadette MacLellan, Oona Landry, and Marie White who were research assistants, to Sylvia Keet who was the medical advisor, and to Sue Adams, Gary Brooks, Tara Callaghan, and Chris Tragakis for their helpful comments on earlier drafts of this chapter. REFERENCES

Adelson, E., & Fraiberg, S. (1974). Gross motor development in infants blind from birth.

Child Development, 45, 114-116.

Andersen, E. S., Dunlea, A., & Kekelis, L. S. (1984). Blind children's language: Resolving some differences. Journal of Child Language, 11, 645-664. Bayley, N. (1969). Bayley scales of infant development. New York: Psychological Corporation. Bertenthal, B. I., Campos, J. J., & Barrett, K. C. (1984). Self-produce locomotion: An organizer of emotional, cognitive, and social development in infancy. In R. N. Emde & R. J. Harmon (Eds.), Continuities and discontinuities in development (pp. 175210). New York: Plenum. Bigelow, A. (1992). Locomotion and search behavior in blind infants. Infant Behavior and Development, 15, 179-189. Bigelow, A. (1991) Spatial mapping of familiar locomotions in blind children. Journal of Visual Impairment and Blindness, 3, 301-310. Bigelow, A. (1987). Early words of blind children. Journal of Child Language, 14, 4756. Bigelow, A. (1986). The development of reaching in blind children. British Journal of Developmental Psychology, 4, 355-366. Bigelow, A. (1983). The development of the use of sound in the search behavior of infants. Developmental Psychology, 19(3), 317-3 21. Brazelton, T. B., Koslowski, B., & Main, M. (1974). The origins of reciprocity: The early mother-infant interaction. In M. Lewis & L. A. Rosenblum (Eds.), The effect of the infant on its caregiver (pp. 49-76). New York: Wiley.

BLINDNESS AND THE SELF

345

Campos, J. J., Hiatt, S., Ramsey, D., Henderson, C., & Svejda, M. (1978). The emergence of fear on the visual cliff. In M. Lewis & L. Rosenblum (Eds.), The development of affect (pp. 149-182). New York: Plenum. Casey, S. M. (1978). Cognitive mapping by the blind. Journal of Visual Impairment and Blindness, 72, 297-301. Cicchetti, D. (1991). Fractures in the crystal: Developmental psychopathology and the emergence of self. Developmental Review, 11,271-287. Cicchetti, D., Beeghly, M., Carlson, V., & Toth, S. (1990). The emergence of the self in atypical populations. In D. Cicchetti & M. Beeghly (Eds.), The self in transition: Infancy to childhood (pp. 309-344). Chicago: University of Chicago Press. Clifton, R., Perris, E., & Bullinger, A. (1991). Infants' perception of auditory space. Developmental Psychology, 27(2), 187-197. Clifton, R., Rochat, P., Litovsky, R., & Perris, E. (1991). Object representation guides reaching in the dark. Journal of Experimental Psychology: Human Perception and Performance, 17(2), 323-329. Dodds, A. G., Howarth, C. I., & Carter, D. C. (1982). The mental maps of the blind: The role of previous experience. Journal of Visual Impairment and Blindness, 76, 5-12. Dunlea, A. (1984). The relationship between concept formation and semantic roles: Some evidence from the blind. In L. Feagans, C. Gravey, & R. Golinkoff (Eds.), The origins and growth of communication (pp. 224-243). Norwood, NJ: Ablex. Flavell, J. H. (1963). The developmental psychology of Jean Piaget. New York: Van Nostrand. Fletcher, J. E. (1980). Spatial representation in blind children. 1: Development compared to sighted children. Journal of Visual Impairment and Blindness, 74, 381-385. Fletcher, J. E. (1981a). Spatial representation in blind children. 2: Effects of task variations. Journal of Visual Impairment and Blindness, 75, 1-3. Fletcher, J. E. (198 lb). Spatial representation in blind children. 3: Effects of individual differences. Journal of Visual Impairment and Blindness, 75, 46-49. Fraiberg, S. (1977). Insights from the blind: Comparative studies of blind and sighted infants. New York: Basic Books. Fraiberg, S. (1968). Parallel and divergent patterns in blind and sighted infants. Psychoanalytic Study of the Child, 23, 264-300. Fraiberg, S., & Adelson, E. (1973). Self-representation in language and play. Observations of blind children. Psychoanalytic Quarterly, 42, 539-561. Freedman, D. A., Fox-Kolenda, B. J., Margileth, D. A., & Miller, D. H. (1969). The development of the use of sound as a guide to affective and cognitive b e h a v i o r - A two-phase process. Child Development, 40, 1099-1105. Gibson, E. J. (1993). Ontogenesis of the perceived self. In U. Neisser (Ed.), The perceived self." Ecological and interpersonal sources of self-knowledge (pp. 25-42). Cambridge, MA: Cambridge University Press. Haith, M. M. (1980). Rules that babies look by. Hillsdale, NJ: Erlbaum. Harris, P. L. (1983). Infant cognition. In P. H. Mussen (Series Ed.) & M. M. Haith & J. J. Campos (Vol. Eds.), Handbook of child psychology, Vol. 2. : Infancy and developmental psychobiology, (4th ed., pp. 689-782). New York: Wiley. Hiroto, D. S., & Seligman, M. E.P. (1975). Generality of learned helplessness in man. Journal of Personality and Social Psychology, 31, 311-327. Hobson, R.P. (1993). Through feeling and sight to self and symbol. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 254279). Cambridge, MA: Cambridge University Press. Hobson, R. P. (1990). On the origins of self and the case of autism. Development and Psychopathology, 2, 163-181.

346

ANN E. BIGELOW

Hollyfield, R. L., & Foulke, E. (1983). The spatial cognition of blind pedestrians. Journal of Visual Impairment & Blindness, 74, 373-380. Horobin, K., & Acredolo, L. (1986). The role of attentiveness, mobility history, and separation of hiding sites on Stage IV search behavior. Journal of Experimental Child Psychology, 41, 114-127. Jones, D., Biringen, Z., Butterfield, P., Henderson, C., Robinson, N., Aman, C., Emde, R.N., & Campos, J.J. (1990, April). Affective development and walking onset. In R. Telzow (Chair), Locomotor experience and psychological development. Symposium conducted at the meeting of the International Conference on Infant Studies, Montreal. Kermoian, R., & Campos, J. J. (1988). Locomotor experience: A facilitator of spatial cognitive development. Child Development, 59, 908-917. Lamb, M. E. (1981a). The development of social expectations in the first year of life. In M. E. Lamb & L. R. Sherrod (Eds.), Infant social cognition: Theoretical and empirical considerations (pp. 155-175). Hillsdale, NJ: Erlbaum. Lamb, M. E. (1981b). Developing trust and perceived effectance in infancy. In L. P. Lipsitt (Ed.),Advances in infancy research. (Vol. 1, pp. 101-127). Norwood, NJ: Ablex. Lockman, J. J., Rieser, J. J., & Pick, H. L. Jr. (1981). Assessing blind travelers' knowledge of spatial layout. Journal of Visual Impairment and Blindness, 75, 321-326. Loveland, K. A. (1993). Autism, affordances, and the self. In U. Neisser (Ed.), The perceived self." Ecological and interpersonal sources of self-knowledge (pp. 237-253). Cambridge: Cambridge University Press. Maurer, D. (1985). Infants' perception of facedness. In T. M. Field & N. A. Fox (Eds.), Social perception in infants (pp. 73-100). Norwood, NJ: Ablex. Millar, S. (1988). Models of sensory deprivation: The nature/nurture dichotomy and spatial representation in the blind. International Journal of Behavioral Development, 11(1), 69-87. Murray, L., & Trevarthen, C. (1985). Emotional regulation of interactions between twomonth-olds and their mothers. In T. M. Field & N. A. Fox (Eds.), Social perception in infants (pp. 177-197). Norwood, NJ: Ablex. Nielsen, L. (1992). Space and self Copenhagen: Sikon. Neisser, U. (1993). The self perceived. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 3-21). Cambridge: Cambridge University Press. Neisser, U. (1991). Two perceptually given aspects of the self and their development. Developmental Review, 11, 197-209. Neisser, U. (1988). Five kinds of self-knowledge. Philosophical Psychology, 1, 35-59. Norris, M., Spaulding, P., & Brodie, F. (1957). Blindness in Children. Chicago: University of Chicago Press. Perris, E. E., & Clifton, R. K. (1988). Reaching in the dark toward sound as a measure of auditory localization in infants. Infant Behavior and Development, 11,473-491. Rieser, J. J., Guth, D. A., & Hill, E. W. (1982). Mental processes mediating independent travel: Implications for orientation and mobility. Journal of Visual Impairment & Blindness, 76, 213-218. Rieser, J. J., Hill, E. W., Talor, C. R., Bradfield, A., & Rosen, J. (1992). Visual experience, visual field size, and the development of nonvisual sensitivity to the spatial structure of outdoor neighborhoods explored by walking. Journal. of Experimental Psychology: General, 121(2), 210-221. Rieser, J. J., Lockman, J. J., & Pick, H. L. Jr. (1980). The role of visual experience in knowledge of spatial layout. Perception and Psychophysics, 28, 185-190.

BLINDNESS AND THE SELF

347

Rosencranz, D., & Suslick, R. (1976). Cognitive models for spatial representations in congenitally blind, adventitiously blind, and sighted subjects. New Outlook for the Blind, 70, 188-194. Seligman, M. E. P., & Maier, S. F. (1967). Failure to escape traumatic shock. Journal of Experimental Psychology, 74, 1-9. Spelke, E. S., & Cortelyou, A. (1981). Perceptual aspects of social knowing: Looking and listening in infancy. In M. E. Lamb & L. R. Sherrod (Eds.), Infant social cognition." Empirical and theoretical considerations. Hillsdale, NJ: Erlbaum. Stack, D., Muir, D., Sherriff, F., & Roman, J. (1989). Development of infant reaching in the dark to luminous objects and invisible sounds. Perception, 18, 69-82. Stern, D. N. (1985). The interpersonal world of the infant: A view from psychoanalysis and developmental psychology. New York: Basic Books. Stern, D. N. (1974). Mother and infant at play: The dyadic interaction involving facial, vocal, and gaze behaviors. In M. Lewis & L. A. Rosenblum (Eds.), The effect of the infant on its caregiver (pp. 187-213). New York: Wiley. Streri, A. (1993). Seeing, reaching, touching: The relations between vision and touch in infancy. (T. Pownall & S. Kingerlee, Trans.). Cambridge, MA: MIT Press. (Original work published 1991) Tronick, E., Als, H., Adamson, L., Wise, S., & Brazelton, T. B. (1978). The infant's response to entrapment between contradictory messages in face-to-face interaction. Journal of the American Academy of Child Psychiatry, 17, 1-13. Uzgiris, I. C., & Benson, J. (1980, April). Infants' use ofsound in search for objects. Paper presented at the meeting of the International Conference on Infant Studies, New Haven, Connecticut. Watson, J. S. (1985). Contingency perception in early social development. In T. M. Field and N. A. Fox (Eds.), Social perception in infants (pp. 157-176). Norwood, NJ: Ablex. Watson, J. S. (1979). Perception of contingency as a determinant of social responsiveness. In E. B. Thoman (Ed.), Origins of the infant's social responsiveness (pp. 33-64). Hillsdale, NJ: Erlbaum. White, B. L. (1969). The initial coordination of sensorimotor schemes in human infants Piaget's ideas and the role of experience. In D. Elkind & J. H. Flavell (Eds.), Studies in cognitive development: Essays in honor of Jean Piaget (pp. 237-256). New York: Oxford University Press. Wishart, U. G., Bower, T. G. R., & Dunkeld, J. (1978). Reaching in the dark. Perception, 7, 507-512.

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

349

CHAPTER 17

Intermodal Origins of Self-Perception LORRAINE E. BAHRICK

Florida International University

Introduction The self is certainly the first and one of the most intriguing sources of stimulation that the infant encounters. We create a diversity of rich and varied types of stimulation to all the sensory systems through our actions in the context of a changing environment. The infant moves her hands to her face and mouth; sees and feels her body moving and hears her own vocalizations; and at the same time sees, hears, and feels people and other objects moving around her. Through our exploratory activities, we discover properties and affordances of objects and events in our environment, and at the same time and in the same way, our sensory systems specify the nature of our own actions and properties of the self. Information about the self accompanies information about the environment, and the two are inseparable. Egoreception accompanies exteroception, like the other side of a coin. Perception has two poles, the subjective and the objective, and information is available to specify both. One perceives the environment and coperceives oneself (J.J. Gibson, 1979, p. 126). In the quote above, Gibson (1979) points out the reciprocal nature of exploring the self and the environment. He suggests that when we look at our environment, we perceive objects and events, and at the same time we obtain information about the position and motion of our head, body, arms, and hands. When we walk, optical information (flow patterns) specifies our changing position in space and at the same time the unchanging positions of objects and surfaces in the environment. We perceive ourself in relation to the environment. In this manner, knowledge of the self and the world develops hand in hand. What are the origins of our unique abilities for self-reflection, possession of a self-concept, and understanding of self as both subject and object? It is suggested here that the self-understanding of adults develops from the perceptual experiences

350

LORRAINE E. B A H R I C K

of infancy. The action systems of the infant interact with the environment to create at first a preconceptual and then a more explicit and accessible basis of knowledge about the self. This chapter will focus on the development of two kinds of knowledge about the self in early infancy: intermodal information relating one's seen and felt motions, and featural information about one's visual appearance. Knowledge about these aspects of self derive from different sources of information and thus may develop separately, but in an interrelated manner. The provocative question of, when do infants know that the stimulation from the self specifies the self, will also be explored in the context of the infant's increasing sensitivity to these two sources of information.

History Accounts of the nature and development of self-perception abound. They are characterized by diverse taxonomies, differing theoretical perspectives, and divergent empirical approaches. James (1890) distinguished between two primary aspects of the self: the existential self, or "I," and the empirical self, or "me." This distinction persists in our current thinking. The "I" is considered the "knower," the agent of action, the self as subject, and is experienced through four kinds of awareness: agency, distinctness, continuity in time, and ability to reflect on the self. The "me" is considered to be the self as object, the sum of all one's parts, including material, social, and spiritual aspects (see reviews by Butterworth, 1990; Damon & Hart, 1982, 1988; Harter, 1983). Along similar lines, Rochat (in press) has recently described the "I" as the "situated self," stemming from perception of the self as a separate entity in the environment, and the "me" as the "identified self," entailing conceptual knowledge about the self. How does this self-knowledge develop? Baldwin (1902), Freud (1922), and, more recently, Mahler and'Furer (1968) describe the infant as lacking the capacity for self-awareness and born into an adualistic state of fusion with the environment. Piaget (1954, 1967) also espoused this view and described the "adualistic confusion" as characterizing much of the first year of life. Infants, according to Piaget, experience no distinction between the self and the not-self. Not until the age of 8 or 9 months do they come to gradually differentiate themselves from other objects and events in the environment. In contrast with these traditional views, there is now a resurgence of interest in the early origins of self-perception in infancy. Recently, a number of investigators have rejected the adualistic view and have posited that infants are capable of differentiating some aspects of self during the first days of life (e.g., Butterworth, 1990, 1992; Gibson, 1993; Lewis, 1979; Meltzoff, 1990; Neisser,

INTERMODAL ORIGINS OF SELF-PERCEPTION

351

1988, 1993; Rochat, in press; Samuels, 1986). For example, Rochat (in press) proposes that the perception of self as a separate and causal agent ("situated self") develops in the first months of life and is a developmental precursor to the conceptual self ("identified self'). The Gibsons (E.J., 1969; J.J., 1966; 1979) provide one of the most well-developed accounts of the origins and development of self-perception, which is based on perceptual experience. Their ecological view rejects the notion of early fusion with the environment and emphasizes the interdependence between perception of the self and perception of the world (as conveyed by the quote above). All the senses have both a propriospecific (specifying self) and an exterospecific (specifying other) function. Thus, the act of perceiving entails both self-perception and perception of the environment at the same time. This enables infants to directly perceive a differentiated self from the beginning. According to this view, the infant comes into the world prepared to detect invariant information specifying the self through all the senses. For example, vision provides powerful information for the self through changes in the optic array that result from one's motion. Posture and locomotion are controlled through vision, even in young infants (see Butterworth, 1990, for a review). Research in the areas of perceptual and cognitive development has proliferated in recent years and has shaped our view of the infant's developing sense of self. It suggests that young infants have a growing awareness of self across the first months of life. Body awareness is demonstrated through patterns of tactual exploration and self-directed behavior in neonates; they systematically explore their own bodies and anticipate the arrival of their hand to their mouth by mouth opening (e.g., Rochat, Blass, & Hoffmeyer, 1988; see Butterworth, 1990, for a review). Even neonates show imitation of facial expressions, demonstrating an intermodal representational system for the body (Meltzoff & Moore, 1977; 1983). Furthermore, young infants correct their imitative responses to gradually approximate the gestures of the model, suggesting that they have access to proprioceptive information (Meltzoff & Moore, 1994). Visually guided reaching is present in neonates (Hofsten, 1980), and young infants can adapt the trajectory of their reach to catch a moving object (Hofsten, 1983) and show anticipatory hand-shaping when reaching for objects of different sizes (Bower, B roughton, & Moore, 1970). Young infants detect the contingency between their leg motion and that of an attached mobile (Rovee-Collier & Fagen, 1981) and quickly narrow down their response to the one limb that is attached to the mobile (Rovee& Rovee, 1969). Infants also use visual information to adapt their posture (e.g., Bertenthal & Bai, 1989; Butterworth & Hicks, 1977; Lee & Aronson, 1974), and even neonates are able to compensate for different gravitational forces when moving their limbs (Van der Meer, 1993). Preliminary results also indicate that 1month-old infants distinguish between a touch on the cheek by their own hand

352

LORRAINE E. BAHRICK

versus a touch by that of another object and show a rooting response only to objects other than the self (Rochat, in press). Furthermore, by 1 to 2 months of age, looming objects elicit avoidance reactions (Ball & Tronick, 1971; Nanez, 1988; Yonas, Pettersen, & Lockman, 1977). These abilities show an impressive awareness of the body in space and time. They demonstrate how infants adjust their actions to environmental change, and they support Gibson's ecological view that exploration involves perception of both the world and the self at the same time, and in the same way. These converging findings are consistent with the view that infants perceive the self as a separate entity in the first months of life and are capable of both accommodating to changes in the environment and acting as agents of change on the environment. The evidence for this conclusion is robust and comes from a variety of domains that tap different response systems and utilize different procedures. However, the evidence is also indirect and therefore must be viewed as a working hypothesis at present.

O v e r v i e w and Definition of "Self' This chapter explores the intermodal bases for self-perception in early infancy. I will review evidence from a series of studies conducted in our lab that suggests that infants make important strides toward perceiving and understanding the self, even during the first half year of life. Early experience serves to establish a foundation of knowledge about the self that is an antecedent to later self-awareness and selfunderstanding. The view of self articulated here is consistent with that developed by the Gibsons. A central tenant of this view is that self-knowledge is rooted in the perceptual experiences of infancy. Further, the infant is seen as a differentiated entity from the start, capable of perceiving the self and the enviromnent in relation to one another through detecting invariant relations. Self-perception provides the basis for knowledge about the self and for the development of a conceptual understanding of the self. The perceptual and conceptual modes of experiencing the self are viewed along a continuum. This has been succinctly pointed out belore: "To perceive the environment and to conceive it are different in degree, not in kind. One is continuous with the other" (Gibson, 1979, p. 258). "Just as perceiving and conceiving the environment are different in degree but not in kind, so are perceiving and conceiving oneself" (Grene, 1993, p. 117). Thus, as Grene suggests, the conceptual understanding of self emerges from the perceptual experiences of infancy. Here, the self is considered to be a constellation of perceptions, beliefs, and knowledge about different aspects, including the self as a differentiated entity: as having a unique identity that persists through time and space, as having a

INTERMODAL ORIGINS OF SELF-PERCEPTION

353

particular visual appearance, a way of moving, sound of one's voice, odor and taste, being a causal agent, a social being and of a certain gender, to name some aspects. Other investigators recently have also emphasized the diverse set of phenomena that fall under the umbrella of "self." Samuels (1986) identified 11 potential bases for self-perception in infancy, including body boundaries, perceptual invariants, and social contingencies. Cer.tainly the diversity of topics represented in this volume attests to the multifaceted nature of our concept of the self, as well as to the resurgence of interest in the topic. Neisser (1988) distinguishes between five kinds of self-knowledge, each based on different sources of information, including the "ecological" and "interpersonal" selves, both based on perceptions; the "extended" self, based on memory; the "private" self, based on conscious experience and feelings; and the "conceptual" self, based on beliefs and assumptions. Thus, Neisser has differentiated aspects of "the self" according to the type of information accessed, that is, perceiving, remembering, experiencing/feeling, or conceiving. According to Neisser (1993, p.3), "each kind specifies a different aspect of the individual and thus implicitly defines a different sort of self." I find it more useful, however, to view the self as having many related aspects or domains defined by content. These domains or aspects cut across the five kinds of self described by Neisser (1993). For example, the sound of my voice, the appearance of my face, or the contingency that specifies I am an agent of action are all aspects that are perceived, remembered, experienced, and conceived. What is perceived today becomes remembered or conceived tomorrow. Given this view and the assumption that perception and conception lie along a continuum, it is likely that knowledge of different aspects of the self develops at different rates. Thus, at one time, the infant may posses only perceptual information about one aspect (e.g., appearance of the face) and a conceptual understanding of another (e.g., I cause things to happen in the world). If the distinction between perception and conception is one of degree, there can be no clear boundary between them, and the point where conceptual understanding begins and perceptual knowledge leaves off cannot beuniformly defined or clearly demarcated. It is therefore not surprising that studies with different procedures and content areas yield divergent criteria and ages of onset for conceptual knowledge of self. To summarize the view presented here, the development of self-knowledge begins with the perceptual experiences of infancy. There is but one self with many facets. Exploration in infancy leads to the progressive elaboration of different aspects of self, including differentiation of aspects of both the "I" (perceptual) and then the "me" (conceptual) within a given domain. Development of knowledge about various aspects of self may proceed at different rates, in different ways, and on the basis of different sources of information. Yet the overall effect is the

354

LORRAINE E. BAHRICK

evolution of a coordinated system of knowledge. This is seen as a gradual process of differentiation of both relatively distinct yet interrelated processes, where knowledge in one domain may inform exploration in another, and differentiation of some aspects precedes and facilitates differentiation of others. Knowledge about the self, of course, also evolves in the context of our general knowledge about the world, and is both influenced by and influences the attainment of general developmental milestones. Much research lies ahead before these processes are identified and their interrelations delineated. This chapter reports a first step in this direction. My focus is on the development of two different and interrelated bases for self-knowledge: visual information about the self and contingent feedback from self-motion. The "visual self" includes what is commonly referred to as one's appearance. It consists of all the visual characteristics of the face and body that distinguish us from others, that is, "what I look like." The "contingent self" entails the perception of a relation between one's own actions and their consequences, which in this case is the resulting visual stimulation. This is an intermodal relation between vision and proprioception. It is also an important basis for perceiving causality and the self as an agent of change in the environment. Before reviewing research on these two topics, let us put this in the context of prior research on self-recognition in children and toddlers.

Studies of Self-recognition In the past, systematic studies of the development of self have primarily focused on the development of self-recognition in toddlers and young children, using behavior in from of the mirror as an index of self-recognition. In a paradigm originated by Gallup (1970) for use with primates, the child's nose is Unobtrusively marked with a rouge spot, and the child is placed in front of the mirror. The age at which children first show mark-directed behavior and attempt to wipe off the spot has been considered the onset of self-recognition. It is assumed that this behavior reflects the existence of a mental representation of self against which the mirror image is compared. Mark-directed behavior typically emerges sometime between 15 and 20 months and is typical by about 2 years of age (Amsterdam, 1972; Johnson, 1982; Lewis & Brooks-Gunn, 1979; Loveland, 1987; Schulman & Kaplowitz, 1977; see Anderson, 1984; Cicchetti, Beeghly, Carlson, & Toth, 1990; Damon & Hart, 1982; and Harter, 1983, for reviews). The attempt to wipe off the rouge spot is inferred to indicate self-recognition on the basis of featural and contingency information. It entails an understanding that the self has stable features that do not include a rouge spot on the nose. It may reflect some degree of conceptual understanding of self (the "me") rather than

INTERMODAL ORIGINS OF SELF-PERCEPTION

355

perceptual knowledge of self (the "I"). Furthermore, this ability is delayed in children who are mentally retarded (e.g., Mans, Cicchetti, & Stroufe, 1978), which indicates its association with cognitive development. Nevertheless, it appears that the rouge-spot task provides a stringent test of self-recognition, given its reliance on a number of complex abilities that must also undergo development. For example, the infant's understanding of the properties of reflecting surfaces develops with age (Loveland, 1986), and the response system used to index self-recognition (mark-directed behavior) may also develop with age and is thus not well suited for use in testing young infants. Prior to the time when toddlers demonstrate self-recognition by their behavior in the rouge-spot mirror task, researchers have observed a progression of diverse behaviors in front of the mirror or video image of self, with varying consistency across studies (see reviews by Damon & Hart, 1982, 1988; and Harter, 1983). Dixon (1957) postulated four stages of behavior in front of mirrors, including: 1) interest in the mother's reflection but not their own (4 months); 2) social behavior toward their mirror image (5-7 months); 3) distinguishing their own image from that of another infant (7-12 months); and 4) avoidance of their image. Amsterdam (1972) and Schulman and Kaplowitz (1977) also found a social stage, and Amsterdam noted an avoidant stage (13-24 months). Because the mirror provides both featural and contingency information for self, Lewis and Brooks-Gunn (1979) attempted to separate these sources of information using video images of the self. They tested recognition on the basis of mirror contingency while controlling for featural information by showing a live video image of self paired with a prerecorded image of self. Discrimination between the two types of images was found at 9 months (see also Amsterdam & Greenberg, 1977). However, it was not until 15 months that infants were able to discriminate a prerecorded film of themselves from that of another child, indicating their ability to use featural information for self-recognition in the absence of contingency. The authors hypothesized that contingency information is an important basis of selfrecognition and that self-knowledge develops in four stages: 1) an attraction to faces of infants; 2) recognition of self through contingency; 3) recognition of one's permanent features; and 4) self defined by categorical features independent of contingency. Butterworth (1992) succinctly summarized the literature in this area and delineated five stages of responding to mirror and video images of self, similar to those proposed by Lewis and Brooks-Gunn: 1) attraction to images of others (0-3 months); 2) contingency detection (3-8 months); 3) awareness of the self as a permanent object (8-12 months); 4) differentiation of the infant's own image from that of others (12-15 months); and finally 5) facial feature recognition (15 months-2 years). Other investigators have focused on delineating the relation

356

LORRAINE E. BAHRICK

between mirror behavior and stages of cognitive development. For example, Bertenthal & Fisher (1978) found a predictable sequence of five behaviors in infants between the ages of 6 and 24 months that correlated with object concept development: touching one's image, using the mirror to locate a hat attached to self, locating a toy, succeeding in the rouge-spot task (at stage 6 of object concept development), and finally naming one's image. However, Loveland (1986) describes the changes in children's behavior in front of the mirror as reflecting a developing understanding of the nature and properties of reflecting surfaces. That is, the mirror is a special tool for mediated perception, whose affordances take years to discover. This word of caution is well taken and highlights the importance of using convergent approaches in exploring complex phenomenon such as the development of knowledge about the self. Only a few studies have empirically tested self-perception in infants younger than 9 months using videos or other visual representations of self. Bennett, Smith, and Loboschefski (1992) recently found that 5-month-olds could discriminate a photo of their own face from those of same-aged peers following preexposure to a moving video display of a peer, but not following preexposure to the self. Papousek and Papousek (1974) found that 5-month-olds discriminated a live video image of self from a prerecorded image of self. Infants showed preferential visual fixation of the noncontingent, prerecorded display. In a subsequent study (Field, 1979), 3-month-olds responded differently to a contingent mirror image of self and a noncontingent presentation of a peer. They looked more to the self, but smiled and vocalized more to the peer. Although both of these studies suggest that discrimination of contingent information elicited by the self emerges in early infancy, both had procedural confounds that made interpretation difficult. First, by recording the noncontingent video film of self under different conditions from that of the contingent film, one introduces the possibility that amount of body motion displayed by infants in the different conditions may have differed and then served as a basis of discrimination. Second, the use of the face creates a potential confound of differential eye contact and eye motion in contingent and noncontingent displays because the mirror image provides constant eye contact. Papousek and Papousek (1974) attempted to separate eye contact and contingency by using video controls. They did find an effect of eye contact; however, the overall effect of contingency was still evident. Bahrick and Watson (1985)conducted a study with 5-month-olds to determine whether they could in fact discriminate a video film of self from that of another infant on the basis of contingency alone, using a method that eliminated the above confounds. This research is discussed in detail in the next section. The research reviewed here has revealed a developmental progression of infants' behavior in front of the mirror, leading to evidence of serf-recognition in the rougespot studies by the age of 15 months. The understanding that the mirror image

INTERMODAL ORIGINS OF SELF-PERCEPTION

357

signifies the self is implicit in the child's meaningful response of attempting to remove the rouge spot using the mirror reflection as a guide. However, one must be cautious in making cross-age comparisons about psychological phenomenon on the basis of a response system that may also undergo development (e.g., see Porges, 1979). The fact that children under the age of 15 months fail to wipe off a rouge spot reflected in the mirror does not necessarily mean that they fail to perceive the image as specifying the self or as related to the self. It may reflect their developing awareness of the properties of reflecting surfaces or a development in the child's motivation to remove a spot on the nose, for example. Further, it should be noted that control conditions assessing the ability of younger children to attempt to wipe off a spot that was directly visible on the self or on another person were not typically included in these studies. Consequently, it appears that this measure is more appropriate for toddlers and young children; therefore, the age onset of self-recognition has been overestimated. Assessing the "meaning" of stimulation to the infant is a difficult task and requires innovative research designs and convergent methods. By using other measures more appropriate for young infants, the origins of self-perception and self-recognition can be more effectively examined at younger ages. Two sets of studies are described in the next sections that answer many of the methodological concerns raised here.

Investigations of "The Contingent Self" The perception of a contingency between one's behavior and its effects on something in the environment is an extremely important accomplishment. It defines the infant's early orientation toward the world with respect to effectineness and competence, on the one hand, and helplessness, on the other. It forms the basis for the infant's sense of self as an agent of action in the environment. Research has demonstrated that young infants are competent perceivers of contingencies. For example, they detect the relation between leg kicks, sucking, or vocalization and various forms of auditory and visual stimulation including lights, tones, and the movements of a crib mobile (e.g., DeCasper & Fifer, 1980; Kalnins & Bruner, 1973; Rovee-Collier & Fagen, 1981; Siqueland & DeLucia, 1969; Watson & Ramey, 1972). They are also sensitive to social contingencies (e.g., Murray & Trevarthan, 1985). However, these types of contingencies differ from those provided by mirror or video stimulation. The former are imperfectly related to the baby's behavior. When the infant engages in a social interaction (e.g., the infant smiles, the mother smiles) the infant's behavior has only a moderate probability of eliciting the mother's behavior, and the mother's behavior occurs with a certain probability in the absence of the infant's behavior (see Watson,

358

LORRAINE E. BAHRICK

1985). Although easily detected by the infant, these kinds of contingent relations are far from perfect. Furthermore, when the infant turns on a light display by kicking his leg (e.g., Watson, 1979), the onset of the visual display is contingent on the frequency of the infant's behavior, but the direction, intensity, and duration of motion does not covary with the light display. When the infant moves his leg, causing an attached crib mobile to move, his behavior is conjugately related to the movement of the mobile. Neither provides stimulation that is perfectly isomorphic with the infant's behavior like that of mirror or video stimulation. Bahrick and Watson (1985) hypothesized that the imperfect contingency tested in these studies specifies to the infant a class of social objects that have potential for interaction. In contrast, the perfect contingency provided by mirror or video stimulation specifies the self, and may serve as an early basis for distinguishing between what is self and what is not self. Stimulation from the mirror provides a perfectly contingent relation between how one moves (proprioception) and the consequent visual stimulation from that motion. The observer can feel his or her own body motions through proprioceptive feedback and can observe the consequent visual stimulation. No object other than the self is capable of providing stimulation that is perfectly correlated with one's felt motions. Consistent with the hypothesis of Lewis and Brooks-Gunn (1979), we suggested that the contingency between visual and proprioceptive information for one's body motion could serve as an important basis for self-perception in early infancy. For example, this kind of information is available each time the infant moves her body; the infant can both see and feel her hand opening and closing. This information is invariant, amodal, and specifies the self (Gibson, J.J., 1966, 1979; E.J., 1969). The proprioceptive-visual contingency is available from birth onward and may potentially provide the basis for one of the earliest forms of self- perception. Perception of this contingency arising from body motion can provide a simple and reliable basis for distinguishing self from not-self. Bahrick and Watson (1985) assessed the ability of 5-month-old infants to make use of this kind of contingency in distinguishing a video of self from one of another infant. In a series of four experiments, we presented infants with a live video film of their own legs moving (contingent display), side by side with a film of another infant's legs, or a prerecorded film of their own legs (noncontingent displays). All infants were fitted with yellow booties prior to filming, and their legs and feet were portrayed in an inverted position on the screen, much like the infant would experience by looking down at his own legs (see Figure 1). Infants received four 60-second trials of the contingent and noncontingent displays side by side. The lateral positions of the two displays were counterbalanced across subjects. The design of the study corrected for several confounds inherent in prior video and mirror studies. The problem of differential eye contact between live

INTERMODAL ORIGINS OF SELF-PERCEPTION

359

versus prerecorded displays or displays of other infants was eliminated by using images of the infant's legs rather than face. Featural differences were minimized across displays of self and other by fitting all infants with yellow booties. The potential for differential amounts of body motion across different types of displays was also eliminated by using a yoked control design. That is, each infant's live film served as the prerecorded film for the next infant. Finally, in other mirror and video studies, the infants had visual access to the motions of their own body; thus, the question of whether infants actually detect proprioceptive information could not be addressed (Field, 1979; Papousek & Papousek, 1974). Visual access permitted detection of an intramodal (rather than intermodal) contingency. That is, a direct comparison between the visual stimulation from the mirror and the visual stimulation from one's body motion could be made. To eliminate this possibility, the infant's direct view of his own body was occluded in Experiments 2-4.

FIGURE 1. An example of the visual displays portraying a live and a prerecorded film of the infant's legs. From Bahrick & Watson (1985). In three separate studies, we found that 5-month-old infants showed a significant visual preference for the noncontingent display. In Experiments 1 and 2, where infants saw a live film of their own legs alongside that of another infant's legs, they showed significant preferences (p <.005, p <.001, respectively)) for the films of the other baby. This occurred even when the infant's view of their own legs was occluded in Experiments 2 and 3. Moreover, in Experiment 3, when featural differences between the contingent and noncontingent displays were eliminated

360

LORRAINE E. BAHRICK

altogether (by presenting two films of the baby's own legs), infants again showed a robust preference (p <.001) for the noncontingent display. That is, they watched a prerecorded film of self more than a live film of self. These findings demonstrated that 5-month-old infants are capable of detecting the perfectly contingent relation between their own body motion and the visual stimulation provided by their live video image: However, they showed more interest in the imperfect contingency provided by a film of another infant's legs. A further study was conducted to assess performance of younger infants in this task. Three-month-old infants viewed a live film of their own legs and a film of another infant's legs side by side, as in Experiment 2. Results, however, indicated no significant preference for one display over the other. Given that 3-month-olds detect contingency under other conditions (e.g., Rovee-Collier & Fagen, 1981; Watson & Ramey, 1972), it seemed unlikely that they failed to detect the contingent relations in this setup. Further analyses revealed a distribution that was significantly bimodal. That is, some infants predominately watched the contingent display of self, whereas others watched the noncontingent display of the other infant. Consistent with other recent findings (Bahrick & Pickens, in press; Hunter & Ames, 1988), it appeared that visual preferences were in transition from familiarity to novelty, yielding a null preference overall. Thus, it may be that 3 months is an age of transition from the infants' preference to explore the perfect contingency generated by their own body motions to the imperfect contingency afforded by social objects. These findings, along with those of Field (1979), which indicate that 3-month-olds look more to their mirror image than to a peer, suggest the possibility that at some point prior to this age, infants may have a general visual preference for the stimulation from self. Together, this research demonstrates that by 5 months, and possibly earlier, infants are sensitive to the intermodal proprioceptive-visual contingency specifying self and self-motion. This ability is viewed as a fundamental basis for knowledge about the self and may underlie early differentiation of self from other. Rochat (Rochat & Morgan, 1995, this volume) recently extended our investigation of temporal contingency to the domain of spatial contingency. Our research manipulated temporal contingency and held spatial contingency constant across the side-by-side presentations. Rochat and Morgan's work complements ours by manipulating spatial contingency and holding temporal contingency constant across the side-by-side presentations. They presented 3- and 4 1/2-monthold infants with two live films of their own legs side by side. The legs were presented from different perspectives, including an ego versus an observer view, with or without a right/left reversal. In one study, infants saw two ego views of their legs, one normal and the other with a right/left reversal. This caused a discrepancy in the spatial mapping and the direction of movement of the video image, with respect to the infant's legs. Results indicated that infants at both ages

INTERMODAL ORIGINS OF SELF-PERCEPTION

361

demonstrated a significant preference for the incongruent display with the right/left reversal. This shows a sensitivity to spatial contingency; in particular, the directionality of the leg motion. A further study found no evidence of sensitivity to spatial orientation per se when right/left was not reversed. These results extend those of Bahrick and Watson (1985) by demonstrating that infants detect the visual-proprioceptive contingency generated by their own motion on the basis of spatial as well as temporal information. A further extension of our research investigated both temporal and spatial contingency and was conducted using films of the infant's' arms and hands (Schmuckler, 1994, this volume). In a study identical to ours except for the body part filmed, Schmuckler presented 5-month-olds with a live display of their own arm and hand alongside a prerecorded fdm of another infant's moving ann and hand, both presented from the same ego view. Infants showed a significant preference for the noncontingent, prerecorded film. This result replicates both our findings of sensitivity to temporal contingency and the direction of this effect. In a second study, which was identical to the first except that both displays were presented with a right/left reversal, no significant preferences were found. This converges with Rochat's (Rochat & Morgan, 1995, this volume) findings and suggests the importance of spatial congruence in the directionality of motion for detection of temporal contingency. Also consistent with Rochat's findings, a third study demonstrated a significant preference for the noncontingent display when the views of both the live and prerecorded displays were novel for the infant (filmed underneath the palm), but right/left orientation was properly aligned. This suggests that as long as the right/left directionality of motion is preserved, spatial orientation per se is not critical for perceiving temporal contingency. Taken together, the results of these three sets of studies show a remarkably consistent pattern. By 5 months, infants are sensitive to proprioceptive information for the temporal contingency and directionality of movement of their bodies with respect to a visual display of that motion, and they show this sensitivity by selectively watching the more novel, noncontingent display. This research raises a provocative question: Does the infant know that the legs/arms displayed by the live video image are their own? That is, does the contingency provided by this stimulation actually specify self to the infant, and if not, at what age does this understanding emerge? It is quite possible for infants to show discrimination of self from other on the basis of contingency or between one view of self and another, without attributing the contingent stimulation to the self. That is, the preferences could be based on familiarity with contingency rather than a full understanding of the meaning of the stimulation. This is an important question that cannot be addressed by the present data, but is the focus of current

362

LORRAINE E. BAHRICK

research in our lab and is revisited below in the context of research on the visual self.

Investigations of "The Visual Self ~'

The visual self can be seen as having two overlapping components, the "visualfeatural self" and the "visual-dynamic" self. The visual-featural self refers to what we typically think of as our visual appearance: "what I look like." It includes all of our permanent features and distinctive visual qualities, such as hair style, body type, facial configuration, etc. It is consistent with our commonly held definition of self-recognition. Mirror and video stimulation also provide information about dynamic visual qualifies. The dynamic information for self consists of distinctive motion patterns, body postures, idiosyncratic gestures or facial expressions, and the relative movement of facial features that are typical of an individual. This information is also visually conveyed and may serve as a basis of self-recognition. Featural and dynamic information overlap in that motion provides excellent information for the visual appearance of an object (Gibson, 1969). For example, through motion, infants can abstract invariant relations that specify the shape of an object more easily than when the object is still (e.g., Owsley, 1983; Kellman & Spelke, 1983). As a first step toward assessing the development of knowledge about the visual self, we conducted a study to determine whether and at what age infants could differentiate between a video film of their own face and that of an agematched peer (Fadil, Moss, & Bahrick, 1993). Although prior research had documented excellent discrimination of both moving and still faces by infants and even neonates (e.g., Barrera & Mauer, 1981; Bushnell, 1982; Fagan, 1972, 1976), no researchers had yet assessed whether young infants could discriminate between the self and a peer solely on the basis of visual-featural information. Results of a study by Lewis and Brooks-Gunn (1979) suggested this ability did not emerge until the age of 15 months, when infants first distinguished between their own pretaped video image and that of another child. Infants of 5 (N= 24) and 8 months (N = 32) were tested in a visual preference test under both a moving and a still condition. The infant's own face and upper body (wearing a yellow bib) was prerecorded while he/she watched an interesting toy presented in four different locations. This generated a film of the infant from the shoulders up, moving naturally while looking toward the left, right, center, and upward directions (or the reverse sequence), following the off-camera toy (see Figure 2). Thus, in the moving condition, consisting of four 30-second trials, the faces of both the self and peer were shown side by side, oriented in the same

INTERMODAL ORIGINS OF SELF-PERCEPTION

363

general direction on each trial. In the still condition, consisting of four 15-second trials, s011 projections of the infants' faces were shown side by side, following the same pose sequence as before. This was accomplished by displaying a frozen image from the prerecorded film. A frame with a representative example of each face was selected, one from each orientation. A yoked-control design was used so that the face of each infant served as the face of the peer for the next infant of the same age. This controlled for any differences between faces (e.g., facial attractiveness, activity level, affect, etc.) with respect to the main variable in question (self vs. peer). The lateral positions of the two images were alternated a c r o s s trials.

FIGURE 2. An example of the visual displays portraying an image of the self side-byside with that of a peer. From Fadil, Moss, & Bahrick (1993). Results indicated that both the 5- and 8-month-old infants spent a significant proportion of their total looking time (p <.01 and p<.05 for 5- and 8-month-olds, respectively) fixating the image of the peer (see Figure 3). Furthermore, when the results were broken down according to condition, both the 5- and 8-month-olds showed a significant looking preference for the peer when the faces were moving, but only the 8-month-olds significantly preferred the peer when the faces were still. Results of the 5-month-olds were in the expected direction, but attenuated. These findings provided clear evidence that both 5- and 8-month-old infants are able to discriminate the visual appearance and/or movements of their own face from that of an age-matched peer. Furthermore, consistent with the view that motion provides

364

L O R R A I N E E. B A H R I C K

an opportunity for abstracting properties of objects (e.g., Gibson, 1969; Owsley, 1983; Kellman & Spelke, 1985), it appears that making the distinction between self and other was somewhat easier when the faces were moving rather than still. The basis for this discrimination must have been visual-featural information at 8 months of age because infants in this group were just as good at discriminating the still as the dynamic displays. At 5 months, dynamic visual information may have played a more important role, or perhaps the moving displays provided better information about the visual features of the faces. In either case, discrimination of the visual-self appears to be present long before the age of 15 months observed by Lewis and Brooks-Gunn (1979).

0.651

8 month-olds

NN

5 month-olds

0.60 11

0.55-~!

55" 55" (.o8)(.o9)

IL

56"

55"

(.11)

'

0.50

. . . . . ~ . : . . . ~

0.40

N

~

' .... All T r i a l s

Moving

Still

* p<.05, two-tailed " p < . 0 1 , two-tailed

FIGURE 3. The proportions and standard deviations of total looking time (PTLT) spent fixating the display of the peer by 5- and 8-month-olds for all trials, and for the moving and still conditions separately. From Fadil, Moss, & Bahrick (1993). When might this ability have developed? We replicated the above study using younger infants aged 2 and 3 months. Results demonstrated no significant preferences for one image over the other at the age of 2 months. However, by 3 months of age there was an emerging preference for the image of the peer by the second block of trials (p <.05). These findings suggest that the ability to recognize one's face and distinguish it from that of another infant emerges between the ages of 2 and 3 months. Although it is unlikely, one cannot rule out the

INTERMODAL ORIGINS OF SELF-PERCEPTION

365

possibility that this ability might emerge even earlier and might be expressed as a visual preference for the sell which would be consistent with our prior speculations about the detection of visual-proprioceptive contingency. Lacking definitive data with younger infants, however, we must conclude that infants' performance at 3 months is the best indicator of the emerging ability to differentiate between the faces of self and other. It is remarkable that infants as young as 3 or 5 months are familiar enough with their own visual appearances and/or motions to recognize their own faces as familiar and to discriminate them from those of other infants of the same age. How might this ability have developed? Prior experience with their mirror image may be an important basis for this ability. Parents completed a questionnaire regarding the extent of their infant's exposure to their mirror image, video images, and photographs of self. Results indicated that all but 2 of our 104 infants had at least weekly experience with their mirror images, and most had daily exposure. At 8, 5, 3, and 2 months, the percentage of infants who had daily exposure to their image in the mirror was 88%, 87%, 83%, and 67%, respectively. However, all infants had negligible experience with photos or videos. Thus, exposure to the mirror may account for infants' familiarity with their visual images. These findings again raise the question of whether the infants understood the meaning of the video stimulation they viewed. Infants could have shown discrimination and recognition of their image on the basis of familiarity, without an understanding that the image belongs to the "self." At what point might the infant come to understand that "this is me!"? Experience with mirror stimulation provides a perfect contingency between visual and proprioceptive stimulation that specifies self. Furthermore, infants have been exposed to this kind of perfect visual-proprioceptive contingency from birth onward in their intermodal explorations of their own bodies. We do not yet know when the ability to detect the relation between visual and proprioceptive stimulation emerges, although research reviewed earlier (e.g., Meltzoff & Moore, 1977; Nanez, 1988; Rochat, in press; Van der Meet, 1993) suggests it may be as early as the first month of life. However, it seems both reasonable and parsimonious to suppose that through mirror contingency, the appearance of the infant's face comes to signify the self. That is, once the perfect contingency between one's body motion and the visual stimulation from that motion specify self, then the face and its features observed in the mirror may also come to specify self because of the perfect visual contingency observed. In contrast, if no contingency were observed or the contingency did not specify self to the infant, the experience of seeing one's face in the mirror might at first be like that of seeing a familiar peer. In essence, without self-recognition, the mirror image of self would be similar to that of a familiar face to the infant. At

366

LORRAINE E. B A H R I C K

what point in development the familiar face seen in the mirror is perceived as belonging to the self is a provocative question for future research. It may occur as soon as the perfect contingency is detected, or it may develop over time and experience with the mirror. One way to get closer to an answer to this question is by distorting the features of the infant's face in a video display and observing the infant's reaction. We can distort the features of the infant's face and the peer's face in the same manner, and assess whether infants respond differently from the way they respond under normal, "undistorted" conditions. Given that infants of 5 and 8 months previously showed a significant looking preference to the peer under a nondistorted condition, if they switched their preference to that of the self under a distorted condition, this would be evidence that their own face is perceived as special or different from that of a peer. This could be taken as preliminary evidence that the face seen in the video display is perceived as specifying the self. Put another way, if infants detect "something different about me," then one would expect that they would now look to the self rather than the peer, even though the peer face is also distorted in the same manner as the self. On the other hand, if infants do not yet attribute the image to the self, then their own image should be treated just like a familiar face. Because both the familiar and novel faces would posses the same distortions, infants ought to prefer the novel face as before because it would s011 be the more novel of the two faces. Following this logic, 5- to 8-month-old infants (N = 14) were tested in a procedure similar to before, but under only the moving condition because it provided the most robust results. They were all marked with two rouge spots, one on each cheek. The rouge was unobtrusively applied by the parent in the waiting room, prior to filming for the study. The infants' faces were then prerecorded as before, but to shorten the filming procedure, only two orientations were used: the fight and left 3/4 views. Then the forced-choice preference test was administered. It was identical to the moving condition of the prior study, except infants viewed trials of their own and another infant's face side by side, both displaying bright rouge dots on each cheek. Results were strikingly different from before. This time infants showed a marginally significant visual preference for the image of self, consistent with our predictions of self-recognition (t (13) = 2.04, p =.06). Further, when compared with the prior results of the 5- and 8-month-olds, infants showed a significant shift in the direction of attention (t(68) = 3.67, p = .0005). That is, previously, when no rouge spots were visible, infants showed a preference for the novel face. However, when rouge spots were visible, they shifted their attention to the familiar face; the self. Given that both the peer and self had the same novel features, it is unlikely that this shift in the direction of the visual preference was due to novelty. Rather, it is more likely that by the age of 5

INTERMODAL ORIGINS OF SELF-PERCEPTION

367

months, infants recognized the image of their face as specifying the self or as somehow different and more special than that of a peer. These findings of an attentional shift from the undistorted to the distorted conditions are only preliminary and require replication. Thus, the conclusions are tentative. However, along with our prior findings of discrimination between video displays of the self and peer by 3-, 5-, and 8-month-olds, the implications stand in sharp contrast to some prevailing views and to the interpretation of data from the self-recognition studies with toddlers. Specifically, our research calls into question two conclusions drawn from the prior body of research: First, that infants do not discriminate their own face from that of another infant until the age of 15 months; and second, that it is not until 15-18 months that the image in the mirror is attributed to self (e.g., Amsterdam, 1972; Butterworth, 1992; Damon & Hart, 1988; Lewis & Brooks-Gunn, 1979). In contrast, the results of our research show that by 3 months of age, discrimination of self from other on the basis of visual information provided by the face is possible. Furthermore, our attentional shift data suggest that some understanding of the meaning of this stimulation occurs well before toddlerhood and possibly by 5 months of age. This is, however, not to suggest that such young infants have a well-developed concept of self. There is a great deal of learning and developmental change that must occur in the months and years ahead (see Damon & Hart, 1988, for a review). The data do suggest that 5to 8-month-old infants may perceive the stimulation generated by the self as different from that generated by others in important ways. Future research will explore the nature of this difference.

Concluding Remarks This chapter summarized research exploring the development of infants' sensitivity to two types of information for self: intermodal proprioceptive-visual contingency and visual-featural information. Several findings and conclusions emerged from these studies: 1. Detection of visual-proprioceptive contingency for temporal relations is present by 5 months of age and perhaps earlier (Bahrick & Watson, 1985; Schmuckler, 1994). Sensitivity to visual-proprioceptive contingency for spatial relations is also present at 5 months (Rochat & Morgan, 1995; Schmuckler, 1994) and emerges by 3 1/2 months (Rochat, 1995). Thus, the potential for perceiving self and differentiating self from other on the basis of temporal contingency and spatial congruence between visual and proprioceptive information is present early in infancy.

368

LORRAINE E. BAHRICK

2. Even by the age of 2 months, infants are frequently exposed to mirrors. Most infants in our samples at each age (2, 3, 5, and 8 months) were exposed to their own mirror images at least daily. This may serve as a basis for familiarity with their own visual appearance. 3. By 3 months of age, infants are able to discriminate between a prerecorded videotape of their own face and that of another infant (Fadil, Moss, & Bahrick, 1993). They show a preference for the face of another baby over their own, demonstrating familiarity with their own visual appearance. This finding stands in contrast to prior conclusions that featural discrimination of one's own face from that of a peer does not emerge until 15 months of age (Lewis & Brooks-Gunn, 1979). 4. Preliminary results suggest an attentional shift when features of the infant's face as well as that of a peer are marked with a rouge spot. Infants aged between 5 and 8 months no longer prefer to watch the face of a peer. Rather, they tend to watch their own face more. This suggests that there is something different or special about stimulation from the self. Thus, evidence of self-knowledge is apparent in two domains in early infancy. Contingency and visual-featural information are different sources of information for self, and thus their development may proceed according to different timetables. However, they also form part of a coordinated system of knowledge about the self. Development in one domain influences development in the other. The perfect contingency generated by visual stimulation from one's body motions is excellent information for self and for differentiating between what is self versus not-self. It may serve as a primary source of information about the self. Through experience with mirror stimulation, sensitivity to this perfect contingency may subsequently facilitate the understanding that one's visual features specify the self. Future research in our lab will unravel the interdependence between development in these two domains by conducting training studies with infants who do not yet show evidence of discriminating the self from another infant in video presentations. As discussed earlier, Gibson's (1979) theory posits that our perceptual systems provide information about the self and about the environment at once. This view provides a point of departure for the developmental process and suggests that infants perceive a differentiated self from the beginning. Although the findings of infant and newborn capabilities outlined in the beginning of this chapter and the data presented here are consistent with this view, the conclusions are still inferential. There is no direct test of whether a preverbal infant knows that the self is a separate object among other objects in the environment, or recognizes that "this is me!" in the mirror. Ascribing meaning to the infant's actions or ability to discriminate between two displays is a difficult task, and conclusions should be

INTERMODAL ORIGINS OF SELF-PERCEPTION

369

viewed with caution. At a minimum, converging evidence from varied approaches is required. This problem is especially apparent if our criteria for self-recognition include conscious awareness of the self as separate (e.g., Bertenthal, 1992). Along similar lines, most recent discussions of the development of self suggest that the infant's understanding of self is at first "preconceptual" or perceptual (e.g., Butterworth, 1990; Meltzoff, 1990; Neisser, 1993; Rochat, in press), and later, knowledge of self becomes "conceptual." We eventually attain a self-concept that is available to conscious awareness. I would agree that self-knowledge progresses from perceptual to conceptual; however, the dichotomy seems arbitrary. At what point does the transition from preconceptual to conceptual knowledge occur and how does this occur? At present, there is no single or best answer to this question. The view elaborated here is that through development, there is a growing awareness of the self as separate. Perceiving and conceiving fall along a continuum and differ in degree, not kind. Knowledge about the self becomes progressively more elaborated, explicit, and available to consciousness in a number of different domains. This development may occur at different rates for information in different domains, and development in one domain may influence development in another. Thus, knowledge in some areas may become more explicitly conceptual prior to knowledge in other areas. This process may even continue into adulthood, as self-understanding evolves and implicit knowledge becomes more explicit. Thus, there is no single point where the system of knowledge about the self shifts from perceptual to conceptual, nor is there a point where knowledge in a single domain shifts from perceptual to conceptual. It is a dynamic, gradual, and ongoing process over the lifespan. In the context of these words of caution, the research reported here challenges the prevailing view that the child's conceptual understanding of the significance of his or her mirror image and other stimulation from the self does not emerge until well into the second year. The attentional shift observed in infants when features of their own faces are distorted suggests that there is already a growing awareness of something special about the visual stimulation from oneself that sets it apart from that generated by others. Before revising our view of infants' understanding of the meaning of their images in the mirror, however, it is important to replicate these findings and to provide converging evidence from other domains and with different procedures. Research is currently in progress in our lab to address these issues. Furthermore, when taken together with the ongoing discovery of new capabilities in infancy for detecting and responding to stimulation arising from the self (see other chapters in this volume), it is clear that our concept of the emergence of selfknowledge requires updating. Because this topic stands at the interface between perception, cognition, and social and personality development, new discoveries

370

LORRAINEE. BAHRICK

about the infant's emerging sense of self promise to have a far-reaching impact on theories in developmental psychology. ACKNOWLEDGMENT

This chapter was supported by a grant from the National Institute of Child Health and Human Development, RO1 HD25669.

REFERENCES

Amsterdam, B. (1972). Mirror self-image reactions before age two.Developmental Psychology, 5, 297-305. Amsterdam, B., & Greenberg, L.M. (1977). Self-conscious behavior of infants: A videotape study. Developmental Psychobiology, 10, 1-6. Anderson, J.R. (1984). The development of self-recognition: A review. Developmental Psychobiology 7, 35-49. Bahrick, L. E., & Pickens, J.N. (1994). Amodal relations: The basis for intermodal perception and learning in infancy. In D. Lewkowicz and R. Lickliter (Eds.), The development of intersensory perception: Comparative perspectives (pp. 205233). Hillsdale, NJ: Lawrence Erlbaum. Bahrick, L.E., & Pickens, J.N. (in press). Infant memory for object motion across a period of three months: Implications for a four-phase attention function. Journal

of Experimental Child Psychology.

Bahrick, L. E., & Watson, J.S. (1985). Detection of intermodal proprioceptive-visual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21, 963-973. Baldwin, J.M. (1902). Social and ethical interpretations in mental life. New York: Macmillan. Ball, W., & Tronick, E. (1971). Infant responses to impending collision: Optical and real. Science, 171, 818-820. Barrera, M.E., & Mauer, D. (1981). Recognition of mother's photographed face by the 3-month-old infant. Child Development, 52, 714-716. Bennett, K.R., Smith, P.H., & Loboschefski, T.W. (1992, May). Five-month-old infants' differentiation of self and other as a function of experience. Paper presented at The International Conference on Infant Studies, Miami Beach, FL. Bertenthal, B.I. (1992). Implicit versus explicit origins of the self. Psychological Inquiry, 3, 112-114. Bertenthal, B.I., & Bai, D.L. (1989). Infants' sensitivity to optical flow for controlling posture. Developmental Psychology, 25, 936-945. Bertenthal, B. I., & Fisher, K.W. (1978). Development of self recognition in the infant. Developmental Psychology, 14, 44-50. Bower, T.G.R. (1974, 1982). Development in Infancy, 2nd ed. San Francisco: W.H. Freeman and Company. Bower, T.G.R., Broughton, J.M., & Moore, M.K. (1970). Demonstration of intention in the reaching behavior of neonate humans. Nature, 228, 51-72. Bushnell, I.W.R. (1982). Discrimination of faces by young infants. Journal of Experimental Child Psychology, 33, 298-308.

INTERMODAL ORIGINS OF SELF-PERCEPTION 371 Butterworth, G.E. (1990). Self-perceptionin infancy. In D. Cicchetti, & M. Beeghly (Eds.), The self in transition: Infancy to childhood (pp. 119-137). Chicago: The University of Chicago Press. Butterworth, G. (1992). Origins of self-perception in infancy. Psychological Inquiry, 3, 103-111. Butterworth, G. E., & Hicks, L. (1977). Visual proprioception and posture stability in infancy: A developmental study. Perception, 6, 255-262. Cicchetti, D., Beeghly, M., Carlson, V., & Toth, S. (1990). The emergence of the self in atypical populations. In D. Cicchetti and M. Beeghly (Eds.), The self in transition: Infancy to childhood (pp. 309-344). Chicago: The University of Chicago Press. Damon, W., & Hart, D. (1982). The development of self-understanding from infancy through adolescence. Child Development, 53, 841-864. Damon, W., & Hart, D. (1988). Self-understanding in childhood and adolescence. New York: Cambridge University Press. DeCasper, A.J., & Fifer, W.P. (1980). Of human bonding: Newborns prefer their mother's voice. Science, 208, 1174-1176. Dixon, J.C. (1957). Development of self-recognition. Journal of Genetic Psychology, 91, 251-256. Fadil, C., Moss, L., & Bahrick, L.E. (1993, March). Infants' visual recognition of their own faces. Paper presented at The Society for Research in Child Development, New Orleans, LA. Fagan, J.F. (1972). Infants' recognition memory for faces. Journal of Experimental Child Psychology, 14, 453-476. Fagan, J.F. (1976). Infant's recognition of invariant features of faces. Child Development, 47, 627-638. Field, T. (1979). Differential behavioral and cardiac responses of 3-month-old infants to a mirror and peer. Infant Behavior and Development, 2, 179-184. Freud, S. (1922). Group psychology and the development of the ego. New York: Norton. Gallup, G. G., Jr. (1970). Chimpanzees: Self-recognition. Science, 167, 86-87. Gibson, E.J. (1993). Ontogenesis of the perceived self. In U. Neisser, (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp.2542). Cambridge, MA: Cambridge University Press. Gibson, E.J. (1969). Principles of perceptual learning and development. New York: Apple ton- C en tur y-Cro fts. Gibson, J.J. (1966). The senses considered as perceptual systems. Boston: Houghton Mifflin. Gibson, J.J. (1979). The ecological approach to visual perception. Boston" Houghton Mifflin. Grene, M. (1993). The primacy of the ecological self. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self- knowledge (pp. 112117). Cambridge, MA: Cambridge University Press. Harter, S. (1983). Developmental perspectives on the self system. In M. Hetherington (Ed.), Handbook of child psychology:Vol. 4. New York: Wiley. Hofsten, C. von. (1980). Predictive reaching for moving objects by human infants. Journal of Experimental Child Psychology, 30, 369-382. Hofsten, C. von. (1983). Foundations for perceptual development. In L. Lipsitt & C. K. Rovee-Collier (Eds.), Advances in infancy research: Vol. 2 ( pp. 241-261). Norwood, NJ: Ablex.

372

LORRAINEE. BAHRICK

Hunter, M.A., & Ames, E.W. (1988). A multifactor model of infant preferences for novel and familiar stimuli. In C. Rovee-Collier & L. Lipsitt (Eds.), Advances in Infancy Research:Vol. 5 (pp. 69-95). Norwood, NJ: Ablex. James, W. (1963). Psychology. New York: Fawcett. (original work published 1890) Johnson, D.B. (1982). Altruistic behavior and the development of the self in infants. Merrill-Palmer Quarterly, 28, 379-388. Kalnins, I.V., & Bruner, J.S. (1973). The coordination of visual observation and instrumental behavior in early infancy. Perception, 2, 307-314. Kellman, P., & Spelke, E. (1983). Perception of partly occluded objects in infancy. Cognitive Psychology, 15, 483-524. Lee, D., & Aronson, E. (1974). Visual proprioceptive control of standing in human infants. Perception and Psychophysics, 15, 529-532. Lewis, M. (1979). The self as a developmental concept. Human Development, 22, 416-419. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of the self. New York: Plenum. Loveland, K.A. (1986). Discovering the affordances of a reflecting surface. Developmental Review, 6, 1-24. Loveland, K.A. (1987). Behavior of young children with Down's syndrome before the mirror: Exploration. Child Development, 58, 768-778. Mahler, M., & Furer, M. (1968). On human symbiosis and the vicissitudes of Individuation. New York: International University Press. Mans, L., Cicchetti, D., & Sroufe, L.A. (1978). Mirror reactions of Down's syndrome infants and toddlers: Cognitive underpinnings of self-recognition. Child Development, 49, 1247-1250. Meltzoff, A.N. (1990). Foundations for developing a concept of self: The role of imitation in relating self to other and the value of social mirroring, social modeling, and self practice in infancy. In D. Cicchetti & M. Beeghly (Eds.), The self in transition: Infancy to Childhood (pp. 139-164). Chicago: The University of Chicago Press. Meltzoff, A.N., & Moore, M.K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 75-78. Meltzoff, A. N., & Moore, M.K. (1983). The origins of imitation in infancy: Paradigm, phenomena, and theories. In L. Lipsitt & C. Rovee-Collier (Eds.), Advances in infancy research: Vol. 2 (pp. 265-301). Norwood, NJ: Ablex. Meltzoff, A.N., & Moore, M.K. (1994). Imitation, memory, and the representation of persons. Infant Behavior and Development, 17, 83-99. Murray, L., & Trevarthen, C. (1985). Emotional regulation of interaction between two-month-olds and their mothers. In T.M. Field & N.A. Fox (Eds.), Social perception in infants. Norwood, NJ: Ablex. Nanez, J.E. (1988). Perception of impending collision in 3- to 6-week-old human infants. Infant Behavior and Development, 11, 447-463. Neisser, U. (1988). Five kinds of self-knowledge. Philosophical Psychology, 1, 3559. Neisser, U. (1993). The self perceived. In U. Neisser, (Ed.), The perceived self: Ecological and interpersonal sources of self-knowledge (pp. 3-21). Cambridge, MA: Cambridge University Press. Owsley, C. (1983). The role of motion in infants' perception of solid shape. Perception, 12, 707-717. Papousek, H., & Papousek, M. (1974). Mirror-image and self-recognition in young human infants: A new method of experimental analysis. Developmental Psychobiology, 7, 149-157.

INTERMODAL ORIGINS OF SELF-PERCEPTION 373 Piaget, J. (1954). The construction of reality in the child. New York: Basic Books. Piaget, J. (1929, 1967). The child's conception of the world. London: Routledge and Kegan. Porges, S.W. (1979). Developmental designs for infancy research. In J.D. Osofsky (Ed.), Handbook of infant development (pp. 743-765). New York: John Wiley and Sons. Rochat, P. (in press). Early development of the ecological self. In C. Dent-Read and P. Zukow-Goldring (Eds.), Changing ecological approaches to development:

Organism-environment mutualities.

Rochat, P., & Morgan, R. (1995). Spatial determinants in the perception of selfproduced leg movements by 3- to 5-month-old infants. Developmental

Psychology.

Rochat, P., Blass, E.M., & Hoffmeyer, L.B. (1988). Oropharyngeal control of handmouth coordination in newborn infants. Developmental Psychology, 24, 459463. Rovee, C.K., & Rovee, D.T. (1969). Conjugate reinforcement of infant exploratory behavior. Journal of Experimental Child Psychology, 8, 33-39. Rovee-Collier, C.K., & Fagen, J. W. (1981). The retrieval of memory in early infancy. In L.P. Lipsitt and C. K. Rovee-Collier (Eds.). Advances in infancy research." VoI. 1 (pp. 225-254). Norwood, NJ: Ablex. Samuels, C.A. (1986). Bases for the infant's developing self-awareness. Human Development, 29, 36-48. Schmuckler, M. (1994, June). Infants' visual-proprioceptive intermodal recognition. Paper presented at the International Conference on Infant Studies, Paris, France. Schulman, A.H., & Kaplowitz, C. (1977). Mirror-image response during the first two years of life. Developmental Psychobiology, 10, 133-142. Siqueland, E.R., & Delucia, C. A. (1969). Visual reinforcement of nonnutritive sucking in human infants. Science, 165, 1144-1146. Van der Meer, A.L.H. (1993, August). Arm movements in the neonate." Establishing a frame of reference for reaching. Paper presented at the Seventh International Conference on Event Perception and Action Vancouver, B.C., Canada. Watson, J.S. (1972). Smiling, cooing, and "the game." Merrill-Palmer Quarterly, 18, 323-339. Watson, J.S. (1985). Contingency perception in early social development. In T. Field and N. Fox, (Eds.), Social Perception in Infancy, NJ: Ablex. Watson, J.S. (1979). Perception of contingency as a determinant of social responsiveness. In E.B. Thoman (Ed.), The origins of social responsiveness (pp. 33-64). New York: Erlbaum. Watson, J.S., & Ramey, C.T. (1972). Reactions to response contingent stimulation in early infancy. Merrill-Palmer Quarterly, 18, 219-227. Yonas, A., Pettersen, L., & Lockman, J.J. (1979). Young infants' sensitivity to optical information for collision. Canadian Journal of Psychology, 33, 268-276.

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

375

CHAPTER 18

Self-orientation in Early Infancy: The General Role of Contingency and the Specific Case of Reaching to the Mouth JOHN S. WATSON

University of California at Berkeley

When and how the human infant differentiates him- or herself from the environment stands as a classic concern of developmental theorists. Speculation has varied greatly, from the socio-emotional stance of Freud, to the cognitiveperceptual stance of Piaget, and to the computational stance of modern connectionist theory. We will begin this paper by reviewing these three proposals plus a fourth that is less theory-specific: the hypothesis of contingency perception. Following the historical overview, we will consider the idea that self/environment differentiation can have at least three levels of functional complexity: selfdetection, self-orientation, and self-conception. The data we will focus on in this paper bear almost exclusively on the second level (self-orientation) as we will define it. The remainder of the paper will be devoted to some speculation about the possible developmental relations between the three levels, including a specific hypothesis of how one form of self-orientation may provide a base for subsequent self-conception.

Four V i e w s of Self/Environment Differentiation The two major developmental theories of this century, Freud's and Piaget's, both incorporate an assumption that normal development in humans requires an initial investment in the task of differentiating the self from its external environment. Both frame the task in terms of its relevance to the infant's acquiring a capacity to act instrumentally on the environment. However, as we will see below, the

376

JOHN S. WATSON

theories differ as to when and how the task is accomplished. Freud proposed a relatively quick accomplishment within the first few months, while Piaget proposed the task would take two or three times as long (Wolff, 1960). As for how the distinction between self and nonself is accomplished, the two theories differ greatly. A third, nondevelopmental, view has been offered by connectionist theorists (Hinton & Sejnowski, 1986). I will argue here, as I have elsewhere (Watson, 1994), that each of these three classic views fails to provide a truly workable means for dissociating self and environment. Yet, as we will see, they help illuminate a more likely possibility as provided in the hypothesis of contingency perception.

Freudian Theory Freud (1911/1946) attended to why we would be motivated to dissociate self from environment. He proposed that as basic need states arise, "primary thought processes" hallucinate goal states; but, given the failure of these to satisfy the underlying physiological needs, "secondary thought processes" arise to cope with reality and the exigencies of the external world. But just how internal and external are distinguished is not adequately provided. There is a proposal that tensioninducing stimuli arising outside the body can be distinguished from those arising within on the basis of their contingent relation to bodily movement. Rappaport (1951) nicely summarizes the proposal as follows: . . . the still helpless organism [has] the capacity for making a first orientation in the world by means of its perceptions, distinguishing both "outer" and "inner" according to their relation to actions of the muscles. A perception which is made to disappear by motor activity is recognized as external, as reality; where such activity makes no difference, the perception originates within the subject's own body w it is not real . . . . In summary, motility first serves as the channel of discharge for tensions due to needs of drive origin; later it becomes the tool of primordial reality-testing by distinguishing between inner and outer sources of stimuli, that is to say, between the "I" and "not I"; finally, it assumes the character of action, altering the external world for the purpose of gratification. (footnotes on pp. 323-333.) This proposal would appear to be seriously deficient on two grounds. First, by defining self in terms of noncontingent stimuli, any external events that are unaffected by the infant's behavior would be categorized as internal. For example, infants are often situated in places where the intensity of sight and sound vary greatly: a pram being walked down a busy city sidewalk; a baby tender next to a TV. These situations generate a cascade of light and sounds that are unrelated to the infant's behavior (i.e., noncontingent). By Freud's rule, however, these stimuli would be categorized as arising within the infant's body (the "I"). Moreover, additional confusion for the categorization of sensory innervations that are intrinsic

SELF-ORIENTATION IN EARLY INFANCY

377

to motor action seems inevitable by this rule. By Freud's proposal, proprioceptive stimuli, being directly affected by motor acts, would be perceived as arising in the external world.

Piagetian Theory Although a major concept in Piaget's (1936/1952) developmental theory is that of egocentrism and its progressive replacement by an objective world view, the theory does not appear to offer a clear mechanism for detecting self versus nonself. For Piaget, the first step in breaking out of a complete solipsistic selfcontainment is with the onset of what he termed "secondary circular reactions." He proposed that as infants later advance to the stage of coordinating secondary circular reactions (at about 8 months), they progressively distinguish their causal action from its environmental effects. One must note, however, that Piaget's definition of the crucial constructive transition from primary circular to secondary circular functioning introduces a logical circularity if one tries to use this distinction as the mechanism for differentiating self from environment. In the following quote, Piaget's (1936/1952) reference to "by convention" may be an effort to avoid such circularity: After reproducing the interesting results discovered by chance on his own body, the child tries sooner or later to conserve also those which he obtains when

his action bears on the external environment. It is this very simple transition which determines the appearance of the "secondary" reactions . . . . Of course, all the intermediaries are produced between the primary circular reactions and the secondary reactions. It is by convention that we choose, as criterion of the appearance of the latter, the action exerted upon the external environment (pp. 154-155). Comparative psychologists and anthropologists recently have centered on this distinction between primary and secondary reactions and have used it to contrast development in various primates (Antinucci, 1990; Parker, 1977). Antinucci has extended the distinction to any behaviors, whether or not they are repeated or circular. However, it should be clear that although this may work well as a comparative developmental milestone, it leaves obscure just how the infant or primate forms the perceptual distinction that assures "the action is exerted upon the external environment."

Connectionist Theory A strikingly simple mechanism for sorting out self from environment has been offered by theorists in the theoretical domain of what is called "new connectionism." This relatively new theoretical perspective is founded on the assumption that human brains function in a manner analogous to computers, with

378

JOHn S. WATSOn

architecture that allows massively parallel computation. The so-called "units" of a computational network (e.g., neurons of the brain, silicon chips of a computer) are viewed as having three basic levels: an input layer, an output layer, and a "hidden layer" that is positioned between the input and output layers. Perceptual and cognitive capacities are viewed as emerging properties of nets that systematically change the pattern of transmission bias (i.e., "weights") across the interconnections of units in the net. The systematic change in weights results from various forms of "learning" as the input layer of units are activated by contact with energy from the environment. Applying this general model, Hinton and Sejnowski (1986) have offered a refinement of an earlier suggestion by Crick and Mitchison (1983), to the effect that REM sleep may provide a control computation for disambiguating stimulation arising from self and environment. Hinton and Sejnowski propose: Our learning algorithm refines Crick and Mitchison's [1983] interpretation of why two phases are needed. Consider a hidden unit deep within the network: How should its connections with other units be changed to best capture regularity present in the environment? If it does not receive direct input from the environment, the hidden unit has no way to determine whether the information it receives from neighboring units is ultimately caused by structure in the environment or is entirely a result of the other weights. This can lead to a folie a deux, where two parts of the network each construct a model of the other and ignore the external environment. The contribution of internal and external sources can be separated by comparing the co-occurrences in p h a s e + with similar information that is collected in the absence of environmental input. P h a s e - thus acts as a control condition. Because of the special properties of equilibrium it is possible to subtract off this purely internal contribution and use the difference to update the weights. Thus, the role of the two phases is to make the system maximally responsive to regularities present in the environment and to prevent the system from using its capacity to model internally generated regularities (pp. 297-298). However, note that "environment" is defined as that which is outside the network. In the standard model of brain function, the outermost layer of a neural net will be input units representing the basic sensory systems. However, this relationship between environment and net does not support a clear distinction between body and world. The problem is that weights derived in dreaming states cannot provide a representation of bodily self-stimulation. The sight, sound, and sensation of one's body doing act X is not produced at the input layer while one dreams of doing X. In other words, while this connectionist algorithm can distinguish net from out-of-net experience, it can not yet distinguish "body experience" from "out-of-body experience." The algorithm of subtracting dream weights from weights obtained during wakefulness will categorize as environmental the experience of self as presented in proprioceptive feedback of

SELF-ORIENTATION IN EARLY INFANCY

379

skeletal and muscular movement and the self-stimulation occurring when the body stimulates its exteroceptive senses (e.g., when seeing its own movement or feeling its touch). The connectionst algorithm should work well to dissociate brain from nonbrain. That would help guarantee that our perception of our body (outside of the brain) will be as veridical as our perception of the environment beyond our body. But the line between our body and the environment is not specified by this algorithm. This is much the same problem as was noted above for the Freudian contingency algorithm. In the present case, however, the confusion would presumably be only in one direction (namely portions of the body being identified as environment), whereas the Freudian proposal allowed confusion in both directions (contingent stimuli arising in the body being identified as environmental, and noncontingent stimuli in the environment being identified as attributes of the body).

The Hypothesis of Contingency Perception There does appear to be at least one mechanism that would work to distinguish the line between body and world. Lewis and Brooks-Gunn (1979) speculate in a manner similar to Gibson (1966, 1979) that the earliest form of self/ other distinction is probably based on some kind of response-contingent stimulation, such as that produced when the infant sees the movements of his hand as he waves it in front of his face or when she feels the proprioceptive stimulation of leg movement as she flexes her legs. Given the substantial amount of evidence on the human infant' s capacity to perceive response-stimulus contingencies by the age of 3 months or earlier, Lewis and Brooks-Gunn set this age period as the probable time of the first primitive discrimination of self and other. There have been a number of studies to date that have attempted to a~ssess the infant's direct discrimination of self versus other in the age span of 3 to 5 months. With the exception of a study by Bahrick and Watson (1985), all these attempts have involved the discrimination of face and upper torso. Bahrick and Watson presented an image not of the face and torso, but of the legs and feet. The immediate effect of this change is that it eliminates the potential artifacts of eye motion and eye contact that exist with facial images. That is to say, in studies comparing the relative attractiveness of mirror (or on-line video) reflection of self versus the view of another infant, looking at self entailed looking at an image with eyes immobile. Furthermore, with the exception of a study by Papousek and Papousek (1974), the immobile eyes were staring at the subject. By contrast, the "other" image had moving eyes that probably seldom fixated the subject. Replacing faces with feet, Bahrick and Watson avoided this apparent confound in the earlier studies between category of target and its stimulus structure.

380

JOHN S. WATSON

In Bahrick and Watson's study, the infant sat facing two video screens. Each screen displayed the dynamic image of a pair of legs and feet. By inverting the video camera, the images produced approximately the retinal distribution of the subjects' normal view of their own legs and feet. That is, the left, right distribution was correct and the legs projected toward the upper portion of the visual field as is so for infants' direct view of their legs and feet. In three of four experiments, the infants' legs were occluded with a screen so that the only visual information for their leg movements was on one of the TV monitors. Three experiments were run to see if 5-month-olds would show discriminative attention to a choice between a live video image of their legs versus those of a peer or, in one experiment, a prior recording of themselves. The results confirmed the expectancy that 5-month-old infants would perceive the difference between the contingent self-image and the noncontingent peer image. Consistent with prior findings with facial images at this age (e.g., Papousek & Papousek, 1974), the subjects in each experiment looked significantly longer on average at the display that was not their contingent self. The authors concluded that these data support the proposal that young infants can use behavior-stimulus contingency as a basis of differentiating self from environment.

Three Levels of Function in Self/Environment Differentiation: Detection, Orientation, Conception There are many ways in which systems of the human body take into account distinctions between features of the body and features of other things. T-cells make something like self/other distinctions long before birth. Temperature regulation involves control of circulation in the extremities relative to the external temperature. Air pressure differences between the environment and body are detected and adjusted to in various ways in different parts of the body (e.g., via the eustachian tube for the eardrum). But these biological functions are not really relevant to the classic concern of psychology. That concern, whether framed in mental or behavioral terminology, focuses on when and how acts of the individual are modulated to some degree by a distinction between attributes of one's body versus attributes of one's environment. There certainly have been important conceptual extensions of this focus (e.g., James' discussion of "materialist" and "spiritual" dimensions of the self). However, I would suggest that the primitive body/environment distinction is virtually universal as a core distinction in psychological discussions of self. For present purposes, it is sufficient for grounding the three levels of function to be proposed here.

SELF-ORIENTATION IN EARLY INFANCY

381

Self-detection Some actions of individuals imply that a relation between body and environment (and thus a distinction between them) is being detected. This arises commonly in spatial relations. Acts are performed in response to change in the relation. For example, a young infant will alter the position of its head and/or blink in response to the sight of an object approaching on a collision course (Dunkeld & Bower, 1980; Pettersen, Yonas, & Fisch, 1980). A toddler will show postural adjustments to visual information that implies a change in body alignment with gravity (Lee & Aronson, 1974). Even pre-toddlers, if not precrawlers, will display increased heart rate when placed onto the deep side of a visual cliff (Bertenthal & Campos, 1984; Campos, Langer, & Krowitz, 1970; Richards & Rader, 1981, 1983 ). In each of these examples, the action (or reaction) of the infant carries the implication that some reference has been made to the state of the body in relation to the environment. These are important indications that the action to some extent involves self-detection.

Self-orientation Some actions imply not only a reference to the body as distinct from its environment, but that the body/environment distinction is to some extent a perceptual/motor target of the act. For example, as noted above, infants younger than about 3 months will selectively fixate a reflection of their own image (Field, 1979; Bahrick & Watson, 1985), and infants of about 4 or 5 months will selectively fixate away from their on-line image to the image of another infant (Papousek & Papousek, 1974; Bahrick & Watson, 1985). It is a common observation that infants reach for their feet and their opposing hand. Although I do not know of any sufficiently controlled study to support the assumption that the attraction to the body part is due to its being detected as "of the selL" were it so, then such acts would be evidence of self-orientation. Later in this paper we will discuss some controlled studies of 5-month-olds reaching to their mouths. In each of these examples, the criterial act not only implies that the infant is making use of a distinction between body and environment (i.e., self-detection) while composing the act, but that the act is oriented toward (or away from) the targeted body portion. The distinction I am trying to make between self-detection and self-orientation is motivated by the intuition that in the case of what I am calling orientation, body parts (or sounds) become perceptually salient as objects of attention. By contrast, I find it easy to imagine that what I am calling acts of self-detection can occur without any change in attention toward or away from the body. Presumably the toddler maintains upright posture by some constant assessment of body versus environmental alignment. But when action is guided toward or away from a

382

JOHN S. WATSON

portion of the body, in that moment something is added to the perceptual salience of at least that body portion. I am using orientation to capture that addition in functional salience. I have no evidence that this distinction between self-orientation and self-detection has any profound theoretical significance. We will consider some possibilities in the concluding section of this chapter. My guess is that at the very least, the distinction may have some temporary heuristic value.

Self-conception Lewis and Brooks-Gunn (1979) present a distinction in self-function that is clearly of theoretical significance. Taking their lead from James (1890/1950), they draw a distinction between what they call the categorical or objective self and the existential or subjective self. The former implies some representational memory and is reflected in some self-referencing behavior that discriminates unique features of the body (Gallup, 1970; Lewis & Brooks-Gunn, 1979). The existential sell by contrast, has to do with a perceptual sensitivity. It implies some detection mechanism and is reflected in some self-referencing behavior that discriminates self from nonself at least momentarily. These roughly overlap with what James termed the "me" and the "I" of self, respectively. The preceding distinction between self-detection and self-orientation is an attempt to formulate a functional distinction within the concept of the existential or subjective self (indeed, it is a distinction within what I previously termed the process of self-detection; Watson, 1994). The remainder of this chapter will focus primarily on self-orientation. The functional aspects of self we are calling selfdetection and self-orientation are probably cognitively less complex than selfconception. Yet there are at least two reasons for scientific interest in the levels below self-conception. One is that they appear to occur before self-conception and thus may play a role in generating the later function (see Bahrick, this volume). The other reason for interest in the levels below self-conception is that self/environment differentiation at this preconceptual level is still quite relevant to a long history of philosophical and psychological speculation. One way of framing the classic philosophical concern is to ask, "When we think we are acting on the world, how can we be sure we are not just acting on ourselves (or in our minds)?" In philosophy this was the challenge of solipsism. In psychology, as noted above, major theories have taken pains to formulate some stance on how some(body) might come to distinguish itself from its environment. The question naturally arises as to whether self-detection and self-orientation mark different stages of self-development. Although, as noted above, I view selforientation as embodying an increase in functional complexity over self-detection, I know of no empirical data to support a claim of developmental sequence for their appearance, nor will the present paper provide such data. At this point, the

SELF-ORIENTATION IN EARLY INFANCY

383

distinction is only taxonomic. What follows is a claim of evidence for instances of the category of self-orientation behavior.

Evidence of Self-orientation

The Contingency Studies The studies referenced above in support of contingency serving as a basis for self/other discrimination meet our criteria for evidence of self-orientation. This seems obvious in the case of Field's (1979) data, where subjects expr6ssed their discrimination by a preferential fixation of their own reflection over the image of another infant. But what of the data from Papousek & Papousek ( 1 9 7 4 ) and Bahrick and Watson (1985), where infants looked longer at other than at on-line self? Were they looking at the other or were they looking away from the self? Only the latter would count as self-orientation (i.e., behavior oriented toward or away from self). Behavior simply oriented toward the other might count as reflecting self-detection, but not self-orientation. In the present case, however, experimental data support an inference of self-orientation, even though the attentional act was toward the image of the other. The inference of self-orientation derives from the fact that these studies introduced controls for factors other than contingency. For example, in two of the Bahrick and Watson studies, a yoked control procedure was employed wherein each infant served as the "other" for another infant in the sample. In their third study, each subject served as its own "other" by using a segment of its own behavior as recorded a few minutes earlier. With these controls, presumably the only stimulus basis available for discriminating the two images was the feature of perfect contingency. That being so, the choice of orienting toward other could only be formed on the basis of detecting the contingency in the self-image and moving away from it. A fourth study reported by Bahrick and Watson helps make the point clearer. This study was carried out with a sample of 3-month-olds. Examination of the distribution of individual percent fixation scores uncovered a significant bimodal distribution. That is, about half displayed strong preference for self and about half displayed strong preference for the noncontingent image. Relatively few showed a marginal preference. The authors concluded that 3-month-olds are capable of discriminating self from other but are in a developmental phase that involves a transition from attentional deployment toward self to attention away from self toward others. This interpretation fits with the findings of Field (1979), whose sample of 3-month-olds viewing faces showed a significant preference for the contingent self-image as opposed to the noncontingent image.

384

JOHN S. W A T S O N

On the basis of the cited data and additional analyses of young infants' reactions to variation in response-reward contingencies, I have developed an hypothesis about the initial phases of human self-orientation as based on contingency perception (Watson, 1985, 1994). Self-orientation is proposed as shifting from an initial stage of self-seeking to a later stage of self-avoidance. Until about 3 months, the infant seeks the highest level of contingency it can detect. At about 3 months of age, it would appear that a bias matures to the effect that perfect (or very nearly perfect) contingency is categorized as "self." This selfobject is perceived as less attractive and arousing than objects whose reactive contingency is high but not perfect. That bias affords a functional discrimination and categorization of self as distinct from other objects. It also orients the infant to engage in the reactive causal properties of external objects in preference to generating recursive cycles of self-stimulation. On first thought, this hypothesis seems farfetched. Why would a system develop an avoidance of itself? Put another way, how could the principle of self-interest in the evolutionary sense be served by a mechanism of self-disinterest in the perceptual/attentional sense? With additional thought, however, the hypothesis would seem to possess high face validity. How could development proceed if it were not true? A continuing bias to attend to and pursue perfect contingency would be self-consumptive in the near-literal sense. The contingencies of the external world could not compete with the perfect behavior stimulation contingencies created by blinking, rubbing oneself, vocalizing, etc. If behavior is to be organized for action on the environment, the organizing mechanisms must avoid getting locked onto the purely bodily contingencies.

A Potential Function of Initial Self-seeking Considering the preceding hypothesis, the question arises as to why there would be an initial phase of self-seeking. If this is not a residual quirk of evolution, what might an initial phase of self-seeking serve? Although focused on a different problem, the connectionist Rumelhart (1990) has done some work that raises an interesting possibility (see also Jordan & Rumelhart, 1991). He has tested various teaching strategies in simulations of a computer net learning to guide a mechanical arm in accurate reaching for objects. He describes the task as involving at least a net for mapping motor activation and arm movements, and another net for mapping of perceived object placement and motor activation. It would appear that this double net can learn the task most efficiently if it begins by organizing the motor net through a series of random self-activations prior to input from the object perception net. Rumelhart notes that Piaget seemed to have a similar strategy in mind in his notion of primitive sensory motor schema becoming organized by primary circular reactions prior to their application to the external environment and the advent of secondary circular reactions. Thus, it would

SELF-ORIENTATION IN EARLY INFANCY

385

seem that an early stage of self-seeking behavior may serve as a preparatory stage for the later capacity of coping with the environment. We will see this theme repeated as we now tum to data on hand-to-mouth behavior.

Reaching to the Mouth: The Second Reach Another source of data that may imply self-orientation involves manual acts directed toward the body. More specifically, we will consider the act of moving the hand to the mouth after reaching and grasping an object. I propose that this behavior is of special interest, in contrast to the object-oriented reaching that usually precedes it. Infants reach for almost anything graspable that is within ann's length. For a contented and alert infant between about 4 and 8 months, this is a very reliable behavior. However, because they will reach for virtually anything, including parts of themselves (e.g., their feet), this reaching outward ("the first reach") does not clearly distinguish object-self from objects in the environment. By contrast, the case of reaching to the mouth ("the second reach") following a grasp of an object implies by its singularity a distinctiveness as the target of action by a hand-that-holds-an-object. An algorithm of the form, "what you grasp is the environment" would be risky at best. But an algorithm of the form, "where you reach to following a grasp is yourself" would be pretty reliable information for the young infant. In a series of three studies, infants from 4 to 6 months of age were tested for their tendency to bring a grasped object to their mouth (Watson, Umansky, Marcy, & Repacholi, 1995). The objects varied from one study to the next in terms of their color, shape, and sound-producing capacity, but the general procedure was basically the same. The subject was seated in a parent's lap and a toy was presented for 5 seconds at midline, about chin level, but beyond ann's length. The object was then moved to within reach for 5 seconds. Finally, if the infant did not grasp the object within the 5 seconds, the experimenter touched the object to the infant's most extended hand. This was seldom necessary because the mean proportions of trials on which subjects reached and grasped the object were .70, .95, and .90 for the three studies, respectively. (In response to a significant effect of age observed in the first study, the second and third studies sampled a slightly older and more restricted age range.) The presentations were repeated over a series of trials with various objects. The number of trials differed across the studies, but in each case a summative score was derived for each subject that expressed the proportion of 30second trials on which a grasped object was subsequently brought to touch the mouth. These average proportions were .62, .72, and .81 in the three studies. A second score was derived in each study based on the infants' reaction to a special procedure of applying tactile stimulation to the infants' hands with no object present. The procedure involved placing a cloth-covered elastic band (a

386

JOHN S. WATSON

commercial "scrunchy" for styling hair) either around the infant's wrist, or at midpalm and free of the thumb (in the first study there was an additional condition of placement mid-palm and across the thumb). Placement at the wrist was a control for general stimulation of the subject and a control for the subject's general inclination to bring the hand to the mouth within the 30-second test trials (mean proportions being .14., 34, .33 in the three studies, respectively). For our purposes here, two findings were notable in these three studies. First, placement of the band across the palm lead to a higher proportion of hand-tomouth movement than did placement of the band on the wrist (the average across the 58 infants of the three studies was .25 for the wrist and .48 for the palm). This effect existed in all three studies and was highly reliable in the first and third. The effect was equally powerful for both the right and left hands. Second, and of particular note, the individual variation in subjects' hand-to-mouth response to the palm stimulation from the band was a significant predictor of the individual variation in subjects' tendency to bring a grasped object to the mouth. In each of the three studies, stepwise regression analysis singled out the reaction to palm stimulation as the only significant predictor of bringing a grasped object to the mouth. The resulting regression coefficients were .65, .57, and .60, for the three studies, respectively. In each study, the predictive association between response to palm stimulation and the response to grasping an object was significmat when the effect of wrist stimulation was partialled out. Thus, we may conclude that the predictive relation, in terms of individual difference, is not attributable to a general tendency to bring the hand to the mouth. The theoretical importance of this finding is that it implies that infants of about 5 months of age do not bring grasped objects to their mouth merely as the completion of a coordinated reach-to-grasp-to-mouth motor plan. Rather, it is more likely that they reach to a seen object, grasp a tactile stimulus to the hand, reach toward the mouth when the palm receives tactile stimulation, and mouth an object that touches on or near their lips. We will refer to this modular pattern of seeingreaching-grasping-reaching-mouthing as the SRGRM pattern. Watson et al. speculate that infants benefit from this modular pattern in that it probably provides a kind of "self-scaffolding," in which infants perform the equivalent of a coordinated secondary circular reaction (i.e., reach/grasp in order to mouth) prior to full development of the cognitive capacity required for such coordination of sensory-motor schemes. The modularity of the SRGRM pattern presumably degenerates over the latter half of the first year. Although we have no direct assessment of this degeneration, we surmise it on the grounds of the general, casual observation that older infants seem capable of manipulating grasped objects without first mouthing them. Thus, we presume that the mouthing module gradually becomes less insular and less

SELF-ORIENTATION IN EARLY INFANCY

387

reflexive during the latter half of the first year. During this degenerative period, we propose that the memory of the experience of modular control serves as a guide to the eventual intentional coordination of the three-step reach-to-grasp-to-mouth sequence that appears evident by 12 months. Let me suggest that there may also be something instructive about having a perceptual representation of an object prior to and then following an act upon it. In the S RGRM pattern, this occurs in the infant's seeing the object prior to the first reach and in the mouthing following the second reach. These may be lessons in representing clear set-goals for behavioral acts. By contrast, consider the complexity of obtaining a similar "lesson" in arbitrary contingencies. Consider the social contingency wherein a fret cry by the infant leads to mom's picking him/her up. One might well propose that after much experience with this contingency, the infant would actually produce fret cries at times because he/she expects maternal contact to follow (Lewis & Goldberg, 1969; Watson, 1966). But, unlike the SRGRM pattern, this contingency experience does not contain a representation of the contingent event prior to the act. We just assume that eventually experience affects the infant so that a representational expectation arises prior to the act. That is quite an assumption, of course (Tolman was willing to make it, but Skinner forcefully resisted it). Perhaps the prior representation of outcome in goal-directed behavior does not arise in a full-fledged manner. Perhaps there are lessons for framing this structure (goal orientation) of behavior within certain earlier natural patterns of behavior. If so, then the SRGRM pattern would seem a likely exemplar. The modularity data carry an additional implication with regard to our present focus on the development of a differentiation between self and environment. From an evolutionary perspective, it seems clear that the SRGRM pattern is designed to negotiate a special form of contact between body and environment. It may be less specialized than rooting and sucking, but perhaps initially no less modular. A common view of the ultimate purpose of the infant's incessant mouthing of objects is that this act affords the detection of information about objects. The infant is not trying to eat, but is trying to understand (trying to satisfy what Dennett [1991] would term epistemic hunger ). This common view is naturally focused on the infant's gaining an understanding of the environment. After all, the action pattern usually involves reaching to contact objects in the environment. But that common view overlooks the potential importance of the second reach: reaching to contact the mouth once an object has been grasped. I would now propose that the second reach may have evolved as much for its consequences on generating self-knowledge as for its assistance in gaining knowledge of the objects in the environment or for its potential service as a guide to goal-directed action.

388

JOHN S. WATSON

Self-knowledge from Crossmodal Subtraction The proposal I would make is inspired by the connectionist view presented above, wherein dreaming serves as a correction procedure that helps a network obtain veridical perception of its environment. That hypothesis has a more elegant simplicity than the one I will propose. I would justify the increa~sed complexity of my proposal by noting that its objective is to provide a basis for perceiving attributes of the self versus to clarify what is outside of the perceptual network. Consider first an odd feature of the common assumption that infants mouth objects that they have seen and grasped in order to investigate them. Given the evidence that young infants possess a remarkable crossmodal capacity between vision and mouthing (Meltzoff & Borton, 1979; Kaye & Bower, 1994), it is not clear what an infant is learning by bringing a seen object to the mouth. Why would infants need to confirm their crossmodal expectancy? Indeed, what would confirmation be other than enactment of the crossmodal comparison? The presence of this capacity in the neonate argues against the idea that SRGRM is a sensorymotor program that teaches crossmodal associations. So then, the knowledgegathering assumption as to why infants mouth objects they have seen and grasped would seem relevant only so far as crossmodal knowledge were deficient. Perfect crossmodal function would mean that seeing the object is equivalent to mouthing it, and so mouthing after seeing would offer little gain in knowledge of the object. Let us more realistically assume that crossmodal perception is less than perfect. That is to say, when the infant sees an object, some of its features map onto a mouthing/tactile representation of the object. We need at least this partialmapping assumption to explain the empirical data wherein experience with some tactile features of an object guides subsequent (as in Meltzoff & Borton, 1977) or concurrent (as in Kaye & Bower, 1994) visual discrimination of some features of the object. But it should be clear that this assumption does not entail a notion that all self features have been disentangled from all object features. As framed in the earlier discussion, the solipsism problem arises for all perceptual activity because the percept is theoretically bound to be a tangle of featural structures, some of which are derived from the environment (the object), and some from the perceptual apparatus itself (the sensory/perceptual pathways of vision and those of the tactile consequences of mouthing). Let us reconsider the functional consequence of the proposed solipsistic entanglement of object features and sensory/perceptual apparatus features. It would be easy to confuse the issues of what is at risk here. Not knowing whether a perceived feature is of the inside or of the outside is not the same as being insensitive to the features of the outside. Consider the perceptual acts depicted in Table 1. The infant mouths object A and then mouths object B. We may presume that if these perceptual experiences differ, they differ because of the difference in the

SELF-ORIENTATION IN EARLY INFANCY

389

features of the objects. Surely it is true that the extent to which the experiences seem similar is potentially based on a tangle of features, some representing similarity of the objects and some representing similarity of the act of perception by mouthing. But that confounded basis of similarity does not confound the basis of contrast. Analogous to subtracting dream weights from awake weights, the difference between A + M (features of object A plus features of mouthing) and B + M (features of object B plus features of mouthing) can be represented as A - B, a difference attributable to the environment. TABLE 1. Theoretical Derivation of Self-attribute Perception via Crossmodal Subtraction. I

I

I

I

,~.~o~c subtraction

Perceptual = Derivative

Mouthing Object B

(M+A) - (M+B)

= (A-B)

Seeing

(M+A) - (S+A)

= (M-S)

Tangle of

Tangle of

obj~

object

Case of:

and Modali~ i

and

Change of Object

Mouthing Object A

Change of

Mouthing ObjectA

Object A

Change of Object and Modality

Mouthing ObjectA

Seeing

(M+A)- (S+B)

Object B

No Change I of Object or ModatiTv

Mouthing ObjectA

Mouthing Object A

(M+A) - (M+A)

Modalitv I I

I

Contrast of Perceived

Objects

Modality

Contrast of Perceptual acts of Self = (M+A-S-B)

Tangle of Objects and ! Perceptual i acts = (0)

No Contrast

I II

Consider now the perceptual process depicted in the second example in Table 1.1 Here we depict the comparison of percepts derived by mouthing object A and seeing object A. The difference between A + M (features of object A plus features of mouthing) and A + S (features of object A plus features of seeing) can be represented as M - S, a difference attributable to the perceptual acts of the self. This derivation of self-based attributes does not occur for the remaining two logical options of comparison between two tangled percepts. The third row of the table shows that when the tangled percepts involve a change of both object and modality of perception, then the contrast remains a tangle of object and modality attributes. Finally, if the comparison involves no change in either modality or object then the theoretical derivative is zero, representing no contrast. Now, one might resist this notion that the infant gains a basis of perceiving featural aspects of self by comparing the tangled percepts of a particular object from two separate sensory modalities. After all, that gain depends on the subject

390

JOHN S. W A T S O N

assuming the two percepts are dependent on the same object. Intuitively, that seems a riskier assumption than the requirement of assuming the same perceptual act of mouthing when, as in the first example presented in the table, we initially resolved a difference between objects A and B. In the present case, why should the infant assume the two percepts are not of two different objects (e.g., that the comparison involves M + A and, say, S + B)? Under that assumption, as in the third example of the table, the derivative would need to be assumed to remain a tangle. There would seem to be at least two reasons why an infant might assume the existence of a constant object in the situation we are considering. The first reason we might imagine that the infant assumes the percepts are of the same object is that he or she has no choice to do otherwise. The infant is endowed with a means of crossmodal mapping, and the crossmodal similarity perceived forces the assumption of same object. This reasoning is implicit in the interpretations of existing data on crossmodal perception cited above. By some evolved mechanism, relational (or so-called "higher order") features of an object can be represented in more than one modality. Infants display sensitivity to this similarity when selectively looking at the visual representation of the tactile shape they have experienced in their mouth. Even were the infant not endowed with a powerful crossmodal mapping device, the temporal contiguity produced by SRGRM would provide a reason to assume that the mouthing experience is based on the same object that the preceding seeing, reaching, grasping, and reaching to mouth were. This proposal should not be taken as a claim that S RGRM serves to construct the crossmodal mapping. It might help do so for older infants, but the data from newborns (Kaye & Bower 1994) imply that much crossmodal perception is untutored. So then, whether by direct crossmodal perception or by inference following a contiguous SRGRM sequence, the infant is in a position to resolve a perceptual difference between the act of seeing and the act of mouthing following what we are calling the second reach. A thought experiment may help give the flavor of what that discrimination of perceptual acts might be like. If I were to ask you to describe the difference between drinking a glass of milk and a glass of lemonade, you would probably not mention that I had poured one into a plastic goblet and the other into a paper cup. However, if I had asked you to describe the difference between drinking two clarets, the difference of containers might quickly become a salient concern (most certainly if I had put the same wine in both containers). Analogously, it is not difficult to imagine that an infant might fail to note the modality of perceptual events until compelled by an assumption of constancy of object identity as produced by perceived crossmodal similarity or by an innate commitment to object constancy within an evolved sensory-motor pattern such as SRGRM.

SELF-ORIENTATION IN EARLY INFANCY

391

Conclusions

I have tried to develop three theoretical points in this chapter. The first is about the potential use of contingency information in guiding the infant to formulate a distinction between his or her body and the external environment. Much of that proposal has been argued previously (Watson, 1985, 1994). The second point of this chapter is a proposed refinement of a two-category distinction in early self/environment differentiation. The previous distinction between self-detection and serf-conception was expanded by introducing self-orientation as a subdivision of what had previously been termed self-detection. The third point of the chapter involved a speculative proposal about how the self-orientation arising in reaching to the mouth with a grasped object may play a special role in the early perception of self features. The unavoidable entanglement of self-based and enviromnent-based perceptual features is proposed to be to some extent untangled by the crossmodal perception of an object, as provided in the course of compulsive S RGRM patterns of activity. In this manner, then, this form of self-orientation may serve as a basis for a beginning awareness of self as perceiver. That, in turn, while not quite an objectification of self as will exist with self-conception, may serve to prime the self system to begin adding objective features (the look of hands detected as one's own, the look of mirror images detected as one's own, the sound of one's voice, etc.) to the self-based, phenomenal/subjective distinction in perceptual activity. NOTE

1. The reader might prefer multiplicative rather than additive notation in order to represent percepts as interactive products of environmental stimulation and perceptual processes. If so, note that substituting division for subtraction and referencing the state at no contrast to 1 versus 0 leaves the major implications of the table intact. ACKNOWLEDGMENTS

The research reported in this chapter was supported by the Institute of Human Development and a grant from the Palm Memorial Fund of Children's Hospital, Oakland, California.

REFERENCES

Antinucci, F. (Ed.) (1990). Cognitive structure and development in nonhuman primates. Hillsdale, NJ: Erlbaum. Bahrick, L. E., & Watson, J. S. (1985). Detection of intermodal proprioceptive-visual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21, 963-973.

392

JOHN S. WATSON

Bertenthal, B. I., & Campos, J. J. (1984). A reexamination of fear and its determinants on the visual cliff. Psychophysiology, 21, 413-417. Campos, J. J., Langer, A., & Krowitz, A. (1970). Cardiac responses on the visual cliff in prelocomotor human infants. Science, 170 (3954), 196-197. Crick, F., & Mitchison, G. (1983). The function of dream sleep. Nature, 304, 111114. Dennett, D. C. (1991). Consciousness explained. Boston: Little Brown. Dunkeld, J., & Bower, T. G. (1980). Infant response to impending optical collision. Perception, 9, 549-554. Field, T. (1979). Differential behavioral and cardiac responses of 3-month-old infants to a mirror and peer. Infant Behavior and Development, 2, 179-184. Freud, S. (1946). Formulations regarding the two principles in mental functioning. In Collected Papers, Vol. IV (pp. 13-21). London: Hogarth. (original work published 1911) Gallup, G. G. (1970). Chimpanzees' self-recognition. Science, 167, 86-87. Gibson, J. J. (1966). The senses considered a,~ perceptual systems. Boston: HoughtonMifflin. Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton-Mifflin. Hinton, G. E., & Sejnowski, T. J. (1986). Learning and relearning in Boltzmann machines. In D. E. Rumelhart & J. L. McClelland (Eds.), Parallel distributed processing, Vol. 1 (pp. 282-317). Cambridge, MA: MIT Press. James, W. (1950). The principles of psychology: Vol. 1. New York: Dover. (original work published 1890) Jordan, M. I., & Rumelhart, D. E. (1991). Forward models: Supervised learning with a distal teacher. Occasional Paper #40, Center for Cognitive Science, Massachusetts Institute of Technology. Kaye, K. L., & Bower, T. G. R. (1994). Learning and intermodal transfer of information in newborns. Psychological Science, 5, 286-288. Lee, D. N., & Aronson, E. (1974). Visual proprioceptive control of standing in human infants. Perception & Psychophysics, 15, 529-532. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self New York: Plenum. Lewis, M., & Goldberg, S. (1969). The acquisition and violation of expectancy: An experimental paradigm. Journal of Experimental Child Psychology, 7, 70-80. Meltzoff, A. N., & Borton, R. W. (1979). Intermodal matching by human neonates. Nature, 282, 403-404. Papousek, H., & Papousek, M. (1974). Mirror-image and self-recognition in young human infants: A new method of experimental analysis. Developmental Psychobiology, 7, 149-157. Parker, S. T. (1977). Piaget's sensorimotor period series in an infant macaque: A model for comparing unstereotyped behavior and intelligence in human and nonhuman primates. In S. Chevalier-Skolnikoff & F. E. Poirier (Eds.), Primate biosocial development: Biological, social, and ecological determinants (pp. 43-112). New York: Garland. Pettersen, L., Yonas, A., & Fisch, R. O. (1980). The development of blinking in response to impending collision in preterm, full-term, and postterm infants. Infant Behavior & Development, 3, 155-165. Piaget, J. (1936/1952). Origins of intelligence. New York: Norton. Rappaport, D. (1951). Organization and pathology of thought. New York: Columbia University Press.

SELF-ORIENTATION IN EARLY INFANCY 393 Richards, J. E., & Rader, N. (1981). Crawling-onset age predicts visual cliff avoidance in infanlxs. Journal of Experimental Psychology: Human Perception & Performance, 7, 382-387. Richards, J. E., & Rader, N. (1983). Affective, behavioral, and avoidance responses on the visual cliff: Effects of crawling onset age, crawling experience, and testing age. Psychophysiology, 20, 633-642. Rumelhart, D. E. (1990). The role of mental models in learning and performance: A connectionist account. Colloquium presented to Berkeley Cognitive Science Group, Fall 1990. Watson, J. S. (1966). The development and generalization of "contingency awareness" in early infancy: Some hypotheses. Merrill-Palmer Quarterly, 12, 123-135. Watson, J. S. (1985). Contingency perception in early social development. In T. M. Field & N. A. Fox (Eds.), Social perception in infants (pp. 157-176). Norwood, NJ: Ablex. Watson, J. S. (1994). Detection of self: The perfect algorithm. In S. T. Parker, R. W. Mitchell, & M. L. Boccia (Eds.), Self-awareness in animals and humans: Developmental perspectives (pp. 131-148). New York: Cambridge University Press. Watson, J. S., Umansky, R., Marcy, S., & Repacholi, B. (1995). Hand-to-mouth module in early reaching. Psychology Department, University of California at Berkeley. Manuscript in preparation. Wolff, P. H..(1960). The developmental psychologies of Jean Piaget and psychoanalysis. Psychological Issues, Vol. II, No. 1 (Monograph 5).

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

CHAPTER

395

19

The Function and Determinants of Early Self-exploration PHILIPPE ROCHAT and RACHEL MORGAN

Emory University

From birth, the body is a primary object of knowledge and exploration. This fact has been rather overlooked by students of infancy, who have focused their research efforts on the origins of perception and the development of early cognition in relation to external (nonself) objects and events. If infants are actively involved in exploring and learning about their environment from birth, they are equally involved in exploring themselves. Not only do they coperceive themselves in the act of perceiving objects, as suggested by Gibson (1979), but they also spend much of their awake time systematically investigating their own body and directly experiencing the consequences of their own actions. Considering that from an early age infants display exploratory activities that appear to be specifically oriented toward the discovery of their own body's features and characteristics, questions remain as to what young infants know about their own body and how they learn about it. This chapter is an attempt to address these questions by reviewing recent research that has begun to investigate the issue of the function and determinants of early self-exploration. It is proposed that a basic function of self-exploration is to specify the fundamental difference between percePtual self- and nonself-events: events that pertain to the self, and those that do not. We discuss the fact that infants from birth perceive events that uniquely specify either the self or other objects in the environment. We propose that at the origins of a discrimination between self- and nonself-stimulation, there are specific intermodal experiences that emerge via early self-exploration. In support of this proposal, recent empirical findings on the temporal and spatial determinants of self-exploration in infancy are presented.

396

P. ROCHAT & R. MORGAN

Early Discrimination Between Self- and Nonself-experiences If there is a perceptual function associated with self-exploration, it is to specify what pertains to the self, in particular to foster the development of a discrimination between sensory experiences originating either from the self or from objects that are external to the self. Considering that this discrimination is very basic, forming the cornerstone of any conceptualization of the self in the context of comparative psychology, developmental psychology, cognitive psychology, and even robotics, self-exploration construed as a process underlying such discrimination is of great theoretical importance. For example, the assumption of a lack of such discrimination by young infants led William James (1890) to initially depict the blooming, buzzing confusion as characteristic of infants' early sensory experience. Aside from the famous quote, James's theoretical assumption also constrained his views on what needs to develop early in life. If infants do not initially discriminate between self- and nonself-stimulation, how do they come to do so, and on what basis? Inversely, if infants are born with a basic capacity to discriminate between self- and nonself-experiences, what is the nature of this capacity, and how does it develop? We propose that infants, from birth, and possibly from the confines of pregnancy, learn to discriminate between self- and nonself-stimulation. This learning is based on self-exploration, which is expressed from birth and from which infants actively experience fundamental perceptual contrasts between stimulation originating from the sell and stimulation originating from external objects and events. Before discussing the nature of the perceptual experiences specifying the self at the origins of development, and because we assume that these experiences are mediated by self-exploration, let us first define what we understand by the term self-exploration. Self-exploration is a class of behavior through which infants are perceptually oriented toward their own body. They touch themselves, listen to self-produced sounds by vocalizing, or bring their hands and other limbs into the field of view for visual inspection. They also get tense and agitated for the apparent sake of getting tense and agitated, engage in circular reactions, and attend visually or auditorily to the traces of their own actions on objects (see Rochat, this volume). Although self-exploration can be identified as a class of early behavior, it is more than a behavioral inventory: It is a specific process through which infants become perceptually attentive to their own body and engage in a perceptual dialogue with themselves. Again, we propose that this process enables infants to specify themselves as entities that are perceptually differentiated from other objects, people, and any external events occurring in the environment. To a certain extent, self-exploration needs to be distinguished from object exploration, as the emphases

EARLY SELF-EXPLORATION

397

of each are by defmition opposite, one specifically oriented toward things that are external to the body, and the other toward the own body. However, both self- and object exploration inform infants about themselves and in particular how they are situated, differentiated, and are agents in the environment. Both processes specify the infant's ecological self (Neisser, 1991; Rochat, in press). Although the focus here is on self-exploration, self-exploration and object exploration are complementary processes at the origins of the perceptual specification, and eventual conceptualization of the self. But what do infants specifically gain from self-exploration? We propose that, primarily, they gain specific intermodal experiences that form the foundations of the perceptual discrimination between self- and nonself-stimulation. From birth, infants experience contrasting perceptual and sensorimotor events that potentially inform them about their own body as an object differentiated from others in the environment. When infants cry, the sound they hear is combined with kinesthetic and proprioceptive feedback. This intermodal combination uniquely specifies their own body. Sounds originating from another person or any other objects in the environment tend not to share the same intermodal invariants. Aside from vegetative sounds such as crying, coughing, or sneezing, infants from birth produce sounds (i.e., comfort sounds that rapidly become more articulated; precursors of speechlike production) and explore the specificity of their own voice and the potentials (or affordances) of their own vocal track (Oller, 1980; Stark, 1980). In addition, newborns show a robust propensity to bring their hands in contact with their face and mouth (Rochat, Blass, & Hoffmeyer, 1988). Some researchers have observed that newborn infants spend up to 20% of their waking hours contacting the facial region with their hands (Komer & Kraemer, 1972). This simple observation might have implications for the perceptual basis of an early experience of the body as a differentiated object. As in the case of self-produced sound, when newborns manually touch their own face, they potentially experience a sensorimotor and perceptual event that uniquely specifies their own body as a differentiated object. This intermodal event is the "double-touch": the experience of the cutaneous surface of the hand contacting the cutaneous surface of the facial region, which could be any other region of the body surface (yon Glasersfeld, 1988; see Figure 4). The baby's contact with any other physical object, surface, or person in the environment will never correspond to a double-touch intermodal event. But when do young infants start to show discrimination of unique perceptual events, which underlies the ability to perceive their own body as an object different from other objects in the environment?

398

P. ROCHAT •

R. MORGAN

Based on a microanalysis of hand-mouth coordination in newborn infants, Butterworth and Hopkins (1988) report that when infants bring their own hand(s) in contact with the face, this cutaneous self-stimulation is not accompanied by any of the rooting responses normally observed when an external object contacts the same facial location. These observations suggest an early ability to discriminate between environmental (single-touch) stimulation, and self-stimulation (doubletouch + proprioceptive stimulation). In an ongoing experiment conducted at the Emory Infant Laboratory, we are assessing whether newborns discriminate between double-touch stimulation and various external cutaneous (tactile) stimulation. Preliminary results indicate that such discrimination might occur very early in development. In this project, newborns' rooting responses were systematically analyzed (i.e., head turn and oral orienting toward a perioral tactile stimulation) in four different conditions. In one condition, the perioral stimulation originated from the index finger of the experimenter, who was positioned behind the infant and intermittently rubbed the infants' left or fight cheek for 20 seconds. At the beginning of stimulation, the infant's head was oriented at the center. In a second condition, the infant was stimulated by a pacifier held by the experimenter. In a third condition, infants were stimulated by their own hand, either left or fight, which was gently held by the experimenter and rubbed against the infant's cheek. In contrast to the other conditions, this condition provided the infant with a double-touch experience. Finally, in a fourth (control) condition, the hand of the infant was brought close to the cheek by the experimenter, without touching it. In this last condition, infants were stimulated only by passive movement of their own arm as it was moved toward the cheek by the experimenter. Preliminary results obtained with six 3 to 4-week-old infants, each stimulated in each condition once to the fight and once to the left side (for a total of 12 instances), revealed that the proportion of observed rooting in the direction of the stimulation appears to depend on the condition. Infants tended to root more toward either the experimenter's finger or the pacifier, compared with their own hand. No rooting was observed in the fourth condition, where the tactile component was absent. These preliminary observations indicate that early on, young infants might potentially express differential responding to self- versus external tactile stimulation: in other words, between double touch, specifying the sell and single touch, specifying external stimulation. Again, it is important to emphasize the importance of haptic stimulation early in life: This stimulation is a potential source of the perceptual specification of the self, and hence becomes a part of the perceptual basis of an early sense of self. Aside from vocalizing and touching themselves, infants from birth experience the contrast between movements that include their own body, and those that do

EARLY SELF-EXPLORATION

399

not. This contrast is based on different intermodal information specifying either egomotion or motion of objects moving in the environment independently of the self. Egomotion is uniquely specified by combined information from the proprioceptive, visual, and vestibular systems, whereas environmental motion is specified by information from the visual system only. It appears that from an early age, however, infants have a propensity to adjust their posture based on visual information specifying egomotion, with no contingent vestibular stimulation (see chapters by Jouen & Gapenne; Bertenthal & Rose, this volume). They manifest postural adjustments that counteract the information from optic flow specifying egomotion. The fact that infants do not appear to discriminate between self and the environment in this particular situation suggests that there may be two separate classes of environmental information that infants use to specify the self: object movement versus movement of the whole surround. However, evidence exists that young infants use self-produced movement as a source of information for differentiating their own bodies from the surrounding environment. In an experiment by Harris, Cassel, and Bamborough (1974), 8- to 28-week-old infants tracked a moving target only when the object moved alone against a background but not when the background moved in conjunction with the object. Infants typically experience themselves as moving if the background does not remain fixed relative to their own position. In addition, a clever study by Kellman, Gleitman, and Spelke (1987) demonstrated that 4-month-old infants do have the capacity to discriminate between egomotion and the independent movements of an object and are sensitive to the presence or absence of contingent visual-vestibular information. Infants perceived a rod occluded at its center as being an incomplete or broken object when it was stationary. When the same rod was moved back and forth behind the occluder, it was perceived as complete or unbroken. In contrast, when the infant was moved horizontally relative to the stationary object, creating the same translated projection on the infant's retina as in the former case where only the object moved, the object once again was perceived by the infant as being incomplete or broken. The differentially perceived object in the self-motion versus the object motion condition despite exactly the same visual information - - is evidence that proprioceptive/kinesthetic information is used by young infants to discriminate self-movement from independent movement of objects in the environment.

Temporal Determinants of Early Self-exploration in the Mirror It is now well established that young infants manifest temporal contingency perception between their own actions and the transformations they cause in the

400

P. ROCHAT & R. MORGAN

environment (Watson, 1979, 1985, 1994). Lewis, Sullivan, and Brooks-Gunn (1985) report that 2-month-olds whose wrists are connected to a mobile hanging above their crib rapidly learn the contingency between their own ann movements and the movements of the mobile. They display an expression of joy in the process of discovering instrumental arm pulls, and an expression of anger during a period of extinction when arm and mobile are disconnected. It appears that very early on, and probably from birth, infants have the propensity to detect information specifying the contingency between their own actions and their visual consequences (Kalnins & Bruner, 1973; Watson, 1979, 1985; Rovee-Collier, 1987; Siqueland & DeLucia, 1969), as well as information specifying the spatiotemporal congruence of events perceived bimodally (i.e., via vision and audition: see Spelke, 1976; Kuhl & Meltzoff, 1982). The perception of temporal contingency is central to early development. It is a source of information that specifies objects and events, as well as the body as an agent in the environment. Lewis and Brooks-Gunn (1979) and other students of early infancy (Guillaume, 1926; Wallon, 1942/1970) have suggested that the origins of self-perception correspond to the discovery by the young infant of the contingency between visual stimuli and proprioceptive feedback from body movements. Self-exploration in a mirror affords the detection of the perfect visual-proprioceptive contingency. Early on, infants appear to attend to it. Few developmental studies have described infant behavior in front of a mirror in the course of the first year (i.e., prior to the first signs of mirror selfrecognition using the mark task). Before 6 months, infants are shown to be actively involved in discovering their specular image, manifesting signs of perceptual discrimination between themselves and others in a mirror. Again, note that this discrimination does not yet imply any self-recognition, but is an expression of self-exploration. Dixon (1957) described a first stage at around 4 months of age in which the infant looks briefly and soberly at herself in the mirror but shows immediate recognition and sustained attention to the mother's specular reflection. Field (1979) demonstrated that 3-month-olds respond differentially to a mirror image of the self versus an infant peer. In this study, the infant's facial expression, manual behavior, visual activity, and cardiac response were recorded systematically. Field's research suggests the existence of a precocious discrimination between the specular image of the self, and of another person. Amsterdam (1972) reported that between 3 and 5 months, infants show little social behavior in front of the mirror (i.e., smiles, laughs, and vocalization), which only becomes prominent by 6 months. In her study, Amsterdam showed that between 3 months and 12 months, the majority of infants spend time observing their own movement in the mirror, exploring the particular visual-proprioceptive correspondence offered by the mirror's reflection. Interestingly, by 14 months,

EARLY SELF-EXPLORATION 401

infants become less interested in their own movement, and by 20 months begin to show embarrassment and withdrawal in front of the mirror (Amsterdam, 1972; Lewis & Brooks-Gunn, 1979). It thus appears that in the context of the mirror situation, self-exploration is particularly prominent during the first year. Aside from the mirror situation, self-exploration is evident starting at 3 months of age, when infants display long episodes of self-examination; in particular, exploration of their hands in motion (Piaget, 1952). Such exploration and visual control of manual activities have been recently documented in newborn infants (Van der Meer, Van der Weel, & Lee, 1995; Van der Meer & Van tier Weel, this volume). However, very few researchers have attempted to isolate the information (visual, proprioceptive, haptic, etc.) that young infants might be sensitive to when engaged in self-exploration. Papousek and Papousek (1974) placed 5-month-olds in front of two different video images of themselves. Based on the preferential looking of the infant, this method allowed the assessment of the discriminating variables between the two video images. Reporting only pilot observations with 11 infants, Papousek and Papousek found that infants preferred to look at the image of themselves in which eye contact was possible. Using a similar procedure but placing 1- to 24-month-old infants in front of two mirrors that are either flat, blurred, or distorted, Schulman and Kaplowitz (1976) showed that prior to 6 months, infants tend to look more often at the clear rather than the blurred image of themselves, and showed less interest in the distorted image compared to the flat, nondistorted mirror image. Interestingly, Schulman and Kaplowitz note that compared to older infants, 1- to 6-month-olds spend more time looking at a particular mirror, although they do not yet show complex behavior such as looking at a particular body part and then immediately inspecting its reflection in the mirror. Using the principle of the choice method introduced by Papousek and Papousek (1974) but presenting the infant with nonfacial images of the self, in particular their legs, Bahrick and Watson (1985) demonstrated the early detection of visual-proprioceptive contingency. On one of the TV monitors, the infant had access to a contingent view of his/her legs, and on another was simultaneously presented a noncontingent, prerecorded view of the baby's own legs or of another baby's leg movements wearing identical booties. Bahrick and Watson showed that 5-month-olds preferentially looked to the noncontingent view. They also observed this phenomenon in a situation where an occluder prevented the infant from seeing his/her legs directly. Three-month-olds showed split preferences, looking either much longer at the contingent, or much longer at the noncontingent, view. Overall, Bahrick and Watson demonstrated that early perceptual discrimination of the self does not correspond only to facial images of the serf, but includes other parts of the body. This is important because it shows that young infants are

402

P. ROCHAT & R. MORGAN

sensitive to visual and proprioceptive contingency in general, and not only to the contingency of eye contact as suggested by previous researchers, who emphasized the social rather than perceptual context in which first discrimination between self and others takes place (Dixon, 1957; Papousek & Papousek, 1974). Nevertheless, questions remain as to what information is relevant to the young infant in his/her discrimination of intermodal proprioceptive-visual contingency. What is actually detected by the young infant.? Is it a temporal contingency, a spatial congruence, or a combination of both, i.e., a spatio-temporal contingency? Recently, we conducted a series of experiments demonstrating that self-produced movements by young infants entail not only the detection of temporal contingency, but the perception of spatial congruence. Thus, these two factors may be viewed as main determinants of self-exploration in early infancy.

Spatial Determinants of Early Self-exploration In exploring their own body, infants develop expectations about the intennodal correspondence between the proprioception of their body moving in space and its visual consequences. Observations of young infants reveal that by 3 months, infants look systematically at their hands and move them in the field of view for long, careful inspections. From an early age (approximately 3--4 months), infants also commonly grab their feet and bring them in front of their eyes for long bouts of visual-proprioceptive exploration. These bouts of self-exploration provide infants with the opportunity to detect the intermodal invariants that uniquely specify their own body. To further understand what determines early self-exploration, we investigated the extent to which young infants are sensitive to the relative spatial congruence of visual-proprioceptive information. Using a video technique analogous to the one used by Papousek and Papousek (1974) and Bahrick and Watson (1985), we systematically manipulated the spatial configuration of visual feedback pertaining to self-produced movements. In contrast to former studies, we maintained the temporal contingency between visual and proprioceptive information perfectly constant, varying only their relative spatial congruence. We review our experiments and their results below, in relation to each spatial aspect manipulated. As a general paradigm, we used infants' preferential looking to different online views of their legs from the waist down. This paradigm is a modified version of the one used by Bahrick and Watson (1985). Infants were placed in front of a large television monitor with a split screen. On either side of the split screen a particular on-line view of the infant's legs was displayed, from separate cameras placed at different angles or with optical characteristics such as a left/right reversal.

EARLY SELF-EXPLORATION

403

To entice the infant to visually attend to the TV display, attractive striped socks were put on the infant. A small microphone was placed under the infant's feet that picked up rustling/scratching sounds each time the infant produced a leg movement. The leg movements' sounds were amplified and were heard by the infant from a speaker placed centrally on top of the TV. A camera was placed under the TV that provided a close-up of the infant's face .for later preferential-looking analysis. This image was synchronized with the audio recording of the infant's leg activity. Blind coders entered in real time the infant's gazing on two channels of a computerized event recorder corresponding to either the right or left side of the split screen. Meanwhile, the synchronized spectrogram of the audio recording of the infant's leg activity was entered in another channel and digitized into 2-second bouts of leg activity (see Rochat & Morgan, 1995, for details). In short, this technique allowed the coanalysis of preferential looking of either view of the legs and for the amount of self-produced leg activity. The rationale for these experiments was that if infants showed discrimination between the two views of their legs, they should look preferentially at one of the views and should produce a differential amount of leg activity while looking at the preferred view.

Overall Directionality Congruence In the first experiment (Rochat & Morgan, 1995), infants were presented with an Observer's view and an Ego view of their own legs (see Figure 1A). Each view was provided by a camera placed either above and behind the infant, or above and in front of the infant. There were two basic spatial differences between the two views: 1) orientation; and 2) overall movement directionality of the legs. Regarding the experimental design, in all experiments infants were recorded for 5 minutes in front of the display. The side of the view was counterbalanced among subjects of each age group (N = 10). Overall, in the first experiment, infants at both ages and from 3 months of age, expressed a differentiation between the two views of their legs: 1) they tended to look significantly longer at the observer's view (i.e., the noncongruent view); 2) after multiple comparisons between the two views, they tended to settle their gaze toward the preferred view as a function of the 5 minutes of testing time; and 3) they generated significantly more leg activity while looking at the observer's view (noncongruent) compared to the ego (congruent) view, expressing an increase in self-exploration in the context of the nonfamiliar view. In order to untangle the confound between differences in spatial orientation and spatial directionality of the two on-line views presented in the first experhnent, we conducted a second experiment in which both views of the legs depicted a similar orientation (two ego views), the two views being only different in relation to leg movements' directionality (see Figure 1B). Inversion of movement directionality

404

P. ROCHAT & R. MORGAN

was obtained by using a camera with a left/right inverted tube. Again, 3- and 5month-old infants were recorded for 5 minutes in front of the display. The side of the view was counterbalanced among subjects of each age group (N = 10). Overall, in the second experiment, infants of both ages continued to express a differentiation between the two views of their legs: 1) they tended to look significantly longer at the reversed ego view (i.e., the incongruent view); 2) following frequent comparisons between the views, they tended to settle their gaze toward this preferred view as a function of the 5 minutes of testing time; and 3) they generated significantly more leg activity while looking at the reversed ego (incongruen0 compared to the ego (congruent) view, expressing an increase in selfexploration in the context of the incongruent view that varied only in terms of the overall directionality of movement. 1A

OBS. VIEW

EGO VIEW

1B

REV. EGO VIEW

EGO VIEW

lC

"RE~/. 'OBS. VIEW

EGO VIEW

~GURE 1. The two views of their own legs as seen by the infant on the TV in the first 3 ;xperiments (Rochat & Morgan, 1995). (A) Observer view vs. Ego view (Experiment ); 03) Reversed Ego view vs. Ego view (Experiment 2); (C) Reversed Observer view vs. ~go view (Experiment 3).

EARLY SELF-EXPLORATION

405

Relative Orientation of the Body To assess the extent to which the relative spatial orientation of the body is a determinant of early self-exploration, we conducted a third experiment in which the two views presented to the infant varied in orientation only (Rochat & Morgan, 1995). Using the same experimental paradigm and procedure, movement directionality was kept congruent with the infant's own movements in both views, but the legs' orientation corresponded either to an Observer's view or an Ego view (See Figure 1C). The view in which movement directionality was congruent, yet inverted in orientation (reversed Observer's view) was obtained by using the camera with the inverted tube. It changed the orientation of the legs but not the overall directionality of their movements. Again, 3- and 5-month-old infants were recorded for 5 minutes in front of the display. The side of the view was counterbalanced among subjects of each age group (N = 10). In contrast to the other experiments, infants at both ages did not show any preference for either of the two overall orientations, nor any settling of their gaze as a function of testing time, nor any significant increase in leg activity while looking at either view. The results of this experiment indicated that infants do not appear to be sensitive to orientation changes of their own legs when overall movement directionality is maintained constant. In relation to the other experiments, the findings suggest that infants as young as 3 months discriminate between congruent and incongruent views of selfproduced leg movements, the spatial determinant of this early discrhnination being movement directionality rather than the global spatial orientation of body parts.

Relative Directionality Congruence Recently, we conducted two new experiments to address the question of whether young infants are sensitive to changes in the relative position of their own legs that they see moving on a screen (Morgan & Rochat, 1994). Again, 3- and 4to 5-month-old infants were tested in a slightly modified procedure: They were presented with a composite, on-line (ego) view of their own legs, in which both the orientation and movement directionality of either leg were kept constant, but their relative position on the screen was altered. In the Normal view, infants saw their legs in their normal relative positions: the fight leg to the right of the screen and the left leg to the left (see Figure 2A). In the Reversed view, the legs' positions were reversed: the left leg to the right and the fight leg to the left side of the screen (see Figure 2B). Both left and fight images of one leg originated from two separate cameras placed behind and above the infant. Infants were shown the normal and reversed conditions in four alternating sequences of 2 minutes. Order of presentation was counterbalanced among subjects of each age group. The rationale of this experiment was the following: If infants perceive the contrast between the normal and reversed conditions, they should tend to look and

406

P. ROCHAT & R. MORGAN

kick differentially across these two conditions. Results of this experhnent showed that infants from 3 months of age manifest differential looking and kicking behavior across the two conditions. For both groups, infants tended to reduce their looking and leg activity when presented with a reversed relative location of their legs on the screen. These results suggest that young infants are sensitive to differences in the relative movements and/or the featural characteristics of the legs (i.e., the relative bending of the legs at the knees and ankles) across the two conditions.

2 A

LEFT

2B

RIGHT

NORMAL VIEW

RIGHT

REVERSED VIEW

2C

LEFT

LEFT

2D

RIGHT

NORMAL VIEW

RIGHT

LEFT

REVERSED VIEW

FIGURE 2. The legs as they appeared to the infant on the TV in both conditions of Experiments 4 and 5 (Morgan & Rochat, 1994). (A) Normal view (Experiment 4); (B) Reversed view (Experiment 4); (C) Normal view with bulky socks (Experiment 5); (D) Reversed view with bulky socks (Experiment 5).

Featural Characteristics of the Body In order to control for the potential determinant of the relative featural characteristics of the infant's legs (the legs' bending) that changed between the normal and reversed conditions of the latter experiment, we conducted another study where features of the legs were maintained constant while relative leg position was varied across conditions (Normal and Reversed). Again, infants were tested

EARLY SELF-EXPLORATION

407

successively in the two conditions, and wore bulky socks to cover the bending of the legs (see Figures 2C and 2D). In contrast to the preceding experiment, the results indicated for both age groups (N = 10 infants in each) no significant difference in looking, gaze switching, and leg activity between the normal and reversed conditions. These negative results, in conjunction with the positive ones obtained in the preceding experiment in which infants did not wear any bulky socks, indicate that featural characteristics of the legs, combined with relative movement directionality, form important spatial determinants in the perception of self-produced leg movements by infants as young as 3 months of age. Results of the second experiment suggest that relative movement directionality alone is not a significant spatial determinant in the perception of self-produced movements. However, the fact that the relative position of the legs in space appears to be discriminated reveals that infants express a calibrated intermodal space of the body as early as 3 months. Overall, the change in the legs' position created a pattern of movement that was very unlike the pattern of movement infants see when looking down at their own legs. However, the preference for the less novel view is difficult to interpret, given the tendency in our other experiments for infants to look and kick more actively at the more novel view. In general, the results of these experiments demonstrate that the visual-proprioceptive experience of the legs moving while in their proper relative positions combined with their usual featural outline in space seem to be important determinants of early self-exploration. In addition, the invariant information of the relative position of body parts detected by infants may be considered to be one of the earliest expression of a body schema.

Goal-orientation and Self-exploration In the final set of studies conducted so far, the question of whether selfexploration serves a unique function relative to other forms of exploration, such as the exploration of sounding objects, was asked. Using the sequential looking paradigm similar to the one described in the previous experiments, we further examined infants' sensitivity to changes in the spatial arrangement of their legs while moving. Specifically, when the seen movements of the legs (based on feedback from a video image) are opposite in direction to how the legs are felt to be moving, will infants show different patterns of looking and leg activity depending on the nature of the task? A new experimental condition was added, in which infants were oriented toward the goal of contacting an object with their foot, which sounded when touched (Morgan & Rochat, in press). This new condition was compared to the behavior infants express when the task is purely self-oriented in nature, i.e., when no object is present (see Figure 3). Again, two groups of infants were tested (at 3 and 4-5 months). Infants were presented successively with

408

P. ROCHAT & R. MORGAN

an on-line image of their own legs. As in the previous experiments, because they were reclined and the monitor was at an angle, infants were unable to see their own legs directly.

3A

3B

EGO VIEW

REv. EGO VIEW

3C

EGO VIEW (WITH OBJi~CT)

3D

REV. EGo VIEW (WITH OB.IECT)

FIGURE 3. The legs as they appeared to the infant on the TV in all 4 conditions of Experiment 6 (Morgan & Rochat, in press). (A) Ego view. (B) Reversed Ego view. (C) Ego view with object. (D) Reversed Ego view with object.

Infants viewed their legs in two different conditions. In the first condition, they were presented with their own legs moving on the TV with no object to kick at (see Figures 3A and 3B). In the second condition, they viewed their legs and an object that sounded when contacted by their foot (see Figures 3C and 3D). Within each condition, infants experienced two different situations for 2 minutes each. In one situation, they viewed a spatially congruent on-line image of their legs from the waist down (Ego view, Figures 3A & 3C). In another situation, a modified camera, placed in the same position, provided the infant with a similar ego view, except that a left/right reversal of the image on the TV screen provided a spatially noncongruent view of the legs (Reversed Ego view, Figures 3B & 3D). The situations within each condition and the side of the object were counterbalanced across trials.

EARLY SELF-EXPLORATION

409

The rationale of the study was that, if infants were able to discriminate between the different conditions, they would show differential looking and leg activity across the object versus the no-object conditions. In particular, in the condition in which the task was self-oriented in nature, infants would look longer and show more leg activity for the Reversed Ego view, as in Rochat and Morgan (1995). In contrast, because of the goal-oriented nature of the task in the condition where the object was present, the opposite pattern of behavior was predicted. The rationale for this prediction was that congruent visual-proprioceptive feedback would facilitate contact with the object, the functional goal of the task. The results revealed that depending upon the condition (object vs. no-object), infants showed significantly different patterns of looking and leg activity for the Ego versus the Reversed Ego views of their legs. Specifically, they looked longer and showed more leg activity when looking at the Reversed Ego view when there was no object present and looked longer at the Ego view when an object was present. At both ages, results confirmed the predicted patterns of looking and leg activity. These results extend the findings of Rochat and Morgan (1995), which showed that infants prefer looking at displays modifying the directionality of their own leg movements. The present results suggest that infants are able to modulate their activity depending on the context and function of the task. Infants preferentially use the congruent view of their legs in the object condition in order to control their attempts to contact the object, a spatially oriented motor task. In contrast, when infants are engaged in the exploration of their own body (no-object condition), the activity pattern is reversed, as there is no specific spatial goal to direct their movements. The results of this last research suggest that young infants can be differentially oriented while self-exploring. They look preferentially at either the image corresponding to the familiar or unfamiliar visual-proprioceptive correspondence of their own legs, depending on whether or not there is a spatial orientation attached to the task (i.e., whether or not self-produced leg movements have a "spatial address"). Early on, the pattern of infants' visual attention to self-produced leg movements is determined by the nature of the task. Overall, these latter results indicate that young infants express flexibility in the functional orientation of selfexploration.

Conclusion: Self-exploration and Intermodal Body Schema in Early Infancy The research reviewed above suggests that early on, infants pick up invariant information specifying their own body as a differentiated entity in the

410

P. ROCHAT & R. MORGAN

environment. This information is intermodal and pertains to the coengagement of proprioceptive and other perceptual systems (e.g., vision, in the research reviewed above). From at least 3 months of age, infants discriminate visual-proprioceptive information that is either consistent or inconsistent with regular perceptual feedback they experience when, for example, feeling, hearing, or seeing moving parts of their own body. Both temporal contingency (Bahrick & Watson, 1985) and spatial congruence (Rochat & Morgan, 1995) of visual, auditory, and proprioceptive feedback determine young infants' exploration of self-produced movements. The reviewed observations suggest that young infants are sensitive to the dynamic features of their own body, in the same way that they are precociously sensitive to the dynamic features of physical objects in their environment (Gibson & Spelke, 1983; Kellman, 1993). However, there is a fundamental difference between self-perception (own body perception) and object perception. Objects come and go; the perceptual array is constantly refurbished with novel objects, events of appearance, reappearance, and disappearance. In contrast, there is a permanence attached to the perception of the own body that is based on the coactivation of proprioceptive and other perceptual modalities. This permanence supports the early and rapid construction of the body as a privileged object of knowledge. From birth, whenever movement is generated, and whether this movement has an external or internal cause, it always carries information about the body as a whole, its present postural configuration, and the relative position of its parts. Young infants and neonates are actively engaged in picking up this information, which they use to calibrate the features and effectivities of their own body. Self-exploration is an important process underlying such calibration. This process, in which infants are perceptually attentive to their own body, entails the coactivation of multiple modalities (visual, tactile, proprioceptive, auditory, etc.) that specify its spatial and temporal organization. To the extent that young infants manifest an intermodal calibration of the body (i.e., detection of temporal contingency and spatial congruence of intennodal feedback accompanying self-produced movements), they possess an implicit, intermodal body schema. This precocious body schema is fundamentally active (dynamic) and perceptually based. It does not imply any explicit representation of the features and organization of the body as in the concept of body image introduced by Schilder (1935). The term body schema is used here in the sense proposed by Gallagher and Meltzoff (in press), and is based on the contrast between body image and body schema discussed by Gallagher (1986): "The body schema, as distinct from body image, is a nonconscious performance of the body ... an active,

EARLYSELF-EXPLORATION 411 operative performance of the body, rather than a copy, image, global model, or conception of the existing parts of the body. The schema is the body a~s it actively integrates its positions and responses in the environment" (Gallagher, 1986, p. 548). In other words, the body schema corresponds to the perception and calibration of the body and its effectivities (Rochat, this volume). It can be construed as an implicit process accompanying any goal-oriented actions and the perception of self-produced movements. It is the process of a dynamic mapping of

the body, the differentiation of its parts, and how these parts relate to one another in space and time. When, for example, an infant brings one hand in the visual field, he does not gain only visual-proprioceptive information about the hand, but also information about the relative situation of this part of the body in relation to all the other parts that are simultaneously perceived via tactile, vestibular, or proprioceptive feedback. In other words, there is no such thing as the discrete perception of a particular body part because they are all linked in the posture and position of the body as a whole. For example, when an infant reaches for an object, this action implies more than the perception of the hand and of the object target. It entails the perception of the body as a whole, with posture scaffolding any particular action (Rochat & Senders, 1991; Rochat & Bullinger, 1994). In goal-oriented action, the body is mapped into an intermodal body schema. The mapping and calibration of the body by young infants implies movement and action, whether this action is oriented toward the body, as in the case of newborns' hand-mouth coordination (Rochat, Blass, & Hoffmeyer, 1988), toward external objects, as in the case of early reaching (Hofsten, 1982), or toward people, as in the case of neonatal imitation (Meltzoff & Moore, 1977). In other words, the body schema corresponds to the active calibration of the body and its effectivities in relation to functional goals (e.g., bringing hand to mouth, contacting an object, or imitating others). Considering that it is inseparable from oriented action, and to the extent that newborn behavior is not merely reflexive but corresponds to goaloriented actions (see Rochat; Gibson; this volume), an implicit body schema is expressed from birth. When bringing their hand to the mouth, and because this action is not random or reflexive (Butterworth & Hopkins, 1988; Rochat, 1993), neonates express the implicit knowledge of their mouth and hand as situated subsystems of their own body. As illustrated in the example of Figure 4, the hand moves toward the mouth, and the mouth opens in anticipation of a manual contact. This coordinated action implies some rudiments of a mapping of bodily space; in particular, some implicit knowledge about the relative situation of hand and mouth as they move toward one another. The existence of an innate body schema is suggested by the report of phantom limbs in some cases of congenital aplasia (Weinstein & Sersen, 1961). Further support for an innate body schema is provided by the demonstration of neonatal

412

P. ROCHAT ~e; R. MORGAN

imitation (Meltzoff & Moore, 1977), which implies that newborns possess some tactile-kinesthetic mapping of their own body that they are capable of matching to the movement they see in another person (see Gallagher and Meltzoff, in press, for a detailed discussion). In the case of imitation, the implicit body schema pertains to both the mapping of the own body in relation to the body of another person (the model), and the intermodal matching of visual and tactile-kinesthetic information. In particular, imitation of facial expression entails the translation of visual information pertaining to the model into the actual tactile-kinesthetic control of the imitative act.

FIGURE 4. Hand-to-mouth coordination by a 10-minute old newborn infant. (A) Beginning of the sequence: The neonate displays anticipatory mouth opening prior to hand-mouth contact; (B) The hand approaches the mouth; (C) The hand touches the mouth, and the infant experiences a "double touch." Photographs by Philippe Rochat of his daughter C16o.

To the extent that a body schema is expressed from birth, what is the value of self-exploration as a potential opportunity to calibrate the spatial and temporal configurations of the body? If early on, infants manifest that they possess an implicit knowledge about their own body, this early expression of a body schema is linked to limited and rapidly developing goal-oriented action systems (orienting,

EARLY SELF-EXPLORATION

413

imitating, feeding, etc.). Rapid changes in the infant's behavioral repertoire, physical growth, and the emergence of new implementations of body parts within novel coordinative structures and action systems require a constant recalibration of the body and of how subsystems of the body relate spatially and temporally to one another. There is indeed a constant and necessary recalibration of the body schema. For example, the emergence of reaching redefines the mapping of the hands and how they relate spatially and temporally to the rest of the body. In order to reach successfully toward an object, the infant is required first to maintain the overall posture and balance of the body (Rochat, 1992). In learning to reach successfully and efficiently, babies learn about new bodily constraints that call for a redefinition of the body schema. Likewise, when infants start to sit on their own, this progress opens up new possibilities of manual action. As self-sitting frees the upper limbs from the encumbrance of maintaining balance, reaching by self-sitting infants entails a larger prehensile space (Rochat & Bullinger, 1994; Rochat & Goubet, 1995). This development is accompanied by the ability to reach with either one or two hands, at greater distances, and with a broader span, redefining the effectivities of the hands in relation to objects in the environment, but also in relation to the rest of the body. Although the relative location of the hands on the body remains invariant, the emergence of new body effectivities in reaching changes the dynamics of the hands in relation to the rest of the body. In other words, the development of new degrees of behavioral freedom and the emergence of new coordinative structures imply a recalibration of the body schema. In addition to serving the essential function of providing young infants with the opportunity to discriminate between self- and nonself-stimulation, it is in the context of such recalibration that self-exploration continues to serve the function of specifying the body and its effectivities, beyond the first months of life. REFERENCES

Amsterdam, B. (1972). Mirror self-image reactions before age two. Developmental Psychobiology, 5, 297-305. Bahrick, L.E., & Watson, J.S. (1985). Detection of intermodal proprioceptive-visual contingency as a potential basis of self-perception in infancy. Developmental Psychology, 21, (6), 963-973. Butterworth, G., & Hopkins, B. (1988). Hand-mouth coordination in the newborn baby. British Journal of Developmental Psychology, 6, 303-314. Dixon, J.C. (1957). Development of self-recognition. Journal of Genetic Psychology, 91, 251-256. Field, T.M. (1979). Differential behavioral and cardiac responses of 3-month-old infants to a mirror and peer. Infant Behavior and Development, 2, 179-184. Gallagher, S. (1986). Body image and body schema: Conceptual clarification. Journal of Mind and Behaviour, 7, 541-554.

414

P. ROCHAT & R. MORGAN

Gallagher, S., & Meltzoff, A. (in press). The earliest sense of self and others: MerleauPonty and recent development studies. Philosophical Psychology. Gibson, J.J. (1979). The ecological approach to visual perception. Boston, MA: Houghton Mifflin. Glasersfeld, E. von, (1988). The construction of knowledge: Contributions to conceptual semantics. Salinas, CA: Intersystems Publications. Guillaume, P. (1926). L'imitation chez l'enfant, Paris: Alcan. Harris, P.L., Cassel, T.Z., & Bamborough, P. (1974). Tracking by young infants. British Journal of Psychology, 65, 3, 345-349. Hofsten, C. von (1982). Eye-hand coordination in newborns. Developmental Psychology, 18, 3, 450-461. James, W. (1890) The principles of psychology. New York: Henry Holt. Kalnins, I.V., & Bruner, J.S. (1973). The coordination of visual observation and instrumental behavior in early infancy. Peception, 2, 307-314. Kellman, P.J. (1993). Kinematic foundations of infant visual perception. In C. Granrud (Ed.), Visual Perception and Cognition in Infancy (pp. 121-174). Hillsdale, NJ: Lawrence Erlbaum. Kellman, P.J., Gleitman, H., & Spelke, E. (1987). Object and observer motion in the perception of objects by infants. Journal of Experimental Psychology: Human Perception and Performance, 13 (4), 586-593. Korner, A.F., & Kraemer, H.C. (1972). Individual differences in spontaneous oral behavior in neonates. In J.F. Bosma (Ed.), Third Symposium on Oral Sensation an Perception (pp. 335-346). Bethesda MD: U.S. Department of Health, Education and Welfare Publication. Kuhl, P.K., & Meltzoff, A.N. (1982). The bimodal perception of speech in infancy. Science, 218, 1138-1141. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self New York: Plenum Press. Lewis, M., Sullivan, M.W., & Brooks-Gunn, J. (1985). Emotional behavior during the learning of a contingency in early infancy. British Journal of Developmental Psychology, 3, 307-316. Meltzoff, A.N., & Moore, M.K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 75-78. Morgan, R., & Rochat, P. (1994, June). Perception of self-produced leg movements by

3- to 5-month-old infants in conditions of conflicting visual-proprioceptive feedback. Poster presented at the International Conference on Infant Studies, Paris,

France. Morgan, R. & Rochat, P. (in press). The perception of self-produced leg movements in self- versus object-oriented contexts by 3- to 5-month-old infants. In B.G. Bardy, R.J. Bootsma, & Y. Guiard (Eds.), Studies in perception and action III. Hillsdale, NJ: Lawrence Erlbaum. Neisser, U. (1991). Two perceptually given aspects of the self and their development. Developmental Review, 11, 197-209. Oiler, D.K. (1980). The emergence of the sounds of speech in infancy. In G. YeniKomshian, J. Kavanagh, & C. Ferguson (Eds.) Child phonology." Vol.1 Production (pp. 93-112). New York: Academic Press.. Papousek, H., & Papousek, M. (1974). Mirror-image and self-recognition in young infants: A new method of experimental analysis. Developmental Psychobiology, 7, 149-157. Piaget, J. (1952). The origins of intelligence in children. New York: International Universities Press.

EARLY SELF-EXPLORATION 415 Rochat, P. (1992). Self-sitting and reaching in 5- to 8-month-old infants: The impact of posture and its development on early eye-hand coordination. Journal of Motor Behavior, 24, (2), 210-220. Rochat, P. (1993).Hand-mouth coordination in the newborn: Morphology, determinants, and early development of a basic fact. In G.J.P. Savelsbergh (Ed.), The development of coordination in infancy (pp 265-288). Amsterdam: Elsevier. Rochat, P. (in press). Early development of the ecological self. In C. Dent-Read & P. Zukow-Goldring (Eds.).,Changing ecological approaches to development: Organism-environment mutualities. APA Press. Rochat, P., Blass, E.M., & Hoffmeyer, L.B. (1988). Oropharyngeal control of handmouth coordination in newborn infants. Developmental Psychology, 24 (4), 459-463. Rochat, P., & Bullinger, A. (1994). Posture and Functional Action in Infancy. In A. Vyt, H. Bloch, & M. Bornstein, M. (Eds.), Francophone perspectives on structure and process in mental development (pp. 15-34). Hillsdale, NJ: Lawrence Erlbaum. Rochat, P., & Goubet, N. (1995). Development of sitting and reaching in 5- to 6month-old infants. Infant Behavior and Development, 18, 53-68. Rochat, P., & Morgan, R. (1995). Spatial determinants in the perception of selfproduced leg movements by 3- to 5-month-old infants. Developmental Psychology, 31 (4). Rochat, P., & Senders, S. J., (1991). Active touch in infancy: Action systems in development. In M.J. Weiss & P.R. Zelazo (Eds.), Infant attention: Biological constraints and the influence of experience. Norwood, NJ: Ablex Publishers. Rovee-Collier, C. (1987). Learning and memory in infancy. In J.D. Osofsky (Ed.), Handbook of infant development (pp. 98-148). New York: Wiley. Schilder, P. (1935). The image and appearance of the human body. London: Kegan Paul. Schulman, A.H., & Karplowitz, C. (1976). Mirror-image response during the first two years of life. Developmental Psychobiology, 10 (3), 133-142. Siqueland, E.R., & DeLucia, C,A, (1969). Visual reinforcement of nonnutritive sucking in human infants. Science, 165, 1144-1146. Spelke, E.S. (1976). Infant's intermodal perception of events. Cognitive Psychology, 8, 533-560. Stark, R.E. (1980). Stages of speech development in the first year of life. In G. YeniKomshian, J. Kavanagh, & C. Ferguson (Eds.), Child phonology: Vot. 1 Production. (pp. 73-90). New York: Academic Press,. Van der Meer, A.L.H., Van der Weel, F.R., & Lee, D.N. (1995). The functional significance of arm movements in neonates. Science, 267, 693-695. Wallon, H. (1942/1970) De l'acte ?t la pens~e: Essai de psychotogie compar~e. Collection Champs Flammarion. Watson, J.S. (1979). Perception of contingency as a determinant of social responsiveness. In E. Thoman (Ed.), The origins of social responsiveness (pp. 33-64). Hillsdale, NJ: Erlbaum. Watson, J.S. (1985). Contingency perception in early social development. In T.M. Field & N.A. Fox (Eds.), Social perception in infants (pp. 157-176). Norwood, NJ: Ablex Publisher. Watson, J.S. (1994). Detection of self: Perfect algorithm. In S.T. Parker, R.W. Mitchell, & M.L. Boccia (Eds.), Self-awareness in animals and humans: Developmental perspectives (pp. 131-148). New York: Cambridge University Press. Weinstein, S.E., & Sersen, E.A. (1961). Phantoms in cases of congenital absence of limbs. Neurology, 11, 905-911.

This Page Intentionally Left Blank

SECTION 3

Social Origins of the Self

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research

P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

419

CHAPTER 2 0

Self/Other Differentiation in the D o m a i n of Intimate Socio-Affective Interaction: Some Considerations DANIEL N. STERN

University of Geneva

This chapter explores some aspects of self/other differentiation as it occurs in the domain of socio-affective interactions. This exploration takes place in the context of several broader questions that are posed in this volume. Do socio-affective exchanges play a special role in the general differentiation of self from other-things-in-the-world? Some argue that interpersonal interactions play a unique and determining role because infants are so competent in these exchanges very early (in the 3rd and 4th months of life), well in advance of a comparable competence in reaching, grasping, auto-locomotion, etc. (e.g., Trevarthen, 1979, 1993; Reed, this volume). Others argue that the infant begins at birth to explore his or her own body moving and acting and that this early exploration lays the initial (and formative) groundwork for the self/other-things distinction, well before or simultaneous with the establishment of interpersonal competencies, (see Van der Meer, this volume; Rochat & Morgan, this volume). This argument, in large part, turns on which comes first, and that may be difficult or impossible to decide. A prior question concerns whether it is useful to think in terms of development starting with a general differentiation of self from other-things-inthe-world or alternatively, in terms of many parallel self/other differentiations going on simultaneously in separate domains, each with a good deal of domain specificity. From the point of view of research strategy, it seems necessary to first know what may be specific about the identification of the self in the different domains. With this in mind, we will be concerned with some of the features of self/other differentiation that are particular to the domain of socio-affective interaction.

420

DANIEL N. STERN

More to the point, we will attempt to describe the "environment" of information available to the infant for specifying self from other during an affective exchange. A first particularity of this domain concerns the relative range of learning opportunities afforded. During the first 6 months, or even the first 12 months of life, most infants have extensive socio-affective exchanges with very few people, often only one. Even when several possible partners are available, the strong tendency in both baby and mother for privileging one among many relationships is considerable. This is especially so with the more intimate, more purely socioaffective exchanges such as face-to-face play or attachment behaviors. In contradistinction, there are many graspable objects in the babies' world and they will be reached for, even if some hierarchy of grasp ability is operating. Similarly, in another domain, once the baby moves or is moved in space, there is a huge set of different arrays of optical information that will specify the nonself. The contribution of the particularities of any one graspable object informing the infant's overall view of self and object in the domain of grasping will be very small. On the contrary, in the domain of interpersonal affective interaction the contribution of the particularities of the primary caregiver will be enormous in forming the infant's overall view of self and others. In brief, the opportunities for generalization are quite different for the ecology of the domain of intimate socioaffective exchange. In fact, the particularities of one or two person's interactive behavior becomes one of the main features of the "environment" against which the infant's "interpersonal self' (see Neisser, 1993) gets defined. Clinical theories have always implied as much. But we are now in a better position to explore this feature, which we will take up again below. A second special feature of this domain is that during the first half year of life (at least), an affectively interacting person is the most captivating of perhaps all possible stimuli. (One could imagine a robot designed to be a supra-stimulus, but it would most certainly end up being an exaggeration of human forms and behaviors.) Not only are persons (especially interacting ones) preferred and most arousing, they are the form of stimulation that most elicits large rapid shifts in the infant's own activation and affect state. These feeling shifts then become an important part of the total environment that can be assigned to the self or not, during intimate interactions. The third particularity of the affective interpersonal domain concerns the perceivable units or events that make up the infants' subjective landscape in this activity. I will argue that from a very early age, the infant perceives intentions in the self and the other, that he or she "sees past" the specific overt behaviors in order to read in them the intentions that organize these behaviors. This argument has been made before in various related forms by Braten (1988), Stem (1985), and Trevarthen (1979, 1987, 1993). The notion that the

SELF/OTHER DIFFERENTIATION

421

infant seeks and identifies intentions which then organize his or her perceptions into functional units is applicable to self and others, but not to inanimate objects. It makes the effective environment of affective exchange a rather different one, consisting of a landscape of intentions (and motives). A closer look at such an environment is called for. And we will begin with a consideration of how these interactions work, i.e., what is happening during them, and what may be "perceivable." Socio-affective exchanges are conveniently seen in terms of large regulatory goals such as joy or security. These large goals are achieved by way of a constant process of micro-regulations. The perceivable intentions that are of interest to us here lie in or behind these micro-regulations. Two examples of relatively "pure" affective interpersonal interactions will suffice to illustrate the situation: free play from about 3-6 months of age and the negotiation of certain attachment tasks around 12 months and later. (These interactions are relatively "pure" in the sense that they depend on no external objects or references external to the behavior within the dyad that is directed toward each other.) In free play, the overall goal is the mutual regulation of joy and activation at an optimal level (for the dyad) that is not overstimulating and not boring (i.e., to have fun; see Stem, 1974, 1977). In certain attachment tasks, the overall goal is the mutual regulation of proximity and attachment behavior to achieve an optimal level of felt security (e.g., Grossman & Grossman, 1991). To achieve these large-scale goals, the process of constant micro-regulation is enacted to guide the activity to its goal. The perceivable environment consists, in fact, of the train of micro-regulation that adjusts the interaction up and down. These micro-regulations are most often of very short duration ~ seconds and split seconds. They are usually mutual. They involve many behaviors in many channels simultaneously. Their form is complex, fluid, and changeable. They pose the problem of how the infant could possibly parse them, identify invariants, make some sense out of them, and assign what to the self and what to the other. An example suffices here to illustrate the problem. Suppose the mother and a 4- to 5-month-old start a face-to-face free play and are for the moment, at the beginning, relatively expressionless: - The mother then opens with a low-keyed behavioral constellation of simultaneous behaviors: looking at baby; head leans toward him; faint smile; mouth a bit open; eyebrow a bit up; soft vocalization. - The baby responds with a slight and transient smile, mild eye widening, and eyebrows lift. His arms start to move about in irregular circles. The mother then leans closer to the baby, progressively exaggerating the facial display already present (i.e., her expression grows toward one of "mock surprise"; Stern 1977), the baby regards her unflinchingly, his eyes open more and then close -

422

DANIEL N.

STERN

some several times over as does his mouth, as he watches the intensity of her facial display grow. Finally, the baby breaks into a moderately intense open mouth smile accompanied by an elongated vocalization, his eyebrows fly up, his head tilts back, his arms flap, and his hands open and close. This constellation lasts about 2 seconds. - 250 msec after the above baby behavior start~s, the mother bursts into a smile, imitating (unaware) the baby's facial expressions and vocalization- such that for the last 1 3/4 seconds of the baby's behavior the mother's behavior is being contoured to match his. - And so on. -

Recall that while all of this is occurring overtly, a script is unfolding mentally in the mind of both partners. It is safe to suggest on the basis of all of the experiments involving violations of natural interactions (e.g., Trevarthen, 1979; Tronick et al., 1978) that the infant (as well as the mother) has expectations about what the other is likely to do at most steps of the way once the sequence has been initiated. In this situation, there is an enormous informational flow, and a complex one from the point of view of identifying self-versus-other invariants. It is helpful to specify some of these obstacles: 9 Information is arriving in many, sometimes all, sensory channels simultaneously. 9This information is changing almost constantly and not necessarily in systematic ways. 9The contingent relationships between the behavior of the parmers shift very freely across a wide spectrum. This includes changes in schedules of reinforcement, timing parameters of response, extent of overlap of behaviors, and shift in and out of simultaneity and isomorphism (imitation) of actions. Accordingly, changes in the environment may seem at times to originate with the infant or with the mother (and at times, acting at a distance and at other times, by way of contact). 9The goals are largely internal states (especially arousal and pleasure) and are only superficially external events. However, even here the attribution of motive or intentions to alter internal states to the self or to the other is complicated by imitation, emotional contagion, and identification. It is most important here to take into account the recent suggestion of Meltzoff (see Meltzoff & Moore, this volume) that, just as manipulation is the infant's means to explore and know graspable inanimate objects, imitation is his or her means to explore and know other people (same for the mother). 9 Certain internal states for an infant (e.g., intense joy with high activation), which are involved in face-to-face play, can only be achieved

SELF/OTHER DIFFERENTIATION 423

during a specific kind of playful interaction with another. During the first year of life, infants do not burst into dynamite smiles or laughter unless thrown into this higher orbit by an interacting human partner. They cannot get there alone, and inanimate things cannot do it for them. Thus, for an important segment of their own affective life, the interactive presence of another is an invariant. A register of feeling within the self requires interpersonal evocation and regulation (Stem, 1985). 9Finally, the signal value of any behavior or constellation of behavior is constantly changing, depending on what leads up to it and what state the dyad is in when it is performed. Although there is a limited range of meanings for any constellation of actions, the exact meaning intended can change considerably, according to its micro- and macro-interactive context. What, then, is a perceivable event for the infant in this situation? What are the units of information that can be attributed to self or to other? First of all, the interactive flow of behavior during these affective exchanges is, for the most part, broken into bursts of constellations of behavior with relative pauses or obviously marked transitions in between. Weinberg and Tronick (1994) recently have done a fine-grained behavioral analysis of these "affective configurations of facial, vocal, gestural, and regulatory behavior" (p. 1503). The production of temporal packages in the form of these constellations makes the infant's task of parsing the behavioral flow much easier (see also Stern et al., 1977). But it still leaves unanswered how the infant will assign which constellation, or which components of jointly performed constellmions, to self or to other. A reasonable way out of this thicket is to assume that from a very early age, infants, begin to perceive the intentions behind these complex and changing constellations of behavior. The many different behaviors in the many channels are organized into constellations by underlying goal-directed motivations. What adults and older children do is to "read through" the overt behaviors so as to identify the organizing intention or motive, just as we "hear past" phonemes to identify words and "hear past" words to capture the meaning of the phrase. Organizing the perception of the dyadic behavioral flow into units of intentions has the great advantages of efficiency, rapidity, and flexibility in reading what is happening. This is an important part of the argument for placing certain aspects of "folk psychology" and "acts of meaning" (Bruner, 1990) at the starting point of our inquires, rather than at the end point. It is also compatible with views put forward by Pierre.Mounoud concerning early grasping (see Mounoud, this volume).

424

DANIELN. STERN

The question then becomes: When do infants engage in a form of (partially learned) direct perception of intentions and motives? I would suggest along with many others (e.g., Trevarthen, 1987, 1993; Braten, 1988) that it begins very early. Most of the micro-regulations that govern an activity such as free play are conducted with largely innate behavioral constellations that have intra-specific meaning. Let us take for an example the configuration that Weinberg and Tronick (1994) have called "social engagement," and that I have previously called "readiness to interact" (Stem, 1977), which consists of a full orientation toward the other, looking them in the eyes, a facial expression that is open and mildly positive and/or interested; vocalizations that are neutral or positive, etc. (as seen in the above example of the interaction). This configuration can result from a family of general intentions that are intuited by the person to whom they are directed, such as: the performer has focused attention on them; is interested in them; is ready to initiate or invite a positively toned interaction with them; and/or will probably do so given the slightest sign of reciprocal readiness, or invitation, etc. What part of all that is innate? Certainly, many of the different elements, acting alone, have separate signal value, e.g., being looked at versus full-face and full-body orientation versus different facial expressions, etc. But it is the constellation itself, taken all together in a context, that reveals the intention. The organization of these separate elements into a behavioral constellation by the performer is learned upon an innate base. The same is likely true for the reception of such complex signals that reveal intentions. In any event, the person receiving such organized behavioral displays from an interacting parmer does not have to put the meaning of the display together piece by piece. It is read as a whole and is perceived, interpreted, and experienced as an intention on the other's part. It is an emergent property of experiencing the behavioral constellation. The micro-regulations that make up free play consist of many basic intentions each of which corresponds to a basic behavioral constellation. These are: 1) readiness to interact/engage; 2) invitation to interact; 3) intention to increase the level (joy and/or activation) of the interaction; 4) intention to decrease the level of the interaction; 5) intention to maintain/prolong the interaction; 6) disengagement/withdrawal (along a spectrum) from the interaction; 7) avoidance of and refusal to reengage; and 8) active protest to change the status quo. Socio-affective exchanges involving free play consist in large part of the dyadic chaining and overlapping of these basic interactive intentions. When the behavioral flow is perceived as a series of intentions, it becomes much easier to explain the essential subtleties, such as the ability to rapidly read what is happening in spite of constant shifts in the meaning of a behavior, depending on where the interaction has just been and where it seems to be heading.

SELF/OTHER DIFFERENTIATION 425

Once this point of view of the direct perception of intentions is adopted, the question of what belongs to the infant (self) and what belongs to the mother (other) becomes transformed. It is no longer so interesting to ask how the infant may know how a smile, or gesture, etc., belongs to him or mother or even whether it was initiated by him or her. (He readily keeps track of the provenance of such separate behaviors by virtue of his having formed a "perceived self" (Neisser, 1993) or a "sense of a core self' (Stem, 1985). What now becomes the interesting question is: How does he know which interpersonal intentions are his or another's? (If this can be figured out by the infant, the behavioral details will fall into place.) Before addressing that question, let us deal with the question of how an infant might "directly" perceive an intention. I have argued elsewhere (Stem, 1994, 1995) that infants may process units of interpersonal, motivated, goal-directed behavior (such as a micro-regulation) in a global fashion as a "proto-narrative envelope." The basic features of such units are as follows. Goal-directed motivated acts (e.g., enacted intentions to micro-regulate the interaction) occur in real time, in a "moment." And during that "moment" there is a shift in arousal, activation, affect, and level of motivation/goal achievement. These shifts, unfolding in time, describe a temporal contour, which I call a "temporal feeling shape." These subjective temporal feeling shapes are best captured by terms such as: crescendos, fading, explosions, growing, attenuations, etc. (Elsewhere I have called these temporal feeling shapes "vitality affects"; Stem, 1985.) These feeling contours provide the temporal architecture of the experience of an enacted intention. Stated differently, as the enacted intention moves in time toward its goal, it generates a subjective dramatic line of tension, a contour of excitement (the temporal feeling shape), which also happens to be an essential feature of a narrativelike structure (Labov, 1972). This goal-directed movement also favors the perception of other aspects of the experience in the light of this progression toward the end state. This results in the subjective organization of another main element of a narrativelike structure, namely a proto-plot consisting of a phase of complicating action, leading toward a crisis or high-point and followed by a phase of resolving action (Labov, 1972). (Or, with more development, what follows is the perception of a coherent progression of agent, action, instrumentality, goal, and context; Burke, 1945). In brief, goal-directed movement leaves in its immediate wake (so to speak) the tendency to experience enacted intentions in terms of dramatic lines of tension (temporal feeling shapes) and proto-plots. The differentiation and elaboration of the different elements of the plot is a developmental matter. The point here is that a global form of primitive plot is perceived early.

426

DANIEL N. STERN

It is the combination of a proto-plot unfolding along the time contour of the temporal feeling shape that I call a proto-narrative envelope. The main suggestion here is that an infant can perceive an intention in the form of a proto-narrative envelope, which captures the unit of an agency acting toward a goal. It is this that is directly perceived. It is directly perceived both as enacted by the self and by another. Temporal feeling shapes (vitality affects) are well suited to being intuited by another via imitation, contagion, or identification. These are essentially the same as the "forms of feeling" that the philosopher S. Langer speaks of in explaining the emotional impact of music (Langer, 1967). Now that we have suggested a way - - or a structure - - b y which the infant might perceive intentions directly and globally in both self and other, we must return to the question of: On what basis does he or she know whose intentions they are? It should not be so difficult for the infant to know which intentions are his or hers on the basis of several sources of information, such as the presence of a plan and the sense of "willing." These will remain good indicators in spite of confusions about agency or synchronicity or those due to imitation. What the infant cannot know, however, is that his or her intentions assume the form they do so as to fit against the complementary intentions of the parent. The infant's intentions are molded by the mother's intentions. This is the crucial point. To the extent that the caregiver's intentions define the environment in most socio-affective exchanges, the infant's intentions and motives are formed to adapt to that environment. And perceptually speaking, the environment is a shifting array of maternal intentions. By analogy to the domain of grasping: For the infant, the mother's intentions during affective exchanges are the counterpart of the object to be grasped, its size, shape, motion, and distance from the infant. And the infant's intentions are the counterpart of how his or her ann movement and hand position get shaped to conform to a specific object. What appears to belong to the infant objectively and subjectively also belongs to the dyad. The "other" is the silent partner. It is only when a different other is present that is becomes obvious that affective intentions have been created for a specific "other." This consideration brings us back to a point raised at the beginning of the chapter, namely, the importance of the particularities of the individual caregiver, because the infant will form his or her repertoire of socio-affective intentions and motives largely on the basis of the caregiver repertoire of complementary intentions and motives. Two examples of this process of the molding of an infant's socio-affective intentions may be helpful here. In attachment research, one of the patterns of

SELF/OTHER DIFFERENTIATION

427

insecure attachment is the avoidant pattern. During the reunion after a short maternal separation, the infant acts as if nothing has happened when the mother returns to the room. She may not even indicate that she noticed that mother returned. She does not run to her (as is usual in the secure pattern). She does not solicit attachment behaviors from her (to be picked up, held, etc.). Yet it is known from physiological and hormonal studies that the infant has registered her mother's return and is, in fact, stressed rather than nonplussed. (Grossman & Grossman, 1991). Recent thinking about this avoidant pattern suggests that the infant does not show positive attachment behaviors toward her mother upon reunion because she anticipates that a solicitation of attachment behaviors from mother is likely to make her either pull back, reject the infant, or get angry. In sum, the infant can maximize her proximity to mother and optimize her felt security by doing nothing and showing nothing. Her intentions and motives when insecure have been molded by the complementary intentions and motives of her mother. One sees the same kind of molding of intentions (and behavior) in free play. For instance, some infants with depressed mothers (who can be reanimated) frequently become great initiators, charmers, with "sparkle aplenty" while with their mother (at least for short stretches), in order to up-regulate the level of joy and activation. In such cases, the intention to up-regulate to an optimal level of joy and activation has been molded to include the intention to act as an antidepressor through charm and inventive guile. It has gone well beyond a simple communication. It may remain as an interactive coping style well into adulthood. Such occurrences of "intention molding" are not only commonplace, they are ubiquitous and constant in socio-affective interactions. They are, in part, what these interactions are about. The counterpart intentions of the caregiver are considered by the infant to be stable parts of the "environment." The result is that parts of the infant's self (intentions, motives, and desires in this domain) are molded against the intentions of an other who is subjectively taken for granted as how the world is. It is this subjective situation that has pushed many clinical theorists to invent concepts such as "self-objects" (Kohut, 1977) to account for those parts of self-experience that exist or are maintained only so long as they are (silently) regulated by the presence and actions of particular others. I have elsewhere used the terms "self-regulatory-other" (Stem, 1985) or "self-with-other" (Stem, 1995) to describe this assumed category of self-experience in infancy in which there is no confusion between the physical self and other, but where certain self-functions (e.g., a high level of joy or felt security) depend on the other's invariant presence and interaction for their existence. In the mutual regulation of joy, other affects, attachment, love, and many meaning systems and beliefs, this entity of a self-regulatory-other becomes a large part of the interactive experience.

428

DANIEL N. STERN

This, then, is the final and perhaps most far-reaching difference between self/other differentiation in the domain of inanimate objects and in the domain of socio-affective interactions with people. In the latter, a third category beyond "self" and "other" is needed, the "self-with-a-self-regulating-other." This third category creates many new problems and questions. Its phenomenological status is often unclear. This category of self-regulating-other can emerge into, or disappear from, awareness. It can also flow (subjectively) into the experience of self, or that of an other. And the acceptance, private or public, and the felt boundaries of this category are in large part a cultural matter. The relative place or importance given to this self-regulating-other entity is determined by the way we narrate our experience to one another. And at that point, the relative distribution of experience into self, other, and self-regulating-other becomes a narrative construct of the culture more than an objective matter. For instance, it is assumed that in ancient Greece self-regulating-others were a common part of the subjective landscape in the form of gods and goddesses who were taken into the person via the breath, making the person "inspired" to experience and enact a basic passion, such as love. In this situation, the core self is never violated, and the sense of agency remains intact within the self. It is only the authorship of the state and the acts that is assigned to another. In modem western cultures, we also have some vague, intermediary categories of experience that sit somewhere between the self and the other or the inanimate world, such as "I know it in my guts," or "my heart tells me .... " Given the current cognitive science view that many thoughts and feelings are emergent properties of extremely complicated and numerous, multiple, simultaneous processes, it becomes impossible to know exactly why or how an experience emerged, or why then. What becomes more important is the (re)constructive, interpretive process that, after the fact, assigns ownership and provenance to an experience. Such considerations move the inquiry from an examination of what information might specify self, other, or self-other to also include the interpretive aspects of such subjective acts of specification.

REFERENCES

Braten, S. (1988). Dialogic mind: The infant and adult in protoconversation. In M. Cavello (Ed.), Nature, cognition and system. Dordrecht: Klewer. Bruner, J. (1990) Acts of meaning. Cambridge, MA: Harvard University Press. Burke, K. (1945). Grammar of motives. New York: Prentice Hall.

SELF/OTHER DIFFERENTIATION 429 Grossman, K.E., & Grossman, K. (1991). Attachment quality as an organizer of emotional and behavioral responses in a longitudinal perspective. In C. M. Parkes, J. Stevenson, J. Hinde, & P. Morris (Eds. ), Attachment across the life cycle (pp. 93-114). London: TavistocMRoutledge. Labov, W. (1972). Language in the inner city. Philadelphia: University of Pennsylvania Press. Kohut, H. (1977). The restoration of the self. New York: International Universities Press. Langer, S.K. (1967). Mind: An essay on human feelings (Vol 1). Baltimore, MD: John Hopkins University Press. Neisser, U. (1993). (Ed.) The perceived self" Ecological and interpersonal sources of selfknowledge. Cambridge, MA: Cambridge University Press. Stern, D.N. (1974). The goal and structure of mother-infant play. Journal of American

Academy of Child Psychiatry, 13,402-421.

Stern, D.N., Beebe, B., Jaffe, J. ,& Bennett, S.L. (1977). The infants' stimulus world during social interaction. In H.R. Schaffer (Ed.), Studies in infant interactions. London: Academic Press. Stern, D.N. (1985). The interpersonal world of the infant: A view from psychoanalysis and developmental psychology. New York: Basic Books. Stern, D.N. (1994). One way to build a clinically relevant baby. Infant Mental Health Journal, 15(1), 9-25. Stern, D.N. (1995) The motherhood constellation: A unified view of parent-infant psychotherapy. New York: Basic Books. Trevarthen, C. (1979). Communication and cooperation and the making of an infant's meaning. In R. Steele & T. Threadgold (Eds.), Language topics: Essays in honor of Michael HaUiday. Philadelphia: John Benjamins. Trevarthen, C. (1993). The self born in intersubjectivity. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 121-173). Cambridge, MA: Cambridge University Press. Tronick. E.Z., Als, H., Adamson, L., Wise, S., & Brazelton, T.B. (1978). The infants response to entrapment between contradictory messages in face-to-face interaction. Journal of Child Psychiatry, 17, 1-13. Weinberg, K.M., & Tronick, E.Z. (1994). Beyond the face: An empirical study of infant affective configurations of facial, vocal, gestural, and regulatory behaviors. Child Development, 65, 1495-1507.

This Page Intentionally Left Blank

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

431

CHAPTER 21

Becoming A Self 1 EDWARD S. REED

Franklin & Marshal College

The Populated Environment of the Human Infant Becoming a self is something one cannot do all on one's own; it is an inherently social process. The loud arguments within psychology over "nature versus nurture" have tended to obscure the plain fact that no human society ever allows most of its children to grow up on their own. The human race's record when it comes to infanticide and abandonment is nothing to be proud of; nevertheless, even in the worst of circumstances, the majority of infants who are born are surrounded by an intensely active populated environment. Moreover, at least some of the people in this environment engage in a considerable amount of activity aimed at structuring the immediate surroundings of the infant. The subject of this chapter is how we become persons in this very special populated environment. As best we can tell at present, the active structuring of infant environments is a universal attribute of all human cultures. However, unlike the traditional concept of "nurture" m centered on person-to-person interaction and focused on the elements of caregiving - - the ecological analysis of the infant's environment casts a wider net. The special environment created for all the babies of a given culture includes selected objects, places, and events, as well as other people. It is also a developmentally structured environment, changing in time in at least rough concordance with the infants' developmental changes. This "developmental niche," as some have called it (Super & Harkness, 1986; Bril, 1993), also includes selective barriers that prevent children from encountering specific objects, places, and events deemed harmful or inappropriate by the caregiver. I find it useful to distinguish three dimensions of structure within this developmental niche:

432

EDWARD S. REED

1. Special persons. Caregivers for infants are always specially identified people within a culture, whether they are the biological parents, siblings, or individuals brought to the task from outside of the biological family. Who takes care of babies varies with culture, geography (e.g., urban versus rural child rearing), and class. There may be no universal patterns here at all except that, for a majority of children in any given culture, some older female individual(s) is the baby's regular caregiver. Male caregivers are not unknown, but everywhere they are in the minority. The caregiver is by no means the only individual who helps the child to learn about the world, but the caregiver(s) is the person whose influence on the baby is felt most intensively, by dint of common and repeated experience. Thus, the caregiver's interests and abilities, as well as her cultural background, play a major role in organizing the child's experience. 2. Special objects, places, and events for babies. Toys, games, nurseries, infant carriers, seats, and other postural apparatus; cribs, playpens, songs, and other such things are very widespread, to say the least. Given human cultural organization, it is a "natural consequence" of linking a baby with a caregiver that certain special places or items or events will also be marked off as the province of babies and caregivers. Some aspects of this segregation are positive (exposing the child to age-appropriate toys and games) and some are negative (preventing exposure to events deemed inappropriate). Cultures typically have extensive folk wisdom about what the proper things for children are, and when children are ready for them. For example, many rural, nonindustrialized cultures emphasize giving young babies a kind of "gymnastics" or exercise, something rarely done in urban, industrialized places (Reed & Bril, 1995). 3. Among the events targeted for babies are a special class of vocal songs and games, which are strikingly and distinctively human. These games begin at least by the time the infant can reliably look and smile at another person (after 6 weeks: see Trevarthen, 1988) and they develop greatly thereafter. All of the games involve what I have characterized as special elements of the "play" action system: exaggerated postures and movements. The exaggerations are in both the temporal and spatial domain: Movements may be made bigger or smaller than usual, often with a heavy rhythmic feel, including use of syncopation in the rhythm. Every element (and more) of adult face-to-face interaction is brought into these games: facial expressions, mouth gestures, and hand movements, including noise-producing ones. Rhythmic movement and sound production are often combined (claps, whistles, pops, slaps, tonguings, etc.). Although the specialized verbal element of this play (so-called "motherese") has been analyzed

BECOMINGA SELF

433

systematically, research on vocal-manual-gestural play is conspicuously absent. All of us thus become persons within some version of this specially structured social environment. It is not an inanimate environment we live in, but an animate and often very animated environment indeed. The richest and most elaborate affordances of the environment are provided by other animals and, for us, other people. These are, of course, detached objects with topologically closed surfaces, but they change the shape of their surfaces while yet retaining the same fundamental shape. They move from place to place...initiating their own movements, which is to say that their movements are animate...subject to the laws of mechanics and yet not subject to the laws of mechanics, for they are not governed by these laws. They are so different from ordinary objects that infants learn almost immediately to distinguish them from plants and nonliving things. When touched they touch back, when struck they strike back; in short, they interact with the observer and with one another. Behavior affords behavior, and the whole subject matter of psychology can be thought of as an elaboration of this fact (Gibson, 1979, p. 135). This complex, ever-changing, mostly (but not always) responsive populated environment is the world within which human babies fend for themselves, and within which they become selves. Let' s see how they do it.

Developing Action Systems Within Interactive Frames What does the newborn baby know about all this "show" being put on for her? The mere existence of the complexly structured social environment would not matter if the newborn were incapable of perceiving anything in that show. What is striking about human neonates is, of course, their apparent incompetence and dependence upon caregivers. Yet despite their truly helpless position as animals in the wider environment, our babies have certain subtle competencies that make them capable of adapting to the human environment. Foremost among neonatal skills are skills of perception and interaction. It is only a slight oversimplification to say that the perceptual and interaction systems completely dominate development in the first half year of life. The traditional Western view of the infant as first learning about things and then coming to understand people is almost completely backward. In general, infants learn about things and what one does with them through other people, or at least within socially structured settings. No doubt early infancy is a time of confusion, but it is not entirely buzzing and blooming; and, more importantly, even the youngest of infants have specific perceptual and interaction skills that make it possible for them to hunt effectively

434

EDWARD S. REED

for meaning in their world. In addition to being fairly well adapted for suckling, the neonate's mouth is a rich source of perceptual information, often becoming involved in cyclic patterns of activity that help to establish contact-based intimacy with a caregiver. Alert and healthy babies are capable of orienting to a caregiver's voice, even when she is out of sight (Alegria & Noirot, 1978). These alert babies are also capable of seeing and imitating distinctive facial expressions (e.g., tongue protrusion and wide-O mouths) a finding that differed so drastically from traditional views that it was hotly disputed until it had been replicated a number of times (Meltzoff & Moore, 1977; Meltzoff, 1993). Mainstream theories of perception, which hold that perceiving emerges from the processing of sensory inputs, not from the detection of information, imply that crossmodal matching is a difficult process that must be learned. Human babies do not seem to know this: They look for a voice in the spot where they hear it, and they expect (heard) voices and voice affect to match (seen) faces and facial expressions (Walker & Gibson, 1984). They expect an object to look like what it feels like, and to feel the same thing in both hand and mouth (Rochat, 1991). Although the implications of these findings are still resisted by many, I think the burden of proof has shifted to those who deny these competencies to young infants. Along with Stern (1985), Trevarthen (1984, 1994), and Fogel (1993), I take the position that human infants have a congenital capacity to learn about the people who take care of them and a concomitantly strong motivation to do so. Perception in the populated environment is a step ahead of perception in the inanimate environment. We do not learn about objects and then discover that some objects are animate. Human infants do not, for example, need to learn about things like voice-face concordance, and from such perceptions of "object properties" draw "inferences" about the nature of animate objects or people. Experiments seem to show that infants know less about inanimate objects than they do about human beings. A newborn can only reach in the direction of an object (Hofsten, 1982), but he or she can actually imitate at least a few facial expressions. By 6 weeks of age, a baby can even imitate a facial expression she saw 24 hours earlier, but her reaching and grasping skills are still highly undifferentiated (Meltzoff & Moore, 1994). One reason why the human neonate seems to be so evolutionarily "specialized" for face-to-face interaction and little else is because she is so helpless when it comes to postural control. Mastery of independent neck and head control take most babies at least a month or two, and mere rolling over usually takes a couple of months more. Human infants have to be held, swaddled, and/or supported for approximately half a y e a r - or else be left to lie all alone. In most cultures except our own "modem, developed" society, this last option tends to be avoided (Rogoff et al., 1994), and infants fred themselves facing caregivers or are

BECOMING A SELF

435

held by caregivers so as to face other people. Because of their relatively advanced perceptual and interactive skills, these face-to-face situations become a basic scene of early development. Fogel (1993) calls these "interaction frames. ''2 The first of these to emerge is reliable face-to-face interaction. An interaction frame can be thought of as a "site" in the populated environment, a site with, as we shall see, unique and intricate behavioral patterning. These patternings exist not within individuals, but across two individuals (it is only later that other objects or other individuals can be brought into the frames). There has been controversy in the literature as to the causal origins of these interactive patterns: Does the child or the adult do more to establish them (Kaye, 1982; Trevarthen, 1994; Fogel, 1993)? For myself, this is a false issue, especially given the inherent variability of these sites. What is important, and which has not been studied, is how these frames act as windows within which the populated environment entrains the perceiving, acting, and knowing capacities of the infant. What I have called the "field of promoted action" (Reed, 1991, 1993; Reed & Bril, 1995) can only operate when such frames are available. The field of promoted action includes all the affordances made available to or emphasized for the child and excludes those forbidden to it. The field of promoted action includes those different actions that are encouraged or even scaffolded for the infant at different times (e.g., aiding to help sit or stand). The field of promoted action is a powerful force in human development, but it cannot shape the infant except through the "windows" of these interactive frmnes.

The Differentiation of Frames The Primal Frame: 0-3 Months Although human neonates have more social skills than previously thought, they cannot reliably enter into and maintain face-to-face interaction. It takes at least 6 to 8 weeks to do so and may take as long as 12 weeks. Neonatal skills serve to ensure that entrance into the face-to-face frame is possible. Although fleeting in form and haphazard in appearance, newborns do emit many recognizable facial expressions (their relationship to affective states is, however, not yet strongly established) and can, even if fitfully, gaze at another's face. According to Trevarthen (1994), sometime in the first 6 weeks these abilities begin to be organized more systematically. The infant who looks at a face quiets his breathing, focuses his vision on the face, and begins to make facial, limb, or bodily gestures. It is when the infant thus combines his simple perceptual and interactive skills that entry into a primary mode of interaction becomes possible. As both Ulric Neisser and Eleanor Gibson argue in this volume, the early self is

436

EDWARD S. REED

characterized by these first steps at coordinated agency. What I am emphasizing is that these steps are taken in a populated environment, and that they lead not only to agency (Gibson) or to an ecological self (Neisser), but to the co-construction of this basic face-to-face frame of interaction. Once the infant is not only observing the caregiver but is also modulating her own action in detectable ways, an interactive loop can be established. "What the other animal affords the observer is not only behavior, but social interaction. As one moves, so does the other, the one sequence of action being suited to the other in a kind of behavioral loop. All social interaction is of this sort..." (Gibson, 1979, p. 42). It is crucial to note that establishment of this loop is as much dependent upon the child as upon the caregiver; it is transpersonal, not subjective. No amount of "reading in" sociality or intention to the activities of a truly randomly behaving baby by a caregiver would produce any consistently patterned face-toface interaction. It may be the case that some human adults "project" meanings that are not there onto their baby's actions (Kaye, 1982). However, if the child's activities truly lacked systematicity, then such projection would necessarily remain wishful thinking and would become as unreliable as that mode of thought always is. It is only to the extent that the infant engages in reliably patterned action that any kind of interaction loop can be established. Once such an interactive loop is established, it takes on extreme affective importance for most caregiver-infant dyads. It is one of the great joys of life to make a baby smile or coo, and for most babies the reverse is true (Stern, 1985). (There are important exceptions to this rule, as with severely depressed caregivers, but I will not address those issues here.) Stern (1985) calls these linkages "affective surges" and points out that they help to introduce both cvclicity and internal phasing into the interaction. Cyclicity is created as either the caregiver or the baby or both seek to reestablish the loop. Internal phasing is created as the pattern of introduction-rise to a peak-denouement-disengage begins to be established. Again, note that the developmental structuralization of the face-toface frame is ecological and transpersonal. The multimodal nature of infant knowledge and action stands them in good stead in establishing this cyclicity and phasing. Infants attend not only to visual and tactile aspects of their interaction, but they show special sensitivity to vocal ones. Kugiumutzakis (1992) has shown that both infants and their caregivers are more likely to respond to vocalizations than, say, to sneezes. Although the evidence is by no means conclusive, there is reason to believe that vocalizations play a major role in the establishment of reciprocity, the crucial element of turntaking in face-to-face interaction. This reciprocity may be maintained by increasingly predictable patterns of phasing and cycling. Trevarthen (1994) goes so far as to claim that this predictability has its basis in the rhythmic structure of

BECOMING A SELF

437

gesture and vocalization: with beats at 1.5/sec., units of 2-3 beats in length, and maximum duration of approximately 20-30 units. Papousek and Papousek (1991) report that by the time babies are 2 months old, speech directed at them tends to be differentiated by function, with different and distinctive prosodic contours, depending upon the nature of the situation. Although this early face-to-face frame is s011 episodic and not sustainable, a great deal of knowledge can be acquired within these 2-minute (or shorter) windows. Here again, I would argue that the knowledge thus gained is predominantly "interactive." The infants begin to understand their own agency and that of others. They begin to form expectations about external events (especially animate ones) and begin to learn to control their own agency (see E. Gibson, this volume). They have moved from merely tracking another's face and occasionally imitating it to the beginnings of genuine interaction. For example, Meltzoff and Moore (1994) report that by 3 months, infants: had developed social expectations about people, their games, and expressive behaviors that were not present in the newborn and 6-week-old...[the older infants] tried out routinized social behavior (smiling, greeting, cooing), which superseded strict imitation (p. 94). By 3 months, infants know enough about (some) other people to act prospectively within a face-to-face frame. They can thus begin to maintain interaction for longer and longer stretches. They have entered the field of promoted action organized by their culture through the window of the caregivers voice, face, and gesture. After approximately 3 months, human infants thus have the skills in place to enter their culture's field of promoted action on an episodic basis. The next phase of interaction involves the effects of the child's learning to control this entry into a face-to-face frame. Hence, there are really two frames that emerge at about 3 months, one the result of a more controlled form of face-to-face interaction, and the other the result of a more controlled exclusion of interaction, which allows the child to act on his or her own. Before approximately 9 months, infants are not capable of reliably coordinating both action on objects and interaction, so these two frames tend to be relatively exclusive. I call the object-centered frame the "performatory" frame, and shall discuss it first here. The other-centered frame might be called "true interaction," as it is in these months that the child extends outward from face-to-face to other sorts of interactions.

The Performatory Frame: 3-9 Months Infants of about 3--6 months begin to show signs of being able to reject offered bouts of interaction. They avert their gaze, shut their eyes, stare past,

438

EDWARDS. REED

and/or get glassy eyed on occasions when caregivers try to gather their attention (Stem, 1985, p. 21). This they do, not out of sleepiness or crankiness, but out of other interests - - interests in objects, as their behavior reveals. The reciprocity needed for face-to-face interaction is broken and is often replaced by interest in other things. It is important to note that object-oriented action at this age is not devoid of affective involvement or phasic structure. There is a tendency to assume that affect tracks interaction only, but that is patently false. The 3- to 4-month-old child seems to have at least some perception of herself as an agent, and this proprioception can be as rewarding as perceiving others. The conjugate reinforcement paradigm used by Carolyn Rovee-Collier (Rovee-Collier & Gekoski, 1979) is a good example of this. In this paradigm, a string is tied from a supine baby's ankle to a visible and attractive mobile. When the baby kicks, the mobile will turn, and harder kicking yields more turning. Four-month-olds learn this skill very fast and appear highly interested in it. Sullivan and Lewis (1989) used a similar setup and looked at the infant's emotional expressions. Half of their infants could control the display and half could not. Only the infants who could control the events showed "affective surges," with excitement, enjoyment, high interest, and vocalization. In contrast, the infants who could not control events by kicking were very fussy. In this object-centered frame of action, the caregiver takes a peripheral role. She may serve to bring objects into the field of view ( and thereby serve to phase actions) and even to demonstrate the functions of objects, but the child's attention and interest, at least in these cases, is decidedly oriented toward inanimate objects. This increased interest in objects is accompanied by some important developments in the perceptual systems. Most importantly, reaching to grab and mouth or to grab and feel is now possible (Hofsten, 1979; Rochat, 1989). It is around 3-4 months that reaching skill begins to be reliable, and that reaching for objects is combined with feeling their properties (whether by use of the mouth or the hand or both). Note that virtually all 4-month-olds are still unable to sit unaided, so the only objects they will reach for are those promoted for them by their caregiver. Thus, although I refer to this frame as object oriented, one should not forget that it is, in fact, an interactive frame. Either the infant is actively rejecting the possibility of interaction with others in favor of acting on objects, or is acting on objects that are in some way promoted to her attention (e.g., mobiles, toys, pacifiers, etc.). Palmer (1989) and others (Rochat & Senders, 1991) have shown that as young as 6 months, infants will selectively explore the affordances of objects they are grasping. Contrary to Piagetian (1952) theory and popular opinion, infants of this age do not just mechanically apply available action schemes to any and all objects

BECOMING A SELF

439

(i.e., bring everything to their mouths). Instead, they show an appreciation of and interest in differences of surface, texture, substance, and shape. Hard objects are banged on other hard surfaces, malleable objects are palpated, fuzzy objects are stroked (and not mouthed) and so on. At this time or a little later, expectancies about objects (as opposed to expectancies about people) begin to appear. Nachman (cited in Stem, 1985, p. 93) presented 6- to 7-month-olds with a puppet show. When these babies were brought back a week later and showed the (unmoving) puppets, they smiled more than babies who had not seen the show. Having already learned to expect patterns of agency in caregivers and now having learned to expect certain objects to have special ("interesting") properties, babies are at an important juncture in their social lives. In addition to being recipients of promotions, they can begin to promote the actions of others in specific ways. It is one thing to cry or fuss and expect to be comforted (a general or nonspecific interaction); it is something else to single out a specific object or event and manage to obtain it or make it happen through the agency of another person. Mosier and Rogoff (1994) have shown that as early as 6 months, children can arrange for their mothers to get them a specific interesting object. The reverse is true as well: 6-month-olds can observe the use of specific affordances by their caregivers and come to model their own behavior on those uses (Lockman & McHale, 1989; Hofsten & Siddiqui, 1993). Starting at approximately half a year, human infants are thus creating what I have called "fields of free action" (Reed, 1993), which define their own autonomous agency, independently of what is promoted for them. (Let us remember that this is still a relative autonomy within a populated environment: These children for the most part cannot even locomote). They are learning about affordances that interest them and have learned both how to get caregivers to help them detect and use those affordances and how to "shut out" the caregivers attempts at interaction, in order to play with objects. The infant is beginning to take an organizing role in the patterning of her daily life. Episodes of interaction are interspersed with episodes of exploration and performance, both of which are accomplished to some degree autonomously. The process of "choreographing" these phases depends on the development of a richer set of interaction skills, to which I now turn.

True Interaction: 3-9 Months Not only do 3- to 6-month-old infants learn how to "shut out" their caregivers, they also learn how to "turn the caregivers on." Infants begin to show signs of reliably reinitiating face-to-face interaction by gazingat another's face, directing smiles, or vocalizing (Stem, 1985, p. 22). Here, any infant will be greatly helped by high responsivity in the caregiver (not always found) and also will necessarily

440

EDWARDS. REED

come to shape his behavior according to local cultural proprieties, for these are what will define how effective his gazes, smiles, and vocalizations are. Thus, not only can infants turn off the reciprocity of early face-to-face interaction, they can also extend it, opening the "window" through which the field of promoted action can be made richer and more effective. Beginning at about 3 months, infants also show signs of regulating their behavior in relation to changes in interactive context. Experimenters who have asked mothers to stop in the middle of interacting with a child and maintain a still face (or who have arranged for this appearance of a still face to be presented to an infant by other means) have found that this "sobers" the babies. Although the infants in such still-face situations do not usually fuss or cry, they decrease their smiling, look away, and show signs of arousal (Field & Fogel, 1982; Gusella, Muir, & Tronick, 1988). Toda and Fogel (1993) found that manual activities were likely to increase in this still-face situation, especially in 6-month-olds, who showed more signs of purposive manipulative activity under these circumstances than did younger infants. Could this be evidence that the child is coming to accommodate the intrinsically on-again, off-again nature of social interaction? Perhaps these infants are increasingly able to turn to the field of free action when access to the field of promoted action is barred. The major development in interaction after 3 months comes from a significant change in caregiver activity. This is the introduction of motherese, or infantdirected speech. Such infant-directed speech begins to be produced reliably by caregivers once the baby has begun to master basic face-to-face interaction (Snow, 1977; Femald, 1991). Even newborns can discriminate infant-directed speech from normal, adult-directed speech, and there is evidence that very young children prefer this mode of speech to others (Cooper & Aslin, 1990). Infant-directed speech is nearly universal in our species and has a consistent crosslinguistic structure. A gentle voice is used with relatively high pitch; utterances are short and are typically repeated. Pitch contours are exaggerated in form (Femald, 1991). It is only after the basic face-to-face frame is mastered that reliable turntaking is observed, in which the caregiver produces bursts of infant-directed speech, interspersed with vocalizations from the child (Kaye & Fogel, 1980). Interestingly, the child's response to infant-directed speech is to increase her "vocal play" (Stark, 1980). In this mode of vocalizing, the infant explores many dimensions of her speech capacity: changes in volume, pitch, contour, duration, and so on. Thus, a kind of protodialogue begins to emerge within and alongside of face-to-face interaction. Again, this evolving interactive frame includes reciprocity and even turn-taking, with actions that have complex internal structure. It is within this interactive frame that babbling emerges, as the child begins to learn how to produce recognizable speech sounds, not just mere vocalizations (Oller,

BECOMING A SELF

441

1980; Studdert-Kennedy, 1987, 1991), perhaps in part because of the caregiver's promotion of these sounds among all the many diverse sounds produced by the infant (Locke, 1993). There is now evidence that by 9 or 10 months, the babbling sounds of infants can be discriminated by ordinary adults as arising within different linguistic environments (Boysson-Bardies, 1990). Thus, the face-to-face frame expands to include vocal interactions. The combined competence at interaction and vocalization creates the conditions for the emergence of a whole suite of important activities we colloquially speak of as games: clapping, rhythm and rhyme games, vocalizing with gestural changes, and more. In all these games, adult vocalizing as well as facial expression seem to play a large role. Gustafson, Green, and West (1979) studied the development of games between 6 months and 1 year. The games played at 6 months did not require extensive action on the infant's part; for example, peek-a-boo, and "I'm gonna get you!" (a game in which the child is the recipient of tickling or other tactile stimulation while the adult regulates the timing and intensity of those events). Nevertheless, even in these early games the child must be able to play some role. For example, one game is just for the caregiver to coax a vocalization from the child, and another is to have the infant take a ball or toy at an appropriate time. By 8 months, games like pat-a-cake emerge, in which the infant plays a much greater structuring role. By 1 year, games in which the infant plays a major role constitute at least half of all games: for example, give-and-take, build a tower, point and name (child does the naming). In the period from 3 to 9 months, the human infant is becoming a complete interactor, one who combines vocalization and bodily movements with face-toface interaction. He is beginning to make his own choice as to whether or not to interact. He is starting to master the intricate art of turn-taking, the first of the complex of reciprocities that are necessary for successful social interaction. He has become a game player, who not only undergoes "affective surges," but does so in a shared context, linking the phases of his actions with those of his caregiver. Throughout these developments, there is reason to believe that the infant is perceiving himself as an increasingly competent social a g e n t - as a self who can both join interaction and engage in action on his own. Finally, and perhaps most importantly, all of these developments follow from the infant's increasing ability to regulate the stream of action. A 9-month-old has an effective field of free activity as well as an effective "window" for receiving the field of promoted action. She can resist influence and perhaps even "set her own agenda." And she can make known her disagreement with the course of events. She is an agent and she is beginning to be s o c i a l i z e d - she is at the threshold of becoming a real self; or, as I prefer to say, a person. To really be a person she has

442

EDWARDS. REED

to become a social agent: to be able to deal with both the world and with other people simultaneously, within a single frame of interaction.

Triadic Interaction One of the major developments at around this age of 9 months is the child's acquisition of independent locomotion skill. The independently crawling baby has both a wider field of free action and a more intense field of promoted action (as caretakers deploy their skills in structuring environments and events to make sure the infant does not hurt herself or get into things forbidden to her.) Independent locomotion underlies what I can only call an ontological shift in the status of the infant. The nonlocomoting child can only see the world from one perspective m unless she is assisted by a caregiver. alternative perspectives or sequences of vistas.

She is unable to explore

She is limited in her ability to

share the environment. The environment surrounding a nonmobile observer is different from that surrounding a mobile observer, as Gibson (1979) noted: the term surroundings is...vague, and this vagueness has encouraged confusion of thought. One such is the question of how the surroundings of a single animal can also be the surroundings of all animals. If it is assumed that no two observers can be at the same place at the same time, then no two observers ever have the same surroundings (p. 43). Now, this assumption is in fact true for nonmobile observers, but false for mobile observers, as Gibson (1979) realized in a passage of such importance it needs to be quoted in full: This seems to be a philosophical puzzle, but it is a false puzzle. Let us resolve it. One may consider the layout of surrounding surfaces with reference to a stationary point of observation...as if the environment were a set of frozen concentric spheres. Or one may consider the layout...with reference to a moving point of observation along a path that any individual can travel. This is much the more useful way of considering the surroundings ....Although it is true that no two individuals can be at the same place at the same time, any individual can stand in all places, and all individuals can stand in the same place at different times .... In this sense the environment surrounds all observers in the same way that it surrounds a single observer (p. 43). The interpenetration of "my" environment with the environment of any and all other people is, in this analysis, manifested as a function of locomotion.

In

particular, what Gibson has in mind is not merely moving to obtain something in the environment, but moving as an exploratory process.

To share parts of the

environment with others and to discover what is persistent and what is changing in the environment, Gibson hypothesized, requires ambulatory perception.

BECOMING A SELF

443

Gibson's hypothesis receives some corroboration from a clever experiment run by Gwen Gustafson (1984). Gustafson took a group of children aged 6.5 to 10 months and divided them into those who had already achieved some mastery of locomotion and those who had not yet discovered mobility. In addition, she made sure that all her infants had had experience in a "walker," a device that allows a nonmobile baby to become "instantly mobile." The experiment consisted simply in measuring changes in infant behavior brought on by being in a walker. However, Gustafson made sure to measure exploratory activity and interaction as well as locomotor activity. Children who were not independently mobile showed a remarkable reorganization of their exploratory action and their interaction in the walker: They changed their looking patterns (scanning much more of the room) and increased their social behaviors, such as gestures, smiles, and vocalizations. She could measure no such change in the infants who were independently mobile. (Perhaps they had already reorganized their exploratory activities and interactions?) If nothing else, the mobile child can "get into" more objects, places, and events than the nonmobile one. This alone probably serves to elicit an altered field of promoted action. For example, West and Rheingold (1978) found that more than half of mothers' utterances to 1-year-olds were comments on or references to the child's exploration. Mobility creates far more opportunity for conflict and cooperation over aspects of the environment and is very likely a major force in helping the child integrate social interaction with object-directed activity. Infancy researchers have come to refer to the 9-month-old's emerging ability to focus on an object or event with a caregiver as "triadic interaction" (two people plus one object). But this conjures up a static image, and one might do better to refer to "dynamic interaction" (two mobile people within a complex environment). Whatever one calls it, the important point is that this new social frame is dynamic and involves the interpenetration of the environment of any one observer with the environment of other observers (see Tomasello, this volume). Thus, the reciprocity characteristic of interaction is now extended even further, and becomes a sharing of the affordances of the environment. Vocalizations and gestures (from either parmer) are used to indicate objects, events, or places, not merely to mark phases of one's own action. From about 6 to 9 months, infants devote much time to playing with objects "but without the instigation of a partner: infants do not routinely share their involvement with objects with a social partner until nearly the end of their first year" (Adamson & Bakeman, 1991, p. 22). After 9 months, when caregivers and infants are playing a game and the caregiver "misses her turn," the baby will quickly respond with a gesture or vocalization (Ross & Lollis, 1987). Thus, the shared environment is acknowledged ~ and the infant, as well as the caregiver, takes steps to regulate

444

EDWARD S. REED

the interaction according to a desired pattern. The infant's field of free action begins to include the ability to promote specific patterns of activity in others. She begins to see herself not just as an agent, but as a social agent, who can affect others as well as alter her relations to objects. Caregivers appear to be sensitive to this change in their children, and shape their actions and vocalizations accordingly. I have already referred to Gustafson and colleagues' study of changes in the kinds of games played between 6 months and 12 months (Gustafson et al., 1979). This research showed that as babies move into this dynamic triadic interaction frame, they increasingly set the pattern of shared games and activities. The infant-directed speech of adults also tracks these changes: The ratio of informative to noninformative (mere commenting) speech undergoes a complete reversal between 3 and 9 months. Only 38% of infantdirected speech at 3 months is informative, but 61% is informative at 9 months (Boerse & Elias, 1994). Not only do caregivers change their infant-directed speech at this time, but they now begin to add to it a variety of actions that serve to gather and direct the child's attention: Objects are held in front of a child and are made to emit noises (either by manipulating them or by "supplying" the noises); objects are loomed up at and/or away from the child's face; patterns of repetition with crescendos and decrescendos increase in vocalizations; and a variety of rhythmic devices are used as well (Zukow & Duncan, 1994).

Conclusion: Becoming a Self The child who has begun to master a dynamic triadic interaction frame has begun to be a self, in the complete sense of the word. This young person has a set of skills and interests, which, however small and underdeveloped, are his own field of free action. And, importantly, among his skills is the ability to share objects, places, and events in his surroundings with others, to jointly attend to the surroundings. Indeed, he can not only share, but has begun to learn how to indicate what it is he is sharing, by gestures and vocalizations, and by emphasizing specific objects, places, or events in the stream of action, so as to promote the attention to an action of his social partners. As I show elsewhere (Reed, 1995, in press), these early interaction frames are part of the child's pathway to language and thought. In addition, the interpersonal abilities that are nurtured in these interaction frames involve powers of selection and choice that begin to define personality, disposition, and interests. Being mobile, the infant can often choose what object he wants, or find a desired place, or engage in a selected event or activity. In addition, he can comprehend (at least up to a point) what the choices

BECOMING A SELF

445

and selections of his caregiver mean, and he can even take steps to accommodate, reshape, or even thwart his caregiver's promotions. Having achieved this much, the child is a real self, and a small, but significant member of his culture. The idea that in development one first perceives one's agency and only then begins to learn about both one's self and other selves is a classic British empiricist thesis (see Thomas Brown, 1820, and Alexander Bain, 1855, who both go so far as to say all cognition of other objects - - e v e n inanimate ones m is based on one's internal awareness of agency). The pragmatist, George Herbert Mead (1934) inverted this theory, arguing that children first perceive other people as agents or selves, and only then come to realize that they are similar to those others and, therefore, are agents and selves. Neisser (1988), in his more recent theory of the self, seems to agree a bit with both of these earlier theories. Certainly, Neisser makes a strong distinction between a perceived or ecological self and a social or interpersonal self (see Neisser, this volume). The theory presented here goes against all these previous accounts. The human neonate's opportunity to experience agency independently of the populated environment is extremely limited. (Indeed, it is almost exclusively limited to being aware of agency in the service of trying to perceive the world.) But infants do not simply "discover" or "learn" that the others around them are agents. On the contrary, the entry of infants into the face-to-face frame is an intricate and precise developmental adaptation within an interactive context. It is within the face-toface frame that infants learn about both their own and others' agency; and, in large part, this learning is a function of the field of promoted action. The human self does not first emerge as a function of ecological proprioception during action on the environment (as Neisser [1988] seems to imply). Instead, the self begins to emerge as a function of proprioception within interactive frames. And as these frames differentiate, one would expect the self to differentiate as well. Thus, I would argue that, for example, what most of us think of as our autonomous or private selves emerges within a very special social context, the field of free agency. For a child to have the social skills needed to create a field of free action (to reliably reject social advances for interaction temporarily, and successfully reconnect later) itself requires at least a minimally differentiated and socialized self. The study of the development of the self, then, must not only be ecological, but it must be ecological in the sense of taking into account the dynamics of the populated environment. This will be a challenge for developmental psychology in the coming years.

446

EDWARDS. REED

NOTES

1. A different version of this chapter will appear in Edward S. Reed, Encounwring the World." Toward an Ecological Psychology (New York: Oxford University Press) under the title "Becoming a Person." 2. The phase interaction frame comes from Gregory Bateson (1972). However, Bateson's idea is framed'in terms of communication and intention. In particular, he believed that such things as face-to-face interaction constitute a form of metacommunication, in which the rules (or framework) of communication are negotiated. Perhaps this is true of adults within a given culturemwhich is what Bateson was studyingmbut this cannot be assumed for faceto-face interactions with babies. What is important for my argument is not the metacommunication, but merely the fact that establishing reliable rhythms and phases of interaction is necessary for face-to-face (or any other kind of) interaction to be a site within which development occurs.

REFERENCES

Adamson, L. B., & Bakeman, R. (1991). The development of shared attention during infancy. Annals of Child Development, 8, 1-41. Alegria, J., & Noirot, E. (1978) Neonate orientation behavior towards the human voice.

Infant Behavior and Development, 1,291-312.

Bain, A. (1855) The emotions and the will. London: Longmans. Bateson, G. (1955) The message 'this is play.' In B. Schaffer (Ed.), Group processes. New York: The Macy Foundation. Boerse, J., & Elias, G. (1994). Changes in the content and timing of mothers' talk to infants. British Journal of Developmental Psychology, 12, 131-145. Boysson-Bardies, D. (1990) Some reflections on sensory-motor organization of speech during the first year of fife. In H. Bloch & B. Bertenthal (Eds.), Sensory-motor organization and development in infancy and early childhood (pp. 457-466). Boston: Kluwer. Bril, B. (1993) Une approche dcologique de l'acquisition d'habiletds motrices. Habilitation, Universite Rene Descartes - Paris V. Brown, Thomas (1820) Philosophy ofthe human mind. Edinburgh: Black. Cooper, R. P., & Aslin, R. N. (1990). Preference for infant-directed speech in the first month after birth. Child Development, 61, 584-1595. Fernald, A. (1992) Human maternal vocalizations to infants as biologically relevant signals: An evolutionary perspective. In J. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind." Evolutionary psychology and the generation of culture. New York: Oxford University Press. Field, T., & Fogel, A., (Eds.) (1982). Emotion and early interaction. Hillsdale, NJ: Erlbaum. Fogel, A. (1993). Developing through relationships. Chicago: University of Chicago Press. Gibson, J. J. (1979). The ecological approach to visual perception. Boston: Houghton Mifflin. Gusella, J., Muir, D., & Tronick, E. (1988), The effect of manipulating maternal behavior during an interaction on 3- and 6-month-old affect and attention. Child Development, 59, 1111-1124. Gustafson, G. (1984). Effects of the ability to locomote on infants' social and exploratory behaviors: An experimental study. Developmental Psychology, 20, 397-405.

BECOMING A SELF

447

Gustafson, G., Green, J., & West, M. (1979). The infant's changing role in mother-infant games: The growth of social skills. Infant Behavior and Development, 2, 301-308. Hofsten, C. von (1983). Developmental changes in the origins of pre-reaching movements. Developmental Psychology, 290, 378-388. Hofsten, C. von, & Siddiqui, A. (1993). Using the mother's actions as a reference for object exploration in 6- and 12-month-old infants. British Journal of Developmental Psychology, 11, 61-74. Kaye, K., & Fogel, A. (1980) The temporal structure of face-to-face communication between mothers and infants. Developmental Psychology, 16, 454-464. Kaye, K. (1982) The mental and social life of babies. Chicago: University of Chicago Press. Kugiumutzakis, G. (1992). Intersubjective vocal imitation in early mother-infant interaction. In J. Nadel & L. Camaioni (Eds.), New perspectives in early communicative development. London: Routledge. Locke, J. (1993). The child's path to spoken language. Cambridge, MA: Harvard University Press. Lockman, J. J., & McHale, J. P. (1989). Object manipulation in infancy: Developmental and contextual determinants. In J. J. Lockman &. N. Hazen (Eds.), Action in social context: Perspectives on early development. New York: Plenum. Meltzoff, A., & Moore, J. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 75-78. Meltzoff, A.,& Moore, J. (1994). Imitation, memory, and the representation of persons. Infant Behavior and Development, 17, 83-99. Meltzoff, A. (1993). The centrality of motor coordination and proprioception in social and cognitive development: From shared actions to shared minds. In G. Savelbergh (Ed.), The development of coordination in infancy. Amsterdam: North Holland. Mosier, C., & Rogoff, B. (1994). Infants' instrumental use of their mothers to achieve their goals. Child Development, 65, 70-79. Oiler, D. (1980). The emergence of the sounds of speech in infancy. In G. YeniKomshian, J. Kavanagh, & C. Ferguson (Eds.), Child phonology, Vol. 1. New York: Academic. Papousek, M., & Papousek, H. (1991). The meanings of melodies in motherese in tone and stress languages. Infant Behavior and Development, 14, 415-440. Piaget, J. (1952) The origin of intelligence in children. New York: Norton. Reed, E. S. (1991). Cognition as the cooperative appropriation of affordances. Ecological Psychology, 3, 135-58. Reed, E. S. (1993). The intention to use a specific affordance: A conceptual framework for psychology. In R. H. Wozniak & K. Fischer (Eds.), Development in context." Acting and thinking in specific environments. Hillsdale, NJ: Erlbaum. Reed, E. S. (1995) The ecological approach to language development. Language and Communication, 15, 1-29. Reed, E. S. (in press) Encountering the world: Toward an ecological psychology. New York: Oxford University Press. Reed, E. S., & Bril, B. (in press) The primacy of action in development. In M. Latash & M. Turvey (Eds.), N. Bernstein's dexterity and its development. Hillsdale, NJ: Erlbaum. Rochat, P. (1989). Object manipulation and exploration in 2- to 5-month-old infants. Developmental Psychology, 25, 871-884. Rochat, P., & Senders, S. (1991) Active touch in infancy. In M. J. Weiss & P. Zelazzo (Eds.), Biological constraints and the influence of experience. Norwood, NJ: Ablex.

448

EDWARDS. REED

Rovee-Collier, C., & Gekoski, M. J. (1979). The economics of infancy: A review of conjugate reinforcement. In H. Reeses & L. Lipsitt (Eds.), Advances inchild development and behavior (Vol. 13, pp. 195-255). New York: Academic. Snow, C., (1977). The development of conversation between mothers and babies. Journal. of Child Language, 4, 1-22. Stark, R. (1980). Stages of speech development in the first year of life. In G. YeniKomshian, J. Kavanagh, & C. Ferguson (Eds.), ChildphonoIogy, Vol. 1. New York: Academic. Stern, D. (1985) The interpersonal world ofthe infant. Cambridge, MA: Harvard University Press. Studdert-Kennedy, M. (1991). Language development from an evolutionary perspective. In Biological and behavioral determinants of language acquisition. Hillsdale, NJ: Erlbaum. Sullivan, M., & Lewis, M. (1989). Emotion and cognition in infancy: Facial expressions during contingency learning. International Journal of Behavioral Development, 12, 221-237. Super, C. M., & Harkness, S. (1986). The developmental niche: A conceptualization of the interface of child and culture. International Journal of Behavioral Development,9, 545-569. Toda, S., & Fogel, A. (1993) Infant response to the still-face situation at 3 and 6 months. Developmental Psychology, 29, 532-538. Trevarthen, C. (1984) How control of movement develops. In H. T. A. Whiting (Ed.), Human motor actions: Bernstein reassessed (pp. 223-263). Amsterdam: North Holland. Trevarthen, C. (1988). Universal cooperative motives: How infants begin to know the language and culture of their parents. In G. Jahoda & M. Lewis (Eds.), Acquiring culture: Cross-cultural studies in cognitive development. London: Croom Helm. Trevarthen, C. (1994). The self bom in intersubjectivity: The psychology of an infant communicating. In U. Neisser (Ed.), The perceived self." Ecological and interpersonal sources of self-knowledge (pp. 121-174). Cambridge, MA: Cambridge University Press. Walker, A., & Gibson, E. J. (1986). What develops in bimodal perception? In L. Lipsitt & C. Rovee-Collier (Eds.), Advances in infancy research (Vol. 4: pp. 171-181). Norwood, NJ: Ablex. Zukow, P., & Duncan, K. R. (1994). An ecological approach to the emergence of the lexicon: Socializing attention. In V. John-Steiner, C. Panofsky, & L. Smith (Eds.),

Sociocultural approaches to language and literacy: An interactionist perspective. New York: Cambridge University Press.

The Self in Infancy: Theory and Research P. Rochat (Editor) 9 1995 Elsevier Science B.V. All rights reserved.

449

CHAPTER 22

Understanding the Self as Social Agent MICHAEL TOMASELLO

Emory University

There is this thing that happens with children: If no one is watching them, nothing is really happening to them. It is not some philosophical conundrum like the one about the tree falling in the forest and no one hearing it; that is a puzzler for college freshmen. No. If you are very small, you actually understand that there is no point in jumping into the swimming pool unless they see you do it. The child crying, "Watch me, watch me," is not begging for attention; he is pleading for existence itself. - M.R. Montgomery

Saying Goodbye: A Memoirfor Two Fathers

Gibson (1979) made a unique proposal about self-perception. He proposed that perception of the world and perception of the self are two different aspects of a single process. As organisms perceive the world and learn about it, they also perceive themselves and learn about themselves. Of special importance in this process are dynamic interactions in which organisms perceive, through various forms of proprioception, their own actions and also, through direct perception, the results those actions attain in the environment. For example, early in ontogeny human infants learn not only about what objects are reachable in the physical world, but also about their own capacities for reaching in various situations that take into account their whole bodily configuration and posture (Rochat, 1993). A number of contributors to this volume (e.g., Butterworth, Reed, Bertenthal & Rose, Rochat) pursue this general line of research, seeking to understand how infants' transactions with the physical world lead to a knowledge of the self as a physical agent. An analogous line of reasoning may be followed with respect to the infant's transactions with the social world and the resulting knowledge of self as a social agent. Mead (1934) proposed a self-other reciprocity in the social domain perfectly analogous to Gibson's proposal for a self-world reciprocity in the physical domain. In this case, however, the way that other persons work is sufficiently different

450

MICHAEL TOMASELLO

from the way that physical objects work that a very different sense of self results. When other social beings are involved, infants must tune into such things as emotions and intentions m and at some point thoughts and beliefs which are not present in their interactions with the physical world. The resulting sense of self is what Neisser (1988, 1993) called the "interpersonal self," but also includes, later in ontogeny, what he called the "conceptual self." In this chapter, I intend to look at the early development of the understanding of self as social agent in the human species, and to briefly discuss its ontogenetic and phylogenetic origins. Toward this end, I first outline what I see as the canonical developmental sequence, focusing especially on the social-cognitive revolution that occurs at around the infant's first birthday. In this I follow the basic outlines of my proposals in Tomasello (1993a). Because this volume is focused on the early ontogeny of self, however, I then go on to investigate in more detail the ontogenetic processes that might lead to this social-cognitive revolution. I conclude with some thoughts on the phylogenetic origins of the human understanding of self as social agent.

The Early Ontogeny of Self-concept Soon after birth, human infants engage in "protoconversations" with adults involving the mutual regulation of emotional engagement (Trevarthen, 1979; Stern, 1985). In these interactions, adult and infant take turns in an orderly manner, monitor the reactions of the other to their own behavior, and, in general, engage in a direct, meaningful, face-to-face interaction. In these protoconversations, the infant perceives not only the behavior and emotions of the adult, but also his or her own behavior and emotions toward the adult and the feedback they receive. This happens in much the same way as in the physical domain when the infant perceives his or her own behaviors toward objects and the resulting feedback from the physical world. We may thus speculate that the perception of self in the immediacy of social interaction leads to a sense of social self that is analogous to the sense of a physical self that arise from interactions with objects. This is the ontogenetically first sense of a social self, the interpersonal self (Neisser, 1988, 1993). Tomasello (1993a) proposed that something happens in human infants' interpersonal relatedness at around one year of age that changes profoundly the nature of their social interactions and the resulting sense of self. In the months immediately preceding their first birthdays, human infants begin to behave in a number of ways that demonstrate their growing awareness of how other persons work as psychological beings. Thus, it is at this age that they first follow into

UNDERSTANDING THE SELF AS SOCIAL AGENT

451

the attention of others by looking where they are looking (joint attention), auempt to determine how others are feeling toward a novel person or object (social referencing), and attempt to do what others are doing with a novel object (hnitative learning). It is also at this time that infants first make attempts to direct the attention of adults to outside entities through various forms of intentional communication. These first attempts to follow into and direct the attention of others to outside entities indicates a kind of social-cognitive revolution in which, for the first time, infants understand that other people perceive the world and have intentions toward it (Tomasello, 1995). What this development means for self-understanding is this: Now that the infant can monitor the attention of others to outside entities, it happens that on occasion, others are focused on the infant herself. The infant then monitors his or her attention to h e r in a way that was not possible before this new understanding of others as intentional agents. At this point, the infant's face-to-face interactions with others ~ which appear on the surface to be continuous with her face-to-face interactions from early infancy ~ are radically transformed. She is now interacting with another p e r s o n who perceives and intends things toward her in a way that was simply not perceived before. When the infant did not understand that others perceive and relate to an outside world, there could be no question of how they perceived and related to m e . After coming to this understanding, the infant now can monitor the adult's intentional relation to the world, including herself. This is essentially the "looking-glass self' of Cooley, the "me" of James and Mead. The ontogentic sequence as I envision it is summarized in Figure 1. Before 9-12 months of age, the infant may interact with a person or with an object, but never does she coordinate the two (Bakeman & Adamson, 1982). Thus, at this early age, even when an adult is looking at the same object the infant is looking at (as in Figure la), the infant is not aware of this; from the child's point of view, there is only the object. The child may also look at the person with no regard to any objects in the immediate surroundings (as in Figure lb); in this state as well, the infant has no awareness of the other person's focus of attention, even though it happens to be on the infant herself, perhaps in a protoconversation. After 9-12 months of age, the situation changes dramatically. In many situations, the infant is not only aware of an object, but she is simultaneously aware of the adult, s attention to the object (and the adult is engaged in the same process, of course); this is called "joint attention" (as in Figure l c). In the special case in which the object of the adult's attention is the child herself at this same age (as in Figure ld), the child's awareness of adult attention is still operative, and thus the object of their joint attention is me. This is the beginnings of the infant's ability to monitor the attention of others to herself, and thus the beginnings of a true self-concept.

452

MICHAELTOMASELLO Before 9 Months of Age

~Infant ]

~ Adult ]

(

Infant

A

I,o,}

)

(a) Infant engagement with object prior to 9 months (adult passive onlooking).

Co) Infant engagement with person prior to 9 months (object not part of interaction).

After 9 Months of Age

(c) Joint attention after 9 months.

(d) Self-perception after 9 months.

FIGURE 1. Object and person engagement before and after 9 months of age. Solid arrows depict perception; dotted arrows depict simulation of other's perspective (pretense). Perhaps of special importance to the ontogeny of a sense of social self is the transformation of the emotional dimension that occurs during this process. Thus, although infants have emotional relationships with others from very early in life, these are based on direct interactions. From around the ftrst birthday, infants can monitor other persons in their emotional relatedness to outside entities (as well as in their perceptual and behavioral relatedness) - - what has been called "social referencing." As in the case of attention and behavior, this new understanding of others leads to an other-centered view of the self, which concerns their emotional relatedness. Infants are social-referencing the attitudes of others to the self

UNDERSTANDING THE SELF AS SOCIAL AGENT

453

(Hobson, 1993). This new understanding of how others feel about me opens up the possibility for the development of shyness, self-consciousness, and a sense of self-esteem (Hatter, 1983). Evidence for this is the fact that within a few months after the social-cognitive revolution at the first birthday, infants begin showing the first signs of shyness and coyness in front of other persons and mirrors (Lewis, Sullivan, Stanger, & Weiss, 1979). It is important to emphasize that what happens at the first birthday is not the sudden emergence of a full-blown self-concept, but is just the opening up of a possibility. That is, what the infant's new-found social-cognitive skills do is to open up the possibility that the infant may now learn about the world from the point of view of others m a capacity I have previously called "cultural learning," in which the infant learns not from another, but through another (Tomasello, Kruger, & Ratner, 1993). One of the things they may learn about through others is themselves. Because in cultural learning infants employ all of the basic learning and categorization processes that they employ in learning about the world directly, their simulations of others regarding them serve to categorize themselves relative to other people in various ways. This categorical component is an important dimension of self-concept as well, especially during the toddler and preschool periods as children understand themselves in terms of concrete categories such as child, male, good at tree climbing, bad at bike tiding, and so forth and so on (Lewis & Brooks-Gunn, 1979). Finally, at around 3-4 years of age, children begin to develop a "theory of mind": the understanding of others as mental agents possessing not just intentions, but thoughts and beliefs. This understanding allows the child's active manipulation of other's beliefs, including their beliefs about her. Thus, for the first time at this age, children begin to actively engage in deception and in the impression management techniques that are at the core of adult social interaction and selfconcept (Goffman, 1959). The result is an understanding that my social self is not just something I can look at from the outside and categorize as I do other things; it is something I can actively control and self-regulate. The final step in this process is children's understanding at around 6 years of age that just as they are managing the impression they are creating with others, others are doing the same thing to them (Perner, 1988). The intersubjectivity characteristic of children of this age, and therefore their self-concept, is immediately reflective and recursive. Selfconcept at this point is thus very much adultlike, although in adolescence the agent whose perspective the child reflectively takes becomes in some cases the society as a whole (Mead's [1934] "generalized other").

454

MICHAEL TOMASELLO

Ontogenetic Origins The question now arises: How do children come to have the understanding of persons as intentional agents, on which this whole account depends? I have proposed different answers to this question in the recent past, depending on the precise way it was asked. On the one hand, it is possible that children need to explain the behavior of other persons, and they come to do so by constructing concepts of intentionality; however this might occur, it is only when this is done at some level that the child perceives herself in this same manner (Gopnik, 1993; Tomasello, 1993b). On the other hand, it is possible that children are exposed to their own workings as intentional agents first, and this is what makes possible their understanding of others as intentional agents (Harris, 1991; Tomasello, 1995). Both accounts imply that from a relatively early age, children are able to identify with others in a manner that allows them to equate their own psychological processes with those of others. I now believe that there is a sense in which both of these accounts are true; it is a question of how one is looking at the process and of what developmental stage is the focus of attention. First is the question of how infants make the equation between self and other persons on which all accounts rest. Meltzoff and Gopnik (1993) argue that this identification is present in nascent form from birth. From the very beginning, infants identify with and imitate o t h e r s - and know when others are imitating t h e m - so that their growing understanding of others is applied to themselves, and, conversely, their knowledge of themselves is applied to others (see also Gopnik & Meltzoff, 1993). On the other hand, citing Anisfeld's (1991) skeptical review of neonatal imitation research, Moore and Corkum (in press) believe that a certain type of social experience is necessary for children to learn the kind of self-other correspondences that Meltzoff and Gopnik posit as innate. Key to their account are situations in which adult and infant take similar intentional stances toward an outside entity, and the infant attends to this convergence; for example, they both look to the same place or fear the same object. It is in such interactions, and only in such interactions, that children have available both first-person information about their own intentional states and thirdperson information about the intentional states of the other (e.g., through their facial expressions and behavior) in the same situation. From these kinds of convergence experiences, infants come to both identify and differentiate first- and third-person perspectives in much the same way they come to differentiate and coordinate their sensory-motor schemes from the world in general (see also Baressi & M tore, 1993). Neither of these theories of how infants come to identify with others says anything directly about how their understanding of others develops and changes,

UNDERSTANDING THE SELF AS SOCIAL AGENT

455

however. With regard to the understanding of intentionality, my view is that all theorists of infant social cognition have missed one important developmental fact. It is at around 8-9 months of age that infants for the first time differentiate in their o w n behavior between the ends and means of instrumental acts, that is, they begin to behave in ways that are clearly intentional. In Piaget's (1952) formulation, infants begin down this path when they begin removing obstacles and performing acts whose function it is to enable other acts (see also Frye, 1991). Such hierarchically organized behavior indicates the infant's understanding of the relativity of means and ends: that an activity which is an end in one context may be a means in another. The essential point is that this dissociation of means and ends enables infants to formulate a goal independently of and prior to actually acting. This new mode of behavior is the child's first experience with a mental entity (i.e., a goal) that is at least somewhat independent of direct sensory-motor action. It is not accidental, I would argue, that very soon after infants begin behaving in this clearly intentional way, they begin seeing the behavior of others as intentional as well. This new understanding of the behavior of others is evidenced by the emergence of the whole complex of joint attention, imitative learning, social referencing, and intentional communication at around 9-12 months (Trevarthen & Hubley, 1978; Tomasello, 1995). My proposal is thus" that it is in their own intentional behavior that infants first experience intentionality, as they formulate independent goals and then act, either attaining them or not. This experience then provides the basis for an understanding that others also act intentionally in this same way. Exactly how it does so is not clear at this time, but it is obviously based in some way on the fact that infants have previously identified with other persons in the months leading up to this new development (by either the Meltzoff-Gopnik or Baressi-Moore developmental pathway). In either case, the proposal is that children's knowledge of their own b e h a v i o r from the inside as it were - - is the impetus for new levels in understanding the behavior of others (Gordon, 1986). It is important to emphasize that this account does not imply that infants understand the intentionality of their own behavior conceptually before they understand that of others conceptually. Infants can use experience of their own intentional activities ~ i.e., of having goals and pursuing them with varied means to simulate directly the experience and behavior of other persons. To do this, they do not need to treat their own way of functioning as an object of conceptual thinking in which the various components are distinct and conceptually manipulable (Gordon, 1992). The hypothesis here is only that when children begin operating in their sensory-motor behavior with a clear differentiation of ends and means, they automatically begin seeing the behavior of others ~ with whom

456

MICHAEL T O M A S E L L O

they have previously identified m in these same terms. As a corollary of this proposal, I should also stress that the primacy of self-understanding at the major ontogenetic transitions in how children understand persons does not mean that within a given level of understanding n for example, after the understanding of others as intentional agents n that the child cannot also develop various understandings about the specific behavior of other persons that she could then apply back to herself. The only claim is that the major transitions in early social cognition m understanding persons in terms of intentions, beliefs, and reflective beliefs w are grounded in children's understanding of their own transactions with the environment. My specific account of the developmental mechanisms that underlie early selfconcept is thus as follows. In the first 9 months of life, children are interacting with other persons regularly. In these interactions, they both match and reciprocate the interactive behaviors of others, and they notice when others match and reciprocate their own interactive behaviors. This leads to identification with, but differentiation from, other persons: I am like them, but I am a separate individual. At around 8-9 months of age, infants begin to produce instrumental behaviors in which means and ends are clearly separated: They have goals "in mind" ahead of time and employ various means for achieving them, modifying those means as necessary in light of the goals. Because of their already existing ability to identify with others, infants begin to use their own new way of interacting with the world intentionally as the basis for their understanding of the intentional activities of other persons. From this point on, they simply "see" others as intentional agents, and thus they see those agents with perceptions of and attitudes toward themselves. Early self-concept is formed as a result of monitoring the intentional attitudes of others toward the self.

Phylogenetic Origins As I have argued previously (Tomasello, 1993a, 1993b), I do not believe that selfconcept is something that was directly selected for in human evolution. The phylogenetic context in which understanding others as intentional agents (hence, self-concept) evolved was that of taking the perspective of and learning about others. For example, perspective-taking and joint-attentional abilities evolved to allow or to facilitate processes of communication, competition, deception, social learning, and cooperation, that is to say, all types of social interaction toward all types of pragmatic ends. Research with a variety of primate species, and a comparison of these to other animal species, demonstrates that nascent forms of these kinds of social interactions are indeed a distinguishing characteristic of

UNDERSTANDING THE SELF AS SOCIAL AGENT

457

primate cognition, and that humans have evolved their own special variants of them (Tomasello & Call, 1994). The important point for current purposes is that interactions of this sort have obvious evolutionary consequences for surviving and passing on genes, making them excellent candidates for natural selection. The formation of a self-concept, on the other hand, would seem to have, by itself, few directly relevant evolutionary implications for survival and procreation. It is an accident of evolution that when I am taking the perspective of the o t h e r - which is useful in many social contexts for many social ends - - the other is sometimes focused on me. It is thus my contention when looking across the animal kingdom that the extent to which an organism understands others as intentional agents determines the extent to which it has a concept of itself as an object in the world. If this is true, it is a useful fact because we have very few reliable or valid ways of assessing the self-concepts of nonverbal creatures with anything other than mirror selfrecognition tasks - - whose interpretation is far from straightforward (see Neisser, this volume; Rochat, this volume). I will say that based on current research in primate social cognition and social learning, it is my belief at present that only h u m a n s - and perhaps to some degree chimpanzees whose attention has been socialized by human beings early in their development ~ are capable of simulating others looking at themselves (and chimpanzees clearly are not capable of reflective simulations, and probably not mental simulations). Other animals may have interpersonal awareness of some sort ~ perhaps as in the direct perception of an interpersonal self or perhaps of a type we cannot envision - - but self-concept, James's concept of "me" characteristic of human children after their first year of life, may turn out to be the exclusive province of human beings. What is new in the current account relative to my previous accounts is that even though self-understanding and self-concept are not the objects of direct selection pressures in phylogeny, the mechanism by which humans come to understand others involves as a precondition identification with others. And so it may be that what is different about humans is the way in which they identify with others from early in life as both the Gopnik-Meltzoff and Moore-Corkum proposals suggest. But whereas both of these sets of theorists are vague on how infants construct a notion of others as intentional agents, I believe that infants' early identification with others can make its contribution only when the child's sensory-motor activity undergoes the reorganization that typically occurs at around 8-9 months, which leads to the dissociation of the means and ends of instrumental activities. As a final note, I would just like to point out that autistic children clearly do not engage in this same process of self-concept formation. The work of Hobson (e.g., 1993) and Loveland (e.g., 1993) demonstrates this quite clearly, and the work

458

MICHAEL TOMASELLO

of Mundy, Sigman, and Kasari (1990) further demonstrates that social-interactional deficits and self-concept deficits are intimately related as well. It is difficult to be specific about autistic children because they vary greatly among one another. However, all autistic children may be seen on a continuum that involves their abilities to simulate the psychological states of others. Autistic children who show deficits in conceiving of others as intentional agents should have deficits in their understanding of themselves as intentional agents - - and this will be evidenced by their inability to imitatively learn from others (which is usually severely retarded in these children). Autistic children who show an inability to conceive of others as mental agents should have problems treating themselves as mental agents, and thus should show impaired abilities at impression management. And those who show an inability to conceive of others as reflective agents should have self-concepts lacking reflective qualities as well. The general point is that autistic children should have self-concept deficits to the extent that they have deficits in their ability to simulate the psychological states of others. Some hints at such a connection are presented by Hobson (1989, 1993), whose views on selfconcept formation are very similar to those presented here.

Conclusion The account of early self-concept presented here has no novel components. The basic idea was put forth decades ago by James and Mead; all I have done is to place the idea in the context of current developmental research. The importance of early identification with others is one that both Gopnik-Meltzoff and Moore-Corkum have proposed: the development of an interpersonal self as foundation has been cogently argued by Neisser (1993). The importance of early intentionality and agency in the development of self-concept has been discussed by Russell (in press). All I have tried to do here is to mix the brew in a slightly new way by proposing that the process of coming to understand one's self as a social agent involves: 1) an early identification with but differentiation from others in the first 6-8 months of life; 2) a clear demonstration of intentionality toward the world in one's own behavior (differentiation of means and ends) at 8-9 months; 3) the combination of these two developments leading to an understanding of others as intentional agents at 9-12 months; and 4) the application of that understanding when others regard the self. This is the developmental foundation for the uniquely human version of self-concept in which the self is understood as a social agent in the midst of other social agents, all of whom are regarding one another simultaneously.

UNDERSTANDING THE SELFAS SOCIALAGENT 459 REFERENCES

Anisfeld, M. (1991). Neonatal imitation. Developmental Review, 11, 60-97. Bakeman, R., & Adamson, L. (1982). Coordinating attention to people and objects in mother-infant and peer-infant interactions. Child Development, 55, 1278-1289. Barrasi, J., & Moore, C. (1993). Sharing a perspective precedes the understanding of that perspective. Behavioral and Brain Sciences, 16, 513-514. Frye, D. (1991). The origins of intention in infancy. In D. Frye & C. Moore (Eds.), Children's theories of mind (pp. 15-38). Hillsdale, NJ: Erlbaum. Gibson, J. (1979). The ecological approach to visual perception. Boston: HoughtonMifflin. Goffman, E. (1959). The presentation of self in everyday life. New York: Anchor. Gopnik, A. (1993). How we know our minds: The illusion of first-person knowledge of intentionality. Behavioral and Brain Sciences, 16, 1-15. Gopnik, A., & Meltzoff, A. (1993). Imitation, cultural learning, and the origins of "theory of mind." Behavioral and Brain Sciences, 16, 521-23. Gordon, R. (1986). Folk psychology as simulation. Mind and language, 1, 158-171. Gordon, R. (1992). The simulation theory: Objections and misconceptions. Mind and language, 7, 87-103. Harris, P. (1991). The work of the imagination. In A. Whiten (Ed.), Natural theories of mind (pp. 310-395). Oxford: Oxford University Press. Harter, S. (1983). Developmental perspectives on the self system. In P. Mussen (Ed.), Carmichael's Manual of Child Psychology (Vol. 4, pp. 275-386). New York: Wiley. Hobson, P. (1989). Beyond cognition: A theory of autism. In G. Dawson (Ed.), Autism: Nature, diagnosis, and treatment. (pp. 22-48). New York: Guilford. Hobson, P. (1993). Through feeling and sight to self and symbol. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 254-279). Cambridge, MA: Cambridge University Press. Lewis, M., Sullivan, M., Stanger, C., & Weiss. M. (1989). Self-development and selfconscious emotions. Child Development, 60, 146-156. Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self. New York: Plenum. Loveland, K. (1993). Autism, affordances, and the self. In U. Neisser (Ed.), The perceived self" Ecological and interpersonal sources of self-knowledge (pp. 237253). Cambridge, MA: Cambridge University Press. Mead, G. (1934). Mind, self and society. Chicage: University of Chicago Press. Meltzoff, A., & Gopnik, A. (1993). The role of imitation in understanding persons and developing a theory of mind. In S. Baron-Cohen, H. Tager-Flusberg, & D. Cohen (Eds.), Understanding other minds (pp. 95-122). Oxford: Oxford University Press. Moore, C., & Corkum, V. (in press). Social understanding at the end of the first year of life. Developmental Review. Mundy, P., Sigman, M., & Kasari, C. (1990) A longitudinal study of joint attention and language development in autistic children. Journal of Autism and Developmental Disorders, 20, 115-28. Neisser, U. (1988). Five kinds of self knowledge. Philosophical Psychology, 1, 3559. Neisser, U. (1993). The self perceived. In U. Neisser (Ed.), The perceived self." Ecological and interpersonal sources of self-knowledge (pp. 3-21). Cambridge, MA: Cambridge University Press.

460

MICHAELTOMASELLO

Perner, J. (1988). Higher order beliefs and intentions in children's understanding of social interaction. In J. Astington, P. Harris, & D. Olson (Eds.), Developing theories of mind (pp. 141-172). Cambridge: Cambridge University Press. Piaget, J. (1952). The origins of intelligence in children. New York: Norton Books. Rochat, P. (1993). Connaissance de soi chez le b~b~. Psychology Franfaise, 38, 4151. Russell, J. (in press). Agency: Its role in mental development. Stem, D. (1985). The interpersonl world of the infant. New York: Basic Books. Tomasello, M. (1993a). The interpersonal origins of self concept. In U. Neisser (Ed.), The perceived self: Ecological and interpersonal sources of self-knowledge (pp. 185-204). Cambridge, MA: Cambridge University Press. Tomasello, M. (1993b). Where's the person? Behavioral and Brain Sciences, 16, 8485. Tomasello, M. (1995). Joint attention as social cognition. In C. Moore & P. Dunham (Eds.), Joint attention: Its origins and role in development. (pp. 103-130). Hilldale, NJ: Lawrence Erlbaum. Tomasello, M., & Call, J. (1994). The social cognition of monkeys and apes. Yearbook of Physical Anthropology, 37, 273-305. Tomasello, M., Kruger, A., & Ratner, H. (1993). Cultural learning. Behavioral and Brain Sciences, 16, 495-552. Trevarthen, C. (1979). Instincts for human understanding and for cultural cooperation: Their development in infancy. In M. von Cranach, K. Foppa, W. Lepenies, & D. Ploog (Eds.), Human ethology: Claims and limits of a new discipline (pp. 6-37). Cambridge: Cambridge Univeristy Press. Trevarthen, C., & Hubley, P. (1978). Secondary intersubjectivity. In A. Lock (Ed.), Action, gesture, and symbol: The emergence of language (pp. 183-230). New York: Academic Press.

AUTHOR INDEX 461 AUTHOR INDEX

Abravanel, E., 76, 90 Accardo, P. J., 285, 298 Acquarone, S., 106, 113 Acredolo, L. P., 128, 135, 233, 236, 337, 346 Adams, A., 128, 135 Adamson, L. B., 132, 139, 328, 347, 422,429, 443, 446, 451,459, Adelson, E., 329, 332, 343, 3 4 4 , 3 4 5 Adolph, K. E., 7, 13, 27, 32, 231, 236, 319, 321 Ajuriaguerra, J. D., 142, 158, 279, 297 Akbarian, S., 279, 297, 283,299 Alegria, J., 434, 446 Alessandri, S. M., 318, 320 Alexander, T., 133, 135 Allen, M. C., 285, 286, 297 Als, H., 132,, 139, 328, 347, 422, 429

Aman, C., 340, 346 Amblard, B., 289, 297 Ames, E. W., 360, 372 Amiel-Tison, C., 288, 297 Amsterdam, B., 4, 13, 30, 75, 90, 161, 163, 171, 172, 188, 193, 212, 310, 320, 354, 355, 367, 370, 400, 401, 413

Andersen, E. S., 329, 343, 344 Andersen, G. J., 293,297, 312, 320 Anderson, J. R., 169, 170, 171, 174, 175, 188, 190, 245, 255, 313, 354, 370

Anderson, M. A., 95, 114 Andre-Thomas, A. S., 279, 297 Anisfeld, M., 454, 459 Anniko, M., 283, 297 Anson, B. J., 284, 297 Antinucci, F., 377, 391 Arbib, M. A., 146, 158 Archibald, Y. M., 306, 322 Aristotle, 75, 113 Armstrong, C. J., 229, 239 Aron, M., 9, 14 Aronson, E., 39, 50, 227, 238, 268, 273, 286, 296, 300, 303, 314, 322, 351, 372, 381,392 Asendorpf, J. B., 172, 188 Ashmead, D. H., 29, 32, 58, 70, 228, 236

Aslin, R. N., 440, 446

Assaiante, C., 289, 297 Atwell, C. W., 286, 300 Auerbach, J., 76, 91 Aureli, F., 53, 171, 190, 204, 205, 210, 2 1 2 , 2 1 3 Bachman, C., 204, 212 Bahktin, M. M., 117, 121, 135 Bahrick, L. E., 11, 13, 29, 32, 41, 46, 49, 51, 75, 81, 87, 90, 174, 175, 187, 188, 189, 224, 225, 237, 246, 255, 311,320, 321, 356, 358, 359, 360, 361, 362, 363, 364, 367, 368, 370, 370, 371, 379, 380, 381, 382, 383, 392, 401, 402, 410, 413 Bai, D. L., 227, 228, 232, 237, 297, 303, 315, 316, 317, 318, 321, 351, 370

Baillargeon, R., 163, 188, 310, 321 Baillet, J. A., 230, 231, 238 Bain, A., 445,446 Bakeman, R., 443, 446, 451,459 Baldwin, D. A., 164, 188 Baldwin, J. M., 37, 41, 49, 62, 69, 75, 90, 350, 370 Ball, W., 40, 49, 352, 370 Bamborough, P., 399, 414 Banks, M. S., 289, 297 Bard, K. A., 169, 171, 190 Baressi, J., 454, 455, 459 Baron-Cohen, S., 106, 113 Barrera, M. E., 362, 370 Barrett, K. C., 131, 133, 135, 318, 321, 337, 339, 344 Bartsch, K., 163, 165, 188 Basen, J. A., 244, 255 Bast T. H., 284, 297 Bastock, M., 204, 212 Bateson, G., 120, 135, 446 Baudonni~re, P-M., 172, 188 Bausano, M., 319, 323 Baxter, L. A., 121, 135 Bayley, N., 332, 344 Beebe, B., 423, 429 Beeghly, M., 329, 343, 345, 354, 371 Beek, P. J., 262, 263, 269, 271, 274, 275

Bega-Lahr, N., 132, 136 Benhar, E. E., 169, 188 Bennett, K. R., 356, 370 Bennett, S. L., 423, 429

462

AUTHORINDEX

Benson, A. J., 278, 283, 297, 336, 344,347 Benton, A. L., 306, 321 Bergman, A., 54, 55, 70, 311,323 Bergstrtim, B., 284, 297 Bermt~dez, J., 74, 90 Bernstein, N. A., 152, 158, 230, 237, 259, 260, 273 Bertenthal, B. I., 128, 135, 172, 177, 188, 226, 227, 232, 237, 239, 286, 294, 295, 297, 303, 307, 310, 315, 316, 317, 318, 321,323, 337, 339, 344, 351, 356, 369, 3 70, 3 70, 381, 392, 399, 449 Berthoz, A., 227, 238, 282, 289, 294, 296, 297, 298, 300 Bierschwale, D. T., 169, 171, 174, 179, 191, 244, 256 Bigelow, A. E., 22, 175, 188, 229, 239, 329, 333, 335, 336, 337, 339, 340, 344, 344 Birch, H., 285, 301 Biringen, Z., 125, 131, 136, 340, 346 Bischof-K/Shler, D., 163, 188 Blackwell, A. W., 268, 2 75 Blass, E. M., 23, 34, 351,373, 397, 411, 415 B loch, H., 284, 299 Bloom, K., 10, 13 Boccia, M. L., 4, 14, 194, 214, 244, 256 Boerse, J., 444, 446 Borel, L., 280, 281, 282, 300 Borton, R., 39, 50, 80, 92,392, 388 Bosma, H. A., 122, 124, 137 Boudreau, J. P., 223, 237 Bower, T. G. R., 80, 87, 90, 106, 113, 265, 273, 337, 347, 351,370, 381, 388, 390, 392,392 Boysen, S. T., 210, 214 Boysson-Bardies, D., 441, 446 Bradfield, A., 340, 346 Branco, A. U., 119, 120, 132, 136 Brandt, T., 227, 237, 279, 282, 283, 298, 298, 313, 321 Braten, S., 420, 424, 428 Brazelton, T. B., 10, 13, 131, 132, 135, 139, 328, 344, 347, 422, 429 Bremner, J. G., 232, 23 7 Bril, B., 431, 432, 435, 446, 447 Broda, L. S., 229, 237 Brodal, P., 289, 298

Brodie, F., 333, 346 Brooks, P. H., 229, 239 Brooks-Gunn, J., 4, 10, 14, 63, 67, 70, 75, 91, 97, 111, 115, 133. 138, 161, 163, 172, 187, 190, 193, 214, 244, 245, 255, 310, 322, 354, 355, 358, 362, 364, 367, 368, 372, 379, 382, 392, 400, 401, 414, 453,459 Broughton, J. M., 265, 273, 351,370 Brown, J., 285, 299 Brown, T., 445, 446 Brownell, C. A., 133, 135, 164, 189 Bruner, J. S., 9, 13, 23, 33, 76, 90, 117, 122, 127, 135, 357,372, 400, 414, 423, 428 Bryant, P. E., 232, 237 Bullinger, A., 39, 51, 296, 298, 318, 321,337, 345, 411,413,415 Bullock, M., 132, 138 Bundy, R. A., 169, 189 Burke, K., 425, 428 Burton, G., 230, 231, 237 Bushnell, E. W., 58, 70, 223, 237 Bushnell, I. W. R., 362, 370 Butler, J., 127, 135 Butterfield, P., 340, 346 Butterworth, G. E., 37, 38, 39, 40, 45, 46, 47, 49, 75, 89, 90, 101, 110, 113, 128, 129, 135, 163, 187, 189, 212, 227, 237, 244, 245, 255, 265, 273, 278, 286, 287, 296, 298, 303, 309, 310, 312, 314, 315, 321, 350, 351, 355, 367, 369, 371,371, 398, 411, 413, 449 Byrne, R. W., 210, 215 Calhoun, S., 169, 189 Call, J., 186, 191,457, 460 Campbell, J., 74, 91 Campbell, R., 162, 163, 164, 191 Campos, J. J., 128, 131, 135, 227, 238, 318, 319, 321, 337, 339, 340, 344,345, 346, 346, 381, 392 Canal, I., 296, 299 Cant, J. G. H., 199, 212 Capute A. J., 285, 286, 297, 298 Cardinalli, N., 285, 298 Carello, C., 127, 138, 230, 231, 237, 296, 301 Carey, D. P., 306, 322 Carlson, V., 329, 343, 345, 354, 371

AUTHOR INDEX 463 Carlton, P. L., 169, 188 Carpenter, E., 68, 69 Carriger, M. S., 133, 135, 164, 189 Carroll, J. J., 22, 32 Carter, D. C., 340, 345 Case, R., 195, 212 Casey, S. M., 340, 345 Cassel, T. Z., 399, 414 Castillo, M., 39, 49 Cazin, L., 288, 301 Chapais, B., 202, 203, 212 Charles, E. R., 304, 323 Chatillon, J., 318, 321 Cheney, D. L., 186, 189, 204, 209, 212,213

Chomsky, N., 106, 113 Cicchetti, D., 39, 40, 329, 332, 343,, 345, 354, 355, 3 7 1 , 3 7 2 Clarke, A. E., 95, 114 Clarkson, M. G., 29, 32, 56, 69 Clay, E. R., 161, 166 Clifton, R. K., 29, 32, 56, 58, 69, 337, 345, 346 Clyman, R. B., 125, 131, 136 Cochran, E., 45, 49 Cohen, D., 41, 76, 91 Cohn, J. F., 132, 135 Connell, S., 174, 177, 178, 191 Cooley, C. H., 73, 91 Cooper, R. P., 440, 446 Corbetta, D., 287, 301 Corcos, D. M., 230, 239 Cordischi, C., 204, 205, 212 Corkum, V., 454, 457, 458, 459 Cornell, E., 229, 237, 238 Cortelyou, A., 328, 347 Corter, C., 132, 137 Coryell J. F, 285, 298 Cozzolino, R., 204, 205, 212 Craver, K. D., 230, 231, 238 Crick, F., 378, 392 Crittenden, P. M., 107, 113 Crook, C. K., 132, 139 Crook, J. H., 211, 213 Cross, M. E., 230, 231, 238 Csikszentmihalyi, M., 113 Curthoys, I. S., 284, 298 Cushman, P., 118, 135 Damasio, A. R., 306, 321 Damon, W., 350, 354, 355, 367, 371

Danek, A., 279, 283, 298 Darby, B. L., 80, 92 Darwin, C. R., 95, 103, 113,309, 321 Day, R. H., 174, 177, 178, 191 Dayal, V. S., 284, 298 De Schonen, S., 177, 191 De Waal, F. B. M., 53, 202, 206, 209, 210, 213 DeCasper, A. J., 20, 357, 3 2 , 3 7 1 Dedo, J., 131, 136 Delorme, A., 228, 237 DeLucia, C. A., 9, 14, 23, 3 4 , 3 7 3 , 357, 400, 415 Demos, E. V., 122, 131, 132, 135, 136

Dennett, D. C., 387, 392 Descartes, R., 4, 74 Deschesne, C., 283, 298, 300 Diamond, G. R., 132, 136 Diamond, R. M., 128, 137 Dichgans, J., 227, 237, 282, 298, 313, 321 Dickson, K. L., 131, 136 Dieterich, M., 279, 283, 298 Dixon, J. C., 355, 3 71, 400, 402, 413 Dodds, A. G., 340, 345 Douglas, S. D., 230, 231, 238 Doxsey, P. A., 229, 239 Dubois, B., 143, 158 Ducker, G., 169, 190, 244, 255 Dugatkin, L. A., 201, 213 Dunbar, C., 230, 231, 239 Duncan, K. R., 444, 448 Dunham, F., 133, 135 Dunham, P., 133, 135 Dunkeld, J., 337, 347, 381,392 Dunlea, A., 329, 343, 3 4 4 , 3 4 5 Dunn, J., 133, 136 Duval, S., 97, 104, 113 Dyer, F. C., 197, 198, 213 Dyre, B. P., 313, 320 Dzendolet, E., 290, 301 Eckerman, C. O., 133, 136 Eckman, P., 64, 69 Eddy, T. J., 171, 173, 174, 189, 191 Edelman, G. M., 36, 49, 48, 277, 298 Eidelman, A., 76, 91 Eilan, N., 74, 90 Eimas, P. D., 23, 32 Elias, G., 444, 446

464

AUTHORINDEX

Elliot, G. B., 284, 298 Elliot, K. A., 284, 298 Elmore, M., 132, 135 Emde, R. N., 118, 125, 131, 136, 340, 346 Engel, A. K., 277, 278, 298, 299 Enright, M., 127, 139 Eppler, M. A., 7, 13, 27, 32 Erikson, C. W., 226, 237 Escudero, P, 283, 300 Evans, D., 233, 236 Evans, S., 171, 191, 244, 256 Eviatar, A., 279, 284, 286, 299 Eviatar, L., 279, 284, 286, 299 Fadil, C. A., 175, 189, 311,321, 362, 363, 364, 368, 371 Fagan, J. F., 127, 139, 362, 371 Fagen, J. W., 127, 136, 351, 357, 360, 3 73 Fantz, R. L., 66, 69 Farkashidy, J., 284, 298 Farrar, J., 20, 34 Fernald, A., 440, 446 Ferrus, A., 296, 299 Field, J., 27, 32 Field, T. M., 41, 49, 76, 91, 132, 136, 356, 360, 371, 381, 383, 384, 392, 400, 413, 440, 446 Fifer, W. P., 20, 32,357, 371 Fiorito, G., 201, 213 Fisch, R. O., 381,392 Fischer, K. W., 111, 114, 133, 138, 172, 177, 188, 310, 321,370 Fisher, D. M., 262, 274, 356 Fivush, R., 17, 19, 32, 33 Flavell, E. R., 19, 32 Flavell, J. H., 19, 32, 164, 165, 189, 337, 345 Fletcher, J. E., 340, 345 Fogel, A., 41, 49, 118, 119, 120, 121, 122, 124, 127, 128, 130, 131, 132, 136, 434, 435, 440, 446, 447, 448 Fontaine, R., 76, 91 Forguson, L., 164, 189 Forssberg, H., 258, 262, 273 Forsstr~m, A., 267, 273 Foster, E. C., 296, 301 Foulke, E., 340, 346 Fox, T., 230, 231, 238 Fox-Kolenda, B. J., 336, 344, 345

Fragaszy, D. M., 210, 215 Fraiberg, S., 23, 32, 268, 273, 329, 330, 332, 333, 335, 336, 337, 339, 342, 343, 344,345 Franco, F., 46, 49 Fraser, J. T., 186, 189 Frederickson W. T., 285, 299 Freedman, D. A., 336, 344, 345 Freud, S., 37, 96, 98, 101, 108, 114, 350, 371, 375, 376, 377, 379, 392 Friedman, M. B., 296, 299 Friesen, W. V., 64, 69 Frigon, J., 228, 237 Frith, U., 106, 113, 114 Fromhoff, F. A., 19, 32 Frye, D., 455, 459 Fung, H., 133, 138 Furer, M., 350, 372 Gallagher, S., 410, 411, 412,413, 414 Gallistel, C. R., 197, 213 Gallup, G. G. Jr., 4, 13, 30,32, 34, 47, 48, 49, 60, 67, 69, 141, 158, 169, 170, 171, 172, 174, 175, 186, 187, 189, 191, 193, 213, 243, 255, 310, 321,354, 371, 382, 392 Gapenne, O., 24, 227, 237, 279, 289, 290, 291, 292, 293, 300, 316, 399 Garner, W. R., 226, 237 Gazzaniga, M. S., 97, 99, 105. 114 Geertz, C., 95, 101, 114 Gekoski, M. J., 39, 14, 18, 127, 139, 323, 438, 448 Geppert, U., 132, 138 Gergen, K. J., 118, 136 Gerrits, N. M., 281,299 Gesell, A., 285, 299 Gianino, A., 130, 136 Gibson, E. J., 6, 7, 8, 13, 22, 27, 32, 33, 51, 87, 91, 195, 210, 222, 227, 229, 237, 238, 240, 271, 273, 310, 312, 319, 321, 327, 345, 350, 371, 410, 411, 434, 435, 437, 448 Gibson, J. J., 6, 7, 13, 22, 26, 33, 37, 38, 49, 60, 69, 80, 87, 91, 127, 128, 136, 141, 158, 222, 229, 238, 259, 263, 265, 267, 273, 277, 286, 295, 299, 303, 304, 312, 321,322, 349, 352, 364, 368, 371, 379, 392, 395, 414, 433, 442, 446,459

AUTHOR INDEX 465 Gielen, C. C. A. M., 227, 313, 320 Ginsburg, G. P., 127, 128, 137 Gleitman, H., 6, 13,399, 414 Goffman, E., 117, 120, 137, 453, 459 Goldberg, S., 387, 392 Goldenberg, D., 38, 50, 128, 129, 137 Goldman, A. I., 74, 91 Goldstein, S., 132, 136 Goodale, M. A., 304, 305, 306, 307, 309, 3 2 2 , 3 2 3 Goodwin, S. W., 128, 135 Gopnik, A., 74, 91, 97, 114, 163, 164, 167, 189, 190, 454, 455, 457, 458, 459 Gordon E., 285, 301 Gordon, R., 455,459 G6tz, K. G., 295, 299 Goubet, N., 27, 34, 58, 70, 4 1 3 , 4 1 5 Gralinski, J. H., 133, 139 Grammer, K., 120, 13 7 Gray, J. E., 95, 114 Gray, J. T., 19, 32 Green, F. L., 19, 32 Green, J., 441, 444, 447 Greenberg, L. M., 355, 370 Greenberg, R., 41, 76, 91 Greenfield, P. M., 157, 158 Gregg, C., 285, 299 Grene, M., 352, 371 Grenier, A., 288, 297, 299 Griffin, D. R., 193, 213 Grosofsky, A., 230, 231, 237 Grossman, K. E., 421,427, 429 Grover, L., 45, 49, 50 Grtisser, O. J., 279, 283, 297, 299 Guedry, E. E., 278, 299 Guillaume, P., 175, 189, 400, 414 Guldin, W. O., 279, 283,297, 299 Gusella, J., 440, 446 Gustafson, G., 441, 443, 444, 446, 447

Guth, D. A., 232, 239, 340, 346 Haffner, M. E., 285,299 Haith, M. M., 328, 345 Hake, H. W., 226, 237 H aken, H., 260, 273 Harcourt, A. H., 202, 213 Harding, V., 95, 114 Harkness, R. D., 197, 215 Harkness, S., 431, 448

Harre, R., 117, 120, 121, 137 Harris, C. S., 22, 33 Harris, P. L., 212,342, 345, 399, 414, 454, 459 Harryman, S., 285, 298 Hart, D., 350, 354, 355, 367:371 Harter, S., 133, 137, 221, 238, 310, 322, 350, 354, 355, 3 7 1 , 4 5 3 , 4 5 9 Hartmann, E. E., 286, 300 Hauert, C. A., 153, 158, 160 Head, H., 142, 158 Heath, C., 120, 13 7 Hecaen, H., 142, 158 Hediger, H., 209, 213 Heft, H., 127, 128, 13 7 Heidegger, M., 211, 212, 213 Heilman, K. M., 25, 33 Heiman, M. L., 232, 240 Heimann, M., 76, 91 Hein, A., 128, 13 7 Henderson, C., 337, 340, 345, 346 Henty C., 296, 298 Herman, J. F., 229, 238 Hermans, H. J. M., 117, 120, 121, 122, 127, 137 Hess, R., 288, 301 Heth, C. D., 229, 237, 238 Heyes, C. M., 210, 213, 246, 255 Hiatt, S., 337, 345 Hicks, L., 49, 227, 237, 287, 298, 303, 314, 315, 321, 351,371 Higgins, C., 227, 238 Hill, E. W., 232, 239, 340, 346 Hill, S. D., 169, 189 Hinton, G. E., 376, 378, 392 Hiroto, D. S., 329, 345 Hobson, R. P., 23, 33, 329, 331, 332, 345, 453, 457, 458, 459 Hoffmeyer, L. B., 23, 34, 351,373, 397, 411, 415 Hollyfield, R. L., 340, 346 Holt, S., 131, 136 Hoogstra, L., 133, 138 Hooker, D., 284, 299 Hopkins, B., 39, 49, 75, 90, 262, 263, 265, 271, 273,275, 398, 411, 413

Hopkins, W. D., 4, 15, 245, 247, 256 Horobin, K., 337, 346 Howard, I. P., 312, 322 Howarth, C. I, 340, 345 Hubel, D. H., 304, 322

466

AUTHORINDEX

Hubley, P., 130, 139, 455,460 Hudson, J. A., 19, 33 Humphrey, N., 35, 50 Hunter, M. A., 360, 372 Hurshman, A., 133, 135

Kamm, K., 262, 273, 287, 301 Kant, I., 211, 213 Kaplan, A. R., 286, 300 Kaplan, H., 60, 71 Kaplowitz, C., 172, 191, 354, 355, 373,401,415

Karmiloff-Smith, A., 145, 147, 153, Ikegami, H., 281, 282, 299 Illingworth, R. S., 285, 299 Ingle, D. J., 304, 322 Inhelder, B., 7, 14 Ishii, T., 132, 138 Itakura, S., 48, 170, 171, 174, 178, 190

Iwata, H., 132, 138 Jackendoff, R., 147, 154, 158 Jacquet, A. Y., 284, 301 Jaffe, J., 423, 429 Jakobson, L. S., 306, 322 James, W., 3, 13, 17, 18, 33, 46, 50, 54, 69, 99, 106, 114, 121, 128, 137, 162, 165, 166, 168, 186, 187, 190, 195, 213, 319, 322, 350, 372, 380, 382, 392,396, 414, 451, 458 Jarrett, N. L. M., 45, 49, 163, 189 Jencks, C., 117, 118, 137 Jennings, S., 133, 138 Jensen, J. L., 262, 273, 287, 301 Jerison, H. J., 197, 213 Johnson, C. B., 172, 190 Johnson, D. B., 372 Johnson, M., 127, 137, 145, 150, 151, 354, 158 Johnson, M. H., 66, 69 Jones, D., 340, 346 Jopling, D., 17, 33 Jordan, M. I., 384, 392 Jouen, F., 24, 39, 50, 227, 237, 238, 279, 284, 285, 287, 288, 289, 290, 291, 292, 293, 299, 300, 316, 316, 322, 399 Judge, P. G., 204, 213 Jusczyk, P. W., 23, 32 Kagan, J., 161, 163, 165, 190, 311, 322

Kaitz, M., 50, 76, 91 Kalnins, I. V., 9, 13, 23, 33, 357, 372, 400, 414

158

Kasari, C., 458, 459 Kawai, M., 202, 213 Kaye, K. L., 65, 70, 130, 137, 388, 390, 392,435, 436, 440, 447 Kekelis, L. S., 329, 343, 344 Kellman, P. J., 6, 13, 362, 364, 372, 399, 410, 414 Kelso, J. A. S., 259, 261, 273 Kempen, H. J. G., 117, 120, 121, 122, 127, 137 Kendon, A., 120, 13 7 Kermoian, R., 227, 337, 238, 346 Kernberg, O. F., 108, 114 Kienapple, K., 132, 13 7 Kisielow, P., 95, 115 Klein, S. J., 127, 136 Knoespel, K. J., 120, 13 7 Kobayashi, N., 132, 138 Koenderinck, J. J., 296, 300 Koenig, E., 227, 237 K6hler, W., 202, 213 Kohut, H., 427, 429 Koitcheva, 294 Kokshanian, A., 284, 298 Kolb, S., 258, 275 Kolers, P. A., 127, 137 Konczak, J., 230, 231, 238 K~nig, P., 277, 278, 298, 299 Kopp, C. B., 95, 114, 133, 139 Korner, A. F., 397, 414 Korner, A. L., 285, 299 Koslowski, B., 10, 13, 131, 135, 328, 344

Kraemer, H. C., 397, 414 Kravitz, H., 38, 50, 128, 129, 13 7 Kreiter, A. K., 277, 298 Krowitz, A., 381,392 Kruger, A. C., 65, 71, 76, 92, 210, 214, 453, 460 Kugiumutzakis, G., 41, 43, 50, 436, 447

Kugiumutzakis, J., 76, 91 Kugler, P. N., 259, 261,273 Kuhl, P. K., 400, 414

AUTHOR INDEX 467 Kuhn, T. S., 112, 114 Kummer, H., 202, 204, 212, 213 Kuperschmidt, J., 10, 14 Labov, W., 425, 429 Lacan, J., 108, 114 Lacour, M., 280, 281, 282, 300 Lagace, C., 228, 237 Lakoff, G., 145, 150, 151, 158 Lamb, M. E., 328, 346 Landau, K., 169, 171, 174, 179, 180, 181, 191, 244, 256 Langer, A., 381,392 Langer, J., 164, 190 Langer, S. K., 426, 429 Langhorst, B. H., 132, 136 Lavigne-Rebillard, M., 283, 300 Lecanuet, J. P., 284, 301 Ledbetter, D. H., 244, 255 LeDoux, J., 99, 100, 104, 114 Lee, D. N., 39, 40, 46, 50, 51, 75, 87, 92, 227, 229, 238, 241, 259, 263, 264, 265, 267, 268, 269, 270, 271, 273,274, 286, 296, 300, 303, 307, 308, 313, 314, 316, 3 2 2 , 3 5 1 , 3 7 2 , 381, 392, 401,415 Legerstee, M., 76, 91, 132, 137 Lepecq, J. C., 279, 287, 289, 293, 299, 300

Leslie, A. M., 106, 113,114, 163, 190, 212

Lestienne, F., 227, 238, 294, 300 Lethmate, J., 169, 171, 190, 244, 255 Levenson, R. W., 64, 69 Levine, L. E., 133, 137 Lewis, M. D., 4, 10, 14, 23, 30, 33, 63, 67, 70, 75, 91, 97, 99, 100, 103, 104, 105, 110, 111, 114,115, 122, 131, 133, 138, 161, 163, 164, 172, 187, 190, 193, 195, 214, 244, 245, 255,'310, 318, 320, 322, 350, 354, 355, 358, 362, 364, 367, 368, 372, 379, 382, 387, 392, 400, 401, 414, 438, 448, 453,459 Liebowitz, H. W., 313, 322 Lillard, A. S., 164, 190 Lin, A. C., 169, 171, 190 Lin, J. P., 270, 274 Lipsitt, L. P., 9, 14 Lisberger, 309

Lishman, J. R., 40, 50, 227, 229 238, 296, 300, 308, 313, 316, 322 Litovsky, R. Y., 58, 69, 337, 345 Livingstone, M. S., 304, 322 Loboschefski, T. W., 356, 370 Locke, J., 441, 447 Lockman, J. J., 58, 70, 157, 159, 229, 373,238, 340, 346, 447, 352, 439 Logothetis, N. K., 304, 323 Lollis, 443 Loveland, K. A., 7, 14, 30, 33, 177, 190, 332, 346, 354, 355, 356, 372, 457, 459 Lucas, D., 127, 139 Lutkenhaus, P., 132, 138 Lyra, M., 121, 138 Maccoby, E. E., 132, 138 MacFarlane, A., 56, 70 MacKinnon, J., 199, 214 MacMurray, S., 41, 50 Macnamara, J., 164, 190 Maestripieri, D., 171, 190 Mahler, M. S., 54, 55, 70, 311,323, 350, 3 72 Maier, S. F., 329, 347 Main, M., 10, 13, 131, 135, 328, 344 Mandler, J. M., 142, 144, 145, 146, 147, 150, 151, 152, 154, 155, 159 Mans, L., 355, 372 Maratos, O., 41, 50, 76, 91 Marcel, A. J., 74, 90, 157, 159 Marchal, P., 170, 190 Marcy, S., 385, 386, 393 Margileth, D. A., 336, 3 4 4 , 3 4 5 Marian, V., 66 Mark, L. S., 27, 33, 230, 231,238 Markova, I., 119, 138 Mart, D., 156, 159, 323, 304 Marshall, J., 306, 308, 324 Marty, R., 284, 301 Mascolo, M. F., 133, 138 Mascot, 294 Mason, W. A., 11, 14 Matusov, E., 131, 136 Maurer, D., 56, 70, 328, 346, 362, 370

Mc Crea, R., 282, 298 McCarrell, N. S., 229, 239 McCarty, M. E., 228, 236 McClure, B. A., 95, 114

468

AUTHORINDEX

McClure, M. K., 169, 189 McCollum, G., 227, 239 McCrea, 282 McGraw, M. B., 258, 262, 273 McHale, J. P., 439, 447 McKenzie, B. E., 174, 177, 178, 191, 229, 239 Mead, G. H., 19, 20, 33, 73, 91, 121, 128, 130, 138, 445, 449, 451, 453, 458, 459 Meeuwsen, J. J., 230, 231,238 Megaw-Nyce, J., 41, 51 Meijer, O. G., 260, 274 Mellier, D., 285, 300 Meltzoff, A. N., 20, 33, 39, 41, 43, 50, 56, 64, 65, 70, 76, 77, 78, 79, 80, 81, 82, 85, 86, 87, 91, 92, 97, 106, 114, 115, 132, 138, 174, 178, 187, 190, 195, 214, 350, 351, 365, 369, 372, 388, 392, 400, 410, 411, 412,414, 422, 434, 437, 447, 454, 455, 457, 458, 459 Melzack, R., 24, 33 Menzel, E. W., 171, 189, 208, 209, 214

Meredith, M. A., 286, 288, 289, 301 Merleau-Ponty, M., 25, 33 Meschulach-Sarfaty, O., 76, 91 Messinger, D., 131, 136 Mestre, D. R., 268, 275 Michaels, C. F., 127, 138, 263, 269, 274

Michalson, L., 105, 115 Michel G., 285, 298 Michotte, A., 41, 50 Miles, L. H., 211, 214 Millar, S., 341,346 Miller, D. H., 344, 345 Miller, P. J., 133, 336, 138 Milner, A. D., 304, 305, 306, 308, 309, 322, 323 Mintz, J., 133, 138 Mishkin, M., 305, 307, 323 Mistry, J., 19, 34 Mitchell, R. W., 4, 14, 133, 138, 171, 172, 175, 177, 178, 190, 194, 210, 211, 214, 243, 244, 255, 256, 317, 323

Mitchell, S. A., 117, 13 8 Mitchison, G., 378, 392 Mittelstaedt, H., 198, 214 Mizukami, K., 132, 138

Montgomery, M. R., 449 Moore, C. 454, 457, 458, 459 Moore, J., 434, 437, 447 Moore, M. K., 20, 33, 39, 41, 50, 56, 70, 76, 77, 78, 79, 80, 87, 91, 92, 106, 115, 132, 138, 178, 190, 195, 214, 265, 273, 351, 365, 370, 372, 411, 412,414, 422 Morgan, R., 29, 34, 75, 81, 87, 92, 149, 224, 225, 226, 240, 360, 361, 367, 373, 403, 405, 407, 409, 410, 414,415, 419 Morongiello, B. A., 9, 14 Morris, M. W., 204, 212, 268, 2 75 Morton, J., 66, 69 Mosier, C. E., 10, 14., 439, 447 Moss, L. E., 175, 189, 311,321, 362, 363, 364, 368, 371 Mounoud, P., 144, 153, 155, 156. 157, 159, 160, 424 Movshon, 309 Moynihan, M., 201, 204, 212, 214 Mueller, M., 198, 214 Muir, D. W., 29, 32, 337, 347, 440, 446

Mundy, P., 458, 459 Munn, P., 133, 136 Murray, L., 10, 14, 20, 21, 33, 44, 50, 66, 70, 328, 346, 357, 372 Nachman, 439 Nagel, T., 211, 214 Nanez, J. E., 352, 365, 372 Nashner, L. M., 227, 239 Neisser, U., 17, 21, 33, 37, 39, 43, 50, 53, 57, 66, 70, 75, 89, 90, 92, 125, 138, 142, 142, 160, 195, 196, 214, 221, 222, 226, 235, 239, 244, 245, 256, 277, 278, 300, 307, 312, 319, 323, 330, 332, 342, 346, 350, 353, 369, 372, 397, 414, 420, 425, 429, 435, 436, 445, 450, 457, 458, 459

Nelson, K. E., 76, 91, 157, 158, 166, 167, 168, 187, 190, 210, 214 Neufield, J., 229, 239 Newell, K., 230, 239 Neyhus, A., 38, 50, 128, 129, 137 Nielsen, L., 336, 343, 346 Niles, D., 262, 274 Nishida, T., 208, 214

AUTHOR INDEX 469 Noirot, E., 434, 446 Norris, M., 333, 346 Nozick, N., 101, 115 Nwokah, E., 131, 136 Oguz6reli, N. M., 295, 301 Ohr, P. S., 127, 136 Oller, D. K., 397, 414, 440, 447 Olson, D. R., 162, 163, 164, 165, 167, 191 Oppenheim, D., 125, 131, 136 Omitz, E. M., 286, 300 Owen, B. M., 296, 300 Owsley, C. J., 41, 51,318, 324, 362, 364, 3 72 Oyama, S., 118, 138 Paillard, J., 289, 300 Palmer, C. F., 229, 239 Palmer, F. B., 285, 298, 438 Papaioannou, J., 289, 300 Papi, F., 196, 214 Papousek, H., 9, 14, 311,323, 356, 359, 372, 379, 380, 381, 383,392, 401, 402, 414, 437, 44 7 Papousek, M., 9, 14, 311,323, 356, 359, 372, 379, 380, 381, 383,392, 401, 402, 414, 437, 447 Parke, R., 132, 138 Parker, S. T., 4, 14, 175, 194, 191, 214, 244, 256, 377, 392 Parpal, M., 132, 138 Patla, A. E., 229, 239 Patterson, F., 244, 256 Pavlov, I., 111 Peiper A., 285, 300 Perenin, M. T., 306, 323 Perilloux, H. K., 180, 181, 191 Perner, J., 19, 34, 164, 1 9 1 , 4 5 3 , 4 6 0 Perris, E. E., 58, 69, 337, 345, 346 Perrone, J. A., 296, 300 Perrucchini, P., 46, 49 Pettersen, L., 352, 3 7 3 , 3 8 1 , 3 9 2 Piaget, J., 7, 8, 9, 14, 29, 36, 37, 50, 54, 62, 63, 70, 73, 74, 76, 92, 106, 109, 115, 141, 142, 144, 146, 147, 148, 149, 150, 151, 152, 153, 154, 157, 160, 195, 214, 259, 274, 350, 373, 375, 376, 377, 384, 392, 401, 414, 438, 447, 455,460

Pick, A., 142, 160 Pick, H. L. Jr., 340, 346 Pickens, J. N., 360, 370 Pillon, B., 143, 158 Pine, F., 54, 55, 70, 311,323 Pipp, S., 133, 138 Poggio, T., 295, 301 Pope, J. M., 286, 287, 296, 300 Pope, M. J., 315, 323 Porges, S. W., 357, 373 Porton, I., 284, 301 Porush, D., 117, 138 Post, R. B., 313,322 Potts, R., 133, 138 Povinelli, D. J., 53, 60, 70, 169, 171, 173, 174, 179, 180, 181, 185, 186, 189, 191, 209, 210, 214, 244, 256 Powell, G. M., 290, 301 Precht, W., 288, 301 Prechtl, H. F. R., 271, 274 Premack, D., 209, 214 Prentice, S. D., 229, 239 Previc, F. H., 284, 301 Preyer, W., 8, 14, 195, 214 Pribram, K. H., 105, 106, 115 Priel, B., 177, 191 Prigogine, I., 260, 274 Prince, M., 95, 115 Proffitt, D. R., 226, 239 Pufall, P. B., 230, 231,239 Pujol, R., 283, 284, 300, 301 Rader, N., 23, 33, 319, 323, 381,393 Radke-Yarrow, M., 163, 192 Ramey, C. T., 10, 15, 318, 323, 357, 360, 3 73 Ramsey, D., 337, 345 Rand, D. T., 259, 273 Rank, O., 108, 115 Rappaport, D., 376, 393 Ratner, H. H., 65, 71, 76, 92, 210, 214, 453, 460 Reddish, P. E., 259, 273 Reed, E. S., 127, 138, 419, 432, 435, 439, 444, 446, 447 Reichardt, W., 295, 301 Reichel, F. D., 230, 231, 237 Reid, T., 37, 38, 51 Reissland, N., 41, 51, 76, 92 Repacholi, B., 385, 386, 393 Rescorla, R. A., 187, 191

470

AUTHORINDEX

Rey, A., 157, 160 Rheingold, 443 Riccio, G., 8, 13, 27, 32, 237, 229 Richards, J. E., 319, 3 2 3 , 3 8 1 , 3 9 3 Richardson, W. K., 245, 247, 256 Rider, E. A., 232, 239, 240 Ridley-Johnson, R., 262, 274 Rieser, J. J., 229, 232, 239, 240, 346 Ristau, C. C., 194, 214 Robinson, C., 229, 239 Robinson, J. A. 174, 177, 178, 191 Robinson, N., 340, 346 Rochat, P., 23, 27, 29, 33, 34, 39, 51, 55, 56, 58, 59, 69, 70, 71, 75, 81, 87, 92, 129, 139, 149, 152, 153, 195, 224, 225, 226, 240, 244, 278, 301, 337, 345, 350, 351, 352, 360, 360, 361, 365, 367, 369, 373, 396, 397, 403, 405, 407, 409, 410, 411, 4 1 3 , 4 1 4 , 4 1 5 , 419, 434, 438, 447, 449, 457, 460 Rodaniche, A. F., 201,214 Roeder, J. J., 170, 188, 201 Roediger, H. L., 127,137 Rogoff, B., 10, 14, 19, 34, 434, 439, 447

Roitblat, H. L., 115 Roland, A., 95, 115 Roman, J., 337, 347 Ronnqvist, L., 58, 69 Rosch, E., 278, 301 Rose, A., 285, 298 Rose, J. L., 316, 317, 318, 321,323, 399, 449 Rose, S. A., 223, 240 Rosen, J., 340, 346 Rosenberg, D., 8, 13, 27, 32, 229, 237

Rosencranz, D., 340, 347 Rosenhall, U., 284, 301 Ross, C. A., 95, 115 Ross, M., 101, 115, 443 Rossetti-Ferreira, M. C., 121, 138 Rovee, C. K., 9, 14, 351,373 Rovee, D. T., 9, 14, 351,373 Rovee-Collier, C. K., 9, 14, 24, 34, 106, 115, 127, 135, 139, 318, 323, 351, 357, 360, 373, 400, 415, 438, 448

Rozin, P., 314, 323 Rubenstein, J. E., 285, 298 Ruff, H. A., 223,240

Rugg, M. D., 308, 323 Rulf, A., 169, 171, 174, 179, 191. 244, 256 Rumbaugh, D. M., 245, 247, 256 Rumelhart, D. E., 384, 392,393 Russell, J., 458, 460 Sacks, O., 25, 34 Salapatek, P., 56, 70, 289, 297 Saltzman, E., 157, 158 Sampson, E. E., 117, 139 Samuel, D., 169, 188 Samuels, C. A., 351,353, 3 73 Samways, M., 229, 239 Sander, M. D., 130, 306, 308, 324 Sans, A., 283, 284, 298, 300, 301 Sarbin, T. R., 117, 120, 122, 127, 131, 139 Sasaki, M., 281, 282, 299 Savage-Rumbaugh, E. S., 245, 247, 256

Savelbergh, G. J. P., 262, 263, 271, 275

Scafidi, F., 132, 136 Schaffer, H. R., 132, 13 9 Schaller, J., 76, 91 Scheibe, K. E., 122, 139 Scherer, K. R., 120, 139 Schiff, W., 22, 34 Schilder, P., 142, 160, 410, 415 Schillen, T. B., 277, 298 Schiller, P. H., 304, 323 Schino, G., 171, 190 Schmid-Hempel, P., 197, 215 Schmidt, R. A., 260, 2 74 Schmuckler, M. A., 8, 13, 27, 32,225, 227, 228, 229, 230, 237, 238, 240, 361, 367, 3 73 Schneider, K, 287, 301 Scholz, J. P., 261, 273 Sch6ner, G., 296, 301 Schulman, A. H., 172, 191, 354, 355, 373,401,415

Scotto, P., 201, 213 Scucchi, S., 205, 212 Sejnowski, T. J., 309, 376, 378, 392 Seligman, M. E. P., 329, 345, 347 Senders, S. J., 55, 70, 411,438, 415, 447

Senjowski, T. J., 323 Sersen, E. A., 411,415

AUTHOR INDEX 471 Seyfarth, R. M., 186, 189, 204, 209, 212,213

Shapiro, B. K., 285, 298 Sherriff, F. 337, 347 Sherrington, C. S., 51, 55, 71 Shotter, J., 117, 118, 127, 139 Shumway-Cooke, A., 228, 240 Siddiqui, A., 439, 447 Sigafoos, A. D., 76, 90 Sigman, M., 458, 459 Simpson, J. I., 288, 301 Singer, J. M., 127, 136 Singer, W., 277, 298, 300 Siqueland, E. R., 9, 14, 23, 32, 34, 357, 373, 400, 415 Sirigu, A., 143, 158 Skinner, B. F., 111, 387 Slaughter, V., 167, 190 Slobin, D., 145, 160 Smith L., 262, 274, 277, 301 Smith, P. H., 356, 370 Smuts, B., 204, 214 Snow, C., 440, 448 Soechting, J., 227, 238, 294, 300 Solomon, Y., 230, 231, 237 Soodak, R. E., 288, 301 Spada, E. C., 53 Spaulding, P., 333, 346 -Spelke, E. S., 6, 13, 106, 115, 151, 160, 223, 240, 310, 323,328, 347, 362, 364, 372, 399, 400, 410, 414, 415

Sperry, R., 44, 51 Spitz, R. A., 54, 65, 71 Sroufe, L. A., 372 Stack, D., 337, 347 Stanger, C., 164, 453, 459 Stark, R. E., 397, 415, 440, 448 Stassen, H., 267 Stein, B. E., 286, 288, 289, 301 Stein, R. B., 295, 301 Stengers, I., 260, 274 Stern, D. N., 44, 51, 61, 66, 71, 75, 76, 92, 95, 107, 108, 109, 115, 118, 124, 125, 126, 130, 131, 139, 195, 214, 328, 347, 420, 421, 423, 424, 425, 427, 429, 434, 436, 438, 439, 448, 450, 460 Stipek, D. J., 133, 139 Stoffregen, T. A., 8, 13, 27, 32, 227, 229, 237, 240 Strata, P., 288, 301

Streri, A., 328, 347 Stroufe, 355 Studdert-Kennedy, M., 441,448 Suarez, S. D., 30, 34, 169, 170, 171, 189, 191

Suddendorf, T., 186, 191 Sullivan, M. W., 10, 14, 63, 70, 164, 318, 320, 400, 414, 438, 448, 453, 459,

Super, C. M., 431,448 Suslick, R., 340, 347 Sveistrup, H., 230, 241, 296, 301, 303, 324 Svejda, M., 337, 345 Swartz, K. B., 171, 191,244, 256 Talmy, L., 145, 160 Talor, C. R., 340, 346 Tamboer, J. W .I., 267, 274 Taormina, J., 8, 13, 27, 32, 229, 237 Thatcher, R. W., 143, 160 Thelen, E., 131, 136, 262, 273,274, 277, 287, 301 Thom, R., 262, 274 Thompson, E., 278, 301 Thompson, N. S., 210, 211, 214 Thompson, R. L., 169, 189 Toda, S., 440, 448 Tolman E., 4, 387 Tomasello, M., 12, 14, 20, 34, 65, 71, 76, 92, 186, 191, 210, 214, 443, 450, 451, 453, 454, 455, 456, 457, 460

Tomkins, S. 124, 131, 132, 139 Toth, S., 329, 343, 345, 354, 371 Trevarthen, C., 10, 14, 20, 21, 33, 43, 44, 50, 51, 66, 70, 71, 76, 92, 130, 131, 139, 195, 211, 215, 328, 346, 357, 371,372, 419, 420, 422, 424, 429, 432, 434, 435, 436, 448, 450, 455, 460 Troisi, A., 171, 190 Tronick, E. Z., 40, 66, 71, 130, 132, 135, 136, 139, 328, 347, 352, 370, 422, 423, 424, 429, 440, 446 Tulving, E., 105, 115 Turkewitz G., 285, 301 Turvey, M. T., 230, 231,237, 259, 260, 261, 273, 2 74, 296, 301

472

AUTHORINDEX

Uchino, Y., 281, 282, 299 Ulrich, B. D., 262, 2 74 Umansky, R., 385, 386, 393 Ungerleider, L. G., 305, 307, 323 Uzgiris, I. C., 336, 3 4 4 , 3 4 7

Wagner, H., 27, 34 Walcher, J. R., 285, 298 Walk, R. D., 27, 33 Walker, A. S., 6, 13, 41, 51, 434, 448 Wallnau, L. B., 169, 170, 189 Wallon, H., 54, 71, 142, 160, 400,

Valenstein, E', 25, 33 Valsiner, J., 119, 139 Van Asten, W. N. J. C, 227, 236, 313,

Walter, D. O., 286, 300 Warren, W. H., 8, 15, 26, 34, 229, 230, 231, 240, 241, 266, 268, 275 Warrington, E., 306, 308, 324 Washburn, D. A, 245, 247, 256 Watchel, R. C., 285,298 Watson, J. S., 10, 11, 13, 15, 29, 32, 46, 49, 61, 71, 75, 81, 87, 90, 92, 97, 100, 101, 110, 115, 174, 187, 188, 224, 225, 237, 246, 255, 311, 318, 320, 323, 328, 347, 356, 357, 358, 359, 360, 361, 367, 370, 373, 376, 379, 380, 381, 383, 384, 385, 387, 391, 392,393, 400, 401, 402, 410, 4 1 3 , 4 1 5 Watson, R. T., 25, 33 Wehner, R., 197, 198, 2 1 4 , 2 1 5 Weinberg, K. M., 423,424, 429 Weinstein, S. E., 411, 415 Weiskrantz, L., 76, 93, 103, 105, 115, 306, 308, 324 Weiss. M., 164, 453, 459 Weist, R. M., 187, 191 Wellman, H. M., 163, 165, 188, 212 Wells-Gosling, N., 200, 215 Werner, H., 60, 71 West, M., 441, 443, 444, 447 Westergaard, G. C., 4, 14, 15 Whang, S., 8, 15, 230, 231, 240, 266,

320

Van der Gon, J. J. D., 227, 313, 320, 236

Van der Meer, A. L .H., 11, 15, 39, 46, 51, 75, 87, 92, 264, 265, 267, 269, 270, 271, 272, 274, 351, 365, 373, 401, 415, 419 Van der Weel, F. R., 39, 51, 75, 87, 92, 264, 265, 267, 269, 270, 271, 272, 274, 401, 415 Van der Werff, J. J., 122, 139 Van Gielen, C. C. A. M., 236 Van Halen, C., 122, 124, 137 Van Langenhove, L., 121, 137 Van Loon, R. J. P., 121, 137 Van Schaik, C. P., 204, 212 Varela, F. J., 128, 278, 3 0 1 , 1 3 9 Vaughn, L. A., 23, 33 Verbeek, P. 53 Vereijken, B., 260, 274 Vidal, P. P, 282, 298 Vighetto, A., 306, 323 Vigorito, J., 23, 32 Vining, E. P., 285, 298 Vinter, A., 41, 43, 76, 92, 153, 156, 160

Visalberghi, E., 210, 215 Vogele, D., 230, 231, 238 Von Bartalanffy, L., 98, 110, 115 Von Boehmer, H., 95, 115 Von der Malsburg, 277, 300 Von Glasersfeld, 397, 414 Von Hofsten, C., 38, 50, 58, 69, 157, 158, 265, 267, 269, 273,274, 275, 307, 322, 351,371, 411,414, 434, 438, 439, 447 Von Hoist, E., 260, 275 Von Uexktill, J., 197, 215 Vynter, A., 51

415

275

White, B. L., 328, 347 Whiten, A., 186, 191, 210, 215 Wicklund, R. A., 97, 104, 113 Wilmot, W. W., 121, 139 Wimmer, H., 19, 34 Wimmers, R. H., 262, 263, 271, 275 Wise, S., 132, 328, 139, 347, 422, 429

Wishart, U. G., 337, 347 Wittgenstein, L., 25, 34 Wolf, D., 133, 139 Wolff, P. H., 376, 393 Wood, R. W., 291, 3 0 1 , 3 1 3 , 3 2 4 Woodruff, G., 209, 214 Woodson, R., 41, 76, 91

AUTHOR INDEX 473 Woollacott, M., 228, 230, 240, 241, 296, 301, 303, 324 Wraga, M., 59, 71 Wyke, B., 284, 301 Wylie, R. C., 97, 115 Yardley, L., 296, 301 Yonas, A., 318, 324, 352, 373, 381, 392

Yoshida, K., 282, 298 Young, D. S., 229, 241, 263, 273 Zahn-Waxler, C., 163, 192 Zazzo, R., 3, 15, 172, 177, 187, 192 Zelazo, N. A., 258, 275 Zelazo, P. R., 258, 259, 262, 275 Zemicke, R. F., 287,301 Zukow, P., 444, 448

This Page Intentionally Left Blank

SUBmCT INDEX 475 SUBJECT

INDEX

Abstract schemas, 155, 156 Action capabilities, 228, 230, 231, 266, 267, 272 coordination, 147, 150, 152, 154, 155 knowledge, 142 schema, 62, 147, 148, 162, 163, 438 systems, 5, 9, 12, 53, 54, 56, 57, 60, 221, 223, 227, 228, 232, 235, 350, 412, 413, 432, 433 Active intermodal mapping, 78 Adualistic, 73, 350 Adualistic confusion, 36, 37, 39, 350 Affective configurations, 423, 429 Affordances, 5, 7, 12, 26, 27, 31, 58, 62, 130, 177, 263, 267, 318, 327, 330, 342, 349, 356, 397, 433, 435, 438, 439, 443 Agency, 5, 8, 23, 31, 40, 63, 65, 118, 122, 125, 165, 173, 174, 176, 180, 184, 185, 186, 318, 327, 350, 426, 428, 436, 437, 439, 445, 458 Aggression, 203, 204, 205, 206, 210 Aggressive intervention, 203, 206 Alliances, 202, 207, 208 Allocentric search, 337, 342 Altruism, 163 Anticipation skills, 37, 58, 129, 209, 269, 271 Ants, 197, 208 Arboreal locomotion, 198, 199 Arm movements, 58, 264, 265, 267, 268, 269, 272, 336, 384, 400, 426 Auditory localization, 22 Autistic children, 329, 332, 457, 458 Autobiographical memory, 168, 186 Automata, 3, 12, 13, 57, 74 Behavioral constellation, 421, 424 Beliefs, 19, 97, 99, 100, 103, 106, 108, 161, 222, 352, 353 Blindness, 22, 209, 268, 329, 330, 331, 332, 333, 335, 336, 337, 339, 340, 341, 342 Blindsight, 76, 105, 306 Blinking, 41, 55, 381, 384

Body centered, 335, 336, 339, 342, 343 effectivities, 57, 58, 59, 60, 152, 153 gravity, 198 image, 3, 4, 303, 410 movement, 22, 65, 76, 78, 81, 82, 170, 228, 229, 231, 232, 233, 234, 235, 278, 313, 400 orientation, 63, 223, 231, 342, 424 position, 221, 222. 223, 227. 231, 232, 234, 235, 236, 278, 285, 336 posture, 81, 226, 227. 228, 281, 362 representation, 154, 155 scaled information, 7 schema, 75, 87, 142, 154, 407, 410, 411, 412, 413 Body effectivities, 413 Capuchin monkeys, 190, 199, 246, 254 Caregivers, 187, 432, 433, 434, 435, 436, 437, 438, 439, 440, 443, 444 Cartesian dualism, 210 Cerebral palsy, 106, 267, 270 Chimpanzees, 4, 30, 47, 162, 169, 170, 171, 172, 173, 178, 179, 180, 185, 207, 208, 209, 243, 244, 246, 247, 248, 249, 250, 251, 254, 310, 457 Circular reactions, 62, 63, 65, 109,377, 384, 386, 396 Coalition formations, 203 Cognitive ethologist, 193, 194 Conceptual primitives, 145, 146, 152 schemas, 145, 146 self, 18, 19, 20, 25, 26, 60, 125, 142, 195, 243, 254, 255, 330, 351, 450 Connectionist theory, 375, 377 Consciousness, 3, 17, 18, 25, 35, 36, 39, 40, 41, 43, 44, 45, 46, 47, 48, 143, 145, 147, 153, 157, 308, 309, 310, 369, 453 Constructive mirror, 54

476

SUBJECTINDEX

Contingency perception, 375, 376, 379, 384, 399 Contingent, 9, 10, 11, 12, 46, 47, 61, 83, 130, 131, 132, 170, 171, 177, 179, 180, 184, 187, 224, 225, 226, 246, 248, 311, 328, 330, 354, 356, 357, 358, 359, 360, 361, 376, 379, 380, 383, 387, 399, 401, 422 Continuous change, 262 process, 119, 259, 264, 271, 272 Control parameter, 262, 263, 287 Coordination, 36, 38, 39, 58, 75, 76, 128, 147, 148, 149, 150, 151, 152, 153, 200, 223, 259, 260, 261, 281, 282, 286, 287, 306, 315, 319, 342, 386, 387, 398, 411, 412 Courting, 201 Crawling, 26, 28, 227, 318, 319, 332, 333, 337, 339, 340, 442 Cultural learning, 453 Culture, 18, 19, 101, 124, 177, 428, 431, 432, 434, 437, 445, 446 Deception, 96, 453, 456 Desires, 96, 163, 165, 166, 173, 176, 180, 254, 329, 336, 343, 427 Dialogical processes, 119 Differentiation, 5, 6, 37, 38, 54, 55, 68, 74, 79, 96, 99, 103, 104, 108, 162, 209, 258, 295, 353, 354, 355, 360, 375, 380, 382, 387, 391, 403, 404, 411, 419, 425, 427, 435, 455, 456, 458 Direct perception, 37, 38, 41, 43, 117, 195, 259, 263, 264, 269, 271 Discontinuous process, 257 Discovery, 8, 10, 12, 57, 60, 62, 64, 118, 153, 282, 329, 341, 369, 395, 400 Dolphins, 197 Double touch, 129, 398, 412 Dynamic systems, 120, 128, 134 Ecological self, 17, 18, 20, 21, 22, 23, 24, 25, 26, 28, 29, 30, 31, 32, 37, 39, 48, 53, 54, 57, 59, 60, 61, 62, 64, 67, 68, 69, 125, 142, 221, 222, 226, 235, 330, 332, 340, 341, 342, 343, 397, 436, 445

Effectiveness, 8, 30, 46, 57, 343 Ego, 4, 22, 37, 45, 108, 360. 361, 403, 404, 405, 408, 409 Egocentrism, 212, 225, 377 Emotions, 10, 18, 23, 37, 41, 43, 44, 46, 54, 56, 60, 61, 62, 63, 64, 65, 66, 67, 68, 95, 104, 105, 107. 117, 122, 123, 126, 131, 133, 134, 163, 195,327, 331, 375, 422. 426, 438, 450, 452 Existential self, 319, 350, 382 Expectations, 5, 10, 12, 61, 130, 194, 328, 336, 387, 402, 422, 437 Exploration actions, 8, 266, 319, 349, 395, 443 function, 265 Exteroceptive information, 303 Featural analysis, 145 Field of promoted action, 435, 437, 440, 441, 442, 443, 445 Figurative instruments, 148 Flow, 6, 22, 24, 27, 28, 31, 39, 40, 43, 123, 127, 128, 142, 150, 229, 265, 286, 287, 289, 291, 293, 294, 295, 296, 312, 313, 314. 315, 316, 317, 340, 349, 399, 422, 423, 424, 428 Frames, 48, 65, 66, 120, 124, 127, 128, 129, 130, 132, 133, 134, 156, 267, 433, 435, 437, 444, 445 Free play, 421, 424, 427 Functionalism, 3 Games, 65, 80, 120, 432, 437, 441, 444 Goal orientation, 54, 55, 387 directed action, 387, 423, 425 Gorillas, 30, 244 Gravity, 24, 198, 265, 278, 279, 285, 286, 287, 295, 381 Head-turning, 9, 163, 285 Homing behavior, 196, 197, 198

SUBJECT INDEX 477 Id, 98, 108 Idea of "me," 53, 62, 66, 67, 97, 98, 102, 103, 104, 105, 107, 109, 110, 113, 195 Identification, 79, 142, 201, 303, 305, 419, 422, 426, 454, 456, 457, 458 Identified self, 47, 350, 351 Identity, 4, 47, 80, 81, 82, 101, 107, 118, 121, 122, 123, 124, 134, 172, 175, 352, 390 Image of the body, 3 Image schemas, 145, 146, 150, 151, 155, 156 Imitation, 20, 39, 41, 43, 56, 57, 64, 65, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 87, 88, 89, 90, 106, 133, 157, 169, 175, 187, 195, 351, 411, 412, 422, 426, 437, 454 Imitative learning, 451, 455 Independent locomotion, 37, 40, 229, 262, 287, 318, 442 Independent walking, 259, 262, 263 Infantile hospitalism, 65 Initial sense of self, 74, 342 Innate structure, 73 Instrumental conditioning, 8, 9 learning, 63 Intentional deception, 209 mapping, 76 Intentional communication, 451, 455 Intentionality, 5, 8, 9, 23, 54, 65, 98, 100, 102, 110, 112, 143, 194, 455, 458 Interactive intentions, 424 Intermodal contingency, 11 Interpersonal relations, 41, 81, 119, 121, 122, 450 selves, 17, 18, 20, 21, 32, 37, 41, 43, 66, 90, 125, 222, 330, 331, 332, 342, 420, 445, 450, 457, 458 Interposition, 206 Interpresonal competencies, 419 Intersubjectivity, 43, 44, 66, 95, 109, 118, 130, 195, 453 Involuntary reflexes, 258

Japanese macaque, 202, 205, 206 Joint attention, 20, 45, 330, 331, 451, 455 Kinesthetic, 6, 12, 18, 29, 37, 38, 171, 178, 187, 222, 223, 224, 227, 232, 245, 246, 303, 397, 399, 412 Labyrinth, 278, 279, 280, 281, 285 Leaping, 199 Limb position, 223, 224 Locomotion, 12, 21, 22, 27, 37, 38, 40, 128, 198, 199, 223, 229, 230, 231, 235, 261, 262, 287, 314, 318, 333, 337, 339, 340, 342, 351, 419, 442, 443 Locomotor experience, 229, 314 Machine movements, 85 Machinery of self, 97, 104, 105, 106 Mark test, 68, 171, 174, 177, 178, 179, 187, 243, 244, 245, 246 Matching mechanism, 77 Matriline, 203, 206, 212 Maturation, 97, 102, 111, 258, 263, 266, 275, 276, 283, 285, 296 Means and ends, 455, 456, 457, 458 Mental states, 48, 95, 98, 100, 102, 103, 104, 106, 107, 108, 109, 110, 111, 112, 164, 165, 166, 168, 173, 175, 185 Micro-regulations, 421, 424, 425 Mirror image, 4, 8, 30, 311, 312, 354, 355, 356, 360, 365, 368, 369, 391, 400, 401 reflection, 4, 357 self-recognition, 3, 4, 30 Modular organization, 304 Modularity, 386, 387 Monologues, 60, 61, 63, 187 Motives, 103, 424, 426, 427 Motor program, 261, 388 Motor skills, 258, 266, 269, 271, 332, 333, 339, 342 Mouthing, 8, 9, 386, 387, 388, 389, 390, 394 Movement coordination, 260, 262

478

SUBJECTINDEX

Movements, 5, 6, 8, 21, 22, 25, 27, 28, 29, 30, 31, 38, 41, 44, 46, 56, 58, 62, 63, 64, 65, 76, 78, 79, 81, 82, 85, 86, 87, 88, 89, 128, 131, 142, 169, 170, 171, 179, 187, 196, 197, 198, 200, 210, 224, 225, 226, 227, 228, 229, 230, 235, 246, 250, 254, 258, 259, 263, 266, 267, 271, 272, 275, 277, 278, 282, 284, 287, 288, 290, 296, 306, 308, 310, 312, 313, 314, 316, 328, 329, 330, 331, 341, 357, 358, 360, 361, 362, 363, 376, 379, 380, 384, 386, 398, 399, 400, 401, 402, 403, 404, 405, 406, 407, 409, 410, 411, 412, 426, 432, 433 Moving room, 39, 40, 43, 227, 229, 287, 314, 315, 316 Multimodal, 6, 12, 61, 290, 436 Multiperson situation, 80 Mutuality, 43, 74, 87 Narrative, 18, 19, 101, 117, 118, 119, 120, 121, 122, 125, 126, 129, 130, 131, 132, 133, 134, 222, 428 Nativist concept, 258 Natural selection, 184, 304, 457 Neural maturation, 258 New World monkeys, 199 Nonequilibrium, 264 Nonhuman, 30, 53, 120, 129, 162, 169, 171, 172, 193, 196, 202, 204, 209, 243, 244, 245, 252, 253, 254, 255, 317 Nonself, 68, 73, 95, 129, 278, 329, 376, 377, 382, 395, 396, 397, 413, 420 Nonverbal communication, 120 Nonvisual exploration, 271 Notion of self, 73, 74, 75, 76, 81, 87, 88, 89, 90, 100, 260, 318 Object concept, 146, 356 Object permanence, 149, 163, 177, 336, 337, 339 Objectification, 53, 54, 60, 61, 62, 64, 65, 66, 67, 391

Objective, 37, 41, 97. 98. 104. 105. 106, 108, 109, 110. 111. 119. 124. 149, 153, 172, 252, 278, 310, 349, 377, 382, 388, 391, 426. 428 Objective self, 98 Optic array, 6, 22, 40, 265, 272, 286, 307, 340, 351 Optic flow sensitivity, 296 Optical flow, 142, 229, 287, 293, 294, 312, 313, 314, 315, 316, 317 Oral exploration, 23 Orangutans, 4, 30, 47. 169. 185, 199. 244, 310 Others, 3, 5, 6, 7, 12, 29. 30. 32. 43, 44, 45, 48, 54, 64, 65, 66. 67, 68, 74, 76, 79, 80, 81, 83, 85, 87, 88, 89, 96, 97, 100, 101, 102, 105, 106, 107, 108, 109, 110, 112, 122, 124, 126, 130, 131, 133. 134, 156, 171, 172, 173, 175. 177. 184. 185, 187, 200, 201, 206, 207, 208, 209, 212, 222, 318, 329, 330, 331. 332, 342, 343, 354, 355, 360, 367, 369, 383, 397, 400, 402, 411, 419, 420, 421, 424, 427, 428, 437. 438, 439, 440, 442, 444, 445, 451, 452. 453, 454, 455, 456, 457 Own agency, 318, 437 body space, 339 perspective, 7 Pain, 25, 26, 382 Parallax, 200 Path calculations, 208 Perceived self, 12, 24, 61, 295, 425 Perceptual abilities, 44, 48, 59, 245, 327, 332 motor coupling, 79, 88, 125 schema, 145, 146, 155 Perfect contingency, 61, 176, 186, 358, 360, 365, 366, 368, 383, 384 Phase transition, 264 Phylogenetic construction, 153 Postmodern perspective, 117, 118 Postural abilities, 296 adjustments, 130, 282, 295, 381, 399

SUBJECT INDEX 479 Postural (continued) development, 8 equilibrium, 279, 313, 316, 318 Postures, 8, 81, 200, 227, 228, 290, 313, 362, 432 Powers, 7, 12, 31, 113, 444 Preconceptual self, 142, 243, 254, 255 Prediction, 77, 169, 179, 180, 181, 182, 204, 208, 272, 307, 366, 409 Prehensile space, 59, 413 Preverbal, 17, 75, 87, 90, 120, 193, 194, 211, 368 Primate society, 203 Proper self, 162, 167, 181 Proprioception, 22, 24, 38, 39, 40, 82, 90, 130, 286, 287, 296, 303, 304, 309, 313, 317, 318, 319, 354, 358, 402, 445, 449 Prospective control, 271, 272, 275 Proto-narrative envelope, 425, 426 Protoconversation, 428, 450, 451 Proximal development, 331 Purposive, 21, 30, 31, 87, 89, 440 Reaching, 6, 26, 29, 38, 57, 58, 59, 79, 128, 172, 173, 175, 198, 202, 258, 263, 264, 266, 269, 270, 271, 272, 273, 274, 275, 277, 282, 288, 304, 306, 308, 312, 319, 329, 332, 333, 335, 336, 337, 340, 341, 342, 351, 370, 375, 381, 384, 385, 387, 390, 411, 413, 419, 427, 434, 438, 449 Reciprocity, 54, 64, 65, 67, 436, 438, 440, 443, 449 Redirection, 204, 205, 206 Reflexes, 55, 77, 81, 107, 155, 257, 258, 281, 284, 285 Reflexive abstraction, 144, 147, 150, 152 Relational self, 122, 125, 126, 133 Representation, 3, 5, 18, 37, 74, 79, 88, 124, 125, 126, 127, 128, 130, 145, 162, 163, 164, 165, 166, 167, 168, 173, 174, 175, 176, 178, 180, 183, 185, 186, 245, 261, 283, 303, 304, 305, 306, 307, 308, 310, 311, 319, 330, 343, 351, 354, 356, 378, 382, 387, 388, 390, 410

Search in the dark, 337 Self avoidance, 384 aware, 194, 195, 210, 211 awareness, 3, 4, 9, 17, 25, 27, 28, 30, 35, 36, 40, 46, 48, 53, 54, 60, 64, 65, 66, 67, 68, 69, 78, 97, 99, 104, 105, 108, 109, 110, 111, 172, 193, 194, 221, 243, 245, 246, 350, 352, 393 concept, 4, 18, 37, 161, 162, 165, 166, 167, 168, 172, 174, 178, 179, 185, 186, 187, 221, 235, 245, 349, 369, 375, 382, 391, 450, 451, 453, 456, 457, 458 conscious, 133, 163, 310 consciousness, 25, 35, 36, 39, 40, 45, 46, 48, 53, 60, 65, 66, 67, 68, 69, 221, 310, 453 decection, 375, 381, 382, 383, 391 exploration, 31, 38, 62, 129, 170, 171, 175, 179, 320, 395, 396, 397, 400, 401, 402, 407, 4O9 knowledge, 17, 18, 19, 21, 25, 30, 32, 35, 36, 37, 40, 41, 53, 58, 59, 60, 64, 65, 66, 69, 141, 142, 144, 146, 155, 161, 173, 179, 180, 187, 221, 222, 223, 226, 227, 228, 231, 232, 234, 235, 236, 245, 309, 318, 330, 332, 350, 352, 353, 354, 355, 368, 369, 387, 388 motion, 128, 142, 232, 279, 280, 283, 286, 289, 295, 303, 312, 313, 314, 315, 316, 354, 360, 399 organization, 128, 131, 277 orientation, 284, 375, 381, 382, 383, 384, 385, 391 perception, 7, 27, 28, 37, 39, 187, 188, 221, 243, 246, 278, 283, 304, 308, 309, 312, 319, 320, 349, 350, 351, 352, 353, 356, 357, 358, 400, 410, 449

480

SUBJECTINDEX

Self (continued) recognition, 3, 4, 46, 47, 68, 75, 76, 81, 95, 133, 141, 161, 162, 163, 168, 169, 170, 171, 172, 173, 174, 175, 177, 179, 182, 186, 187, 193, 196, 210, 221, 243, 244, 245, 246, 309, 310, 311, 312, 317, 320, 354, 355, 356, 357, 362, 365, 366, 367, 369, 400, 457 reference, 141, 193, 194, 195, 201, 382 representation, 162 seeking, 384, 385 system, 98, 99, 100, 101, 102, 110, 125 Selfless, 73 Sensory-motor schemes, 386, 454 Separating intervention, 206, 207, 2O8 Sitting, 59, 102, 180, 182, 207, 208, 224, 231, 288, 316, 317, 319, 332, 413 Situated self, 350, 351 Smiles, 10, 63, 76, 98, 163, 357, 400, 423, 439, 440, 443 Social agent, 21, 441, 442, 444, 449, 450, 458 field, 202, 204 interaction, 10, 18, 19, 41, 43, 44, 54, 64, 65, 73, 195, 202, 222, 328, 331, 357, 436, 440, 441, 443, 450, 453, 456 mirror, 54, 64, 67, 82 referencing, 331, 451, 452, 455 self, 18, 450, 452, 453 Socio-affective exchanges, 419, 420, 421, 424, 426 Solipsistic, 36, 126, 377, 388 Somatic information, 8 Somatosensory, 9, 279, 283, 303 Spatial awarness, 333, 339, 340 congruence, 361, 367, 402, 410 knowledge, 208, 339, 341 Spatial awareness, 23 Standing, 120, 202, 227, 228, 231, 257, 312, 313, 317, 332 Structural congruence, 83

Structure, 5, 11, 22, 73, 83. 97, 101, 125, 148, 149, 150. 152, 153, 154. 155, 186, 194, 272, 280, 282. 286, 304, 312, 378, 379. 387. 425, 426, 431, 436, 438, 440 Subjective self, 98, 104. 111. 382 Sucking, 9, 23, 56, 62, 63. 77, 149, 150, 357, 387 Superego, 98 Symbolic play, 163, 329 Theory of mind, 4, 48, 185, 187. 209, 453 Timing, 67, 168, 194, 228, 263, 269, 270, 271, 422, 441 Tool 7, 9, 25, 57, 81, 123, 125, 126, 162, 202, 243, 356, 376 Touch, 4, 10, 11, 24, 25, 28. 36, 39, 59, 107, 123, 128, 129, 169, 172, 175, 266, 306, 310, 312. 328, 333, 336, 337, 343, 351, 352, 356. 379, 385, 386, 396, 397, 398. 407, 412, 433 Triadic awareness, 202, 204, 212 interaction, 442, 443, 444 social relationships, 202 Umwelt, 197, 210 Unconsciousness, 98. 103 Undifferentiation, 54, 55 Unduplicated self, 161. 162 Upright orientation, 227. 228 Verbal mediation, 193 Vestibular sensory system. 278, 279, 286 Vision, 22, 24, 36, 38, 39. 88. 123. 197, 268, 271, 278, 280, 286. 288, 296, 303, 304, 312, 313, 327, 328. 329, 330, 331, 333, 340, 341, 351. 354, 388, 400, 410, 435 Visual deficit, 268 proprioception, 39. 40. 268. 286, 287, 289, 304, 317. 318, 319, 358, 365, 367. 400, 401, 402, 407, 409, 410, 411

SUBJECT INDEX 481 Visual (Continued) system, 8, 45, 278, 284, 287, 304, 306, 307, 327, 328 Visuo-postural coupling, 293, 296 Visuomotor program, 319 Vitality affects, 61, 425, 426 Vocalizations, 10, 44, 349, 424, 436, 440, 443, 444 Walking, 26, 157, 198, 229, 230, 231, 257, 259, 262, 263, 267, 271, 332, 333, 337, 339, 340, 342

This Page Intentionally Left Blank

Advances in Psychology Volume 112 The Self in Infancy: Theory and Research

Advances in Agronomy, Volume 112

Progress in infancy Research: Volume 2 (Volume in the Progress in Infancy Research Series)

Advances in Psychology Research Volume 44

Advances in Psychology Research Volume 44

Advances in Psychology Research volume 80

Advances in Psychology Research. Volume 72

Advances in Psychology Research Volume 57

Advances in Psychology Research Volume 57

Progress in infancy research, Volume 1

Advances in Research and Theory

Constructing the Self (Advances in Consciousness Research, Volume 73)

Advances in Virus Research, Volume 46 (Advances in Virus Research)

Advances in Virus Research, Volume 54 (Advances in Virus Research)

Advances in Virus Research, Volume 47 (Advances in Virus Research)

Advances in Virus Research, Volume 47 (Advances in Virus Research)

Advances in Virus Research, Volume 58 (Advances in Virus Research)

Research Advances in Rabies (Advances in Virus Research, Volume 79)

Self-Reference and Self-Awareness (Advances in Consciousness Research)

The Psychology of Learning and Motivation, Volume 53: Advances in Research and Theory

The Psychology of Learning and Motivation, Volume 54: Advances in Research and Theory

The Psychology of Learning and Motivation, Volume 51: Advances in Research and Theory

The Psychology of Learning and Motivation : Advances in Research and Theory, Volume 31

Perceptual Learning (The Psychology of Learning and Motivation: Advances in Research and Theory, Volume 36)

Advances in Psychology Volume 40 Current Issues in Theoretical Psychology

Media effects: advances in theory and research

Advances in Psychology Volume 20 Trends in Mathematical Psychology

Advances in Imaging and Electron Physics (Volume 112) (Advances in Imaging and Electron Physics)

Advances in Psychology Research, Vol. 78

Psychology of Learning and Motivation, Volume 39: Advances in Research and Theory

Psychology of Learning and Motivation, Volume 44, Advances in Research and Theory

Advances in Psychology Volume 112 The Self in Infancy: Theory and Research

Advances in Agronomy, Volume 112

Progress in infancy Research: Volume 2 (Volume in the Progress in Infancy Research Series)

Advances in Psychology Research Volume 44

Advances in Psychology Research Volume 44

Advances in Psychology Research volume 80

Advances in Psychology Research. Volume 72

Advances in Psychology Research Volume 57

Advances in Psychology Research Volume 57

Progress in infancy research, Volume 1

Advances in Research and Theory

Constructing the Self (Advances in Consciousness Research, Volume 73)

Advances in Virus Research, Volume 46 (Advances in Virus Research)

Advances in Virus Research, Volume 54 (Advances in Virus Research)

Advances in Virus Research, Volume 47 (Advances in Virus Research)

Advances in Virus Research, Volume 47 (Advances in Virus Research)

Advances in Virus Research, Volume 58 (Advances in Virus Research)

Research Advances in Rabies (Advances in Virus Research, Volume 79)

Self-Reference and Self-Awareness (Advances in Consciousness Research)

The Psychology of Learning and Motivation, Volume 53: Advances in Research and Theory

The Psychology of Learning and Motivation, Volume 54: Advances in Research and Theory

The Psychology of Learning and Motivation, Volume 51: Advances in Research and Theory

The Psychology of Learning and Motivation : Advances in Research and Theory, Volume 31

Perceptual Learning (The Psychology of Learning and Motivation: Advances in Research and Theory, Volume 36)

Advances in Psychology Volume 40 Current Issues in Theoretical Psychology

Media effects: advances in theory and research

Advances in Psychology Volume 20 Trends in Mathematical Psychology

Advances in Imaging and Electron Physics (Volume 112) (Advances in Imaging and Electron Physics)

Advances in Psychology Research, Vol. 78

Psychology of Learning and Motivation, Volume 39: Advances in Research and Theory

Psychology of Learning and Motivation, Volume 44, Advances in Research and Theory

Recommend Documents