Advances in the Study of Behavior, Volume 6

Advances in

THE STUDY OF BEHAVIOR VOLUME 6

Contributors to This Volume P. P. G. BATESON BENNETT G. GALEF, JR. SARAH BLAFFER HRDY J. B. HUTCHISON PAUL ROZIN GEORGE N. WADE

Advances in

THE STUDY OF BEHAVIOR Edited by JAY S. ROSENBLATT Institute of Animal Behavior Rutgers University Newark, New Jersey

ROBERT A. HINDE Medical Research Council Unit on the Development and Integration of Behavior University Su b-Department of Animal Behavior Madingley, Chmbridge, England

EVELYN SHAW Department of Biological Sciences Stanford University Stanford, California

COLINBEER Institute of Animal Behavior Rutgers University Newark, New Jersey

VOLUME 6

ACADEMIC PRESS

New York San Francisco London 1976 A Subsidiary of Harcourt Brace Jovanovich, Publishers

COPYRIGHT 0 1976, BY ACADEMIC PRESS,INC. ALL RIGHTS RESERVED. NO PART O F THIS PUBLICATION MAY BE REPRODUCED OR TRANSMITTED IN ANY FORM OR BY ANY MEANS, ELECTRONIC OR MECHANICAL, INCLUDING PHOTOCOPY, RECORDING, OR ANY INFORMATION STORAGE AND RETRIEVAL SYSTEM, WITHOUT PERMISSION IN WRITING FROM THE PUBLISHER.

ACADEMIC PRESS, INC.

11 1 Fifth Avenue, New York, New York 10003

Wniled Kingdom Edition published by ACADEMIC PRESS, INC. (LONDON) LTD. 24/28 Oval Road. London N W l

LIBRARY OF CONGRESS CATALOG CARD NUMBER: 64-8031 ISBN 0- 12-004506-0 PRINTED IN THE UNITED STATES O F AMERICA

Contents

..................................... Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contents of Previous Volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of Contributors.

ix

xi xiii

Specificity and the Origins of Behavior P. P. G . BATESON

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

1 2

Determinants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. The Nature of “Relevant” Experience, . . . . . . . . . . . . . . . . . . . V. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8 12 17 18

I. Introduction..

11. Initial Determinants in the Development of Behavior 111. Classification of Behavior in Terms of Developmental

The Selection of Foods by Rats, Humans, and Other Animals PAUL ROZIN

I. Solutions to the Food Selection Problem . . . . . . . . . . . . . . . . . 11. Rats: An Example of Successful Generalists . . . . . . . . . . . . . . . 111. Food Selection in Humans . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21 27 52 67

Social Transmission of Acquired Behavior: A Discussion of Tradition and Social Learning in Vertebrates BENNETT G. GALEF, JR.

I. Introduction.,

.................................. V

77

vi

CONTENTS

I1. I11. I v. V.

Field and Associated Laboratory Studies . . . . . . . . . . . . . . . . . Learning and Conditioning Paradigms . . . . . . . . . . . . . . . . . . . Problems of Terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

81 88 92 95 97

Care and Exploitation of Nonhuman Primate Infants by Conspecifics Other than the Mother SARAH BLAFFER HRDY I. I1. I11. IV V. VI

. .

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Male Care vs . Exploitation of Infants . . . . . . . . . . . . . . . . . . . . Nurture vs . Abuse-Male and Female Roles . . . . . . . . . . . . . . . . The Pros and Cons of Aunting . . . . . . . . . . . . . . . . . . . . . . . . Selective Pressures on the Infant ....................... Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

101 104 118 120 142 148 150

Hypothalamic Mechanisms of Sexual Behavior. with Special Reference to Birds J . B. HUTCHISON Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Localized Steroid Effects in the Brain .................... Biochemical Factors in Androgen Action . . . . . . . . . . . . . . . . . Variable Hypothalamic Sensitivity to Androgen . . . . . . . . . . . . . Hypothalamic Androgen Concentration and the Structure of Courtship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I. I1. I11. IV . V.

159 160 165 173 185 190 194

Sex Hormones. Regulatory Behaviors. and Body Weight GEORGE N .WADE I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1. Influence of Sex and Reproductive Condition . . . . . . . . . . . . . . 111. Activating Effects of Sex Hormones: Gonadectomy and Replacement Therapy in Adults ....................... IV . Site and Mechanism of Action of Estradiol and Progesterone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

201 203

207 215

CONTENTS

V . Development of Responsiveness to Ovarian Steroids and Effects of Lactation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Sex Differences in Neuroendocrine Regulation of Body Weight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII . Hormonal Effects on Taste Preferences and Dietary Self-Selection ................................... VIII . Hormones and Weight Regulation in Nonrat Species . . . . . . . . . . IX . Conclusions and Directions for Future Research . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Subject Index

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

vii

237 243 253 260 264 267

281

This Page Intentionally Left Blank

List of Contributors Numbers in parentheses indicate the pages on which the authors' contributions begin.

P. P.G. BATESON, Sub-Department of Animal Behaviour, University of Cambridge, Cambridge, England ( 1 ) BENNETT G. GALEF, Jr., Department of Psychology, McMaster University, Hamilton, Ontario, Canada ( 7 7 ) SARAH BLAFFER HRDY, Peabody Museum, Harvard University, Cambridge, Massachusetts ( 1 01) J.B. HUTCHISON, MRC Unit on the Development and Integration of Behaviour, University Su b-Department, Madingley, Cam bridge, England ( 159) PAUL ROZIN, Department of Psychology, University of Pennsylvania, Philadelphia, Pennsylvania (21) GEORGE N. WADE,Department of Psychology, University of Massachusetts, Amherst, Massachusetts (201)

ix

This Page Intentionally Left Blank

Preface The study of animal behavior is attracting the attention of ever-increasing numbers of zoologists and comparative psychologists in all parts of the world, and is becoming increasingly important to students of human behavior in the psychiatric, psychological, and allied professions. Widening circles of workers, from a variety of backgrounds, carry out descriptive and experimental studies of behavior under natural conditions, laboratory studies of the organization of behavior, analyses of neural and hormonal mechanisms of behavior, and studies of the development, genetics, and evolution of behavior, using both animal and human subjects. The aim of Advances in the Study of Behavior is to provide workers on all aspects of behavior an opportunity to present an account of recent progress in their particular fields for the benefit of other students of behavior. It is our intention to encourage a variety of critical reviews, including intensive factual reviews of recent work, reformulations of persistent problems, and historical and theoretical essays, all oriented toward the facilitation of current and future progress. Advances in the Study of Behavior is offered as a contribution to the development of cooperation and communication among scientists in our field.

xi

This Page Intentionally Left Blank

Contents of Previous Volumes

Volume 1 Aspects of Stimulation and Organization in ApproachlWithdrawal Processes Underlying Vertebrate Behavioral Development T. C. SCHNEIRLA Problems of Behavioral Studies in the Newborn Infant H. F. R. PRECHTL The Study of Visual Depth and Distance Perception in Animals RICHARD D. WALK Physiological and Psychological Aspects of Selective Perception GABRIEL HORN Current Problems in Bird Orientation KLAUS SCHMIDT-KOENIG Habitat Selection in Birds P. H. KLOPFER and J. P. HAILMAN Author Index-Subject Index

Volume 2 Psychobiology of Sexual Behavior in the Guinea Pig WILLIAM c. YOUNG Breeding Behavior of the Blowfly V. G. DETHIER Sequences of Behavior R. A. HINDE and J. G. STEVENSON The Neurobehavioral Analysis of Limbic Forebrain Mechanisms: Revision and Progress Report KARL H. PRIBRAM

xiii

XiV

CONTENTS OF PREVIOUS VOLUMES

Age-Mate or Peer Affectional System HARRY F. HARLOW Author Index-Subject Index

Volume 3 Behavioral Aspects of Homeostasis D. J. McFARLAND Individual Recognition of Voice in the Social Behavior of Birds C. G. BEER Ontogenetic and Phylogenetic Functidns of the Parent-Offspring Relationship in Mammals LAWRENCE V. HARPER The Relationships between Mammalian Young and Conspecifics Other Than Mothers and Peers: A Review Y. SPENCER-BOOTH Tool-Using in Primates and Other Vertebrates JANE van LAWICK-GOODALL Author Index-Subject Index

Volume 4 Constraints on Learning SARA J. SHETTLEWORTH

.

Female Reproduction Cycles and Social Behavior in Primates T. E. ROWELL The Onset of Maternal Behavior in Rats, Hamsters, and Mice: A Selective Review ELIANE NOIROT Sexual and Other Long-Term Aspects of Imprinting in Birds and Other Species KLAUS IMMELMA" Recognition Processes and Behavior, with Special Reference to Effects of Testosterone on Persistence R. J. ANDREW Author Index-Subject Index

CONTENTS OF PREVIOUS VOLUMES

Volume 5 Some Neuronal Mechanisms of Simple Behavior KENNETH D. ROEDER The Orientational and Navigational Basis of Homing in Birds WILLIAM T. KEETON The Ontogeny of Behavior in the Chick Embryo RONALD W. OPPENHEIM Processes Governing Behavioral States of Readiness WALTER HEILIGENBERG Time-sharing as a Behavioral Phenomenon D. J. McFARLAND Male-Female Interactions and the Organization of Mammalian Mating Patterns CAROL DIAKOW Author Index-Subject Index

xv

This Page Intentionally Left Blank

Specificity and the Origins of Behavior P.P.G.BATESON SUB-DEPARTMENT OF ANIMAL BEHAVIOUR UNIVERSITY OF CAMBRIDGE CAMBRIDGE, ENGLAND

I.

Introduction

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

............ 111. Classification of Behavior in Terms of Developmental Determinants . . . . Iv. The Nature of “Relevant” Experience ..................... V. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Initial Determinants in the Development of Behavior.

I.

1

2 8 12 17 18

INTRODUCTION

What factors during development determine the special ways in which an individual animal eventually will behave? What decides the specific form and patterning of its behavior? What gives a behavior pattern its unique character, making it different from other behavior patterns? It would be useless to pretend that the attempts to answer these questions about the ontogeny of behavior bring widespread agreement. Nor is there harmonious consensus among those who study behavior as to the ways these questions should be answered or even about the nature of relevant evidence. The debate about the best ways to study behavioral development has, of course, been extensive (see Barnett, 1973; Beach, 1955; Dawkins, 1968; EiblEibesfeldt, 1961, 1970; Ewer, 1971; Hailman, 1967; Hebb, 1953; Hinde, 1968, 1970a; Jensen, 1961; Konishi, 1966; Kuo, 1967; Lehrman, 1953, 1970; Lehrman and Rosenblatt, 1971; Lorenz, 1961, 1965; Moltz, 1965; Schneirla, 1956, 1966; Thorpe, 1956, 1963; Tinbergen, 1963). It would be quite wrong to suggest that nothing has been achieved as a result of this debate. In particular, many of the disagreements have been shown to arise from differences in interest and emphasis. Those ethologists influenced by Lorenz have been primarily interested in the origins of behavioral adaptiveness, whereas others studying behavior, par-

2

P.P. G . BATESON

ticularly those who were influenced by the writings of Kuo, Schneirla, and Lehrman, have been principally concerned with development in the individual animal. Even though this point was clarified many years ago (eg., Tinbergen, 1963), the controversy has rumbled on. Despite frequent announcements of the death of the natureimrture dichotomy of behavior, a distinction between activities that are learned and those that are not is still widely used. In part this has been because classifications of the origins of behavior have been frequently muddled with classifications of behavior itself. To state that inheritance and the environment determine the characteristics of behavior is not the same as urging that all behavior patterns can be divided into those that are inborn and those that are environmentally determined. As I shall point out later, a residual confusion between the sources of behavioral distinctiveness and the origins of its adaptiveness is still found in the literature. I believe, however, that the reasons for the persisting, wide and often bitter differences of viewpoint are much more deeply seated than could be explained by mere errors of logic. In this chapter the possibility is explored that different people perceive the same body of data in different ways. Where some see sharp discontinuities, others see smooth gradations, and, accordingly, classifications differ. In order to develop the argument, I shall first consider factors in development that are responsible for the distinctiveness of behavior. 1 believe that when these sources of difference are scrutinized, it becomes much easier to understand why classifications of behavior in terms of developmental origins have generated so much heated argument.

11. INITIAL DETERMINANTS IN THE DEVELOPMENT OF BEHAVIOR It is customary now to distinguish between the factors that control behavior from moment to moment and those that are responsible for its development (e.g., Hinde, 1970a). The distinction may not always be easily drawn in practice since a factor responsible for the development of a behavior pattern may lie close in time to the occurrence of that behavior. In general, though, sources of behavioral distinctiveness usually lie considerably farther back in time from the behavior they affect than controlling conditions. Developmental determinants are initiating agents that lastingly give a behavior pattern its peculiar characteristics differentiating it from other types of behavior; of course, a lasting effect is not necessarily irreversible under all conditions. Once one starts to trace back through the nexus of events that precede a behavior pattern, there might seem no obvious stopping point. However, what is usually meant by a developmental determinant of an individual’s behavior is a factor that was responsible for the distinctiveness of the individual’s behavior

SPECIFICITY AND THE ORIGINS OF BEHAVIOR

3

TABLE I A CLASSIFICATION OF DEVELOPMENTAL DETERMINANTS OF BEHAVIOR Determinants with specific effects

Determinants with general effects

Inherited

A

B

Environmental

C

D

Determinants

and which operated at some point in the life of that individual. Wherever I refer to “determinant” in this chapter I use it in this special sense. Few people would disagree nowadays that part of the initial determinants of behavior are already present in latent form within the fertilized egg; some determinants are, perhaps, present as cytoplasmic factors, but most are represented in the nucleus of the zygote-presumably in genetically coded form. An important semantic issue is at what stage a gene is to be treated as a developmental determinant. I believe a gene would generally be regarded as a determinant at the time of its activation. However, t o discover the actual moment when gene expression occurs for the first time is an extraordinarily difficult task for embryology, and most statements about inherited determinants will be based on inference rather than evidence. There is also widespread acceptance that other necessary conditions for the development of any pattern of behavior lie in the environment in which the animal grows up. Difficulties and disagreements arise, however, because both the inherited and the environmental determinants of behavior can be further subdivided into those that exert specific effects and those that have general effects. 1. General and Specific Effects of Determinants

It is important to ask whether it is possible to draw a sharp line across the continuum that runs from those determinants affecting only one pattern of behavior to those having such general effects they are necessary for life itself. In principle, though, the determinants of behavior could be placed somewhere in the matrix shown in Table I. An example of A might be the gene affecting the hygienic behavior in honeybees (Apis mellvera) that involves the uncapping of hive cells containing diseased larvae (Rothenbuhler, 1967). A representative of B might be a gene that affects the responsiveness of Drosophilu melanogaster to light (Benzer, 1967); loss of responsiveness to light, not surprisingly, has a widespread effect on all visually guided activities. An example of C might be the experience of chicks (Callus gullus) that have pecked at small objects painted with bitter-tasting substances; as a result, they develop a selective aversion for pecking at these objects (e.g., Lee-Teng and Sherman, 1966). Finally D might be

4

P. P. G.BATESON

early experience of crowded conditions by locusts (Locusra migratoria) subsequently leading them to become migratory (Ellis, 1964). The distinction between specific and general effects of determining events poses a number of difficulties. How can we ever be certain that a determining event has only one outcome? Any determinant that seems to have a highly specific effect on behavior is in danger of being reclassified as having more general consequences after further study. For example, further analysis of the honeybee may show that the genes affecting hygienic behavior have pleiotropic effects on other dissimilar behavior patterns. Even after the most convincing demonstration that differences between one animal and another in the way they make nests, say, is dependent on differences in the way they were reared, an experimenter is in no position to claim that other differences in behavior will not subsequently be found. On the other hand, if he finds that the experimental operation is the source of differences in nest-building, aggressive behavior, and feeding, he would probably not even wish to claim that it had highly specific effects. Therefore, it might seem that the categories of determinants with specific outcomes are liable to be eroded by the collection of fresh evidence, and individual cases will tend to move t o the right in the matrix shown in Table I. However, if a determinant affects a number of apparently different types of behavior, does it necessarily mean that its consequences are general? Could not those categories be thought of as having some special feature in common? Perhaps the determinant imposes some constraint on the way the animal’s head can be moved and this limitation shows up most noticeably when the animal is making a nest, threatening another individual, or feeding. Alternatively the nonspecific effects on behavior may themselves turn out to be consequences of a highly specific behavioral outcome of a developmental process. The point is, then, that the placing of a particular determinant in the matrix shown in Table I is always subject to alteration in either direction as fresh evidence becomes available. A related point is that a decision on how to classify a determinant may depend critically on the level at which its consequences are assessed. For example, phenylketonuria is a hereditary disease which, among other things, results in rather general disorders of behavior. However, the disease is caused by a specific deficiency of the liver enzyme phenylalanine hydroxylase (Hsia, 1967). Does the classifier utilize this knowledge about the specificity of the genetic determinants of the disease at the biochemical level? Or does he, as seems more logical, consistently apply behavioral criteria throughout and classify the determinants of phenylketonuria as having general effects?

2.

The Problem of Behavioral Units Another issue impinges crucially on the distinction between specific and general consequences. How should behavior patterns themselves be divided up? Are

SPECIFICITY AND THE ORIGINS OF BEHAVIOR

5

there obvious units that would provide a basis for the distinction between one behavior pattern being affected by some preceding event and many patterns being affected? It is an important question, but, once again, there is little agreement about the answer to it. The traditional response of many ethologists has been to argue that “natural” units of behavior become apparent to anyone who knows and loves his animals. On this view it is possible to assemble an ethogram-a complete inventory of behavior patterns shown by a species. However, thoughtful reviewers of the field have pointed out that selection of evidence is inevitable in the study of animal behavior as in everything else and that any ethogram will reflect the interests and preconceptions of its compiler (see, e.g., ‘ Marler and Hamilton, 1966, pp. 71 1-717; Hinde, 1970a, pp. 10-13). Furthermore, many difficulties remain even when it is possible to obtain agreement about the ostensive definition of a behavior pattern after pointing it out as it occurs or after detailed description. For instance, the same display given in two different contexts may serve two different biological roles in communicaton; although the message is the same the meaning is different in each case (e.g., Smith, 1968). For purposes of classification, do we have two behavior patterns or one? Another illustration is provided by the great tit (Pants major) which hammers with its bill in exactly the same way when it is feeding and when faced with a stimulus that evokes attack. Blurton-Jones (1968) argued that the behavior patterns are different because h a found that one increased in frequency after food-deprivation but the other did not. His experiment did not settle the matter, as Andrew (1972) points out, because the motor pattern of hammering may be controlled by the same stimulus in both cases. The food-deprived great tit may hammer more frequently at food because, as a result of its own searching behavior, it sees more food than objects evoking attack. So we are left with the dilemma whether or not we should split or lump bill-hammering in the two situations. Yet another difficulty is that, even with the most unequivocal items of behavior for inclusion in a classic ethogram, the temporal pattern of occurrences may be such that different measures of the behavior yield different results. For instance, the “chink” call given by chaffinches (Fringillla coelebs) when mobbing potential predators first increases in frequency and then declines gradually. Now, when Hinde (1960) measured the response of chaffinches to a stuffed owl and a toy dog, he found that on three measures the owl was more effective than the dog; the chaffinches called more at the owl than at the dog during the first 6 minutes of presentation; they responded more rapidly to the owl; and their calling at the owl waned more slowly. However, the time taken t o reach the peak rate of calling was shorter when the birds were presented with the dog; the birds’ calling in response to the dog apparently warmed up more quickly than was the case with the owl. In order to account for results such as these, it is necessary to postulate a number of underlying processes that interact to produce the temporal pattern of calling (Hinde, 1970b). Where does that leave the treatment of “chinking” as a unitary end product of development?

6

P. P. G.BATESON

Whatever way one chooses to deal with this particular example, it serves to warn that the types of measure chosen may have a profound effect on the interpretation of how the behavior is controlled and initially determined. It is easy to lose patience with arguments such as these on the grounds that, despite some imprecision, most people know what they are talking about. But how public are the rules that each of us uses? The difficulties in communication are not trivial and, indeed, present a major problem to philosophers. The issue is stated succinctly by Goldman (1970, p. 1) at the beginning of a book devoted t o the topic. He writes: Suppose that John does each of the following things (all at the same time): ( I ) he moves his hand, (2) he frightens away a fly, (3) he moves his queen to king-knightseven, (4) he check mates his opponent, (5)he gives his opponent a heart attack, and ( 6 ) he wins his first chess game ever. Has John here performed six acts? Or has he only performed one act, of which six different descriptions have been given?

The relevance of this problem to my argument is that the way in which behavior is divided up into units is very much a matter of opinion which, in turn, is 3 reflection of what questions about behavior are considered t o be important. The relative weights given by the classifiers to factors involved in development and control, to context, to consequences of behavior, and to its biological function differ from one school of thought t o the next. Classifications of behavior depend very much on the interests of the compiler and what may seem a natural unit from one vantage point may not even be noticed from another (cf. Hinde, 1970a). A decision about how finely behavior should be divided or about what features of behavior are important would obviously have profound effects on the placing of determinants on the specific-general scale. For example, if a gene affects all aspects of migratory behavior in a bird, its effects would be treated as specific if migration is regarded as a single pattern of behavior and nonspecific if the different aspects of migration were regarded as separate activities.

3. A Continuum in Effects of Determinants A final difficulty that threatens a simple division of determinants into those with specific outcomes and those with general ones is the likelihood of continuity. If one category of conditions affects single patterns of behavior and another category of conditions affects all the behavior patterns in an animal’s repertoire, every type of intermediate between these two extremes is possible in principle. In practice, intermediates are posing difficulties for simple dichotomies. For instance, an important criterion used to characterize conditions responsible for learning is that the lasting consequences on behavior of these training conditions are limited in extent. If environmental conditions have diverse effects on behavior persisting for a long time, those effects are not ordinarily attributed to learning. For example, when a rat is handled early in infancy and subsequently its behavior is found to be affected in a whole variety of

SPECIFICITY AND THE ORIGINS OF BEHAVIOR

7

different ways, it is not thought to have learned anything as a result of the handling. Nevertheless, the line of demarcation is arbitrary. Again, when kittens are exposed to vertical or horizontal lines at a particular stage in development, the kittens are subsequently said to be unresponsive to lines placed at right angles to the familiar orientation (Blakemore and Cooper, 1970; Blakemore, 1973). In some ways these effects are rather similar to those of imprinting in which a bird's social responsiveness is narrowed down'to the familiar object. However, the birds have no difficulty in detecting unfamiliar conspicuous objects which they actively avoid, whereas the kittens appear to be unable to detect lines of unfamiliar orientation. Consequently, all behavior patterns dependent on the detection of lines placed at right angles to the familiar orientation would presumably no longer occur in the kittens, and the effect of their early experience would have much more general consequences than that of the young birds. Most people would now want to treat imprinting as an example of learning, but the effects of restricted visual experience on the kittens is much less easily classified. It is worth noting that even the effects of imprinting are relatively nonspecific in as much as the learning process affects the subsequent occurrence of nonsocial behavior such as feeding and grooming by narrowing the range of objects with which the bird associates. In the absence of the mother or her substitute, the birds will generally abandon all other activities while they search for her. Furthermore, imprinting has marked facilitating and constraining effects on what the animal can subsequently learn (Bateson, 1973). Both Schneirla and Lehrman were concerned about the arbitrary way in which ethologists and experimental psychologists alike have so neatly demarcated the conditions necessary for learning from other types of experience. Lehrman (1970, p. 32) illustrated the conceptual problem facing us by sketching in the stages between environmental conditions having very general effects and those having highly specific effects. He listed the following points on the continuum : 1 . Effects on neural development of nonbiological conditions (temperature, light, chemical conditions in the environment). 2. Nonspecific effects of gross stimulus input. 3. Developmental effects of practice passively forced during ontogeny. 4. Developmental effects of practice resulting from spontaneous activity of the nervous system. 5. Links and integrations between behavioral elements resulting from early, nonfunctional partial performances. 6. Interoceptive conditioning resulting from inevitable tissue changes and metabolic activities. 7. Simple conditioning to stimulation resulting from spontaneous movements. 8. Simple instances of conventional conditioning and learning.

8

P. P. G . BATESON

Where does all this take us, then? A classification of determinants into those that have specific effects and those that have general effects is likely to be revised as fresh evidence is collected. Furthermore, it assumes a classification of behavioral units or types about which there may not be widespread agreement. Finally, it cuts arbitrarily across a continuum. None of these points render such a classification useless but they do mean that a sharp distinction between determinants with specific and general effects may create conceptual difficulties when attempts are made to unravel the processes involved in development. 111. CLASSIFICATION OF BEHAVIOR IN TERMS OF

DEVELOPMENTALDETERMINANTS So far I have tried to outline the difficulties inherent in one classification that rests in part on the nature of longlasting effects on behavior. It is now useful to reverse the procedure and consider a classification of behavior patterns in terms of developmental determinants. Naively it might be supposed that correspondence can be found between the two classifications. Indeed, preformationist views have from time to time slipped into ethological discussions of the origins of behavior. Behavior patterns are sometimes thought of as encapsulated in latent form in the fertilized egg; they are like Japanese flowers that will unfurl under the right environmental conditions. But even the most ardent preformationist does not insist that the blueprint for behavior, to use Lorenz’s metaphor, is the same as bricks, mortar, and a work force. In other words, even for the extreme nativist, a host of environmental conditions will obviously be necessary if the behavior pattern is to develop. Therefore it is not necessary to consider a class of behavior patterns that might be determined by a single factor alone. A much more plausible class is one in which the determinants of the behavior patterns are of the type shown in Fig. 1. In this case, a behavior pattern can be determined by one or more factors specifically affecting it as well as by one or more determinants that have general effects. In the example given in Fig. 1, each letter could represent many determinants each of which had the long-term inDETERMINANT A

<

BEHAVIOR 1

DETERMINANT B

BEHAVIOR 2.3.4

FIG. 1 . Determinant A has a specific effect on Behavior 1. Determinant B has a nonspecific effect on Behavior 1 and many other patterns. The arrows indicate that the determinants are necessary for the development of the behavior patterns to which they point.

9

SPECIFICITY AND THE ORIGINS OF BEHAVIOR TABLE I1 CLASSIFICATION OF BEHAVIOR IN TERMS OF DEVELOPMENTAL DETERMINANTS WITH SPECIFIC EFFECTS ON BEHAVIOR Environmental Inherited

No determinants with specific effects

At least one determinant

No determinants with specific effects

E

F

At least one determinant

G

H

fluence indicated by the arrow. Determinant A is necessary for Behavior 1 alone, whereas B is riecessary for 1 as well as many other behavior patterns. Inasmuch as many classifications of behavior have been concerned exclusively with developmental determinants that have specific outcomes, such as A, they have rested on a distinction, which is usually implicit, between determinants with specific effects and those with general effects. As we have already seen, this distinction raises a number of difficulties; even so it is worth following the logic of this classification. Table I1 shows the various categories of behavior available if we concentrate on determinants that have specific effects on the development of behavior. In some terminologies, behavior patterns in category G would be called “innate.” For example, Tinbergen (1951, p. 2) represented most ethologists at the time when he wrote: “Innate behaviour is behaviour that has not been changed by learning processes.’’ Tinbergen has changed his views a great deal since then, but some ethologists still cling to the old definition. For example, although admitting a preference for the term “endogenous,” Ewer (1971) thought “innate” could be usefully applied to behavior that matures without practices or example. She took this to be Lorenz’s position, although he (Lorenz, 1965) had changed his explicit definition of innate and now uses it as a synonym for “phylogenetically adapted.” According to this concept, the specific details of the behavior that adapt the animal to its natural environment were selected during the evolution of the animal’s species. Now, as has frequently been emphasized, natural selection acts on phenotypic outcomes not on the genotype. So the distinction between phylogenetic and ontogenetic sources of adaptiveness is not the same as the distinction between inherited and environmental determinants. Lorenz (1965) made this point strongly himself and argued that the outcomes of learning processes, such as imprinting, would have been selected during evolution. In other words, the learned preferences of birds for members of their own species are innate in the sense of being phylogenetically adapted. Despite this valuable clarification, old habits die hard. A dichotomy of origins of adaptation is all too easily used to justify once again a dichotomy of behavior

10

P. P. G. BATESON

and, to compound the muddle, also to refute the existence of behavior patterns specifically affected by both inherited and environmental determinants (category H in Table II).The confusion is evident even in Lorenz’s (1965, p. 71) book in which he wrote: “I strongly doubt that the motor co-ordination of phylogenetically adapted motor patterns are at all modifiable by learning.” One can only suppose that by an unconscious association of ideas, he was using “phylogenetically adapted” as a synonym for “innate” in the old sense, namely for behavior that is not changed by learning processes. Behind the inconsistent and inaccurate terminology lies a coherent point which bears directly on the matrix shown in Table 11: in the preceding quotation, Lorenz was in effect denying the existence of behavior patterns specifically affected by both inherited and environmental determinants (category H). He was still thinking in terms of his old notion of the ‘‘intercalation’’ of inborn and learned components of behavior. This idea of “instinct-learning intercalation” was also pursued energetically by Eibl-Eibesfeldt (1 970) who argued strongly against the view that blended intermediates constitute the majority of behavior patterns. Among other examples, he cited his own study of squirrels (Sciums uulgutis) opening nuts in which a complex sequence can be analyzed into components some of which are learned and some of which are thought t o develop without specific opportunities for practice. However, he seems t o suggest that because some behavioral sequences can be analyzed in this way, all behavior can be. Is it really possible to break up the fully developed song of an experienced male chaffinch into components, some of which are specifically affected by experience and some of which are not? Even though we know that many factors have been responsible for the detailed specification of the song (Thorpe, 1961), it does not follow that somehow these factors will correspond to constituents of the final behavioral product. Rather than liken the development of such behavior to the insertion of days into an existing calendar (intercalure), I suggest a more appropriate analogy would be the baking of cake. The flour, the eggs, the butter, and all the rest react together to form a product that is different from the sum of the parts. The actions of adding ingredients, preparing the mixture, and baking all contribute to the final effect. The point is that it would be nonsensical to expect anyone to recognize each of the ingredients and each of the actions involved in cooking as separate components in the finished cake. For similar reasons, I think those cases in which a simple relationship can be found

DETERMINANT C /

BEHAVIOR 6.7.8

FIG. 2. The special properties of Behavior 5 arise from developmental determinants with many other effects on behavior. The arrows indicate that the determinants are necessary for the development of the behavior patterns to which they point.

11

SPECIFICITY AND THE ORIGINS OF BEHAVIOR

between the determinants of behavior and the behavior itself will be exceptional. Behavior patterns that are affected by both inherited and environmental determinants with specific effects will lie in category H in Table 11. On the face of it, category E in Table I1 should be empty. However it is logically possible and, indeed, rather likely that the necessary conditions for the development of a behavior pattern are frequently those shown in Fig. 2. Behavior 5 is determined by B or C both of which have other effects as well. A hypothetical example might be provided by a Drosophila mutant whose reduced rate of courtship was known to be due to the general effects of a single gene on its visual system. If this gene only expresses itself when such Drosophila are reared at a certain temperature and the environmental condition also affects other patterns of behavior, then the distinctive courtship would, indeed, be an example of behavior falling into category E. Such cases would be particularly interesting because they would lie outside the framework in which the origins of behavior are conventionally treated. Two other points are worth making about the classification shown in Table 11. First, the cell in which a behavior pattern is placed will depend critically on what is meant by a “determinant with a specific effect.” If, on the one hand, a liberal view of specificity is taken and the line is drawn toward the general end of the specificgeneral scale, the major proportion of behavior patterns will, of course, be classified as being affected by both inherited and environmental determinants; if, on the other hand, stringent criteria are used to define specificity, the behavior patterns will be more evenly distributed in the matrix. The second point is that if we were omniscient and were able to quantify all the determinants exclusively affecting any given behavior pattern occurring at a particular stage of development, it would be possible to build up a scatter diagram such as is shown in Fig. 3. I cannot, of course, justify the relative

MANvl I

INHERITED DETERMINANTS WITH

SPECIFIC

rn

HUMAN LANGUAGE

CHAFFINCH SONG

rn HYGIENIC

EFFECTS

BEHAVIOR IN BEES

NONE

J

TYPEWRITING I N HUMANS

NONE

MANY ENVIRONMENTAL DETERMINANTS WITH SPECIFIC EFFECTS

FIG. 3. A scatter diagram showing hypothetical points that might be placed on it if all the developmental determinants with specific effects were known.

12

P. P. G.BATESON

positions of the four entirely hypothetical dots placed on the scatter diagram which is unsatisfactory, in any event, because it misrepresents the dynamics of behavioral development. Any one diagram can be nothing more than a snapshot of a changing scene. The positions of some behavior patterns would, doubtless, move more during development than others. Many would move to the right on the scatter diagram as the behavior patterns became increasingly enriched and differentiated by experience. Some might move upward or diagonally as fresh genes affecting the details of already established behavior patterns became activated during development. Although lability of behavior is, in general, taken as evidence for the influence of environmental factors, it would clearly be a mistake to assume that this was always the case. In any event, lability of a behavior pattern means that it might have t o be moved around in the matrix shown in Table 11. All of this might be taken to suggest that any kind of classification of behavior based on origins ‘is useless. I think that to adopt such a view would be unduly purist since many people evidently d o find it helpful to break up diverse and extensive material into manageable units so that they can think about it more easily. Rapid pigeon-holing of the evidence may frequently be misleading, but it certainly helps to unclutter the mind. Who is to say when it is better to disregard rather than focus on the relations and continuities between conventional categories? I shall consider this question in the next section. IV. THE NATURE OF “RELEVANT” EXPERIENCE 1 have tried to show that differences of opinion about the classification of behavior in terms of origins stems from different perceptions of the evidence. Lorenz has drawn a sharp distinction between factors responsible for the detailed characteristics of behavior (on which its adaptiveness to particular environmental conditions depends) and those factors necessary for continuity in development. Schneirla and Lehrman have objected to this formulation and where Lorenz saw two discrete categories, they perceived a spectrum of determinants. It is common enough in any science for different people to classify the same body of data in totally different ways. But it is worth while to ask whether some type of evidence can be found that would break the apparent impasse. Some progress in this direction may be made by looking more closely at the thinking underlying the experimental strategy proposed by Lorenz (1 965). Lorenz strongly argued for an experimental approach in which it would be possible t o identify internal mechanisms responsible for the adaptiveness of behavior by systematically excluding likely sources of environmental “information.” The isolation experiment, as it is called, clearly has been of service in eliminating possible explanations for the determination of some behavior pat-

SPECIFICITY AND THE ORIGINS OF BEHAVIOR

13

terns. It has suggested hypotheses that are fruitful in the sense that they can be tested. On the other hand, isolation experiments cannot provide direct tests of the hypotheses they propose. In order to demonstrate rigorously that a suspected source of variation does, indeed, have the effect it is supposed to have, that factor must be manipulated directly (see Hinde, 1968). If that cannot be done, progress may still be made by watching what happens when the suspected source of variation fluctuates spontaneously. Either way, the isolation experiment can usefully precede but does not replace direct analysis of behavioral determinants. As a strategy, Lorenz’s approach has the great merit of being positive and directed. Rather than bother about possible unknown sources of variation, the prescription to the experimenter is straightforward: if you consider something as the source of variation, then remove it. However, there are difficulties in this general approach which bring us to the nub of the whole problem. How does the experimenter know when he has excluded everything that is important? As Schneirla and Lehrman frequently asked: Can the experimenter tell the difference between “relevant” and “irrelevant” experience? Even when considering experience that has a specific effect on behavior, it may be very difficult to know in advance when an animal is likely to generalize the effects of one kind of training to a novel situation. Can we really be so certain that we know what are equivalent types of experience for an animal? The potential importance of this question, which is discussed by Schneirla (1 966) and Gottlieb (1973a), is easily underestimated. However another mattter polarizes opinion even more sharply. As we consider experiences with decreasingly specific outcomes at what point do we suddenly say that they are no longer providing relevant information? For Lorenz (1965, p. 37) this was not a problem and he took the following no-nonsense approach in his book: No biologist in his right senses will forget that the blueprint contained in the genome requires innumerable environmental factors in order to be realised in the phenogeny of structures and functions. During his individual growth, the male stickleback may need water of sufficient oxygen content, copepods for food, light, detailed pictures on his retina, and millions of other conditions in order to enable him, as an adult, to respond selectively to the red belly of a rival. Whatever wonders phenogeny may perform, however, it cannot extract from these factors information which simply is not contained in them, namely, the information that a rival is red underneath.

Lorenz saw a clear difference between experiences that produce their adaptive effects on behavior through learning and those experiences that are required for normal development and, when witheld, damage the animal in some way. Lorenz may have been led to this position, because many of the early experiments on the effects of sensory deprivation did, indeed, have pathological effects inasmuch as they resulted in degeneration in the deprived sensory modality (see Riesen, 1966).

14

P. P. G . BATESON

More recent work had suggested that nonspecific experience can have facilitating effects on development which are not easily predicted in advance. A wide body of evidence indicates that the development of functional connectivity of many neurons in the central nervous system can be markedly changed by stimulation (e.g., Jacobson, 1969; Horn et al., 1973; Riesen, 1975). Examples at the behavioral level of unexpected effects of stimulation are also beginning to appear in the literature. For instance, exposure of domestic chicken eggs to light before hatching had a marked effect on the responsiveness of the chicks to conspicuous objects after hatching (Dimond, 1968; Adam and Dimond, 1971). Similarly, relatively short periods of exposure to constant white light after hatching markedly enhanced the responsiveness of one-day-old domestic chicks to a visually conspicuous object (Bateson et al., 1972; Bateson and Wainwright, 1972; Bateson and Seaburne-May, 1973; Kovach, 1971). After exposure to constant light for as little as 18 minutes, chicks approached a flashing, rotating light more rapidly than those kept in the dark (Fig. 4), and the effects persisted for at least 12 hours and probably much longer (Bateson, unpublished data). The differences between chicks exposed to light and those kept in the dark could not be attributed to difference in handling or differences in the temperature at which the chicks were kept, and the likelihood that the lightexposed chicks were generally aroused and, therefore, approached rapidly did not appear so attractive after the effects of stimulation in other modalities were examined. Prior exposure to tape recordings of loud peep calls in the dark made the chicks less responsive to a conspicuous visual stimulus (Bateson and Seaburne-May, 1973). Similarly, Graves and Siege1 (1968) found that after gentle stroking in the dark domestic chicks took longer to approach a moving object than unstimu400.

1

200. APPROACH TIME

IN

loo :

50-

SEC.

"I

10'

0

3

7

18

46

120

EXPOSURE TIME IN MIN.

FIG. 4. The effects of varying exposure to a constant light on time taken to approach a flashing tight by domestic chicks. Medians and interquartile ranges are given for the time taken to approach. Each group consisted of 8 birds. Both scales are logarithmic. (From Bateson and Seaburne-May, 1973.)

SPECIFICITY AND THE ORIGINS OF BEHAVIOR

15

lated chicks. These results suggest that the birds must be stimulated in the visual modality if visually guided behavior is to be facilitated. Indeed, one-day-old chicks that have previously been exposed to light for an hour are much more accurate when pecking at millet seed than dark-reared birds (Vauclair and Bateson, 1975). The difference in accuracy was obtained when the chicks were unable to move their heads during the period of exposure to light. The difference might be attributed to deterioration in performance in the darkreared birds rather than to improvement in the lightexposed ones. However, in a careful study, Cruze (1935) reared and fed chicks in the dark for varying amounts of time before giving them an opportunity to peck at millet seed. He showed that the accuracy of pecking in naive birds continued to improve over the first 5 days after hatching. Although this improvement can probably be attributed, at least in part, to increasing motor coordination (e.g., Bird, 1933), it seems unlikely that visual acuity could have been markedly declining over the first 5 days after hatching. Eventually, of course, prolonged rearing in the dark does lead to deterioration of pecking performance and Padilla (1935) had great difficulty in eliciting any pecking from chicks reared in the dark for 14 days from hatching. It may be useful, therefore, to distinguish between the effects of light that influence the initial development of visually guided behavior and the effects of light that are necessary for the maintenance of the behavior once it is already established. The distinction is illustrated in Fig. 5. Light seems to have a remarkably similar effect on the development of depth perception in hooded rats. Tees (1974) found that although the performance of dark-reared rats on the visual cliff initially improved with age, the rate of improvement was not as rapid as in light-reared rats. Up to around 60 days of age, then, light seemed to have a facilitatory effect on development. However after 80 days of age the performance of the dark-reared rats sharply deteriorated whereas that of the light-reared rats remained stable; in the older animals light appeared to serve a maintenance function. DEVELOPMENT

MAINTENANCE

PECKING

AGE

FIG. 5. Schematic diagram of effects of light- and dark-rearing on chicks’ pecking accuracy at different ages.

16

P. P. G . BATESON

Returning to chicks, the explanation for the relatively nonspecific effects of light on approach behavior and pecking in chicks may be that activation of the visual pathways by mere use enables visual stimulation to elicit visually guided behavior more readily. The visual systems of young dark-reared birds are not, on this view, damaged or functionally degenerate but are less well developed than the previously stimulated animals. A similar explanation may account for some remarkable results obtained by Cottlieb (197 1). By devocalizing Peking duckling embryos (Anus plafyrhynchos) between 24 and 25 days after the beginning of development and just before the embryos penetrated the air space, Gottlieb seriously disrupted the preference of ducklings for the maternal call of their species after hatching. If the same operation was done immediately after the ducklings had broken into the air space, when they vocalize more and presumably can hear much better, the operated animals performed just as well as normal animals, strongly preferring the maternal call of their species to the chicken call (Fig. 6). This evidence strongly suggests that sounds the duckling emits itself shortly after it has broken into the air space play a part in the normal development of its auditory preferences. However generalized the outcome of stimulation in the visual and auditory modalities of young birds, the effects on the development of their social relations with their natural mothers would undoubtedly be adaptive. It would seem, then, that relatively nonspecific stimulation can provide “information” in Lorenz’s sense. If that point is accepted, the sharp distinction between determinants of behavior that have specific outcomes and supply relevant “information” and those that have general outcomes and are irrelevant begins to

1 i:

SPECIES

PERCENT PREFERRING MATERNAL CALLS

‘

0-

CHICKEN &

DAY 24 O

DAY 25

L

AGE OF DEVOCALIZING

FIG. 6. Auditory preferences of devocalized Peking ducklings. The embryos were devocalized before (day 24) or after (day 25) they had broken into the air space in the egg. At 48 hours after hatching, each duckling was given a 5-minute choice test between the maternal call of its own species and the maternal call of the chicken. (From Gottlieb, 1971.)

SPECIFICITY AND THE ORIGINS OF BEHAVIOR

17

evaporate. It would be absurd, though, to use such evidence as a last-ditch defense of environmentalism, since an identical argument can be mounted in favor of analyzing inherited determinants with relatively nonspecific outcomes. The essential point is that if factors with nonspecific effects are disregarded, the chances of unraveling the variety of conditions necessary for the detailed determination of any behavior pattern will be greatly reduced. It may prove helpful to distinguish, as Gottlieb (1 973b) has done, between “facilitative precursors” and “determinative precursors.” This is a distinction between factors in development that have quantitative effects and those that have qualitative effects. Of course, in the grey area between the two categories, it is probably difficult t o decide whether a factor facilitates a process that has already been established or is responsible for the development of a new one. In any event the distinction is not the same as the one between specific and general effects. Facilitative factors may have highly specific effects on development, and, conversely, determinative factors may have general consequences.

V. CONCLUSION In this chapter, I have argued that a classification of behavior patterns in terms of their developmental determinants depends critically on a sharp division between determinants with specific outcomes and those that have general effects. Where the line is drawn is very much a matter of opinion, and it is hardly surprising that many people have regarded the classificaton of behavior into “innate” and “acquired” as an unwarranted abstraction. Even if the distinction between specific and general is accepted as a matter of convenience, four categories of behavior rather than two are needed. A third category is needed because many behavior patterns are likely to be affected by both inherited and environmental determinants with specific outcomes. The suggestion that such patterns can invariably be unscrambled into intercalating innate and acquired behavioral components is not convincing. The fourth category is needed for cases where the distinctiveness of the behavior arises from the interaction of inherited and environmental determinants both having general effects on a wide variety of behavior patterns. The four-part classification has its uses inasmuch as it helps many people to think more easily about complex and diverse material. Furthermore it does provide a focus for research into the sources of behavioral distinctiveness. Environmental determinants that have specific outcomes are, almost by definition, rriediated by learning processes, and it is undoubtedly a useful tactic in anyone’s strategy for studying behavioral development to deny an animal particular opportunities for learning. However, aids to thought at one stage of analysis can become shackles at the next and eventually hinder further understanding.

18

P. P. G. BATESON

Certain types of evidence, such as that provided by the development of social preferences in young chicks and ducklings, do not fit easily into a framework in which experience is either “relevant” or “irrelevant.” Therefore, when study moves from preliminary sorting of complex material to detailed and comprehensive analysis, it becomes increasingly necessary t o recognize the assumptions underlying a classification of behavior patterns in terms of origins. In this chapter I have attempted to uncover these assumptions in order to prepare the way for an integrated approach to the study of behavioral development. Acknowledgments The problems discussed in this chapter are steeped in controversy; therefore, I have shown drafts to a large number of friends in different disciplines. Whereas I benefited enormously from their comments, it must not be assumed, of course, that my views are necessarily theirs. In any event, I am greatly indebted to the following for their help: G. Barlow, C. Beer, R. Dawkins, C. Erickson, Ariane Etienne, G. Gottlieb, R.A. Hinde, N.K. Humphrey, P. Leyhausen, A. Manning, R. Rappaport, Amelie Rorty, J.S. Rosenblatt, B.A.O. Williams. References Adam, J., and Dimond, S. A. 1971. The effect of visual stimulation at different stages of embryonic development on approach behaviour. Anim. Behuv. 19.5 1-54. ‘Andrew, R. J. 1972. The information potentially available in mammal displays. In “Nonverbal Communication” (R. A. Hinde, ed.), pp. 179-206. Cambridge Univ. Press, London and New York. Barnett, S. A. 1973. Animals to man: the epigenetics of behavior. In “Ethology and Development” (S. A. Barnett, ed.), pp. 104-124. Spastics Int. Med. Pub., London. Bateson, P. P. G. 1973. Internal influences on early learning in birds. In “Constraints on Learning: Limitations and Predispositions” (R. A. Hinde and J. Stevenson-Hinde, eds.), pp. 101-1 16. Academic Press, New York. Bateson, P. P. G., and Seabume-May, G. 1973. Effects of prior exposure to light on chicks’ behaviour in the imprinting situation. Anim. Behuv. 21, 720-725. Bateson, P. P. G., and Wainwright, A. A. P. 1972. The effects of prior exposure to light on the imprinting process in domestic chicks. Behuviour 42, 279-290. Bateson, P. P. G., Horn, G., and Rose, S. P. R. 1972. Effects of early experience on regional incorporation of precursors into RNA and protein in the chick brain. Bruin Res. 39, 449465. Beach, F. A. 1955. The descent of instinct. Psychol. Rev. 6 2 , 4 0 1 4 1 0 . Benzer, S . 1967. Behavioral mutants of Drosophila isolated by counter current distribution. Proc. Nut. Acud.Sci. U.S. 5 8 , 1112-1119. Bird, C. 1933. Maturation and practice: their effects upon the feeding reaction of chicks. J. Comp. Aychol. 16,343-366. Blakemore, C. 1973. Environmental constraints on development in the visual system. In “Constraints on Learning: Limitations and Predispositions” (R. A. Hinde and J. Stevenson-Hinde, eds.), pp. 5 1-73. Academic Press, New York. Blakemore, C. and Cooper, G. F. 1970. Development of the brain depends on the visual environment. Nature (London) 228,477478. Blurton-Jones, N. J. 1968. Observations and experiments on causation of threat displays of the Great Tit (Purusmajor). Anim. Behav. Monogr. 1, 74-158.

SPECIFICITY AND THE ORIGINS OF BEHAVIOR

19

Cruze, W.W. 1935. Maturation and learning in chicks.J. Comp. Psychol. 19, 371-408. Dawkins, R. 1968. The ontogeny of a pecking preference in domestic chicks. 2. Tierpsychol. 25, 170-186. Dimond, S . J. 1968. Effects of photic stimulation before hatching on the development of fear in chicks. J. Comp. Physiol. Psychol. 65, 320-324. Eibl-Eibesfeldt, I. 1961. The interactions of unlearned behaviour patterns and learning in mammals. In “Brain Mechanisms and Learning” (J. F. Delafresnay, ed.), pp. 53-73. Blackwell, Oxford. Eibl-Eibesfeldt, 1. 1970. “Ethology: The Biology of Behavior.” Holt, New York. Ellis, P. E. 1964. Marching and colour in locust hoppers in relation to social factors. Behaviour 23, 177-192. Ewer, R . F. 1971. Review of “Animal Behaviour,” 2nd Ed., by R. A. Hinde. Anim. Behav. 19,802-807. Goldman, A. 1970. “A Theory of Human Action.” Academic Press, New York. Gottlieb, G. 1971. “The Development of Species Identification in Birds.” Univ. of Chicago Press, Chicago, Illinois. Gottlieb, G. 1973a. Neglected developmental variables in the study of species identification in birds. Psycho[. Bull, 79,362-312. Gottlieb, G. 1973b. Introduction to behavioral embryology. In “Studies on the Development of Behavior and the Nervous System. Vol. 2. Behavioral Embryology (G. Gottlieb, ed.), pp. 3 4 5 . Academic Press, New York. Graves, H. B., and Siegel, P. B. 1968. Prior experience and the approach response in domestic chicks. Anim. Behav. 16, 18-23. Hailman, J. P. 1967. The ontogeny of an instinct. Behaviour, Suppl. XV. Hebb, D.O. 1953. Heredity and environment in mammalian behaviour. Brir. J. Anim. Behav. 1 , 4 3 4 7 . Hinde, R. A. 1960. Factors governing the changes in strength of a partially inborn response as shown by the mobbing behaviour of the chaffinch (Fringilla coelebs): 111 The interaction of short-term and long-term incremental and decremental effects. Proc. Roy. Soc., Ser. B 153, 398-420. Hinde, R. A. 1968. Dichotomies in the study of development. In “Genetic and Environmental Influences on Behaviour” (J.M. Thoday and AS. Parkes, eds.), pp. 3-14. Oliver & Boyd, Edinburgh. Hinde, R. A. 1970a. “Animal Behaviour: A Synthesis of Ethology and Comparative Psychology,” 2nd Ed. Maraw-Hill, New York. Hinde, R. A. 1970b. Behavioural habituation. In “Short-Term Changes in Neural Activity and Behaviour” (G. Horn and R. A. Hinde, eds.), pp. 3-40. Cambridge Univ. Press, London and New York. Horn, G., Rose, S. P. R., and Bateson, P. P. G . 1973. Experience and plasticity in the central nervous system. Science 181,506-514. Hsia, D. Y.-Y. 1967. The hereditary metabolic diseases. In “BehaviorGenetic Analysis” (J. Hirsch, ed.), pp. 176-193. McGraw-Hill, New York. Jacobson, M. 1969. Development of specific neuronal connections. Science 163, 543-547. Jensen, D. P. 1961. Operationism and the question “IS this behavior learned or innate?” Behaviour 1 7 , l d . Konishi, M. 1966. The attributes of instinct. Behaviour 27, 316-328. Kovach, J. K. 1971. Interaction of innate and acquired: color preferences and early exposure learning in chicks. J. Comp. Physiol. Psychol. 75,386-398. Kuo, Z. 1967. “The Dynamics of Behavioral Development.” Random House, New York. Lee-Teng, E., and Sherman, S.M. 1966. Memory consolidation of one-trial learning in chicks. Proc. Nut. Acad. Sci. US.56,926-931.

20

P. P. G. BATESON

Lehrman, D. S. 1953. A critique of Konrad Lorenz’s theory of instinctive behavior. Quart. Rev. Biol. 28, 337-363. Lehrman, D. S. 1970. Semantic and conceptual issues in the nature-nurture problem. In “Development and Evolution of Behavior” (L.R. Aronson, E. Tobach, D. S. Lehrman, and J. S. Rosenblatt, eds.), pp. 17-52. Freeman, San Francisco, California. Lehrman, D. S., and Rosenblatt, J. S. 1971. The study of behavioral development. In “The Ontogeny of Verbebrate Behavior” (H. Moltz, ed.), pp. 1-27. Academic Press, New York. Lorenz, K. 1961. Phylogenetische Anpassung und adaptive Modifikation des Verhaltens. Z. Tierpsychol. 18, 139-187. Lorenz, K. 1965. “Evolution and Modification of Behavior.” Univ. of Chicago Press, Chicago, Illinois. Marler, P. R., and Hamilton, W. J. 1966. “Mechanisms of Animal Behavior.” Wiley, New York. Moltz, H. 1965. Contemporary instinct theory and the fixed action pattern. Psychol. Rev. 12,2747. Padilla, S . C. 1935. Further studies on the delayed pecking of chicks. J. Comp. Psychol. 20, 413443. Riesen, A. 1966. Sensory deprivation. h o g . Physiol. Psychol. 1, 117-147. Riesen, A.H. 1975. (Ed.) “The Developmental Neuropsychology of Sensory Deprivation.” Academic Press, New York. Rothenbuhler, W. C. 1967. Genetic and evolutionary considerations of social behavior of honey bees and some related insects. In “Behaviorgenetic Analysis” (J. Hirsch, ed.), pp. 61-106. McGraw-Hill, New York. Schneirla, T. C. 1956. Interrelationships of the “innate” and the “acquired” in instinctive behavior. In “L’lnstinct dans le Comportement des Animaux et de I’Homme” (P.-P. Grasse, ed.), pp. 387452. Masson, Paris. Schneirla, T.C. 1966. Behavioral development and comparative psychology. Quart Rev. Biol. 41,283-302. Smith, W. J. 1968. Message-meaning analyses. In “Animal Communication” (T. Sebeok, ed.), pp. 44-60. Indiana Univ. Press, Bloomington. Tees, R. C. 1974. Effect of visual deprivation on development of depth perception in the rat. J. Comp. Physiol. Psychol. 80, 300-308. Thorpe, W. H. 1965. “Learning and Instinct in Animals.” Methuen, London. Thorpe, W. H. 1961. “BirdSong.” Cambridge Univ. Press, London and New York. Thorpe, W.H. 1963. Ethology and the coding problem in germ cell and brain. Z. flerpsychol. 20,529551. Tinbergen, N. 1951. ‘The Study of Instinct.” Oxford Univ. Press, London and New York. Tinbergen, N. 1963. On aims and methods of ethology. Z. Tierpsychol. 20,410433. Vauclair, J., and Bateson, P.P.G. 1975. Prior exposure t o light and pecking accuracy in chicks. Behaviour 52,196-201.

The Selection of Foods by Rats,

Humans, and Other Animals PAULROZIN DEPARTMENT OF PSYCHOLOGY UNIVERSITY OF PENNSYLVANIA PHILADELPHIA, PENNSYLVANIA

Solutions to the Food Selection Problem . . . . . . . . . . . . . . . . . . . 21 A. TheSpecialists.. 24 B. The Generalists or Omnivores 27 11. Rats: An example of Successful Generalists 27 A. The Specialist within the Generalist .................... 28 B. The Rat as a Generalist 34 111. Food Selection in Humans 52 A. Biological Factors in Human Food Selection 53 B. Specific Hungers in Adult Humans 56 C. Biological Basis of Ethnic-Racial Dietary Differences 58 D. Culture and Cuisine 62 References 67 I.

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

I.

SOLUTIONS TO THE FOOD SELECTION PROBLEM

Feeding and the search for food are probably the predominant ,activities of most animals. For some, clams or cattle, for instance, feeding occupies almost all waking time. Through adaptive radiation, animals have managed t o exploit just about every source of nutrition in the world. The pressure for survival is too great to leave a potential food source untouched. Even the most inaccessible nutritional riches have been compromised. Consider the clam, its rich meat so well protected by a thick shell that can seal tight. Even such an impregnable beast has been compromised in remarkably diverse ways: certain mollusks gain access t o the clam by slowly drilling a hole through the shell with a raspy organ, the radula; starfihh do the job by attaching their feet to both halves of the shell and exerting a steady but powerful force to separate the halves; herring gulls fly 21

22

PAUL ROZIN

up with a clam in their beak and drop it over rocks, shattering the shell; otters find a rock and break the shell with it; and man often makes a mess of it, but ends up with the meat. A paper such as this could easily succumb to the “gee whizzery” of adaptive radiation of feeding mechanisms, but the task at hand is quite different. It is to describe the mechanisms of food selection-how food is recognized and how choices are made. Food selection implies food ingestion. Food ingestion implies the presence of food. Therefore, background for the study of food selection includes the food search process: search images and search mechanisms for finding appropriate food stimuli in the environment. Honey bees (von Frisch, 1967) provide fine examples of a highly developed food search system. Food selection also implies the ability to obtain or capture food, and to assimilate it, for which many often exotic mechanisms have been evolved (Jennings, 1965). The presence of food, however, is not a sufficient condition for food ingestion. Food must ordinarily be accompanied by the organism’s inclination to eat it, at any particular time. In the absence of strong competing stimuli or drives, it is normally assumed that some aspect of the internal state of the organism determines whether or not it will eat a particular food. This state can be described as a “detector” that facilitates or inhibits ingestion. If the internal state or detector controls ingestion rather tightly, so that the internal signal is held within a narrow range, the process can be described as a regulation of food intake. In this sense, most animals seem to have some internal system, directly or indirectly responsive to energy balance, that modulates food intake (Rozin, 1964). Given the presence of potential food, ingestion then usually depends on an internal state or detector indicating a “need” for the particular food or class of foods, and a recognition of the potential food as food. In a few cases, such as some filter feeders (for other examples, see Rozin, 1964), internal state may play a minimum role; feeding always occurs in the presence of adequate stimuli. In many cases, the issue is simplified, because a species may consume only a rather small set of nutritionally exchangeable foods, such as the larger fauna of the African savannah which serve as food for lions. In this case, only one detector system is in principle necessary: any food source will serve to correct the internal, presumably energy-deficient state. In some cases, the single detector may be linked to a simple food recognition system, if the class of foods can be easily categorized (e.g., small moving things). As the range of food for a species is enlarged, as it approaches omnivory, the problems of both detection and recognition are vastly increased. There is no simple way of separating the class of potential foods from inedible or harmful substances. Also, it is likely that the numerous acceptable foods are not nutritionally equivalent, so that a single detector mechanism will not suffice. The specialists in the animal kingdom, who eat a narrowly circumscribed group of foods, are likely t o have detection and recognition of foods under tight genetic control. The generalists or omnivores require a much more plastic

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

23

system. This paper focuses on the complex problems, especially in food recognition and choice, in the omnivores or generalists. The complex interaction of genetically determined and experiential influences on food selection will be constantly in view, since there are both clear advantages and disadvantages to a heavy reliance on either nature or nurture in solving the problem. Omnivores, such as rats and humans, faced with an enormous number of potential foods, must choose wisely. They are always in danger of eating something harmful or eating too much of a good thing. Although there are some helpful internal mechanisms, such as poison detoxification, nutrient biosynthesis, and nutrient storage, the major share of the burden for maintaining nutritional balance must of necessity come from incorporation of appropriate nutrients in the environment and, hence, behavior. Curt Richter, the great Hopkins psychobiologist, demonstrated in the 1930s and 1940s that behavior was equal t o this task and that, in rats, metabolic homeostasis could be maintained by adaptive selfselection of nutrients (Fig. 1). The concern here is to describe and extend Richter’s work, by looking further into mechanisms, and looking at the food selection of man. When omnivores are examined closely, their resemblance to specialists becomes greater. In some respects, an omnivore is simply a number of specialists combined in one organism. To explore the “specialist within the generalist” and the role of built-in programming in food selection, the specialists are considered

FIG. 1. Example of self-selection behavior of rats on a “cafeteria” regime. Selections from among a variety of mineral sources are not included in this figure. The left portion of the figure shows self-selection with fully adequate diet components available. At about day 148, yeast, a source of B complex vitamins is removed, but rats are allowed to consume feces, which normally contain B complex vitamins. Selection of feces averts a vitamin deficiency. When the feces are removed, the rats gradually become B vitamin-deficient. Note the marked decrease in carbohydrate intake with B vitamin deficiency and the predominant role of fats as a calorie source. Note also the stability in day-to-day selections prior to deficiency. (From Richter and Rice, 1945.)

24

PAUL ROZIN

first. This approach leads to clear parallels with food selection in rats and, possibly, in humans. A. THE SPECIALISTS

The specialists survive by being especially good at finding, catching, and eating their special food. The price is complete dependence on one or a family of foods. Since the food category is limited and usually homogeneous, the problem of food recognition is easily solved with fixed circuitry i.e., easily programmed genetically. The problem of food choice rarely arises since the world is pretty well categorized as food and not food; natural selection rather than the individual organism makes the significant food choices. The problem of deficiency is not a behavioral one, since each individual food in the narrow range of foods is ordinarily nutritionally complete in itself. Carnivores, for example, rely on their prey to regulate intake qualitatively: only vitamin-deficient zebras can produce vitamindeficient lions. The extreme form of specialists are those that consume only one type of food (monophages). Examples are the koala bear, surviving on eucalyptus leaves, and the caterpillar of the monarch butterfly, which eats only milkweed. Omnivores may become monophagous-as almost happened to the Irish peasant who relied almost completely on the potato. The dangers of monophagy are clearly illustrated here by the disastrous potato crop failures in the mid-nineteenth century and the resultant famine. Animals that restrict their food intake to a rather well-defined category of. foods can “solve” the food selection problem at the receptor level. Certain patterns or sets of patterns of receptor responses can define acceptable foods (Dethier, 1967, 1973). This linkage can be permanent and unmodifiable by experience. For example, in some species of frogs that limit their fare to insects, there is a special visual receptor and central nervous system processing system that responds to small, convex, dark moving objects-bugs to be sure (Lettvin et al., 1959). Such a recognition system could be easily wired into the motor side of the feeding system. Among planteating insects, which often rely on chemical stimuli for food identification, approach is often guided via the olfactory system, whereas ingestion per se is under the control of contact chemoreception. Although there is some evidence for control of behavior by specific “token” substances in acceptable or unacceptable foods, in most cases feeding seems to be controlled by the combined response of several receptors to multiple constituents (Dethier, 1973). The chemicals and receptors may be described with some precision (Dethier, 1967, 1973), and identification by the experimenter of acceptable foods by electrophysiological response in olfactory or contact chemoreceptor nerves is a real possibility.

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

25

A limited definable food category, then, is associated with a characteristic receptor response pattern, probably connected directly to the motor side of the feeding system, with some sort of control over the linkage by what may often be a simple, single, internal detector system. Such systems are often genetically programmed and also unmodifiable. All that is necessary is a way of locating the food, which can be accomplished by a hierarchy of “search images,” often olfactoly or visual to contact chemoreceptor (gustatory) (Dethier, 1967, 1973). For virtually all animals, the mouth is the final checkpoint before entry into the sacred precincts of the body. The final criterion for the acceptable-unacceptable “judgment” is made here, mediated by the taste response. These taste judgments may override the previously acceptable food signals coming from other receptors (Dethier, 1969). Vomiting is one of the few available defenses once food is in the stomach, and rats, at least are incapable of vomiting. Staying within the range of specialists, one can see increasing levels of complexity by the concatenation of specialized systems. The best example I know of this, which is also the best example I know of the detailed analysis of any significant behavior, is the work by Dethier and his colleagues (Dethier, 1969) on feeding in the blowfly. The adult blowfly, in its rather brief life, needs only an energy source (e.g. carbohydrate) and water, except for the female’s need for protein during the stages of egg development. Flies identify foods primarily on a chemical basis, ultimately by contact chemoreception. The chemosensory hairs contain four to five nerve fibers-one particularly sensitive t o sugars, one to water, one t o salt, with the remaining one or two difficult to categorize. Highprotein foods lead t o a characteristic response pattern across receptors. Potential food encounters a hierarchy of chemoreceptors: first olfactory receptors, then chemoreceptors on the legs when the fly alights on the food, and, subsequently, two sets of receptors in the oral area, if the receptors previously stimulated indicate acceptable food. The energy control system seems to be built around the sugar receptor. A fly with an empty gut is particularly responsive to sugar solutions. When the insect steps in such a solution, receptor discharge leads to proboscis extension and sucking. Sucking continues, with a gradual rise in the sugar concentration that will maintain it, until there is sufficient adaptation so that no significant signal is received by the critical part of the central nervous system. A sort of regulation occurs, since the receptor input is attenuated (probably centrally) by the presence of solutions sensed by interoceptors (probably mechanoreceptors) in the foregut (Gelperin, 1966). The supply of sugar solution in the foregut is maintained over modest periods by periodic squirting of stored, recently ingested sugar solutions, from the crop into the foregut. When the crop is empty and, hence, there is no sugar in the foregut, input from “sugar” receptors again drives feeding effectively; there is an effective drop in the sugar response threshold (Dethier, 1969). Note that this energy control system does not directly regulate

26

PAUL ROZIN

the amount of energy intake, since the prime determiners of the amount ingested are sweetness of solution and amount of solution in the gut. In the presence of dehydration, sensed probably as a reduction in blood volume (Dethier, 1969), flies show an increased tendency to ingest water, i.e. water receptor input effectively drives sucking. The sodium receptor activity seems to turn off sucking; that is, it seems to be a food-avoidance mechanism. Gravid females show an adaptive increase in protein and decrease in carbohydrate intake during the period of egg development (Dethier, 1969). The explanation of this protein-specific hunger in terms of internal metabolic and/or hormonal changes is not yet at hand. The blowfly solves the problem of ingesting three types of substances with three specific systems, involving specific receptors or receptor complexes and detectors. Each system is quite inflexible and not susceptible to change via learning. The blowfly is, in effect, a small bundle of specialists. Even among species with a rather narrow set of potential foods, there is quite a bit of evidence for experiential influences on food selection. Unlike intraspecies recognition, where imprinting seems to be a common mechanism for determining selection, irreversible effects of early contact with a particular food are not common. However, there is a general tendency for animals to prefer familiar foods, Thus, Jermy et al. (1968) demonstrated preferences for feeding on particular plants in lepidopteroid larvae selectively exposed t o these plants 1 or 2 instars prior to testing. Similarly, Fuchs and Burghardt (1971) showed that within the narrow set of potentially acceptable food stimuli, young garter snakes would develop a selective preference for fish or worms on the basis of prior exposure. Unlike imprinting, this familiarity effect was reversible. Similarly, Burghardt (1967) reported a preference effect lasting over 1 week for a food (meat or worms) offered t o snapping turtles in their first meal, indicating both a familiarity and primacy effect. Finally, Hess (1964: see also Hogan, 1973a) reported evidence for a critical period for chicks in acquisition of preferences for stimuli that, when pecked, led t o food reward. Chicks receiving such an experience on days 3 and 4 of life showed a continued preference for this stimulus, whereas, if the critical reinforced experience occurred before day 3 or on day 7 , little effect was seen. The role of familiarity in food acceptance seems almost universal. It is clearly present in primates. Weiskrantz and Cowey (1963) studied the response of rhesus monkeys to new foods in the laboratory. They found that though monkeys tended to sample new foods, immediately (e.g., black currant juice or a chocolate malted drink), many would consume very little on the first few days, and later increase their intake considerably over a period of weeks. They noted that visual exposure t o other monkeys ingesting the new food facilitated increased acceptance by “nonconsumer” monkeys. Imitation or observation also seemed critical in acceptance of new foods by free-living Japanese macaques (Itani, 1958). Some new foods were rapidly accepted by the troop studied, e.g.,

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

27

wheat, summer oranges, whereas others, such as apples, were accepted more gradually. In a detailed study of the acceptance of candy, to which the monkeys did not initially respond with enthusiasm, Itani showed a clear pattern of social transmission of candy-eating from young and infant monkeys t o older siblings and mothers to the rest of the troop. The key role (resulting from minimal neophobia) of the young animals was indicated by the fact that, although less than 10% of adults ate candy on the initial presentation, 50% of 3-year-old and younger monkeys did. Furthermore, a year after initial introduction, 100% of 1-year-olds were candy eaters, compared to 5 1% of adult and young females and 32% of adult and young males. We shall again see the critical importance of this novel-familiar dichotomy as we consider selection in rats and humans. B.

THE GENERALISTS OR OMNIVORES

Versatility and flexibility in choice of foods is a great asset. In a changing environment, it is a much less Spartan solution than massive deaths resulting from natural selection against specialists whose food is on the decline. Omnivory allows a change in preferred basic food to occur within an organism’s lifetime rather than over many generations. A true omnivore such as cockroach, man, or the rat, considers anything of potential nutritional value as a possible food. The problem is that experimentation with new foods can be dangerous, since such substances can be harmful. On the one hand, the omnivore should be familiar with and in touch with the various food sources in its environment; on the other hand, this involves risks, particularly needless risks, if there is already adequate familiar food. One sees in some omnivores, particularly the Norway rat (Rattus norvegicus) a fascinating conflict arising from these opposite forces: a distinct exploratory tendency, coupled with an often powerful avoidance of new things (neophobia). The optimal solution to the omnivoral problem involves devoting quite a bit of brain circuitry to the food problem, and employing multiple mechanisms. Thus, we see instances of built-in programming, modification through “general experience,” more traditional learning, imprinting, social interactions, and culture or tradition all playing a role in food selection. The object is to explain the great success of at least some omnivores such as roaches, rats, and humans as indicated by their incredible numbers and resistance to annihilation. After building up a picture of food selection in rats, I shall turn to man, to see what common and what new principles are needed to account for man’s diverse food habits and cuisines.

11. RATS: AN EXAMPLE OF SUCCESSFUL GENERALISTS Richter’s classic demonstration (Richter, 1942-1943, 1955) (see Fig. 1) of the “cafeteria” seems a most effective starting point. Left to their own devices, with

28

PAUL ROZIN

a large variety of purified ingredients, most rats self-select wisely and grow about as well as rats on nutritionally balanced laboratory diets. Some of this might be accounted for as a generally broad sampling pattern since acceptable ranges for most nutrients are quite wide. Whatever contribution such nonspecific factors make, the basic validity of Richter’s work is demonstrated by the challenges he presented to the rats, within the cafeteria self-selection situation. Induction of an increased need for sodium via deficient diets or adrenalectomy led to appropriate adaptive changes in sodium intake. Similarly, vitamin deficiencies resulted in increases in intake of the appropriate vitamin; parathyroidectomy, with the induced Ca2+ loss, led to increased Ca2+ intake; and diabetes mellitus resulted in a shift from reliance on carbohydrate to greater reliance on fat and protein. We must attempt t o explain these behaviors, often called specific hungers. Moreover, we need t o examine how, in the wild, rats discover and test new foods and how they strike the balance between exploration and neophobia. Most of the work discussed involves domesticated rats as subjects, but the major phenomena are present in both wild and domesticated animals. A. THE SPECIALIST WITHIN THE GENERALIST

The list of substances required by the Norway rat (or, with little modification, man) is large and impressive. It consists of thirty to forty different components, including water, nine amino acids, a few fatty acids, at least ten vitamins, and at least thirteen minerals, and involves, in some instances, critical levels of these (National Academy of Sciences, 1962). For three required “nutrients,” each of extraordinary importance, rats behave as specialists, with a rather fixed, largely genetically determined, selection system. These three substances are oxygen, water, and sodium. I will not discuss oxygen intake here, since it is not quite a form of ingestion, but it does share important features with the other two. The main difference is that breathing virtually guarantees adequate oxygen, since oxygen is quite uniformly distributed in air, so that no specific recognition system is needed: only a detector hooked into the motor side. A fourth special partly built-in system regulates calories or energy intake. It is more complex than the other systems because it involves, in varying degrees, almost the full range of acceptable foods. All four of these systems have the characteristic of being absolutely basic and of representing “substances” that must be present in the organism within a limited density range. In short, all four are rather precisely regulated. Compared to other essentials, such as vitamins, a severe lack, imbalance, or cutoff of these four components can lead rather quickly to death. [A fifth possible candidate might be protein. There is evidence for regulation of protein intake, although little is known about recognition of protein sources (Rozin, 1968b).]

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

29

No wonder, then, that relatively foolproof, rather fixed solutions have been found. The importance of these substances to survival is such that much of internal metabolic physiology is devoted to their defense-hence the existence of liver and kidney. However, since all four components are dissipated over time, internal homeostasis cannot do the job by itself, and ingestion and, hence, behavior, must be harnessed to the system. Richter was the first scientist t o clearly demonstrate this link. To illustrate the specialist within the generalist, I will describe briefly the systems involved in water, sodium, and calorie selection and regulation.

1.

WaterHunger Rats and probably most other mammals come equipped with detector systems that indicate the state of body fluids vis-his the need to ingest water. There appear to be two detector systems, that sense some aspects of the tonicity of intracellular fluid and the volume of the intravascular space (Stricker, 1973; Epstein, 1973). The result of a signal indicating water deficit is presumably a unique sensation that we call thirst. Of course, other factors contribute to the sensation, such as temperature or dryness in the throat. The thirst sensation arouses exploratory behavior. The question is whether the target for this search, water itself, is prewired into the organism or whether water’s ability to reduce the thirst sensation is discovered through experience. In other words, Is there built-in specificity both in terms of unique internal state with its own detectors and in terms of a system for identifying the target substance? Surprisingly, we know very little about water recognition in rats. Rats do not drink water until just before the time of weaning (Teitelbaum et al., 1969), suggesting that if it is built in, water recognition is late in maturing. At the moment, there is no simple way of distinguishing between absence of thirst sensation (e.g., absence of functioning internal state detectors) or failure to recognize water. However, the existence of a specific water-recognition mechanism is suggested by reports of a characteristic water response in taste receptors (Zotterman, 1956; Bartoshuk, 1972), although the response seems to vary markedly depending on the state of adaptation (Bartoshuk, 1972). Whatever the specificity for recognition of water in the mouth, it is hard to imagine how visual recognition of water could be prewired, given the various visual forms that water may take. (It is anecdotally reported that following removal of congenital cataracts, a human patient was unable to immediately recognize water.) There is almost certainly an important role for experience in the “distal” recognition of water. The relative simplicity of chemical as opposed to visual recognition of water is indicated by the fact that chicks seem to have built-in water-taste recognition, but must rapidly learn its appearance (Morgan, 1894; see also discussion in Section 11, B, 6).

30 2.

PAUL ROZIN

Sodium Hunger

Sodium hunger is the example, par excellence, of a genetically determined specific hunger. Its properties were first described by Richter in the late 1930s (Richter, 1936; for reviews, see Richter, 1956; Denton, 1967; Nachman and Cole, 1971). Richter showed that rats normally have a preference for low concentrations of NaCl in water. This preference was enhanced, and extended to both lower and higher concentrations, by induced sodium deficiency. Similarly, in natural periods of increased sodium utilization, such as pregnancy, increases in intake and preferences occurred (Richter and Barelare, 1938; for a review, see Richter, 1956). Richter believed the increased sodium preference in the face of increased sodium need was innate, expressed in part by a drop in the absolute threshold for detection of NaCl, since the minimum preferred NaCl concentration was lower in deficient animals (Richter, 1939). The preference was specific to sodium: it appeared with various sodium salts (e.g., the chloride, phosphate, or lactate) but not for the equival$nt salts of other anions (e.g., potassium) (Richter and Eckert, 1938; Nachman, 1962). Richter’s theorizing about both the innateness and the sensory threshold drop has formed the focus of research in the field since his early work. On the issue of innateness, Richter’s original explanation has been strongly confirmed. The detection of sodium deficit seems to be a part of the prewired, body fluid regulation system. Since sodium is the major extracellular electrolyte, this should not be too surprising. Changes in electrolyte concentration, including hyponatremia, hypovolemia, and changes in mineralocorticoid levels have all been implicated as triggers of sodium hunger (see Stricker, 1973, for a discussion of the physiological conditions necessary and sufficient to release sodium appetite). I t is the innateness of sodium recognition, rather than internal detection, that is of particular interest here, and the data are impressive. The evidence indicates that sodium recognition is mediated by the sense of taste (Richter, 1956). I will describe here only three of many lines of evidence indicating that the increased, taste-mediated sodium appetite accompanying sodium deficiency is innate. First, there are taste receptors that are especially sensitive to sodium salts. Sodium ions, and sodium chloride in particular, seem to define a basic taste modality and produce characteristic electrophysiological responses (Pfaffman, 1959; Bartoshuk, 1972). What could the function of these receptors be except to signal the presence of this critical element? Sodium specificity of taste receptors must be related to the ability of sodium-deficient rats to show a preference for a variety of different sodium salts. And given the existence of sodium receptors, how easy it would be to connect them into an existing internal sodium-detection system. Second, sodium hunger appears immediately upon exposure to solutions containing sodium. The critical point here is that the first time a rat is made

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

31

sodium-deficient, it shows a preference for sodium salts in solution in less than a minute (Nachman, 1962; Handal, 1965; Quartermain et ul., 1967). Since the deficiency and the solutions were not experienced before and since the preference occurs before postingestional effects could occur, the argument for innateness is strong. Third is the exception that really proves the rule. There is one way to fool sodium-deficient rats and that is to offer them lithium salts. Rats behave toward these salts as they do toward sodium salts. In an elegant series of experiments, Nachman (1963a,b) has turned this exception into a telling proof of the innateness of sodium hunger. Sodium-deficient rats cannot prefer lithium for its effects, since, although it has similarities to sodium as an electrolyte, it produces rapid toxic effects. In fact, LiCl has emerged as the poison of choice in experiments on rat poison avoidance (see below). To humans, LiCl and NaCl taste almost identical. For this reason, LiCl was used for a while as a salt substitute. If lithium salts tasted like sodium salts to the rat, then an innate mechanism could be triggered into making a dreadful mistake by consuming a poison that tastes like the built-in “target” substance. Nachman was able to demonstrate that lithium and sodium salts do taste almost the same to rats. He poisoned rats after they consumed a sodium salt solution, so that they avoided this solution on future encounters. He now tested for generalization of the avoidance or aversion by measuring intake of other solutions in brief exposures. The results were that sodium chloride aversion generalizes completely to lithium chloride, and much less so to potassium, ammonium, or other chloride salts. [Under appropriate circumstances, rats can discriminate sodium from lithium salts (Harriman and Kare, 1964; Balagura et ul., 1972).] There is plenty of opportunity for experience to supplement and modulate sodium hunger. Rats must still learn where to find sodium. Krieckhaus and Wolf (1968; Wolf, 1969) trained normal rats in a two-lever box, where one lever delivered a low-concentration sodium solution as a reward and the other water. These rats had never been sodiumdeficient. After establishing baseline pressing rates for the two levers, rats were made sodium-deficient. On returning to the box, under extinction conditions (no fluid delivery), an enhanced preference for the sodium lever was shown. The rats had learned where sodium taste could be found. With deficiency and enhanced preference, they applied their learning immediately. Rats can also learn t o avoid foods that are sodium-deficient. They will reduce their food intake rather than continue to consume a sodium-deficient diet and will show an immediate preference for a new food over the sodium-deficient food they had previously been eating (Rodgers, 1967a,b). When sodium hunger is pitted against training to avoid sodium, the strength of sodium hunger becomes apparent. Sodium hunger appears even in rats that had a previously established sodium aversion by association of sodium solution intake with poison (Stricker and Wilson, 1970). Furthermore, sodium-deficient rats will

32

PAUL ROZIN

not abandon their sodium preference even when poisoned after drinking it (Frumkin, 197 1). Since 1 shall argue that the majority of what are called specific hungers are basically acquired, the question arises as to why sodium (but not calcium, vitamin B1, etc.) hunger should be innate. I can offer four somewhat related reasons. 1. Sodium is of particular importance in body fluid homeostasis, and a foolproof mechanism of regulated ingestion would have high adaptive value. 2 . There are significant shortages of sodium in some geographic areas, so the likelihood of deficiency is rather high. The existence of animal salt licks and salt mines exploited by man testify t o the unequal distribution of salt in the environment. 3. Probably as a consequence of reasons 1 and 2 , there exist both sodiumsensitive taste receptors and internal detectors, which would neatly serve an innate recognition system. 4. It is just possible that learning about the positive effects of sodium would be difficult, since it might be that the initial effects of sodium ingestion by a sodium-deficient animal might be negative. Rodgers (1967b) was unable to induce a preference for a neutral substance in sodium-deficient rats, when ingestion of this substance was immediately followed by intragastric intubation of NaC1. Since intragastric delivery of food or water can serve as a reinforcement (Miller and Kessen, 1952; Epstein and Teitelbaum, 1962), this failure suggests that the concentrations of NaCl used may produce negative effects in the gut. Richter’s second supposition, that sodium hunger was mediated by lower absolute threshold, has met a more uncertain fate. This mechanism, if it did exist, would only explain part of the phenomenon. It would not, of itself, account for increased preferences for higher concentrations. The upshot of a number of experiments on the existence of lowered absolute thresholds has been that it is the preference rather than the absolute threshold that is lowered in sodium deficiency. Normal and sodium-deficient rats, when tested appropriately, show the same absolute thresholds for NaCl (Carr, 1952; Harriman and MacLeod, 1953; Koh and Teitelbaum, 1961). There was also no difference in the absolute thresholds determined electrophysiologically (Pfaffman and Bare , 1950). The difference between them was that deficient rats start preferring sodium soltuions at the absolute threshold level, whereas normals are indifferent to the lowest sodium levels. However, there is evidence that there may, indeed, be increased sensitivity to low concentration of sodium in deficient rats or humans (Yensen, 1959; McBumey and Pfaffman, 1963; Henkin et al., 1963). This apparent increase appears to be related to the drop in sodium levels in the saliva during deficiency. This changes the adaptation level to sodium and increases sensitivity (McBurney and Pfaffman, 1963). The adaptation level notion has interesting implications for the detection of

SELECTION OF FOODS BY RATS,HUMANS, AND OTHER ANIMALS

33

needed substances, because it implies that solutions will taste salty only when they have a higher concentration of salt than the existing body level, as represented in the saliva. Thus, the adaptation would neatly provide a characterization of all those concentrations of biological utility at the moment (Desor, personal communication). It is suggested that it is not so much a change of threshold as a change in “classification” of the stimuli: concentrations of sodium below those in normal saliva, which might have been discriminable but were not “saltyyyin taste to normals, would taste salty in the sodium-deficient organism, since they would represent higher sodium levels than those in deficient saliva. All of these interesting matters may have little to do with sodium hunger. As pointed out above, changes in threshold or classification for very low concentrations cannot account for most of the characteristics of sodium hunger, which are clearly present in reactions to high concentrations. In summary, in the case of sodium hunger or appetite, there is a built-in sodium receptor, a built-in system for detecting the body’s state of need for sodium, and some built-in linkage between them (presently not understood), in which the motivational value (or preference for) the incoming sodium signal is modulated by the report of the internal detector.

3.

Calorie Hunger

Calorie hunger, or what is usually called just plain hunger, has a rather anomalous position as a specific hunger. There is an elaborate machinery, only partly understood, to detect energy imbalance in mammals, and evidence that similar control systems are at work in other groups, such as fish (Rozin and Mayel, 1961, 1964; Rozin, 1964). There have been suggestions that this system operates by detecting levels or amount of utilization of specific substances, such as glucose (Mayer, 1955). The presumably prewired regulation system is modified in very significant ways by experience; otherwise, it would be difficult to explain, for example, adjustments in meal size made when caloric density of food is varied. It is also possible that some eating may be viewed as a way of avoiding a hunger signal (see Le Magnen, 1971, for a general discussion of the role of experience in the regulation of food intake). The major significance of regulation of energy balance is indicated by the fact that it takes precedence over acquisition of sufficient amounts of many other essentials; for example, rats will not overeat calories in order to obtain adequate amounts of protein (Andik et d., 1963) or water (Bruce and Kennedy, 195 1). The peculiar feature of the specific calorie system is that there is no simple way of specifying what substances in the environment are adequate sources of calories, i.e., no possible equivalent to a sodium or water receptor. Given a limited range of foods, as with the great cats, specification of the class of foods can be accomplished and, hence, preprogrammed. However, true omnivores base their success on their ability to tap the widest range or sources of nutrition.

34

PAUL ROZIN

Thus, omnivores must basically learn what is food, which usuauy means calories, and what is not food. Undoubtedly, there are biases in exteroceptors, most especially in chemoreceptors, that relate to food recognition. Sugar-sensitive receptors, common in a wide variety of animals, undoubtedly serve such a function (although their presence in carnivores is a bit puzzling). Thus, the complexity and variability of the food (calorie)-recognition problem requires a great deal of plasticity. In this case, then, there is an elaborate and substantially prewired regulation-detection mechanism and a rather loosely constrained, plastic recognition component.

B. THE RAT AS A GENERALIST

Extension of specific innate mechanisms to handle the full range of food selection in rats is unthinkable, both in fact and in principle. One should have to postulate the equivalent of a full table of nutritional essentials in the rat’s head. For each component, there would have to be a unique specific sensory message and a unique central state characterizing the deficiency and sensed by a specific detector. The incredible amount of machinery needed for this would, for the most part, remain unused during the lifetime of the animal, since it is probable that a given animal does not experience most specific nutritional deficiencies in its lifetime. Moreover, the selection behavior of animals deficient in most dietary essentials does not show the certainty, directedness, and rapidity seen in sodium or water deficiency. In particular, poison avoidance cannot, in principle, be explained with a Specific, built-in mechanism, since some poisons successfully avoided by rats are man-made and were synthesized or made available for the first time during the to-be-poisoned rat’s very own lifetime. Clearly, here we must assume an ability t o learn about dangerous foods. In the past, the main argument in favor of innate-type mechanisms for most specific hungers was the inability to find a reasonable way of explaining how they could be learned. In principle, according to this argument, these specific hungers could not be learned because the interval between ingestion and consequences was too long (at least 30 minutes) and violated the basic law of temporal contiguity for associations. Of course, the great advantage of a mechanism through which a rat learns what foods make it sick and what foods make it feel better is that with one basic system, the whole host of nutritional and poison-avoidance problems can be solved. This simplicity apparently appealed to mother nature.

1. Poison Avoidance - Background Rats and man have been locked in a fierce battle of wits probably since the earliest days of civilization. In the twentieth century, on man’s side, has been an

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

35

experimental attempt to find ways to eradicate local rat populations through poisoning. A critical problem in the field has been to get rats to ingest enough poison. A wide variety of poisons has been tried (see Chitty and Southern, 1954, for a general review). Workers in this field, nonpsychologists on the whole, have assumed that rats learn to avoid poisons. Their evidence was overwhelming in amount, although critical experiments were not done. Wild rats (Rzoska, 1953), when faced with a new food, become extremely “shy” and “suspicious” and may avoid it for long periods. When they finally ingest it, they take a very small amount and go away, as it were, to “test” the food. Should the food contain poison, the mild effects of the small amount of poison lead t o a learned aversion to the food, and the rat does not return. For this reason, rat exterminators often use a procedure called “prebaiting.” The rat is first offered the vehicle (diet) in which the poison will later be placed. The rat’s initial neophobia to the vehicle thus dissipates, and it becomes an accepted part of the diet. Now, a poison is added. To the extent that the poison is potent, tasteless, and odorless and does not change the texture or appearance of the food, it may be accepted in large quantities and be successful. The method has had fair success, but the eradication of local rat populations is still outside man’s capabilities. The considerable literature on poisoning has been unknown to most psychologists, until very recently. Ironically, the “father” of specific hungers, Curt Richter, is one of the few psychologists who made contact with it. Richter has done some of the finest experiments in the area and participated in development of a major rat poison (Rchter, 1950, 1953). He also showed (Richter, 1953) the extent to which the wild rat’s aversion to new things, especially foods (neophobia) can be augmented by poisoning experiences. He produced some rats so suspicious of new foods, on account of successive poisonings, that they starved to death rather than try additional new foods. The work of Richter (1950, 1953), Barnett (1956, 1963), and Rzoska (1953), all with some significant exposure t o the psychological community, raised fundamental issues regarding the psychology of learning and suggested the critical importance of neophobia and responses to novel vs. familiar stimuli in the rat world. The work was not assimilated into psychology (as indicated by its absence from texts through the 1960s and into the 1970s). I believe this occurred (a) because the work dealt primarily with wild rats, organisms avoided by psychologists (the feeling was probably mutual), (b) because the work challenged some dearly held beliefs in the psychology of learning, and (c) because the work was not presented in “mainstream” psychology journals. However, the situation has now changed, as two lines of research within psychology have converged on the major issues raised by poison avoidance: the existence of special learning processes and the importance of the novelty-familiarity dimension. These two approaches-the further explanation of specific hungers and the analysis of poison avoidance within experimental psychology, together have led to reconsideration of some previously accepted views in the psychology of learning.

36 2

PAUL ROZIN

Poison Avoidance

- Recent

Work

Parallel to the rat poisoning studies, and closer t o the center of experimental psychology, were a series of investigations on the effects of X-irradiation on behavior. This work included reference to the fact that rats tended t o avoid foods whose consumption was followed by X-irradiation. [see Garcia e t a l . (1961) and Smith (1971) for reviews]. It was from the basic work of Garcia, Smith, and their colleagues that the ground was laid for findings which would have vastly more generality than the confines of X-rays and their effects on food preferences. Two critical experiments by Garcia and his colleagues solved the basic problem of how learning principles could explain X-ray- or poison-induced aversions. The experiments were simple and incredibly brief (a total of 4 pages in all), which is fitting for studies of major importance. One problem for a learning interpretation of poison avoidance is that foods and feeding must be selectively associated with poisons, even though other behaviors (running, sleeping) and their consequences were as closely associated in time with poisoning as the feeding. How could the rat “know” what was relevant? On the basis of what was known in the psychology of learning, the answer was that the rat could not and that poison avoidance was therefore a true mystery. Garcia and Koelling (1966) did a simple but powerful experiment demonstrating that rats do “know” what is relevant (although it was not explained how they knew). Thirsty rats were given “bright, noisy, and tasty” water to drink; that is, they were given flavored water, and each time they licked it a light flashed, a buzz occurred, and, of course, the taste was experienced. Following a brief drinking session, half the rats were poisoned by injection or X-irradiation and the other half were punished by strong electric shock to the feet. On a subsequent day, rats were tested t o see whether they had developed an aversion to the taste and/or the light or sound. Rats that had received poison or Xirradiation would not drink the flavored water, but would drink plain water, when licks of this were accompanied by the light and sound. Conversely, the shocked rats avoided “bright, noisy” water but did not avoid the taste. Garcia and Koelling described this important finding as an instance of “belongingness.” [This idea along with relevant data are present in a less clearly defined form in prior work by Capretta (1961) and Braveman and Capretta (1965).] According t o this notion, certain stimuli preferentially associate with certain others; in particular, tastes and possibly smells associate selectively with a set of internal visceral events that include gastrointestinal disturbances. h i s particular linkage is obviously perfect to handle food selection, since food enters by the mouth (hence taste) and produces gastrointestinal and metabolic consequences. The important taste-visceral link and the more general “belongingness” principle have been further amplified and extended (Garcia and Ervin, 1968). (Shettleworth (1972) has reviewed a wide variety of examples of specificity in associations,

SELECTION OF FOODS BY RATS,HUMANS, AND OTHER ANIMALS

37

which she calls “constraints on learning.” Seligman (1970) has extended the notion to a general principle of learning, called “preparedness” [see also the recent volumes: “Constraints on Learning” (edited by Hinde and Stevenson-Hinde, 1973) and “Biological Boundaries of Learning” (edited by Seligman and Hager, 1972)] .) To date, the limits of the taste-visceral system have not been defined. There is very little work on the relevant visceral field [e.g., Would pain in the chest be an unconditioned stimulus (US) for a taste conditioned stimulus, (CS)?] . Tastes appear more associable with gastrointestinal events than smells, although there is clear evidence for “smell-aversion learning” (Garcia and Koelling, 1967; Pain and Booth, 1968; Lorden etul., 1970; Domjan, 1973). It is possible, however, to “associate” exteroceptive cues with internal malaise, but the conditioning process is much longer and the results less impressive (Garcia et ul., 1961 ;Rozin, 1969a). Widening the horizon beyond rodents, it appears that the general principle here is that animals tend to associate food-related stimuli with the type of consequences that foods produce (Rozin and Kalat, 1971, 1972). In the case of rats, this leads naturally to the taste-gastrointestinal linkage; in the case of quail, which identify food visually, it leads to preferential association of visual characteristics of food with gastrointestinal consequences (Wilcoxon er al., 1971). Coupled with the major finding on belongingness and taste-visceral linkages, Garcia etal. (1966) published another paper, at about the same time, which is leading to major revision in our view of learning and to great advances in our understanding of mechanisms of food selection. They showed clearly for the first time that rats could leam to avoid a solution (CS) if its gastrointestinal consequences (US) occurred an hour of more after ingestion. This long-delay learning was only demonstrable with taste (or smell) CSs and gastrointestinal or metabolic USs (Garcia et ul., 1972; Rozin, 1969a). This is, of course, as it should be, since only in the feeding system are the initial events (tastes) separated significantly in time from their (metabolic) consequences; the gut induces an inherent delay. In the rest of life, e.g., predator avoidance, causes and consequences follow rapidly in time. Garcia, Ervin, and Koelling originally demonstrated the long-delay learning by using saccharine solutions as a CS and apomorphine injection as a US. Previous studies by J. C. Smith and his colleagues (reviewed in Smith, 1971) and Garcia’s group had all flirted with this fundamental new phenomenon. However, Garcia, Ervin, and Koelling were the first to physically separate CS and US by long time intervals, so that it could not be argued that there were early, subtle immediate effects of the US (as when a poisoned solution is drunk) that provided temporal contiguity. The long-delay finding is of such significance for both learning theory in general and food selection that it has been subject to rather intensive scrutiny and study, particularly by Revusky and Garcia (1970; Revusky, 1971) and by

38

PAUL ROZIN

Rozin and Kalat (1971, 1972; Kalat and Rozin, 1973). The following is a summary of what is known about this system. 1. Long-delay learning has been clearly demonstrated for intervals as long as 8 (Revusky, 1968) or 12 hours (Smith and Roll, 1967). There is evidence that with anesthesia administered for the period between CS and US, the CS-US interval could be extended indefinitely (Rozin and Ree, 1972). 2. Long-delay learning is limited to taste and smell CSs and an unknown class of visceral USs that include gastrointestinal stimuli. 3. Longdelay learning occurs rapidly-in most cases in one trial. 4. Long-delay learning lasts a long time but can be extinguished rather easily. The adaptive fit of this and the previously stated features with the problems of poison avoidance and food selection should be obvious. 5. Long-delay learning cannot be explained as a peripheral phenomenon, e.g., aftertaste contiguous with sickness. Evidence against t h i s view has been reviewed (Rozin and Kalat, 197 1 ; Revusky and Garcia, 1970) and includes the following: (a) aftertastes would hardly be likely 6 hours or more after drinking (Revusky, 1968; Smith and Roll, 1967); (b) rats can quickly learn t o avoid a particular concentration of a solution, which would be hard to d o on the basis of aftertastes (Rozin, 1969a); ( c ) quail show long-delay learning for food-related visual stimuli, in which aftertaste cannot be involved (Wilcoxon el al., 197 1). There is at present n o clearly correct explanation of the mechanism of longdelay learning. Three theories have been suggested. In brief, a trace decay notion, asserting, simply, that the associability of taste memory traces decays less rapidly than other memory traces (Rozin and Kalat, 1971, 1972); an interference theory (Revusky and Garcia, 1970) asserting that limits on the length of the CS-US interval are always produced by retroactively interfering stimuli (in the taste-visceral system, with only tastes as relevant stimuli, very little taste interference occurs); and a learned safety approach (Rozin and Kalat, 197 1, 1972; Kalat and Rozin, 1973), asserting that what rats really learn is what is safe and that they learn this gradually over time. There is some evidence for each of these views, which are, in fact, mutually compatible.

3. Specific Hungers a. Thiamine-Specific Hunger. I shall now describe the mechanisms of specific hungers and relate them to poison avoidance. Specific hungers exist in two forms: (1)the self-selection of adequate diets by healthy rats and (2)the adaptive selection of specific nutrients by animals deficient in those nutrients. I will concentrate on the latter, as it is better understood and may help t o explain selection in healthy rats. The prototypical, simple specific hunger experiment involves raising an animal o n a diet, D, deficient in element X, and, when deficiency signs appear, offering the animal a choice between diet D and diet D + X. More complex versions involve using more choices. I will focus on the simple setting, and discuss the specific hunger for vitamin

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

39

B1 (thiamine), as the best-investigated example. Thiamine deficiency, in young rats, produces clear deficiency signs of anorexia and weight loss within a few weeks after thiamine is removed from the diet. Classic studies by Richter et ul. (1937) and Harris etal. (1933) clearly demonstrated a preference for thiaminerich foods by thiamine-deficient rats (Fig. 2). These studies were subsequently Hat N o I

70-

"

- 4

*

a

.

E

2

. 3

%

Rat

N O 1 (el

C

I BAS

4

I

5

0

-I-[

-

191

MAR.

I

10days

indiosta t l u t thr diet

WM

o J d during the period ahom.

rsprnonts the muountd~of the dirt m k n b y 11y d y .

4

r

w

=

-

diets conhining adequacy of the ritcunin. diet. devoid of the dtnmin.

FIG. 2. Preference of 4 vitamin B complexdeficient rats for a diet (Marmite) containing B vitamins. (From Harris et a/., 1933.)

40

PAUL ROZIN

confirmed and extended (Scott and Quint, 1946a; Scott and Verney, 1947; Rozin et ul., 1964). Work by Scott and his colleagues (Scott and Verney, 1947) suggested that there was no specific recognition of thiamine per se. When they added a distinctive anise flavor to the diet with thiamine, the rats developed a preference for the thiamine-anise diet. The anise flavor was then switched to the deficient diet, and the rats now preferred the anise-deficient diet. The upshot of the early work is that thiamine hunger appears reliably and rather rapidly (within 1 day). In its simplest and historically most commonly accepted form, a learning explanation of specific hungers assumes the following sequence. A rat is deficient in X and, presumably, feels sick. It encounters, among other foods, a food containing X, and eats some. It starts feeling better and is, thus, reinforced for eating X. Hence, a preference for X develops. In spite of experiments by Harris et al. (1933) and Scott and Verney (1947) demonstrating something like this, the conflict with basic learning principles was too great to convince psychologists that some specific hungers were learned. Specific hungers, when discussed at all in elementary textbooks were mentioned only under the heading of motivation. In addition to the serious long-delay problem, there was a problem in explaining how foods (as opposed to light or sounds, grooming, etc.) would specifically be associated with their consequences. The notion that the positive reinforcement of beneficial consequences following on ingestion of enriched food explains specific hungers had two other serious shortcomings: (a) rats failed to show a vitamin Bl-specific hunger when the choice was water vs. vitaminenriched and flavored water, even though vitamin-enriched water produced the same recovery as vitaminenriched food (Rozin el ul., 1964); and (b) rats that had recovered from deficiency by injection of thiamine showed a preference for thiamine-rich foods when they were presented for the first time after recovery. Under these circumstances, the vitamin in the preferred choice should have no particular positive effects (Rozin, 1965). Thus, we have absence of specific hunger when the positive reinforcement condition is fulfilled (a), and presence of the specific hunger in the absence of need, i.e., no positive reinforce men t ( b ) Observation of rats while they became deficient and in subsequent choice situations led to a resolution of this puzzle (Rozin, 1967a). In the deficiency period, when the standard deficient food is presented after a period of food privation, rats avidly approach the food cup, sniff at it, and then either spill the food with their paws or walk away and chew on something inedible in the cage. This is the same type of behavior as is shown by normal rats when they are offered a highly unpalatable, quinine-adulterated diet. This “displacement” or “redirected” behavior suggests that the deficient diet is aversive to the rats. If given a new diet, these same rats consume it avidly. The aversion conception gains force with the observation that when rats made deficient on diet A, and

.

SELECTION OF FOODS BY RATS, HUMANS,AND OTHER ANIMALS

41

recovered on diet 9, are offered diet A again, they will not eat it, even if food deprived. When a hungry rat prefers eating nothing to a particular diet, it seems fair to call that diet aversive. But, if there are aversions to deficient diets, then the following conclusions may be drawn. 1 . Vitamin-deficient diets are like slow poisons, and the literature on poisoning is relevant to specific hungers [in fact, it was this specific aversion experiment that connected diet deficiency to the researches on poisoning (Rozin, 1967a,b)]. 2. The critical learning apparently takes place during exposure to the deficient diet, not at the time of choice. 3. The preference first manifested after recovery (Rozin, 1965) is not a problem, since it can be seen as a retained aversion to deficient diet (Rozin and Rodgers, 1967). 4. Given that rats do not stop eating per se, but rather stop eating a particular diet, what has been learned appears to fit better into a classical-aversive paradigm than into the suggested operant-positive reinforcement scheme. This is supported by the fact that taste-aversion learning may occur by association of taste o r smell with gastrointestinal upset in the absence of ingestion (Domjan and Wilson, 1972a,b; Bradley and Mistretta, 1971). 5. With points 1 4 , the Garcia experiments (which appeared concurrently with and independently of these specific hunger experiments) seem to provide a basis for a learning explanation. The belongingness principle takes care of the food-illness association. In fact, independently of Garcia, Rozin (1967a) showed that deficient rats, although avoiding their deficient food, did not avoid the food cup or its location-only its contents. In essence this is another, but less elegant, version of the Garcia belongingness effect. Thus t o some extent at least, thiarninespecific hunger can be described as an ' aversion to thiamine-deficient diet, learned with the special belongingness and long-delay abilities of the feeding system. A full understanding of thiamine and related specific hungers involves considerably more than this, however. We must yet consider (1)the critical novel-familiar distinction-its potency as a food classifier and importance in determining specific hungers; (2) the possibility that, in addition to learning what is bad for them, rats can learn what is good; (3)mechanisms through which (moving toward cafeteria situations) rats select enriched food in complex multichoice situations; and (4) the extent to which what we have described for thiamine deficiency also holds for other essential nutrients. b. The Novel-Familiar Dimension. Research on specific hungers (Rodgers and Rozin, 1966; Rozin, 1968a) has clearly highlighted the importance of past experiences with foods on rat behavior. In a very real way, the food world of a rat consists of those foods never before tried (novel) and those previously

42

PAUL ROZIN

sampled (familiar). The familiar category subdivides into three subcategories: harmful, neutral, and “beneficial” foods. The marked negative response of wild rats to new events or objects in their environment, especially foods, is well known among rat exterminators. Rats, most especially wild rats, tend to stick with familiar foods. The way to get a clear preference for a new food, in rats, is to offer a choice between a familiar aversive diet and a new food (Rodgers and Rozin, 1966; Rodgers, 1967a). Poisoning (or deficiency) experiences are almost by definition situations in which a new food (new by virtue of presence o f poison or absence of essential nutrient) is associated with aversive consequences. Rats appear to become more and more neophobic, the more experiences of this type they have (Richter, 1953; Rozin, 1968a). Rats that have been poisoned or deficient show an increased preference for old “safe” familiar foods, and an increased avoidance of new ones (Rozin, 1968a). This is true for both wild and domestic rats, the difference between them being simply a more generally neophobic base line for wild rats (Rozin, 1968a). The novel-familiar dichotomy allows an important stimulus selection principle to operate: rats tend to associate new events (e.g., new CSs) with new consequences (e.g., new USs). Hence, in a confounded situation in which a new and a familiar food are both consumed prior to poisoning, only the new food acquires a significant aversion (Revusky and Bedarf, 1967; Wittlin and Brookshire, 1968; Kalat and Rozin, 1973). Only one prior experience with a previously new food, followed by neutral or positive consequences, suffices to make that food strongly resistant to becoming aversive (Kalat and Rozin, 1973). However, a few minutes of exposure t o a food is required for it to become effectively familiar (Domjan, 1973). I might add, parenthetically, that the novelty effects make obvious sense in an adaptive framework, as do the belongingness and long-delay capacities. c. Learning Preferences as Well as Aversions. The original formulation of specific hungers was in terms of learning about the beneficial consequences of an enriched food. We have seen that, in fact, the major phenomenon appears to be learning about aversive consequences of deficient foods. The question remains: Can rats learn about the positive effects of foods? The answer seems to be yes. A number of investigators (Garcia et ai., 1967; Zahorik and Maier, 1969; Revusky, 1967) have reported enhanced preferences for substances whose ingestion is followed by an improvement in “physiological state” (thiamine or caloric repletion). The effect is usually relatively small and appears only over a number of conditioning trials in contrast to poison avoidance. One study (Seward and Greathouse, 1973) directly comparing “positive” learning about recovery from thiamine deficiency with aversion learning when thiamine deficiency is the aversive event found much more rapid and marked learning in the aversion paradigm. Recent evidence (Simson and Booth, 1973) also suggests that long delay intervals may be more characteristic of learned aversions to toxins than t o learning with “nutritional reinforcers.” These authors raise the important

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

43

question of whether the fundamental distinction should be between positive and negative reinforcers, or “nutritional” (part of normal metabolic function) and toxic events. The question is whether or not and under what circumstances rats discriminate between safe and particularly beneficial foods. In other words, does the food world of rats consist of three categories-novel, familiar-dangerous, and familiar-safe-or four, with the last category bifurcating into familiar-safe and familiar-beneficial. To distinguish between safe or neutral and beneficial foods, the food that is a candidate for positive preference (e.g., associated with recovery from thiamine deficiency) must be tested against a familiar-safe food. Until very recently the only study supporting such a positive preference has involved caloric repletion as the reward (Revusky, 1967). However, Zahorik et d.(1974) have just demonstrated a clear preference for foods associated with recovery from thiamine deficiency over familiar safe foods, indicating four functional food categories. The positive preference has obvious adaptive value. However, there is no doubt that rats, at least, are strongly biased toward learning effectively and rapidly what makes them sick, and rather poor at learning what makes them well. This makes sense for an animal whom everyone is trying to poison-paranoia in wild rats is consistent with contact with reality. d Selection among Foods in Complex Situations. Returning to our thiamine-deficient rat faced with a new garbage can, we must ask (with an obvious eye toward cafeteria-type experiments), how he uses the capacities described up to this point to, as it were, “find the good stuff.” Long-delay learning makes it possible, belongingness effectively limits the candidates for dangerous or beneficial things to foods, and the novel-familiar dimension further restricts the number of “suspect” foods to the new ones. But, faced with a variety of new foods, how does the rat solve the problem? The first thing to realize is that under such circumstances, rats are often not successful (Harris et al., 1933; Rozin, 1969b). Harris et al. (1933) showed that, if vitamin B-deficient rats were offered a choice among ten new foods, only one of which contained the vitamin, they typically did not show a selective preference for the enriched source. It was usually necessary to “educate” them by exposing them exclusively to the enriched choice for a period of days, following which they would show a maintained enriched food preference. However, such educational guidance is not to be expected in the real world. In the absence of educational guidance, and with a smaller number of new food choices, many rats do develop a preference for a single enriched food source (Harris et al. , 1933;Rozin, 1969b). How, then, does the rat learn which of the variety of foods present produces significant positive, or at least, nonnegative effects? How, if the rat consumes a number of foods at a time, can it specifically learn about the consequences of a particular food? First of all, as mentioned above, the novelty principle simplifies the situation to some extent. Some of the available choices may have already been associated with deficiency, and will not be ingested. If there are, in addition, some familiar,

44

PAUL ROZIN

neutral (safe) or beneficial foods, say from the days before the deficiency, rats will explicitly prefer these. In their absence, the rat is faced with novel foods, among which it must choose. Another factor, which may simplify the task, is that all novel foods do not have an equal probability of being associated with toxic consequences. Some tastes are more likely to become aversive by conditioning than others (Kalat and Rozin, 1970; Brackbill etul., 1971). We do not know, at this time, what characteristics make a particular taste more salient than others vis-&vis poison association. There is evidence (Kalat, 1974) that novelty per se is an important determiner of salience. Salience could be of significant adaptive value if it was tied to the real world probabilities of given tastes being associated with natural poisons. For example, it appears that bitter tastes are often characteristic of poisonous plants, and the general negative response toward bitter tastes may well be, in fact, a built-in danger recognition mechanism. The primary burden of selection among multiple food sources is borne by the rat’s natural feeding pattern itself. Rats d o not eat randomly in time. Their feeding is clustered into short bouts, appropriately dubbed meals, separated from one another by periods of 30 minutes of more. Observation of rats faced with multiple new food sources (Rozin, 1969b) (Fig. 3) indicates that any given meal tends to involve only one food source. In other words, rats seem t o sample new foods one at a time. This pattern is exaggerated in deficient animals. The consequence is that each food can be evaluated in an uncontaminated fashion. The rat’s natural feeding behavior simplifies a complex situation (see also Barnett, 1956). Observation of deficient rats faced with a number of new food choices suggests that it may be a period of days before a rat samples a single enriched source, but one or two meals from this source seem to be sufficient to establish a stable preference for it (Fig. 3). In nature, rats are social animals. There are numerous possible avenues of social interaction which could facilitate some transfer of information about foods in the environment from one animal to another. Galef (see first chapter in this volume) has done some elegant studies demonstrating parent-child interaction in domestic rat colonies. In outline, if parent rats learn to avoid diet A and to eat diet B, their infants, at the time of weaning and separation from the parents will show the same preference pattern (Calef and Clark, 1971). FIG. 3. Meal patterns of 3 thiaminedeficient rats faced with a choice of four foods. The rats were allowed t o feed freely for 8 hours each day, with intake recorded every half hour or hour. Each rat, following a deficiency period on one diet, was offered a choice of this diet and three new diets. One of the new diets (indicated by a 8 on the figure) was enriched with thiamine. The figure shows the intakes for 3 rats over a 3-4 day period, during which they “discovered” the vitamin-rich diet. This occurred during the first 3-4 days of testing for rats 239 and 243, and during days 4-6 for rat 241. Subsequent to these days, each rat ate the enriched choice almost exclusively for a number of days. Note the tendency for the rats to consume only one food at a time, and to show a maintained preference for the enriched food once they have eaten an isolated meal of it. (From Rozin, 1969b.) Copyright 1969 by the American Psychological Association. Reprinted by permission.

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

45

The preference for B over A in the young weanling rats has been shown to be explainable in terms of neophobia. Young rats will tend to eat familiar safe foods, and for them, diet B is familiar and safe for two reasons. First, before total weaning, young rats go out and feed in the environment along with their RAT 239

4

oe 3

5

31

1 d

RAT 241

4

0

38

S

RAT 243

DAY 2-3

4

oe 3

tiE HOURS

46

PAUL ROZIN

parents. Since their parents feed on diet B, they will also and, thus, this diet becomes familiar (Galef and Clark, 1971, 1972). Second, salient characteristics of the diet eaten by the mother appear in the mother’s milk (LeMagnen and Tallon, 1968; Galef and Sherry, 1973). Nursing pups experience chemical stimulation that resembles parental food, making it familiar and safe and, hence, preferable (Galef and Henderson, 1972). No doubt there are other social interactions among adults as well as between parent and child that are of great significance (Barnett, 1956). e. Other Adaptations to Thiamine Deficiency. We have yet to exhaust the multiple mechanisms available to handle thiamine shortage. Vitamin B1-deficient rats, as well as rats deficient in some other substances, show a marked increase in feces ingestion or coprophagy (Richter and Rice, 1945; Barnes, 1962; Rozin, 1967b). This has adaptive value, since the flora of the hindgut synthesize many vitamins, which can only be utilized by the host through feces ingestion. This was first demonstrated, it should by now be n o surprise, by Richter and Rice (1945). They showed that rats on a cafeteria regime would ingest the fecal output of 2-4 rats per day, when this was their only source of B vitamins, and would remain healthy on this regimen (see Fig. 1). Removal of the feces led to onset of deficiency symptoms. Rats normally consume about 3540% of their feces, even when kept on screen floors. Feces ingestion frequently rises to 100% in the case of vitamin deficiency (Barnes etal., 1957,1960; Barnes, 1962). Rats that are B complex-deficient develop specific dietary habits which have the effect of prolonging survival. Fats spare thiamine; rats survive longer on a high fat than on a high carbohydrate or protein diet when placed on a thiaminefree diet (Scott etal., 1950a). Clearly, a shift away from the normally high carbohydrate diet toward more fat would spare thiamine, and just such a shift has been observed in B complex deficiency (Richter etal., 1938; Richter and Hawkes, 1941) (see Fig. 1) and with pure thiamine deficiency (Scott etal., 1950a). The avoidance of sucrose could well be accomplished through the mechanisms already described, with fat emerging as the primary calorie source on the grounds that it produced the least aversive consequences. fi Summary of Thhiamine-Specific Hunger. When they are thiaminedeficient, rats solve the problem of obtaining thiamine-rich foods by means of the following adaptations. They quickly learn, via long delay and belongingness, that certain foods make them sicker, and avoid these foods. This leads to a preference for old, safe foods, which are likely to contain thiamine. In their absence, rats systematically sample new foods, testing them, as it were, for consequences. They certainly learn quickly which new foods have aversive consequences and also learn which foods have positive effects. This selection among new foods is aided by salience and social information transfer. Related adaptations, mitigating the effects of deficiency, include increased coprophagy and increased reliance on fat as a calorie source.

SELECTION OF FOODS BY RATS,HUMANS, AND OTHER ANIMALS

47

g. Explanation of Other Specific Hungers. Are the principles described adequate to account for the wide variety of specific hungers that have been reported? On the whole, the answer appears to be yes, with the exception of sodium and a few serious remaining problems. A sure indicator that tasteaversion learning is at work in a particular deficiency syndrome is the presence of anorexia, which characterizes most deficiencies. Since anorexia can be described as food avoidance, it may be that in most cases it is a learned response to deficiency. It probably has adaptive value in some cases, since lower food intake may conserve the scarce essential nutrient. Anorexia, probably for this reason, is a general consequence of malaise. The critical determiner of whether the anorexia is learned is in the response t o new foods or old safe foods. Thiamine-deficient rats avidly ingest such foods (Rodgers and Rozin, 1966), showing that their anorexia is specific to the foods associated with illness. Insofar as this is the case in other syndromes, there is evidence for learned aversion, and the stage is set for a learned specific hunger. All the other components are there already in the food-selection system of the rat. Anorexia appears as a prominent symptom in a number of other B-complex vitamin deficiencies and is associated with clear learned aversions (Rozin and Rodgers, 1967). Specific hungers for riboflavin and pyridoxine (Scott and Quint, 1946a; Rozin and Rodgers, 1967) have been demonstrated. Rodgers (1967a) demonstrated that in each of these cases, as with thiamine, there was no tendency to prefer the flavors of these vitamins, per se, but purely a learned preference and/or aversion. The case of pantothenate is instructive, since Scott and Quint (1964a) failed to find a clear defined specific hunger, in spite of obvious anorexia. The explanation is simple. They showed that pantothenate is virtually tasteless. When a distinct flavor was used either in the deficient or enriched diet, a clear adaptive preference emerged. Specific hungers for vitamins A and D have been difficult to demonstrate (Harris etal., 1933; Young and Wittenborn, 1940; Rodgers, 1967b). This is n o doubt related to the fact that anorexia does nor appear as a prominent symptom in these deficiencies. The built-in mechanism for sodium regulation has already been discussed. Whether rats can learn to prefer sodium-rich foods is not clear. However, it is established that they can learn to avoid sodium-deficient foods (Rodgers, 1967a). Concordant with this finding, anorexia is a component of sodium deficiency. k c h t e r and his colleagues demonstrated a variety of specific hungers for other minerals, both in specific testing situations and in the cafeteria setting (see Richter, 1942-1943, for a review). There is an especially clear-cut increase in calcium consumption with increased need (e.g., parathyroidectomy, lactation) (Richter and Eckert, 1937), and mixed evidence for specific hungers for other minerals, such as potassium (Richter, 1942-1943; Richter and Helfrick, 1943; Scott er al., 1950b). In most of these studies, the critical experiments to indicate

48

PAUL ROZIN

whether or not there is an innate component have not been performed. In the case of calcium deficiency, Rodgers (1967a) demonstrated a learned aversion to a calcium-deficient diet. Adam (1973) has shown that potassium hunger is not potassiumspecific, so that novel diet preferences in potassium-deficient rats cannot be overcome by presence of potassium in a familiar diet, as would be the case with sodium. The case of magnesium is sufficiently bizarre t o merit mention. Scott et d. (1950b) demonstrated an inverse magnesium-specific hunger: magnesiumdeficient rats seem to avoid magnesium-rich diets (confirmed by Rodgers, 1967a,b). This surprising finding might be explained on the assumption that magnesium deficiency produces a “high” (hyperactivity and hyperirritability are symptoms) and that ingested magnesium initially brings the animal “down” from its high. An alternative explanation (E. M. Stricker, personal communication) assumes specific negative consequences, including nausea, following ingestion of magnesium by magnesium-deficient animals. This could come about because of the critical role of magnesium ions in the catecholamine systems. In magnesium deficiency, levels of catecholamines would be expected to be low, with consequent receptor supersensitivity. Magnesium ingestion might result in an initial overresponse of this system, which could involve the nausea center in the brain. Rats have a minimum protein requirement and also must ingest a reasonably balanced mixture of 9 or 10 essential amino acids (Harper, 1964). Insufficient protein or an absence or excess of particular amino acids lead to anorexia, a picture with which we are already familiar. Harper and his colleagues have shown (Rogers and Harper, 1970) that rats will select a food source high in an amino acid in which they are deficient, avoid a diet with an amino acid imbalance, and choose a balanced over an imbalanced amino acid diet ( b u n g et al., 1968; Zahler and Harper, 1972). The positive and negative responses of rats to diets of varying degree of amino acid adequacy can be related to the blood amino acid pattern generated by these diets (Zahler and Harper, 1972). Work by Booth and Simson (1971, 1974; Simson and Booth, 1974) has clearly demonstrated that these protein and amino acid preferences can be attached to arbitrary olfactory or taste cues, indicating that they can be explained within the framework of learned preferences and aversions described above. In the case of protein intake, for which there would appear to be both upper and lower limits, we have evidence of a rather precise modulation of intake according to needs. Thus, Richter’s many cafeteria experiments (Richter, 1942-1943, 1955) show appropriate increases or decreases in protein intake in cases such as lactation, although others have reported a substantial number of rats that fail to select enough protein (Scott, 1946; Scott and Quint, 1946b; Pilgrim and Patton, 1947). Direct evidence for protein regulation comes from a cafeteria experiment (Rozin, 1968b) in which a liquid protein source (casein hydrolysate in water) was offered in varying concentrations, with appropriate changes in volume ingested, so that the total amount of protein ingested remained approximately constant (see also Booth, 1974).

SELECTION OF FOODS BY RATS,HUMANS, AND OTHER ANIMALS

49

Richter’s (1942-1943, 1955) classic work demonstrated that over a wide variety of metabolic conditions, rats self-selecting on a cafeteria behaved adaptively, compensating by their food choices (behavioral homeostasis) for disturbances in internal homeostasis produced by pregnancy, lactation, thyroidectomy, parathyroidectomy, diabetes mellitus, etc. In general, these adjustments in food selection seem t o be consequent upon metabolic disturbances and, thus, can be seen as resulting from learned aversions and preferences, plus the built-in sodium-specific hunger and possibly some unknown other built-in mechanisms. The adaptive choices of normal rats on cafeterias (Richter, 1942-1943,1955; Young, 1944), resulting in excellent growth rates on lower caloric intake than with standard mixed diets, presents a more difficult problem. It is hard t o imagine this adaptive selection being based on incipient deficiencies alone, since such a situation would almost certainly result in a significantly lower growth rate. This classic demonstration of specific hungers remains, then, the most fascinating (Richter, 1955), although some rats (Pilgrim and Patton, 1947; Scott, 1946) fail to thrive on the cafeteria regime, primarily due to inadequate protein intake. Sampling tendencies (Rozin, 1969b) and a tendency to alternate preferences among familiar foods (Holman, 1973; Morrison, 1974), the learning mechanisms already described, possibly a fortunate selection of basic cafeteria choices by the experimenter (especially protein source), plus some presently unknown factors, could together explain the phenomenon (see Lat, 1967).

4. Nature of Taste-Aversion Learning Taste-aversion learning appears t o be a very low-level phenomenon-its impressive characteristics notwithstanding. It is probably widespread among the vertebrates. Its fundamental importance and high reliability suggest rather tight wiring. Seligman (1970) has described it as “prepared” learning and, hence, “primitive” or subcortical (see also Rozin and Kalat, 1971; Garcia et al., 1970; Seligman and Hager, 1972); the evidence supports this view. Most notably, tasteaversion learning can occur in an anesthetized animal (Roll and Smith, 1972), where, presumably higher centers are selectively depressed. Furthermore, the Kamin (“blocking”) effect, what might be considered a higher-order learning effect seen with exteroceptive stimuli, is difficult to obtain with taste-aversion learning (Kalat and Rozin, 1972). Since common sense and human experience seem, retrospectively, to have been very good guides in this field, we might risk a prospective look in this direction. The verdict here is clear: human taste aversions seem to be independent of cognitive control. I have heard of a number of cases in which the situation leading to a specific food aversion is known by the “affected” individuals and it is known that the illness following ingestion of the now aversive food was not caused by the food (for example, often others ate the same food and did not become sick). Yet, a deep-seated aversion remains, uninfluenced by a contrary cognitive “overlay.” This again’suggests a low-level system.

50

PAUL ROZIN

Furthermore, there is a question as to whether the classical conditioning paradigm‘ is appropriate for taste-aversion learning. Most modern investigators of exteroceptive conditioning consider the CS as conveying information about the probability of a US. “Stimulus substitution” is not the preferred formulation. But the behavior of animals and humans toward aversive foods suggests that the foods themselves may arouse strong emotions of disgust. Food aversions have an immediacy of affect usually lacking in exteroceptive CSs. Pfaffman (1960) has pointed out that the taste system itself has an unusual affective loading compared to other systems (see also Young, 1948), and this may be carried over to the taste-aversion paradigm. The recent demonstration of clear taste projections in the rat into the hypothalamus as well as the traditional projection pathways supports this view (Norgren and Leonard, 1973). At any rate, we may be dealing with a new (or rather, very old) kind of learning here. I t seems quite likely that the tastes in taste-aversion learning acquire some of the affective qualities of the US, in contrast to the usually employed exteroceptive CSs, which serve as signals for USs. [Gleitman (1974) has recently suggested methods to determine in animals whether a CS serves as a signal or acquires the properties of a US. As he points out, these methods could usefully be applied to this problem.] 5. Domestication

The great majority of specific hunger experiments have been done on domestic rats. The adaptive food selections shown by these creatures is remarkable, given that they have been raised for fifty or so generations on laboratory chowwithout any selection pressure to maintain their exquisite food-selection abilities. Indeed, great changes in appearance, physiology, anatomy, and behavior have occurred during the domestication process (e.g., see Barnett, 1963; Kavanau, 1964; Richter, 1954, 1959). Some may directly affect food choices, such as decreased neophobia in domesticated rats (Richter, 1953; Rozin, 1968a; Galef, 1970) or changes in the adrenal gland and salt tolerance, which might account for the changes in salt preference (Richter and Mosier, 1959; Richter, 1959). Yet, in spite of all this, the domesticated rat seems surprisingly able to deal behaviorally with nutritional stresses. Surely, there must have been some basic changes in food preferences. With this in mind (and with tongue in cheek), some years ago I initiated a search into the food habits of domestic rats for some sign of the decadence and frailties of so many generations in what Richter (1959) has called a “welfare state.” The results from actual experiments are clear, as shown in Fig. 4. When 4 domestic male rats were offered binary choices between Hebrew National and Genoa salami, they clearly preferred the former (Fig. 4, left). Subsequently, when offered a choice between gefulte fish and shrimp, they showed a massive preference for the gefiilte fish, averaging around 50 gm a day of this delicacy (Fig. 4, center). To further consolidate the view of an emerging preference for kosher foods, a final test choice between Mogen

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

51

David and Virginia Dare wines was arranged. As can be seen in Fig. 4 (right) the kosher hypothesis was strongly supported. I can only hope that such preferences are not present in wild rats-I have not put this to the test. 6.

The Chicken

-

Some Interesting Parallels

Some very old and very recent studies on the genesis of food and water recognition in chicks, omnivores with a somewhat limited food range, provide an instructive comparison with the work on rats described above. Classic work by C. Lloyd Morgan (1894), extended and confirmed by Hunt and Smith (1967), showed that chickens had to learn to identify water visually. The taste of water and the regulation of water intake (Stricker and Sterritt, 1967) are apparently preprogrammed, but neither can come into play until water is visually recognized in the outside world and ingested. Young chicks, virgin with respect to water and rather dehydrated, would run through water puddles without recognizing them. Chicks have a built-in tendency to peck at small irregular objects (e.g., grains). When this happened to occur at an irregularity in water, the association between visual water and "prewired" water taste was rapidly made (Morgan, 1894), so that the chicks immediately began drinking and drank an amount approximately equal to their water deficit (Stricker and Sterritt, 1967). From that moment on, water was recognized visually. Thus, everything but the visual recognition of water is prewired here. The same tendency to peck at small irregularities initiates the development of KOSHER EXPERIMENT

!IO/ 6o

0 Hebrew National Genoa

WINE

FISH

SALAMI

0 GefGlte

0 Mogen David

Shrimp

Virginia Dare

zz

In

40

0

In

2

I

a

0

s

0

FIG. 4. Preferences of 4 albino male rats for kosher versus nonkosher foods. The intakes plotted represent the mean intake in grams per day, with each pairing of foods being represented exclusively over a period of a few days. Sidney died between the fish and the wine. (From P. Rozin, unpublished observations.)

52

PAUL ROZIN

food preferences which have been analyzed by Hogan (1973a,b,c). He notes that, given certain innate stimulus constraints on peckable items, young chicks are directed to useful foods by the following mechanisms. 1. Chicks tend to eat in proximity to the mother hen, whose food call has a directing effect on their pecks. 2. Chicks learn quickly t o reject foods with bad tastes or other irritating properties. 3. Chicks can learn about the consequences of food ingestion by a long-delay learning mechanism similar to that described for rats. 4. Such learning commences on day 3, when yolk sac reserves are down to one-half and the need for an environmental food source is becoming imminent. Long-delay learning does not occur prior to day 3. 5. Initially, what the chick seems to learn is that pecking is desirable, in that pecking followed by metabolic repletion leads to increased pecking. But this pecking increase is also shown for nonnutritive sand, even though edible grains provided the initial repletion experience. 6. Later on, specific acquisition of pecking to edible stimuli occurs. In other words, the first effect of delayed reinforcement is a generalized increase in pecking, which is later discriminated. This may involve further maturation of learning abilities and/or be related to development of sampling patterns, where foods are tried one at a time. Hogan reports that separate experiences with a nutritive and nonnutritive source facilitate discrimination, possibly through a combination of delayed learned aversion and preference. This work is interesting not only for comparative purposes, but for its emphasis on the ontogeny of food recognition, a subject little studied in mammals. Very little is known about how mammals come to recognize food and water.

111.

FOOD SELECTION IN HUMANS

The most striking parallel between human and rat feeding is in the neophobia seen in both. The mouth is the final voluntay checkpoint on the route into the body and, thus, the last opportunity (other than vomiting and diarrhea) to reject dangerous foods. Probably for this reason, strong likes and strong aversions t o tastes or smells under minimal higher or cognitive control appear in rats and humans. Appropriately, as mentioned above, both taste and smell systems have rather direct projections to the hypothalamus, and/or limbic system, which mediate emotional responses (Norgren and Leonard, 1973; Pfaffman, 1960). Even with the enormous overlay of culture in humans, one can clearly see this neophobia at work: indeed, it is at the heart of the conservatism of cuisines. Many observers have remarked that ethnic food habits in minority groups are the last vestiges of the old culture to disappear. The best way to tell the ethnic

SELECTION OF FOODS BY RATS, HUMANS,AND OTHER ANIMALS

53

origins of a particular minority ethnic group is to go into the kitchen. Long after accents and costumes are assimilated in the culture, food habits remain. I shall return to this issue in the later discussion of cuisine. For the moment I would like to discuss the multiple determinants of food selection in man. These can be divided into biological factors and effects of individual experience, on the one hand, and cultural influences, on the other. A.

BIOLOGICAL FACTORS IN HUMAN FOOD SELECTION

There is compelling reason to believe that we have descended from apelike forest dwellers with an omnivorous, but primarily vegetarian diet (Pfeiffer, 1969). Somewhere around 3 million years ago, our apelike ancestors ventured gradually out of the jungle and onto the savannah, motivated very likely, by the possibility of exploiting new food sources. Thus began a shift from a diet probably dominated by fruits and other plant materials, with occasional insects or very small game, to a primarily carnivorous pattern (Pfeiffer, 1969; Morris, 1967). If we assume, as seems reasonable, that our ancestors (Reynolds, 1967; Jolly, 1972) had food habits similar to present-day chimpanzees, then we can assume a marked preference for sweet things such as fruits. [The sweet preference is strikingly illustrated by fruit-eating spider monkeys, which are reported to bite Yocoyena fruit skin when unripe, but not consume the fruit. This accelerates ripening, so that the monkey can return in a day or two to eat the ripe fruit (Jolly, 1972). However, van Lawick-Goodall’s (1971) recent observations suggest that a substantial portion of the chimp’s diet is made up of insects, meat, and nonfruit plant material.] In spite of millions of years on a substantially carnivorous existence, possibly supplemented by a modest ingestion of seeds, the sweet taste is clearly with us. Our taste system, both psychologically and physiologically, to some extent (Pfaffman etal., 1971) consists of the four basic submodalities: sweet, salt, bitter, sour. The sweet system seems to be tied directly into an acceptance o r pleasure system, appropriately given the sweetness of mother’s milk and our ancient beginnings as fruit eaters. Infants prefer sweet solutions to water @esor et al., 1973) and show characteristic “positivey’ facial expressions on their first contact with sweet substances (Steiner, 1973). Conversely, the bitter system seems to have the opposite affective loading. Bitter tastes lead to body and facial movements of rejection (Steiner, 1973) and, in at least some studies of infants, appear to be avoided (see Maller and Desor, 1973, for a summary). Adaptively, this is probably related to the bitter tastes in many naturally occurring poisom, e.g., alkaloids and glycosides (Shallenberger and Acree, 1971; Richter, 1950). How this basic bitter aversion becomes transformed in adults of some societies, into a strong preference for bitter substances such as quinine water or coffee remains a mystery. Possibly, bitter sensitivity decreases around the time of puberty. With these minimal biological constraints (and possibly some additional

54

PAUL ROZIN

taste or olfactory tendencies), the human infant begins life on its all-milk diet. The question is, Are the various special recognition or general learning mechanisms described above present in the human and can they be studied in the face of the vast cultural overlay? After all, the smells in the home and taste and smell of the mother’s milk reflect the food eaten by adults in the family. The possibilities for definitive research o n the uncontaminated basic biological system are limited, but we are fortunate that one remarkable series of experiments on self-selection in human infants, free from most sources of contamination, is available. These are the classic experiments of Clara Davis in the late 1920s and 1930s (Davis, 1928,1935,1939). In the original and most thoroughly documented study, 3 children, weaned in the hospital, were immediately placed on a cafeteria diet in the hospital for the following 6 months to 1 year. Prior t o weaning, 1 had only had milk, 1 milk and orange juice, and the third milk, orange juice, and cod-liver oil. Only natural foods, raw or simply cooked without seasoning, were used. Children were presented with a tray containing about twenty different foods. They indicated selections by pointing to a food, which was then offered by a nurse. Children rotated through three different meal selections each day, with milk, lactic milk, and sea salt available at all meals, as well as two cereals, some meats (including organ meats), and fruits and/or vegetables. Results with these children, plus an additional group (Davis, 1939), were extremely successful. Children sampled rather broadly at first but later narrowed down to a rather stable selection of a narrower range of foods in the 1939 study, and higher variability in 1928. Binges (self-terminating) were reported in 1928. Davis reports that appetite nicely anticipated state of health-dropping 24-48 hours before signs of frank illness and picking up 12-24 hours before other signs of recovery. In the second group of 15, in which the children were on the regime for 1 to 4% years, on the average 17% of calories were taken as protein (range 9-20%)-just in the recommended range. In the initial study (N = 3), Davis (1928) reports greater than average weight gains. No deficiencies appeared on the self-selection regimen. It is hard to evaluate the possibility that deficiencies could have developed, with nutritious foods such as milk, whole-grain cereals, sea salt, and meat making up the majority of the choices, and the absence of highly sweet artificial foods, such as candy or cake. Nonetheless, the reported stability of choices and thriving on the regime is impressive. Davis (1928) provides some sample daily intakes, from which one can estimate relative food preferences. By weight, from the samples provided, milk was most preferred, accounting for 2 6 to 53% of the total weight of ingested food, with fruits next (1449%) and then cereals (10-16%). Meats (organ and muscle) were eaten in smaller quantities, as were vegetables (34%). One can make a case for sweetness being the prime determiner of infant choice, with milk and fruit the most popular sources. Furthermore, the preferred food by all children, at the one meal a day in which it was offered, was prototypical chimpanzee food-raw bananas. The percent of calories taken as fruit is under-

SELECTION O F FOODS BY RATS, HUMANS, AND OTHER ANIMALS

55

estimated, since one meal of the three did not contain any fruit, so that milk was the only sweet choice. Even with this limitation, one child took 41% of calories as fruit. If we can discount subtle influences from the cooperating nurses, the culturally more-or-less unspoiled human infant does seem t o show an adaptive pattern of food preferences. Davis (1935) was sufficiently encouraged that she operated an orthopedic ward for 3- to 12-year-olds on the self-selection principle with good results. It would be highly desirable to repeat these classic studies with some control for “demand” characteristics and some less nutritionally adequate choices. However, the studies stand as a major contribution. Turning now to the correction of nutritional imbalance (specific hungers), there are two striking studies on children. One of the children in the Davis (1928) study had ricketts (vitamin D deficiency) on admission. Cod-liver oil was offered as a choice, along with an addition of cod-liver oil to the milk. Over a period of 101 days, the child consumed 178 cc of pure cod-liver oil (plus 80 cc more in milk). Davis reported that when the child’s blood calcium and phosphorus returned t o normal and X-rays were normal, the cod-liver oil appetite ceased. These are suggestive data at best, especially since recovery from vitamin D deficiency is slow, and vitamin D hunger has been hard to demonstrate in animals. Again, one can worry about demand chracteristics and the effects o n cod-liver oil preference of mixing it with milk. The classic example of specific hungers in children comes, not surprisingly, from the work of Wilkins and Richter (1940). It concerns a 3Yi-year-old boy with the primary symptom of marked development of secondary sexual organs. He was admitted t o the hospital, ate very little of the food, and died suddenly 7 days after admission. Postmortem revealed that death was due to adrenal cortical insufficiency. The child had had a great craving for salt and had eaten salt in large quantities from the age of 12 months. The hospital diet did not give him the opportunity t o ingest enough sodium t o maintain electrolyte balance and probably caused his death. The following is part of the remarkable letter written to Wilkins and Richter by the parents some time after the child’s death (Wilkins and Richter, 1940): When he was around a year old he started licking all the salt off the crackers and always asked for more. He didn’t say any words at this time, but he had a certain sound for everything and a way of letting us know what he wanted. This was the first we had noticed his wanting the crackers or salt. Finally he started chewing the crackers; but he only chewed them until he got the salt off, then he would spit them out. He did the same with bacon, but he didn’t swallow the pieces. When he was about sixteen months old, crackers were the first food he chewed and swallowed; but it was quite a while after that before he would chew up and eat a whole cracker. He would usually just make a mess of them eating the salt off. In an effort to try to find a food that he would like well enough to chew up and swallow, we gave him a taste of practically everything. So, one evening during

56

PAUL ROZIN supper, when he was about eighteen months old, we used some salt out of the shaker on some food. He wanted some, too. We gave him just a few grains to taste, thinking he wouldn’t like it: but he ate it and asked for moR. This was the beginning of his showing that he really craved salt, because this one time was all it took for him t o learn what was m the shaker. For a few days after that, when I would feed him his dinner alone a t noon, he would keep crying for something that wasn’t on the table and always pointed to the cupboard. I didn’t think of the salt, so I held him up in front of the cupboard to see what he wanted. He picked out the salt at once; and in order t o see what he would do with it, I let him have it. He poured some out and ate it by dipping his fmger in it. After this he wouldn’t eat any food without having the salt, too. I would purposely let it off the table and even hide it from him until I could ask the doctor about it. For it seemed t o us like he ate a terrible lot of plain salt, But when I asked Dr. about it, he said, “Let him have it. It won’t hurt him.” So we gave it to him and never tried to stop it altogether. Afte: we gave it to him all the time he usually didn’t ask for it with his dinner; but he wouldn’t eat his breakfast or supper without it. He really cried for it and acted like he had to have it. Foods that he ordinarily wouldn’t touch he would eat all right if I added more salt to them. He would take the shaker and pour some out on his plate and eat it with his finger, but we always tried to keep him from getting what we thought would be too much for him. He never did care much for zwieback, toast or bread or for cooked potatoes, but he did like raw potatoes, raw carrots, celery, tomatoes, lettuce and different other foods if he could dip them in salt. If I didn’t give it to him, he always asked for it. At eighteen months he was just starting to say a few words, and salt was among the first ones. We had found that practically everything he liked real well was salty, such as crackers, pretzels, potato chips, olives, pickles, fresh fish, salt mackerel, crisp bacon and most foods and vegetables if I added more salt.

-

B.

SPECIFIC HUNGERS IN ADULT HUMANS

There is abundant folklore, or anecdotal evidence, and few hard data on changes in human food habits in response t o nutritional deficiencies or excesses. The best that can be done here is to give an impression of the kind of information available, realizing full well that we may sometimes be in a situation in which cultural traditions operate in opposition to sound nutritional practice. There is an association between pica (clay, starch, or earth eating) and irondeficiency anemia (Cooper, 1957). This widespread practice is most common in children and pregnant women, i.e., during periods of high nutritional demands. We may be dealing here with a mix of deficiences, often including iron, and a cultural tradition which, under some circumstances, has adaptive value, since some of the clays eaten are high in essential minerals. A puzzling increased frequency in ice eating (pagophagia) is also seen in association with iron deficiency in rats and humans (Woods and Weisinger, 1970). The high calcium demands of lactation present special problems in cultures where the prime sources of calcium, milk and milk products, are not consumed. China is a prime example. In this light, the Chinese custom (de Castro, 1952; Simoons, 1961) of

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

57

chewing on sweet and sour spareribs by lactating mothers makes great sense. The acidity (vinegar) tends to render some of the calcium in the bones utilizable. Humans have a sorry record in adjusting food intake to alleviate vitamin deficiencies. In most cases, in Western civilization at least, it was the scientific segment of culture that ultimately brought to light the proper corrective measures. In many cases this did not occur until the development of modem nutritional sciences in this century. Beri-beri has been endemic in many parts of Southeast Asia because of the custom of milling off the vitamin-rich coating of rice before storing and ingesting it. During the age of exploration, vast numbers of sailors died of scurvy, because the simple nutritional cure, vitamin C, ordinarily via fresh fruit, was not understood. Columbus left some men (at their request) on a Caribbean Island t o die of severe scurvy (de Castro, 1952). They ate fruit, recovered, and were picked up by a boat some months later. The island received the name Curacao-meaning cure in Portuguese-from this happening. Cartier, in 1535, was told by the Indians of the cure but didn’t believe them (Lowenberg et ~ l . ,1968). Lind in 1753 did some experiments on scurvy treatment with fresh fruit on his ship with striking results. However, it took the Navy 50 years to accept and implement these results, finally putting to an end massive deaths on the long sea routes. The fresh fruit treatment of scurvy is the origin of the term “limey,” applied t o British seamen. The discovery of the treatment of other vitamin deficiencies occurred even later-the wisdom of the body appears surprisingly fragile. There is spotty evidence for at least interesting diet changes during periods of nutritional stress produced by pregnancy and lactation or disease. Most striking, though utterly puzzling, are the strange cravings and aversions associated with pregnancy. In studies in England (Trethowan and Dickens, 1972), cravings or aversions were found in one-half to two-thirds of the pregnancies, with most cravings directed toward fruits (30%of the cases). These cravings and aversions tend to appear in the first trimester. At the moment, it is difficult to map these onto the orderly adaptive behavior seen in rats in Richter’s (1955) work. One can only speculate that hormonal-metabolic changes of pregnancy, which, in “simpler” organisms trigger adaptive behavior sequences, in humans interact with habits and culture in a peculiar way. All we see in the human food habits is that there is a change inside the organism. We have established at least some biological basis for human appetites. In addition to the weak evidence for biological constraints influencing food choice, we can be quite confident that taste-aversion learning over long delays occurs in humans. The anecdotal evidence in this case is overwhelming and there are also some data from the food-poisoning literature. Garb and Stunkard (1974) in a questionnaire given to over 600 American subjects of varying ages, found reports of aversions in 36% of subjects, 88% of these associated with gastrointestinal upset.

’

58

PAUL ROZIN

C. BIOLOGICAL BASIS OF ETHNIC-RACIAL DIETARY DIFFERENCES

Human food habits are characterized by their diversity in different parts of the world. This diversity is explainable in large part, by variations in the availability of foods on different parts of the earth and, of course, by cultural influences. Nonetheless, one may ask whether any of the substantial genetic differences among races or ethnic groups contribute a biological determinant to food practices or vice versa. I will discuss three examples here: carbohydrate metabolism in Eskimos, phenylthiocarbamide tasting, and lactase deficiency. I expect that there are many more examples to be discovered in the existing literature or through research. The high-protein and -fat and extremely low-carbohydrate diet of Eskimos is unique in the world. Going along with this dietary pattern of long standing is a parallel metabolic adaptation. Schaeffer (1969a,b) has noted a very high incidence of abnormal glucose tolerance curves to orally administered glucose in Canadian Eskimos. Twenty-five percent of normal Eskimos showed abnormal curves, but most had n o other signs of diabetes. Tolerance to intravenously administered glucose was much higher. Schaeffer postulates that a gastrointestinal hormone that stimulates insulin secretion in the presence of carbohydrate in the gut is often absent in Eskimos. He suggests that this is a genetic adaptation, although the possibility remains that it is produced within the lifetime by the very low-carbohydrate diet itself. Alternatively, he suggests, release of the hormone in Eskimos may be under the control of amino acids in the gut. In either case, the suggestion is that a particularly unusual diet is associated with major metabolic changes.

1. Tasting of Phenylthiocarbamide The ability to taste phenylthiocarbamide (PTC) represents a more widespread trait, which has been studied as one of the best examples of a single-allele effect with behavioral implications. Phenylthiocarbamide is a synthetic thyroid antagonist, closely related biochemically and physiologically to natural goitrogens such as thiourea. Natural goitrogens are found in a number of edible plants, including cabbage, turnips, and peas. Phenylthicarbamide and related compounds taste bitter to most persons, but a significant number of people are unable to taste it at concentrations producing a strong bitter sensation in tasters. The nontaster trait is under the control of a single recessive allele (Kalmus, 1971). There is a polymorphism for the trait, with different frequencies at the two modes, depending on the racial-ethnic population studied. Richter and Clisby (1941) were among the first to point to the bimodality inPTC thresholds,in both rats and man. Thirty percent of Caucasians (West European and North American) are nontasters, compared to 10.6% of Chinese, 6.4% of American Negroes, and 1.9% of natives of highland Peru (Greene, 1974). The general notion is that nontasting is maladaptive in areas where there is a combination of the presence o f natural

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

59

goitrogens in the diet and shortage of iodine in the environment (Fischer, 1967; Greene, 1974). The incidence of nontasters is higher in some thyroid diseases, suggesting either a link between the defect producing thyroid dysfunction and the taste defect or that undiscriminating ingestion of bitter goitrogens produces thyroid pathology (Fischer, 1967). (There is no direct evidence of thyroid pathology produced by ingesting natural goitrogens in normal quantities.) Greene (1974) has recently reported work on 6- to 15-year-olds in two Andean communities in Equador which suggests a behavioral (goitrogen-avoidance) interpretation. Both communities have endemic goiter, a level of cretinism around 7%, and presence of natural goitrogens in foods normally included in the diet. In one of the communities, all youngsters were given iodine supplementation, in the other they were not. Greene reports a significant correlation between PTC taste sensitivity and “neurological maturation” (Bender Gestalt) only in the population not protected with iodine. He suggests that the PTC sensitivity has protective value since tasters are less likely to ingest bitter-tasting foods containing goitrogens. This is likely to be a direct behavioral effect rather than a direct effect of the allele on thyroid function (with taste changes as a pleiotropic effect) since PTC sensitivity is correlated with sensitivity to other bitter substances (quinine), suggesting taste as the salient selective factor. In order to explain the polymorphism, Greene points out some advantages accruing to nontasters-including a significant tendency toward hyperthyroidism in adulthood in some tasters. In summary, the hypothesis is that tasters can successfully avoid bitter tasting goitrogens, which has adaptive value in geographic areas where goitrogens are potentially significant in the diet and iodine is in short supply. Greene (1974) claims that the distribution of nontasters in the world is consistent with this hypothesis. 2. Lactose Intolerance

There is one biological factor that has profoundly affected ethnic-racial food pattern differences. This is the substantial decrease in levels of the enzyme lactase, which digests milk sugar, after infancy. As shall be seen shortly, the incidence of this enzyme deficiency that relates to a very basic food has had a significant impact on some world cuisines, especially oriental cuisines. Lactose, a disaccharide made from galactose and glucose, is the only carbohydrate in milk (Kretchmer, 1972). It constitutes 6.5-776 of fresh human milk, and 4.5% of cow’s milk by weight, and accounts for 4% of the calories in human milk (McCracken, 1971; Nelson etal., 1969). Lactose is broken into absorbable constituents-galactose and glucose, by the intestinal enzyme lactase. Lactase is obviously present in the gut of virtually all human infants-otherwise breast or other milk feedings would be disastrous. In very rare cases, the enzyme is absent at birth (genetically controlled, primary lactase deficiency), requiring a milk substitute or specially processed milk diet from birth. We are concerned

60

PAUL ROZIN

here with secondary lactase deficiency, a notable decrease of lactase in early childhood. This decrease usually occurs in the age range of 2 to 4 years for humans (Kretchmer, 1972). Lactase levels drop off rapidly in other mammals that have been studied, the shaq drop in the rat being in the range of 2 to 4 weeks of age. Adult lactose-tolerant humans seem to have the same lactase as infants, so that the difference between deficient and nondeficient subjects appears to be whether the lactase produced in infancy continues to be produced in significant amounts. Lactose cannot be efficiently digested and absorbed by intolerant adults. When eaten in significant amounts, it draws excess water into the gut and produces gas by a fermentation process m the gut. Characteristlc symptoms are thus gastrointestinal upset, bloating, and diarrhea (Kretchmer, 1972; McCracken, 1971). Note that the response to lactose is not an allergic type reaction, so that a substantial intake is required to produce significant symptoms. Typically, one or more glasses of milk produce symptoms in 30 to 90 minutes (Simoons, 1969). These gastrointestinal symptoms must form the basis for a learned aversion. There is extensive data on the incidence of secondary lactase deficiency in groups all over the world. Lactose tolerance is determined either by report of symptoms following lactose ingestion or measurement of blood glucose after orally administered lactose. Blood glucose will rise only if the glucose breakdown product of lactose gets into the blood. There is an enormous range in tolerance across different racial or ethnic groups. Only 10-20% of Caucasians of North European origins are lactose intolerant, compared to high levels in the range of 80 to 99% intolerance for many Oriental and African groups. It is not surprising that there is a relationship between lactose tolerance and amount of milk in the diet. Thus, in Nigeria (Kretchmer, 1972), the Yoruba, originating in the Congo, and without a tradition of keeping cattle are 99% intolerant. In contrast, the Fulani, a pastoral nomadic group with a long history of milk drinking show only 22% intolerance. The general relationship between milkdrinking habits and lactose intolerance is displayed in Table I. Given the subject under discussion, it seems appropriate that we should have to ask, Does milk drinking induce lactase production or does lactase deficiency lead to milk avoidance? The weight of evidence appears to favor the latter-a genetic control of lactase inactivation (Simoons, 1969; Kretchmer, 1972; McCracken, 1971). There is a suggestion (McCracken, 1971) that lactase is under the control of a single locus, with possibly two separate alleles-one programming production of lactase in infancy and the other throughout the lifetime. Although it would at present be hard to establish this point, the following facts strongly support genetic determination in general, with implications regarding the underlying mechanism. 1 . With one lactose-intolerant parent, 45% of offspring are intolerant (McCracken, 1971).

61

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

2. In marriages between Ibos or Yorubas (almost entirely intolerant) with Europeans, most children are tolerant (Kretchmer, 1972). 3. There are other documented, genetically based, disaccharide-splitting enzyme deficiencies. 4. Most animal studies have shown little or n o success in inducing or markedly prolonging lactose tolerance in young animals by milk or lactose administration (Simoons, 1969; Leichter, 1973). 5 . Thai children in institutions in which they were regularly fed milk over the first 2 years became intolerant at about 2 years (Simoons, 1970). 6. Members of ethnic groups with high intolerance, when transplanted to milkdrinking cultures, continue to show high intolerance (e.g., American Blacks). However, this is not entirely compelling, given the apparent popularity of milk among current-day Japanese and, on the animal side, the frequent ingestion of milk by pets such as cats. Assuming a genetic basis, we now ask, What came first, adult lactase deficiency or adult lactase production? Which trait is primitive? The answer here is fairly clear, with the arguments mustered by Frederick Simoons (1969, 1970), an outstanding cultural geographer. Data and common sense support lactase deficiency as primitive, since there is n o need for an ability to digest milk in any adult mammal other than man-and for man only in the last eight or nine thousand years. The argument is, then, that with domestication, there would be an adaptive value for adult lactase, and thus an appropriate selection pressure. What is known about the origin of dairying and its dissemination fits, in general, with the lactose-intolerance incidence: nonmilking groups today show low tolerance, along with migrants from these areas (Simoons, 1970). TABLE I SUMMARY OF LACTASE DEFICIENCY BY CULTUREa Table No.

9 10 11 12 ~~

Table title Herders Hunters and Gatherers Cultures without Dairy Animals Cultures with Dairy Animals but No Adult Milk Consumption (Omitting children) Cultures with Low Milk Production, Variable Consumption Cultures with High Milk Production, Generally High Consumption North American Caucasian Total ~

~~

‘Data from McCracken (1971).

No. subjects

No. deficient

% Deficient

56 16 171 694

21 61 144 566

38 80 84 82

(351) 58

(87) 66

1,810

247

14

582 3,417

91 1,188

16 34

62

PAUL ROZIN

One might ask why, in the primitive state, lactase production was inactivated. I can think of two reasons. First, production of a useless enzyme is a waste of energy-there are a number of examples of metabolic processes that are phased out after infancy, presumably on these same grounds. Second, it is possible that lactase phase-out is a means of promoting weaning in mammals (Marcia Levin, personal communication). The phase-out seems to occur around weaning time in the species studied, and it might be of considerable adaptive value to facilitate this difficult but necessary separation by imparting some aversive character to the milk. Some contemporary mothers do this with quinine on the nipple. The serious problem remaining is to suggest a sequence that would have led to strong selection for adult lactase in some populations. Simoons (1970) hypothesizes that milk was first drunk, in small nontoxic amounts, as a cult offering, and that this formed a base for a gradually increasing milk-drinking habit, coevolving with milk tolerance. I think it is also possible that milk just tasted good and may have been sampled by either adults or very young children prematurely weaned. But whatever the initial scenario, there are serious problems. First, note that many milk products, such as yoghurt or many cheeses, contain very little lactose (Simoons, 1970). Thus, a milk culture could develop before raw milk tolerance. It is hard to imagine a culture that would not have discovered souring or fermenting of milk, since both techniques are widespread in the world, and milk, by its nature, is likely to end up that way. And what would be the selective advantage for a cheese- or yoghurt-based culture to move toward milk? Second, assuming either unavailability of cheeses or yoghurt, or a culture based on them, why would lactose tolerance have a clear selective value? Surely, if a small minority of a group was able to digest lactose as adults, it would not be in a small society’s interest to stop killing animals in order to get milk for a few. Going along with Simoons, we would have to argue that strong selection for lactose tolerance would occur in what must have been a relatively rare group that had not accidentally discovered cheese, butter, yoghurt, or fermented milk and that was under some special food privation such that milk became a particularly valuable food source. Under these circumstances, families who were tolerant (and, of course, it would run in families) would have a much greater chance for survival. The lactase story is obviously of fundamental importance to the understanding of man’s food habits. Many questions, especially pertaining to the origin of lactose tolerance, remain. However, it is already clear that it is a major shaper of man’s food habits: the total absence of milk products from Chinese cuisine, in spite of the fact that the Chinese have been heavily exposed to a culture using milk in the form of the neighboring and invading Mongols, must be in part attributable t o lactase deficiency. D.

CULTURE AND CUISINE

We finally arrive at what is distinctively human about food selectionorganized bodies of knowledge and tradition. But it is not quite uniquely

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

63

human. Recent work by a number of Japanese zoologists has introduced examples of cultural tradition concerning food in nonhuman primates (Macaca fwscata) (Kawai, 1965). A troop of these monkeys, living on Kashima Island in Japan, has been under observation for many years. In 1953, a 1%-year-old female (Imo) initiated a new feeding practice called sweet potato washing by the observers. A sweet potato was held under the water with one hand, and brushed (presumably to remove sand) with the other hand. Over the next 4 or 5 years, this behavior spread as a characteristic way of treating sweet potatoes prior to eating. The route of transmission was primariIy through lineage and playmate relationships. By 1958 about 80% of monkeys aged 2-7 had acquired it and 18% of the adults. All monkeys over 5 that acquired it in this early stage were females. Starting about 1958, the mode of transmission of this “tradition” shifted to a mother-to-child pattern. Young monkeys, eating while their mother was eating, would be directly exposed to the potato-washing ritual, and pick it up quite naturally. Almost all children born after 1958 acquired sweet potato washing. Sweet potato washing took an interesting turn around 1958, when some monkeys began washing the potatoes in the ocean, rather than the brook previously used. By 1961, all monkeys that washed potatoes did at least some in salt water; this probably resulted both from the easier availability of salt water and a preferred taste for the salted potatoes. Subsequent to the shift toward salt water, some monkeys commenced a new variant of washing, called “seasoning.” In this case, between bites, the potato was continually dipped into the salt water. This behavior appears to be supported by taste enhancement, although there is no direct evidence for this. As of 1965, some monkeys did a substantial amount of “seasoning” while others remained straight “washers.” We have in this fascinating work unique observations of the origin and propagation of food traditions. Ritualized food preparation and seasoning have already been demonstrated. In addition, in another case involving the same monkey troop, a trait called “wheat washing” has spread through a population. This involves taking a handful of wheat and sand (the wheat having been thrown on the sand by the observers) and throwing it in the water. This rapidly separates the wheat and sand, making sorting of the wheat much quicker than would otherwise be possible. Imo, again, was the originator of this practice. One can only wait with anticipation for further developments from this exciting line of investigation. Most of man’s daily feeding choices are made within rather severe constraints, which can be defined as the cuisine of the culture in question. In fact the cuisine, coupled with limited availability of types of foods at particular times, forces the eater’s hand, or mouth as it were, making the individual choice for a meal in a nonaffluent society rather meager. However, we should not be too hasty in ascribing this lack of choice t o limited basic ingredients. As my wife discovered when we were going through graduate school together on a low budget, there are lots of ways of preparing tuna fish (masquerading under such

64

PAUL ROZIN

appetizing names as “tuna dream” or “tuna whiz”). The fact is, with a small number of ingredients, some utensils, and fire, one can make a wide variety of dishes. Yet only two of the world’s major cuisines, French and Chinese, seem to be almost limitless in tapping very many combinations, but selection is highly constrained even within these two cuisines. The fact is that man himself has introduced constraints that severely limit acceptable dishes. Recipes are just such constraints, whether written or passed on, appropriately enough, by mouth. I propose, in this last portion of the paper, to try to describe briefly the types of rules or constraints that characterize man’s cuisines and, then, in keeping with earlier material, to try to relate the nature of cuisine to some basic determiners of food selection. Many of the ideas and examples 1 will discuss come from Elisabeth Rozin’s (my wife’s) work in attempting to extract the principles or distinctive features of cuisines and understand their nature (Rozin, 1973, 1975). Are there any universals of man’s cuisines? The answer would seem to be no-but this should not discourage an attempt at understanding general underlying processes. For the unique species that can make do near the Arctic circle, in forbidding deserts, on coral reefs, at elevations over 10,000 ft, it is not surprising that geographic or ecological constraints may overwhelm natural tendencies. The fact remains that when cultures highly deviant in this sense are sometimes excluded (Eskimos being common candidates for exclusion), certain characteristics emerge: (a) almost all cultures practice cooking; ( b ) almost all cultures either prepare an alcoholic beverage or explicitly prohibit it (the exception that proves the rule); (c) almost all cultures have a cuisine, i.e., a set of rules about what to eat and how to prepare it; (d)almost all cultures have characteristic staple foods, characteristic methods of preparation, and characteristic flavors used with their foods. It is surprising how simply one can define a particular cuisine, if the purpose is t o get at the core of the cuisine and encompass its principal dishes. E. Rozin (1 973) has analyzed cuisine into four components. 1. Basic foods-the basic nutrient sources. These arelwere clearly selected largely on the basis of local availability and accessibility of nutrients. Thus, in southern China, rice, eggs, chicken, pork, fish, shellfish, and a few vegetables dominate the cuisine. On the other hand, in the Middle East, wheat, barley, lamb, and goat meat are dominant. 2. Manner of preparation. Again, with a wide variety of possibilities, in this case not so limited by nature, each cuisine selects a few methods for repeated use-both simplifying and giving character to the cuisine. Thus, for southern Chinese cuisine, brief rapid heating (stir fry) is most common, with steaming and deep frying also common, whereas Middle Easterners rely primarily upon stewing (see Rozin, 1973, for further examples). 3. Flavor principles. E. Rozin has argued that the most distinctive definer of the character of a cuisine is the characteristic flavor combinations regularly

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

65

placed on (in a sauce) or mixed with the basic foods. She calls these combinations flavor principles, and attempts to abstract them, as a set of distinctive features, from the corpus of typical foods in a cuisine. The proof of the pudding for the centrality of these principles is that, when applied t o a basic food not characteristic of a particular cuisine, the product turns out to seem to be an “instance” of that cuisine. Thus, potatoes, which are rarely used in China, if stir-fried with soy sauce and gingerroot will taste Chinese, since this is the basic flavor principle of Chinese cuisine (supplemented with garlic, sesame oil, sugar, vinegar, and a few other ingredients on occasion). By contrast, the Middle Eastern flavor principles include a lemon-parsley combination and an olive oil, tomato, and cinnamon combination. Each of the world’s major cuisines has been so characterized (Rozin, 1973). 4. Cuisines also involve a host of rules having to do with who can eat when, what foods can or cannot be mixed with particular other foods, and so on. The points I wish to make here are that cuisines can be defined in rather simple terms and that they are extremely stable and resistant to change. E. Rozin’s research (personal communication) shows that some cuisinesMexican, Indian (subcontinent), Chinese, Middle Eastem-have remained basically the same over many thousands of years. Simplicity and stability surely suggest basic important underlying determinants at work. What are they? Why do we have cuisines? I would like to suggest four answers, and spend some time on the two that have direct relevance to the subject of this paper. Let me begin with a brief mention of two functions of cuisine that are extrinsic to the fundamental nature of food per se and, hence, of only peripheral interest here. First, cuisines can become an art form and, hence, a means of expression and esthetic satisfaction for man, in the same sense as music or art. Second, cuisine as a characteristic of a culturally coherent group, serves as one of many means of identifying that group, setting it off from others, and also making social distinctions within members of the group, or distinctions among occasions within the group. All this may involve overt or symbolic relationships (Levi-Strauss, 1964, 1966). The racial identity function of cuisine may itself be a major determiner of food practices. In part, it may explain some of the conservatism of cuisine. It may account for the origin of some kosher food practices as ways for the ancient Jews to set themselves apart from surrounding tribes. The other two functions of cuisines are intrinsic, that is, directly related to eating and nutrition. Thus, as a third function, cuisines embody some of a culture’s accumulated wisdom about foods. This is not to say that some traditional food practices are not nutritionally maladaptive; however, many practices are clearly functional. The institutionalization of cooking in virtually all cuisines would be such an advantage. Cooking serves to kill some potentially dangerous

66

PAUL ROZIN

microorganisms, render food easier to chew, and in many cases increase its nutritional value by making a larger proportion digestible (see Renner, 1944, for a discussion of this and related issues). Whether any of these factors directly contributed to the origin of cooking is problematical. An improvement in taste may be the critical factor. But the maintenance of cooking practices is almost certainly related t o some of these advantages. The avoidance of milk institutionalized in Chinese cuisine is no doubt in part a cultural adaptation to high lactose intolerance. In this case, cultural mechanisms may have overresponded, since Chinese cuisine also rejects milk products (e.g., cheese) low in lactose. Manioc (cassava) is a staple carbohydrate source in many parts of tropical South America and West Africa. A principal variety is quite toxic, containing what can be dangerous levels of cyanide. Tradition among Brazilian natives in preparation of manioc for ingestion includes crushing, rinsing, and pressing. In this way, most cyanide is washed away, and a rich, easy to grow and resistant carbohydrate source is made available by a traditional preparation procedure (Jones, 1959). Interestingly, when manioc was introduced t o Africa by Portuguese trades in the late 15OOs, they brought the preparation techniques with them and, hence, the detoxification procedure. Cassava is now a staple in parts of Africa, and in many areas the toxic variety is the preferred form (Jones, 1959). Many groups, undoubtedly through trial and error, have developed combinations of staples that complement each other and form the basis of an adequate diet. Thus, for example, corn and beans form the basic protein sources in traditional Mexican cuisine. Although the proteins of corn and beans are each deficient in a different essential amino acid, together they provide a reasonable amino acid balance. We can be quite confident that, in general, there is strong selection in cuisines to provide adequate nutrients, since literal survival is at stake. Again, as with Southeast Asian habits of removing the vitamin-rich rice hull, strong tradition can often act in opposition to nutritional wisdom on occasion, but the pressures for nutritional adequacy of cuisine are strong and must usually have the upper hand. Now, of course, with synthetic or highly refined vitamins and other nutrients, the pressure for nutritional adequacy of cuisine per se is relaxed. A fourth, somewhat speculative, function of cuisines concerns their role as modulators of food neophobia. I have mentioned the drama of ingestionallowing a foreign substance to pass into the body. There is a real and present danger in the act of eating, and the stranger the food the more frightening is the experience. Eating can be both nerve-racking and satisfying. One way, in accordance with our animal heritage, of reducing the tension of ingestion, is to add a characteristic, distinctive, and familiar taste to one’s food. In this respect, familiarity breeds content.

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

67

Flavor principles are then seen as ways for clothing foods in familiarity. And, paradoxically, the characteristic sauce(s) of a cuisine may become the vehicle for successfully incorporating new staple foods, as may happen occasionally, into a cuisine. The familiar flavors blunt the neophobic edge. If this view is true, the flavor principles should be the most conservative aspect of cuisine. This seems true, on the whole, although the wholesale adoption of the tomato as a flavoring element in Mediterranean cuisines would be a glaring exception. Man’s food habits, selections, and cuisines have just been “braised” in this discussion. We have yet to understand the basic forces behind the evolution of man’s foods over the course of the history of civilization. That food has been a potent enough force to have caused wars and formed a basis for the wealth or poverty of nations is unquestioned (Tannahill, 1973). That, in the form of the spice trade, that mysterious search for seasonings, it was a major force in the history of the world over a period of many hundreds of years cannot be denied. That food selection plays a significant role in health, and especially obesity, is certain. I have tried here only to raise some questions, suggest some solutions, and whet my own and hopefully the reader’s palate for more answers. Acknowledgments 1 thank Elisabeth Rozin for development of some of the ideas presented in the latter part of this paper, and my former students, Willard Rodgers, Bennett Galef and James W. Kalat for contributing to some of the formulations presented in the earlier part of this paper. I also thank Jeanette DeSor, Elisabeth Rozin, and Edward M. Stricker for valuable comments on this manuscript and Curt Richter for opening up this field and setting an example of the highest quality of research. Much of the author’s research described herein was supported by the National Science Foundation. A collection of a number of articles relevant to the issues raised in this paper may be found in Kare and Maller (1967).

Refelences Adam, W. R 1973. Novel diet preferences in potassium-deficient rats. J. Comp. PhysioL PsychoL 84,286-288. Andik, I., Donhoffer, S., Farkas, M., and Schmidt, P. 1963. Ambient temperature and survival on a protein-deficient diet. Brit. J. Nutr. 17,257-261. Balagura, S., Brophy, J., and Davenport, L. D. 1972. Modification of learned awxsion to LiCl and NaCl by multiple experiences with LiCI. J. Comp. Physiol. Psycho/. 81, 2 12-219.

Barnes, R. H. 1962. Nutritional implications of coprophagy.Nutr. Rev. 20,289-291. Barnes, R. H., Fiala, G., McGehee, B., and Brown, A. 1957. Prevention of coprophagy in the rat. J. Nutr. 63,489498. Barnes, R H., Kwong, E., Delany, K., and Fiala, G. 1960. The mechanism of the thiaminesparing effect of penicillin in rats. J. Nutr. 71,149-155. Bamett, S.A. 1956. Behaviour components in the feeding of wild and laboratory rats. Behaviour 9 , 2 4 4 3 . Bamett, S. A. 1963. “The Rat. A Study m Behaviour.” Methuen, London. Bartoshuk, L. 1972. The chemical senses. 1. Taste. In “Woodworth and Schlosberg’s Experi-

68

PAUL ROZIN

mental Psychology, Vol. 1 : Sensation and Perception” (J. W. Kling and L. A. Riggs, eds.), 3rd Ed., pp. 169-191. Holt, New York. Booth, D. A. 1974. Food intake compensation for increase or decrease in the protein content of the diet. Behav. Biol. 1 2 , 3 1 4 0 . Booth, D. A. and Simson, P. C. 1971. Food preferences acquired by association with variations in amino acid nutrition. Quart. J. Exp. Psychol. 23, 135-145. Booth, D. A. and Simson, P. C. 1974. Taste aversion induced by an histidine-free amino acid load. Physiol. Psychol. 2, 349-351. Brackbill, R. H., Rosenbush, N., and Brookshire, K. H. 1971. Acquisition and retention of conditioned taste aversions as a function of the taste quality of the CS. Learn. Moriv. 2,341-350.

Bradley, R. M., and Mistretta, C. M. 1971. Intravascular taste in rats as demonstrated by conditioned aversion to sodium saccharin. J. Comp. Physiol. Psychol. 7 5 , 186-189. Braveman, N. S., and Capretta, P. J. 1965. The relative effectiveness of two experimental techniques for the modification of food preferences in rats. Proc. Annu. Conv. Amer. Psychol. ASS.73,129-130. Bruce, H. M., and Kennedy, C. C. 195 1. The central nervous mntrol of food and water intake.Proc. Roy. SOC.,Ser. B 138,528-544. Burghardt, G. M. 1967. The primacy effect of the first feeding experience in the snapping turtle.Psychon. Sci. 7,383-384. Capretta, P. J. 1961. An experimental modification of food preferences in chickens. J. Comp. Physiol. Aychol. 54,238-242. Carr, W. J. 1952. The effect of adrenalectomy upon the NaCl taste threshold in the rat. J. Comp Physwl. PsychoL 45,377-380. Chitty, D., and Southern, H. N. 1954. “Control of Rats and Mice.” Oxford Univ. Press, London and New York. Cooper, M. 1957. “Pica.” Thomas, Springfield, Illinois. Davis, C. M. 1928. Self-selection of diets by newly weaned infants: an experimental study. Amer. J. Dis. Child. 36,65 1-689. Davis, C. M. 1935. Self-selection of food by children. Amer. J. Nurs. 35,402410. Davis, C. M. 1939. Results of the self-selection of diets by young children. Con. Med. Ass. J. 41,257-261.

de Castro, J. 1952. “The Geography of Hunger.” Little, Brown, Boston, Massachusetts. Denton, D. 1967. Salt appetite. In “Handbook of Physiology, Sect. 6: “Alimentary Canal” (C. F. Code and W. Heidel, eds.), Vol. 1, pp. 5 4 3 4 5 9 . Amer. Physiol. SOC., Washington, D.C. Desor, J. A, Maller, O., and Turner, R E. 1973. Taste in acceptance of sugars by human infants. J. Comp. Physid. Aychol. 84,496-501. Dethier, V. G. 1967. Feeding and drinking behavior of invertebrates. In “Handbook of Physiology, Sect. 6: Alimentary Canal” (C. F. Code and W. Heidel, eds.), Vol. 1, pp. 79-96. Amer. PhysioL SOC.,Washington, D.C. Dethier, V.G. 1969. Feeding behavior of the blowfly. In “Advances in the Study of Behavior” (D. S. Lehrman, R k Hinde, and E. Shaw, eds.), Vol. 2, pp. 112-266. Academic Press, New York. Dethier, V. G. 1973. Electrophysiological studies of gustation in lepidopterous larvae. 11. Taste spectra in relation to food-plant discrimination. J. Cornp Physid. 82,103-134. Domjan. M. 1973. Role of ingestion in odor-toxicosis learning in the rat. J. Comp. Physiol. PsychoL 84,507-521. Domjan, M., and Wilson, N. E. 1972a. Specificity of cue to consequence in aversion learning in the rat.Psychon. S c i 26,143-145. Domjan, J., and Wilson, N. E. 1972b. Contribution of ingestive behaviors to taste-aversion learning in the rat. J. Comp. Physiol. Psychol. 8 0 , 4 0 3 4 12. Epstein, A. N. 1973. Epilogue: retrospect and prognosis. In “The Neuropsychology of

SELECTION O F FOODS BY RATS, HUMANS, AND OTHER ANIMALS

69

Thirst” (A. N. Epstein, JZ. Stellar, and H.! Kissileff, eds.), pp. 315-332. Winston, Washington, D.C. Epstein, A. N., and Teitelbaum, P. 1962. Regulation of food intake in the absence of taste, smell, and other oropharyngeal sensations. J. Comp. Physiol. Psychol. 55,753-759. Fischer, R. 1967. Genetics and gustatory chemoreception in man and other primates. In “The Chemical Senses and Nutrition” (M. R Kare and 0. Maller, eds.), pp. 61-81. Johns Hopkins Press, Baltimore, Maryland. F ~ m k i n K. , 1971. Interaction of LiCl aversion and sodiumspecific hunger in the adrenalectomized rat. J. Comp. Physiol. Psychol. 75, 3240. Frumkin, K. 1972. Sodium and Calcium Specific Hungers: Similarity of Response to Preand Postoperative Taste Aversions. Ph.D. Thesis, McCill Univ., Montreal. Fuchs, J. L., and Burghardt, G. M. 1971. Effects of early feeding experience on the responses of garter snakes to food chemicals. Learn. Motiv. 2,271-279. Galef, B. G. 1970. Aggression and timidity: responses to novelty in feral Norway rats. J. Comp. Physiol. Psychol. 70,370-38 1. Galef, B. G., and Clark, M. M. 1971. Social factors in the poison avoidance and feeding behavior of wild and domestic rat pups. J. Comp. PhysioL PsychoL 75,341-357. Galef, B. G., and Clark, M. M. 1972. Mother’s milk and adult presence: two factors determining dietary selection by weanling rats. J. Comp Physiol. Psychol. 38,220-225. Galef, B. G., and Henderson, P. W. 1972. Mother’s milk: a determinant of the feeding preference of weanling rat pups. J. Comp. PhysioL Psychol. 78,213-219. Galef, B. G., and Sherry, D. F. 1973. Mother’s milk: A medium for transmission of cues reflecting the flavor of mother’s diet. J. Comp. Physiol. Aychol. 83,374-378. Garb, J., and Stunkard, A. 1974. Taste aversions in man. Amer. J. Psychiat. 131, 12041207. Garcia, J., and Ervin, F. R 1968. Gustatory-visceral and telereceptor-cutaneous conditioning-adaptation to internal and external milieus. Commun. Behav. Biol. , Part A 1,389-415. Garcia, J., and Koelling, R A. 1966. Relation of cue to consequence in avoidance learning. Psychon. Sci 4,123-124. Garcia, J. and Koelling, R A. 1967. A comparison of aversions induced by X-rays, toxins, and drugs in the rat. Radat. Res, SuppL 7,439-450. Garcia, J., Kimeldorf, D. J., and Hunt, E. L. 1961. The use of ionizing radiation as a motivating stimulus. Psychol. Rev. 68,383-395. Garcia, J., Ervin, F. R, and Koelling, R. A. 1966. Learning with prolonged delay of reinforcement.Psychon. Sci 5,121-122. Garcia, J., Ervin, F. R, Yorke, C. H., and Koelliig, R A. 1967. Conditioning with delayed vitamin injection. Science 155,716-718. Garcia, J., Kovner, R., and Green, K. F. 1970. Cue properties versus palatability of flavors in avoidance learning. Psychon. Sci. 20,313-314. Garcia, J., McGowan, B. K., and Green, K. F. 1972. Biological constraints on conditioning. In “Classical Conditioning Two: Current Theory and Research” (A. Black and W. F. Prokasy, eds.), pp. 3-27. Appleton, New York. Gelperin, A. 1966. Investigations of a foregut receptor essential to taste threshold regulation in the blowfly. J. Insect Physiol. 12,829-841. Gleitman, H. 1974. Getting animals to understand the experimenter’s instructions. Anim Learn. Behav. 2,l-5. Greene, L S. 1974. Physical growth and development, neurological maturation, and behavioral functioning in two Ecuadorian Andean communities in which goiter is endemic. 11. PTC taste sensitivity and neurological maturation. Manuscript. Handal, P. J. 1965. Immediate acceptance of sodium salts by sodium deficient rats. Psychon S C 3, ~ 315-316. Harper, A. E. 1964. Amino acid toxicities and imbalances. In “Mammalian Protein Metab-

70

PAUL ROZIN

o h m ” (H. N. Munro and J. B. Allihn, eds.), Vol. 2, pp. 87-134. Academic Press, New York. Harriman, A. E., and Kare, M. R 1964. Preference for sodium chloride over lithium chloride by adrenalectomized rats. Amer. J. Physiol. 207,941-943. Harriman, A. E., and MacLeod, R B. 1953. Discriminative thresholds of salt for normal and adrenalectomized rats. Amer. J. PsychoL 66,465-47 1. Harris, L. J., Clay, J., Hargreaves, F., and Ward, A. 1933. Appetite and choice of diet. The ability of the vitamin B deficient rat to discriminate between diets containing and lacking thevitamin.Proc. Roy. Soc., Ser. E 113, 161-190. Henkin, R. I., Gill, J. R,Jr., and Bartter, F. C. 1963. Studies on taste thresholds in normal man and in patients with adrenal cortical insufficiency: the role of adrenal cortical steroids and of serum sodium concentration. J. Clin. Invest. 42, 727-735. Hess, E. 1964. Imprinting in birds. Science 146,1128-1139. Hinde, R A, and Stevenson-Hinde, J., eds. 1973. “Constraints on Learning. Limitations and Predispositions.” Academic Press, New York. Hogan, J. A. 1973a. Development of food recognition in young chicks. I. Maturation and nutrition. J. Comp. Physiol. PsychoL 83,355-366. Hogan, J. A. 1973b. How young chicks learn to recognize food. In “Constraints on Leaming” (R. A. Hinde and J. Stevenson-Hinde, eds.), pp. 119-139. Academic Press, New York. Hogan, J.A. 1973c. The development of food recognition in young chicks: 11. Learned associations over long delays. J. Comp. Physiol Psychol. 83, 367-373. Holman, E. W. 1973. Temporal properties of gustatory spontaneous alternation in rats. J. Comp. PhysioL PsychoL 8 5 , 5 36-5 39. Hunt, G.L., and Smith, W. J. 1967. Pecking and initial drinking responses in young domestic fowl. J. Comp. Physiol. PsychoL 64,230-236. Itani, J. 1958. On the acquisition and propagation of a new food habit in the natural group of Japanese monkeys at Takasaki Yama. J. Primatol. 1,84-98. Jennings, J. B. 1965. “Feeding, Digestion, and Assimilation in Animals.” Pergamon, Oxford. Jermy, T., Hanson, F. EL, and Dethier, V. G. 1968. Induction of specific food preference in lepidopterous larvae. Entomol. Exp. Appl. 1 1 , 2 11-230. Jolly, A. 1972. “The Evolution of Primate Behavior.” Macmillan, New York. Jones, W. 0. 1959. “Manioc in Africa.” Stanford Univ. Press, Stanford, California. Kalat, J. W. 1974. Taste salience depends on novelty, not concentration, in tasteaversion learning in the rat. J. Comp. Physioh Avchd. 86,47-50. Kalat, J. W., and Rozin, P. 1970. “Salience:” A factor which can overcome temporal contiguity in taste-aversion learning. J. Comp. Physiol. Psychol. 7 I , 192-1 97. Kalat, J. W., and Rozin, P. 1972. You can lead a rat to water but you can’t make him think. In “Biological Boundaries of Learning” (M. E. P. Seligman and J. Hagex, eds.), pp. 115-122. Appleton, New York. Kalat, J., and Rozin, P. 1973. “Learned safety” as a mechanism in longdelay taste-aversion learning in rats. J. Comp. PhysfoL PsychoL 83, 198-207. Kalmus, H. 1971. Genetics of taste. In “Handbook of Sensory Physiology, IV: Chemical Senses. 2: Taste” (L. Beidler, ed.), pp. 166-179. Springer-Verlag, Berlin and New York. Kare, M. R, and Maller, 0. 1967. “The Chemical Senses and Nutrition.” Johns Hopkins Press, Baltimore, Maryland. Kavanau, J. L. 1964. Behavior: confinement, adaptation, and compulsory regimes in laboratory studies. Science 143,490. Kawai, M. 1965. Newly acquired precultural behavior of the natural troop of Japanese monkeys on Koshima islet. Primates 6, 1-30. Koh, S. D., and Teitelbaum, P. 1961. Absolute behavioral taste thresholds in the rat. J. Comp. Psychol. AychoL 54,223-229. Kretchmer, N. 1972. Lactose and lactase. Sci. Amer. 227,7l-78.

SELECTION O F FOODS BY RATS, HUMANS, AND OTHER ANIMALS

71

Krieckhaus, E. E., and Wolf, G. 1968. Acquisition of sodium by rats: Interaction of innate mechanisms and latent learning. J. Cornp. Physiol. Psychol. 64, 197-201. Lat, J. 1967. Self-selection of dietary components. In “Handbook of Physiology, Sect. 6: Alimentary Canal” (C. F. Code and W. Heidel, eds.), Vol. 1, pp. 367-386. Amer. Physiol. SOC.,Washington, D. C. Leichter, J. 1973. Effect of dietary lactose on intestinal lactase activity in young rats. J. Nutr. 103, 392-396. LeMagnen, J. 1971. Olfaction and Nutrition. In “Handbook of Sensory Physiology, IV: Chemical Senses, 1 : Olfaction” (L. Beidler, ed.), pp. 467-482. Springer-Verlag, Berlin and New York. LeMagnen, J., and Tallon, S. 1968. Pr6f6rence alimentaire du jeune rat induite par l’allaitment maternel. C. R. SOC.Biol. 162, 387-390. Lettvin, J. Y.,Maturana, H. R., McCulloch, W. S., and Pitts, W. H. 1959. What the frog’s eye tells the frog’s brain. Proc. IRE 47, 1940-195 1 . Leung, P. M. B., Rogers, Q. R, and Harper, A. E. 1968. Effect of amino acid imbalance on dietary choice in the rat. J. Nutr. 95,483-492. LeviStrauss, C. 1964. “Le Cru et le Cuit: Mythologiques I.” Plon, Paris. [Engl.trans., 1969, “The Raw and the Cooked: Introduction to a Science of Mythology, I” (J. Weightman and D. Weightman, translators). Harper & Row, New York.] Levi-Straws, C. 1966. Le triangle culinaire. Arc (Aix-en-Provence)26, 19-29. (Engl. trans., 1966. The culinary triangle. New SOC.Dec. 22,937-940.) Lorden, J. F., Kenfield, M., and Braun, J. J. 1970. Response suppression to odors paired with toxicosis. Learn. Motiv. 1, 391-400. Lowenberg, M. E., Todhunter, E. N., Wilson, E. D., Feeney, M. C., and Savage, J. R. 1968. “Food and Man” Wiley, New York. McBumey, D.H., and Pfaffman, C. 1963. Gustatory adaptation to saliva and sodium chloride. J. Exp. PsychoL 65,523-529. Meracken, R D . 1971. Lactase deficiency: an example of dietary evolution. Curr. Anthropologist 12,479-511. Maller, O., and Desor, J. A. 1973. Effect of taste on ingestion by human newborns. In “Fourth Symposium on Oral Sensation and Perception: Development in the Fetus and Infant” (J. F. Bosma, ed.), pp. 279-291. U.S. Gov. Printing Office, Washington, D.C. Mayer, J. 1955. Regulation of energy intake and the body weight: the glucostatic theory and the lipostatic hypothesis. Ann N. Y. Acad Sci. 63, 15-42. Miller, N. E, and Kessen, M. L. 1952. Reward effects of food via stomach fistula compared with those of food via mouth. J. Comp. Physiol. Aychol. 45,555-564. Morgan, C. L. 1894. “An introduction to Comparative Psychology.” Walter Scott, London. Morris, D. 1967. “The Naked Ape.” McGraw-Hill, New York. Momson, G. R. 1974. Alterations in palatability of nutrients for the rat as a result of prior testing. J. Cornp PhysioL Psycho1 86,s 6-6 1. Nachman, M. 1962. Taste preferences for sodium salts by adrenalectornized rats. J. Cornp. PhysioL Psychol. 55,1124-1 129. Nachman, M. 1963a. Learned aversion to the taste of lithium chloride and generalization to other salts. J. Comp. Physiol. AychoL 56,343-349. Nachman, M. 1963b. Taste preferences for lithium chloride by adrenalectomized rats. Amer. J. Physiol. 205,219-221. Nachman, M., and Cole, L. P. 1971. Role of taste in specific hungers. In “Handbook of Sensory Physiology, IV: Chemical Senses, 2: Taste” (L. Beidler, ed.), pp. 337-362. Springer Verlag, Berlin and New York. National Academy of Sciences, 1962. “Nutrient Requirements of Laboratory Animals.” N.A.S., N.R.C. Publ. No. 990. Nelson, W. E., Vaughan, V. C., and McKay, R. J. 1969. “Textbook of Pediatrics.” Saunders, Philadelphia, Pennsylvania.

72

PAUL ROZIN

Norgren, R., and Leonard, C. M. 1973. Ascending central gustatory pathways. J. Comp. Neurol. 150,217-237. Pain, J. F., and Booth, D. A. 1968. Toxiphobia for odors. Psychon Sci. 10,363-364. Pfaffman, C. 1959. The sense of taste. In “Handbook of Physiology, Sect. 1: Neurophysiology” (H. W. Magoun, ed.), VoL 1, pp. 507-533. Amer. Physiol. SOC., Washington, D.C. Pfaffman, C. 1960. The pleasures of sensation. PsychoL Rev. 67,253-268. Pfaffman, C., and Bare, J. K. 1950. Gustatory nerve discharges in normal and adrenalectomized rats. J. Comp. Physiol. Psychol. 43, 320-324. Pfaffman, C., Bartoshuk, L. M., and McBumey, D. H. 1971. Taste Psychophysics. In “Handbook of Sensory Physiology, lV: Chemical Senses, 2: Taste” (L. Beidler, ed.), pp. 75-101. Springer-Verlag, Berlin and New York Pfeiffer, J. E. 1969. “The Emergence of Man.” Haper, New York. Pilgrim, F. J., and Patton, R. A. 1947. Patterns of self-selection of purified dietary components by the rat. J. Comp. PhysioL PsychoL 40,343-348. Quartermain, D., Miller, N. E., and Wolf, G. 1967. Role of experience in relationship between sodium deficiency and rate of bar pressing for salt. J. Comp. Physiol. PsychoL 63,417-420. ReMer, H. D. 1944. “The Origin of Food Habits.” Faber & Faber, London. Revusky, S. H. 1967. Hunger level during food consumption: effects on subsequent preference.Psychon Sci. 7,109-1 10. Revusky, S. H. 1968. Aversion to sucrose produced by contingent X-irradiation: Temporal and dosage parameters. J. Comp Physid. Psychol. 65,17-22. Revusky, S. H. 1971. The role of interference in association over a delay. In “Animal Memory” (W. Honig and H. James, eds.), pp. 155-213. Academic Press, New York. Revusky, S. H., and Bedarf, E. W. 1967. Association of illness with prior ingestion of novel foods.Science 155,219-220. Revusky, S. H., and Garcia, J. 1970. Leamed associations over long delays. In “The Psychology of Learning and Motivation: Advances in Research and Theory” (G. H. Bower, ed.),Vol. 4, pp. 1-84. Academic Press, New York. Reynolds, V. 1967. “The Apes.” Dutton, New York. Richter, C. P. 1936. Incrkased salt appetite in adrenalectomized rats. Amer. J. PhysioL 115, 155-161. Richter, C.P. 1938. Changes in fat, carbohydrate, and protein appetite in vitamin B deficiency. Amer. J. Physiol. 124,596-602. Richter, C.P. 1939. Salt taste thresholds of normal and adrenalectomized rats. Endocrinology 24,367-371. Richter, C. P. 1942-1943. Total self regulatory functions in animals and human beings. Harvey Lect. Ser. 38,63-103. Richter, C. P. 1945. Self-selection studies on coprophagy as a source of vitamin B complex. Amer. J. PhysioL 143,344-354. Richter, C. P. 1950. Taste and solubility of toxic compounds in poisoning of rats and man. J. Comp. Physiol. Psychol. 43,358-374. Richter, C.P. 1953. Experimentally produced reactions to food poisoning in wild and domesticated rats. Ann. N. Y.Acad. Sci. 56,225-239. Richter, C.P. 1954. The effect of domestication and selection on the behavior of the Norway rat. J. Nat. Cancer Inst. 15,727-738. Richter, C. P. 1955. Self-regulatory functions during gestation and lactation. 72uns.. Conf. Gestation, 2nd, Princeton, N.J. pp. 11-93. Richter, C. P. 1956. Salt appetite of mammals: its dependence on instinct and metabolism. In “L’Instinct dans le Comportement des Animaux et de I’Homme” (Fondation Singer

SELECTION OF FOODS BY RATS, HUMANS, AND OTHER ANIMALS

73

Polignac, ed.), pp. 577-629. Masson, Paris. Richter, C. P. 1959. Rats, man, and the welfare state. Amer. Psychologist 14, 18-28. Richter, C. P., and Barelare, B., Jr. 1938. Nutritional requirements of pregnant and lactating rats studied by the self-selection method. Endocrinology 23, 15-24. Richter, C. P., and Clisby, K. H. 1941. Phenylthiocarbamide taste thresholds of rats and human beings. Amer. J. Physid. 134, 157-164. Richter, C. P., and Eckert, J. F. 1937. Increased calcium appetite of parathyroidectomized rats. Endocrinology 21,5044. Richter, C . P., and Eckert, J. F. 1938. Mineral metabolism of adrenalectomized rats studied by the appetite method. Endocrinology 22,214-224. Richter, C. P., and Hawkes, C. D. 1941. The dependence of the carbohydrate, fat, and protein appetite of rats on the various components of the vitamin B complex. Amer. J. Physiob 131,639-649. Richter, C. P., and Helfrick, S. 1943. Decreased phosphorous appetite of parathyroidectomized rats. Endocrinology 33,349-352. Richter, C. P., and Mosier, H. D., Jr. 1959. Maximum sodium chloride intake and thirst in domesticated and wild Norway rats. Amer. J. Physiob 176,213-222. Richter, C. P., and Rice, K. K. 1945. Self-selection studies on coprophagy as a source of vitamin B complex. Amer. J. PhysioL 143, 344-354. Richter, C. P., Holt, L. E., and Barelare, B., Jr. 1937. Vitamin B1 craving in rats. Science 86, 354-355. Richter, C. P., Holt, L. E, Barelare, B., Jr., and Hawkes, C. D. 1939. Changes in fat, carbohydrate, and protein appetite in vitamin B deficiency. Amer. J. Physiol. 124,596-602. Rodgers, W. 1967a. Specificity of specific hungers. J. Comp. Physiol. Psychd. 6 4 , 4 9 4 8 . Rodgers, W. L. 1967b. Thiamine Specific Hunger. Ph.D. Thesis, Univ. of Pennsylvania, Philadelphia. Rodgers, W. L., and R o b , P. 1966. Novel food preferences in thiamine-deficient rats. J. Comp. Physiol. Psychol. 61, 1 4 . Rogers, Q. R., and Harper, k E 1970. Selection of a solution containing histidine by rats fed a histidineimbalanced diet. J. Comp. Physiol. Psychol. 72,66-71. Roll, D. L. and Smith, J. C. 1972. Conditioned taste aversion in anesthetized rats. In “Biological Boundaries of Learning” (M. E.P. Seligman and J. Hager, eds.), pp. 98-102. Appleton, New York. Rozin, E. 1973. “The Flavor Principle Cookbook.” Hawthorne, New York. Rozin, E. 1975. A lunch in Brooklyn: The conservatism of cuisine. Manuscript. Rozin, P. 1964. Comparative biology of feeding patterns and mechanisms. Fed. Proc., Fed. Amer. SOC.Exp. Biol. 2 3 , 6 0 4 . Rozin, P. 1965. Specific hunger for thiamine: recovery from deficiency and thiamine preference. J. Comp. Physiol. Psychol. 59,98-101. Rozin, P. 1967a. Specific aversions as a component of specific hungers. J. Comp. Physiol. Psychol. 64, 237-242. Rozin, P. 1967b. Thiamine specific hunger. In “Handbook of Physiology, Sect. 6: Alimentary Canal” (C. F. Code and W. Heidel, eds.), Vol. 1, pp. 411-431. Amer. Physiol. SOC., Washington D.C. Rozin, P. 1968a. Specific aversions and neophobia as a consequence of vitamin deficiency and/or poisoning in half-wild and domestic rats. J. Comp. Physiol. Psychol. 66,82-88. Rozin, P . 1968b. Are carbohydrate and protein intakes separately regulated? J. Comp. Physiol. Psychol. 65,23-29. Rozin, P. 1969a. Central or peripheral mediation of learning with long CS-US intervals in the feeding system. J. Comp. Physiol. Psychol. 67,421-429. Rozin, P. 1969b. Adaptive food sampling patterns in vitamin deficient rats. J. Comp.

74

PAUL ROZIN

Physiol. Psychol. 69, 126-1 32. Rozin, P., and Kalat, J. 1971. Specific hungers and poison avoidance as adaptive specializations of learning. Psychol. Rev. 78,459-486. Rozin, P., and Kalat, J. 1972. Learning as a situation-specific adaptation. In “The Biological Boundaries of Learning’’ (M. E. P. Seligman and J. Hager, eds.), pp. 66-97. Appleton, New York. Rozin, P., and Mayer, J. 1961. Regulation of food intake in the goldfish. Amer. J. Physiol. 201,968-974. Rozin, P., and Mayer, J. 1964. Some factors influencing short-term food intake of the goldfish. Amer. J. Physiol. 206, 1430-1436. Rozin, P., and Ree, P. 1972. Long extension of effective CS-US interval by anesthesia between CS and US. J. Comp. Physiol. Psychol. 80,43-48. Rozin, P., and Rodgers, W. L. 1967. Novel diet preferences in vitamin deficient rats and rats recovered.from vitamin deficiencies. J. Comp. Physiol. Psychol. 63,42 1-428. Rozin, P., Wells, C., and Mayer, J. 1964. Thiamine specific hunger: Vitamin in water versus vitamin in food. J. Comp. Physiol. Psychol. 57, 78-84. Rzoska, J. 1953. Bait shyness, a study in rat behavior. Brit. J. Anim. Behav. 1, 128-135. Schaeffer, 0. 1969a. Glucose tolerance testing in Canadian Eskimos: A preliminary report and a hypothesis. Can. Med. Ass. J. 99,252-262. Schaeffer, 0. 1969b. Carbohydrate metabolism in eskimos. Arch. Environ. Health 18, 144-147. Scott, E. M. 1946. Self selction of diet. I. Selection of purified components. J. Nutr. 31, 397405. Scott, E. M., and Quint, E. 1946a. Self selection of diet. 111. Appetites for B vitamins. J. Nutr. 32, 285-291. Scott, E. M., and Quint, E. 1946b. Self selection of diet. IV. Appetite for protein. J. Nutr. 32,293-30 1 . Scott, E. M., and Verney, E. L. 1947. Self selection of diet. VI. The nature of appetites for B vitamins. J. Nurr. 34,47 1 4 0 . Scott, E. M., Verney, E. L., and Morissey, P. D. 1950a. Self selection of diet. XII. Effects of B vitamin deficiencies on selection of food comp0nents.J. Nutr. 41, 373-381. Scott, E. M., Verney, E. L., and Morissey, P. D. 1950b. Self selection of diet. XI. Appetites for calcium, magnesium and potassium. J. Nutr. 41, 187-202. Seligman, M. E. P. 1970. On generality of the laws of learning. Psychol. Rev. 77,406-418. Seligman, M. E. P., and Hager, J., eds. 1973. “The Biological Boundaries of Learning.” Appleton, New York. Seward, J. P., and Greathouse, S. R. 1973. Appetitive and aversive conditioning in thiaminedeficient rats. J. Comp. Physiol. Psychol. 83, 157-167. Shallenberger, R., and Acree, T. E. 1971. Chemical structure of compounds and their sweet and bitter tastes. In “Handbook of Sensory Physiology, IV: Chemical Senses, 2: Taste” (L. Beidler, ed.), pp. 221-277. Springer-Verlag, Berlin and New York. Shettleworth, S. 1972. Constraints on learning. In “Advances in the Study of Behavior” (D. S. Lehrman, R. A. Hinde, and E. Shaw, eds.), Vol. 4, pp. 1-68. Academic Press, New Y ork. Simoons, F. J. 1961. “Eat Not This Flesh.” Univ. of Wisconsin Press, Madison. Simoons, F. J. 1969. Primary adult lactose intolerance and the milkdrinking habit: a problem in biological and cultural interrelations. I. Review of the medical research. Amer. J. Dig. Dis. 14,819-836. Simoons, F. J. 1970. Primary adult lactose intolerance and the milkdrinking habit: a problem in biological and cultural interrelations. 11. A cultural-historical hypothesis. Amer. J. Dig. Dis. 15,695-710.

SELECTION O F FOODS BY RATS, HUMANS, AND OTHER ANIMALS

75

Simson, P. C. and Booth, D. A. 1973. Effect of CS-US interval on the conditioning of odour preferences by amino acid loads. Physiol. Behov. 11,801-808. Simson, P. C. and Booth, D. A. 1974. The rejection of a diet which has been associated with a single administration of an histidine-free amino acid mixture. Br. J. Nufr. 31, 285-296. Smith, J. C., 1971. Radiation: its detection and its effects on taste preferences. In “Progress in Physiological Psychology” (E. Stellar and J. M. Sprague, eds.), pp. 53-118. Academic Press, New York. Smith, J. C., and Roll, D. L. 1967. Trace conditioning with X-rays as an aversive stimulus. Psychon. Sci. 9, 11-12. Steiner, J. 1973. The Human gustofacial response. In “Fourth Symposium on Oral Sensation and Perception: Development in the Fetus and Infant” (J. F. Bosma, ed.), pp. 254-278. U S . Gov. Printing Office, Washington, D.C. Stricker, E. M. 1973. Thirst, sodium appetite, and complementary physiological contributions to the regulation of intravascular fluid volume. In “The Neuropsychology of Thirst” (A. N. Epstein, E. Stellar, and H. Kissileff, eds.), pp. 73-98. Winston, Washington, D.C. Stricker, E. M., and Sterritt, G. M. 1967. Osmoregulation in the newly hatched domestic chick.Physiol. Behov. 2,117-119. Stricker, E. M., and Wilson, N. E. 1970. Salt-seeking behavior in rats following acute sodium deficiency. J. Comp. Physiol. Psychol 72,416420. Tannahill, R. 1973. “Food in History.” Stein & Day, New York. Teitelbaum, P., Cheng, M.-F., and Rozin, P. 1969. Development of feeding parallels its recovery after hypothalamic damage. J. Comp. Physiol. Psychol. 6 7 , 4 3 0 4 4 1 . Trethowan, W. H., and Dickens, G. 1972. Cravings, aversions and pica of pregnancy. In “Modern Perspectives in Psycho-Obstetrics” (J. S. Howells, ed.),pp. 25 1-268. Brunerl Mazel, New York. van LawickGoodall, J. 1971. “In the Shadow of Man.” Houghton, Boston, Massachusetts. von Frisch, K. 1967. “The Dance Language and Orientation of Bees.” Belknap Press, Cambridge, Massachusetts. Weiskrantz, L., and Cowey, A. 1963. The aetiology of food reward. Anim. Behov. 11, 225-234. Wilcoxon, H. C., Dragoin, W. B., and Kral, P. A. 1971. Illness-induced aversions in rat and quail: relative salience of visual and gustatory cues. Science 171,826-828. Wilkins, L., and Richter, C. P. 1940. A great craving for salt by a child with corticoadrenal insufficiency. J. Amer. Med. Ass. 114,866-868. Wittlin, W. A., and Brookshire, K. H. 1968. Apomorphine-induced conditioned aversion to a novel food. Psychon. Sci. 17,309-310. Wolf, G. 1969. Innate mechanisms for regulation of sodium intake. In “Olfaction and Taste” (C. Pfaffman, ed.), VoL 3, pp. 548-553. Rockefeller Univ. Press, New York. Woods, S. C., and Weisinger, R. S. 1970. Pagophagia in the albino rat. Science 169, 1334-1 336. Yensen, R. 1959. Some factors affecting taste sensitivity in man. 11. Depletion of body salt. J. EXP.Psychol. 11,230-238. Young, P. T. 1944. Studies of food preference, appetite and dietary habit. 11. Group selfselection maintenance as a method in the study of food preferences. J. Comp. Psychol. 37,371-391. Young, P. T. 1948. Appetite, palatability and feeding habit: a critical review. Psychol. Bull. 45,289-320. Young, P. T., and Wittenbom, J. R. 1940. Food preferences of rachitic and normal rats. J.

76

PAUL ROZIN

Comp. Psychol. 30,261-276. Zahler, L. P., and Harper, A. E. 1972. Effects of dietary amino acid pattern on food preference behavior of rats. J. Comp. Physiol. Psychol. 81,155-162. Zahorik, D., and Maier, S. 1969. Appetitive conditioning with recovery from thiamine deficiency as the unconditioned stimulus. Psychon. Sci. 17, 309-310. Zahorik, D. M., Maier, S. F., and Pies, R. W. 1974. Preferences for tastes paired with recovery from thiamine deficiency in rats: appetitive conditioning or learned safety. J. Comp. Physiol. PsychoL 87,1083-1091. Zotterman, Y. 1956. Species differences in the water taste. ActuPhysiol. Scund. 37,60-70.

Social Transmission of Acquired Behavior: A Discussion of Tradition and Social Learning in Vertebrates BENNETT G. GALEF, JR. DEPARTMENT OF PSYCHOLOGY MCMASTER UNIVERSITY HAMILTON, ONTARIO, CANADA

................................. .................. A. Spatial Utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Feeding and Predatory Behavior ..................... C. Predator Avoidance . . . . . . . . . . . . . . . . . . . . . . . . . . . . D. Bud Vocalizations ............................. Learning and Conditioning Paradigms, .................... Problems of Terminology. . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

I. Introduction.. 11. Field and Associated Laboratory Studies.

111.

IV.

V.

I.

77 81 82 84 87 88 88 92 95 97

INTRODUCTION

There are, broadly speaking, essentially three nonindependent means by which the behavior characteristic of a population may remain constant from one generation to the next. First, adaptive behavior in population members may be largely endogenously organized and genetically transmitted as propensities influencing ontogeny. Second, similar patterns of behavior in successive generations of a population may result from similar histories of individual transaction with the physical environment. And, third, long-term homogeneity of behavior may result from the transmission of patterns of behavior from individual t o individual within a population as a consequence of social interaction (for a similar analysis, see Klopfer, 1961). The assumption has often been made that in most species the adaptive behavior acquired independently by an individual as a result of its transactions with the physical environment is not readily transmitted either to others of its genera77

78

BENNETT G. GALEF. JR.

tion or to members of future generations. In this view, although the genetic material influencing the behavior of an individual which allowed i t to acquire some pattern of behavior may be preserved and disseminated within a population via the mechanisms of Mendelian recombination and neo-Darwinian natural selection, the specific responses acquired by any individual are lost in every generation with the death of their acquirer. The logical extension of this position is that homogeneity in the behavior of members of a population must reflect either common genetic material or similar histories of individual organism-environment transaction in all population members, or both. The human species, and t o a lesser extent the other primates, are treated as exceptional in their ability to desseminate throughout a population and project into future generations, individually acquired patterns of behavior. There is, however, a large but scattered body of literature both on the observation of free-living groups of animals and the study of a few species under controlled laboratory conditions, suggesting that intraspecific interaction resulting in the transmission of acquired patterns of behavior from one individual t o another within a population is a relatively common and important mode of adaptation in both primate and nonprimate vertebrate organisms. The survival value of the ability of organisms t o acquire patterns of behavior as a result of interaction with conspecifics, as well as from transactions with nonsocial aspects of the environment, are relatively straightforward. If laboratory learning paradigms are, in fact, accurate analogs of learning as it occurs in natural habitats, then the trial and error processes necessary for the acquisition of adaptive patterns of behavior must often be both energy-consuming and error-filled undertakings for the acquirer. A young animal, newly recruited to a population, must face particularly acute environmental challenges requiring rapid acquisition of behaviors necessary for survival within the particular area in which it achieves physiological independence. The need to locate areas suitable for survival and reproductive activities, t o find and learn to ingest necessary dietary constituents, to learn to escape or avoid potential predators, and to behave appropriately with respect t o conspecific individuals must place considerable demands on the young organism’s capacities for behavior acquisition during a time when it is highly vulnerable to environmental stress and when errors in response can have serious consequences. Although the naive animal may have the capacity to acquire the learned adaptive behavior of more mature and experienced individuals by repeating their histories of transaction with the physical environment, it would clearly be advantageous to the young if they could in some way incorporate into their own behavioral repertoires the learned adaptive behavior of more experienced conspecifics through some process less cumbersome than de nova trial and error learning. Similarly, adult organisms living in unstable environments could benefit appreciably from the direct acquisition of conspecific patterns of behavior. In the absence of such acquisition, each individ-

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

79

ual would have to discover for itself the existence or novel distribution of important elements in the environment following any environmental change. Again, although each individual might have the capacity to learn its way about in a changed environment, direct acquisition of conspecific patterns of behavior could facilitate behavioral adaptation to changing circumstances. Viewed in a broader context, the social transmission of acquired behavior may be seen as providing an alternative to the, genetic transmission of behavioral propensities, allowing a population to maintain established patterns and to incorporate behavioral novelty into its repertoire rapidly (Mainardi, 1970, 1973). The most readily observable result of social transmission processes would be the existence of different modes of behavior within different geographic subpopulations of a species uncorrelated with gene or resource distribution. Before reviewing examples of patterns of behavior apparently transmitted among conspecifics, it is important to define the range of phenomena to be considered. The task of definition requires that transmitted behaviors be distinguished from other observable changes in behavior resulting from interaction among conspecific individuals. The aim of the definition proposed here is to restrict consideration to instances in which organisms acquire specific patterns of behavior as a result of direct transaction with the environment and increase the probability of other species members exhibiting similar patterns of behavior as a result of interaction with them. Three criteria, discussed below, seem sufficient appropriately to limit examples to be considered. First, our concern here will not be with cases in which social interaction is a necessary condition for the ontogeny of a pattern of behavior. Thus, excluded from consideration will be phenomena such as the development of normal species-specific sexual preference in the zebra finch (Immelman, 1972), the acquisition of species typical song in the white-crowned sparrow (Marler and Tamura, 1964), and the development of normal maternal behavior in rhesus monkeys (Harlow and Harlow, 1965), which are expressed in the behavioral phenotype of only those individuals experiencing crucial social interactions during development. Rather, we will consider only those instances in which social interaction is a sufficient condition for behavior acquisition and provides an alternative or optional route to direct transaction with the nonsocial environment in the development of behaviors in question. The decision to limit discussion to cases in which social interaction is sufficient but not necessary for behavior development results from consideration of apparent differences in the functions of necessary and sufficient social interactions in the ontogeny of behavior. Organisms often require exposure t o specific environmental conditions for the development of a given behavior pattern. If the environmental condition is a social one, as for example interaction with a parent, it is possible to confuse a social exposure necessary for normal development with a social transmission process. In the former case the result of

80

BENNE’IT G . GALEF, JR.

social interaction is normal development of a relatively invariant, species typical behavior. In the latter, idiosyncratic patterns acquired by the transmitter as a result of its history of transaction with the environment may be introduced into a population repertoire. The somewhat conservative approach adopted here is to assume that this qualitative difference in the function of social interactions necessary and sufficient for the development of behavior exists and to restrict discussion to those cases in which social transmission is facultative rather than obligate for behavior development. Second, the change in behavior resulting from interaction among conspecifics should be in the direction of homogeneity rather than heterogeneity of behavior between interactants. This restriction serves to exclude from consideration social interactions, such as dominance hierarchy formation or territorial division of a species range, that produce changes in the behavior of interactants but in which the particular pattern of behavior of one organism is not acquired by another. Third, I wish t o consider only those cases in which the increased homogeneity of behavior extends temporally beyond the period of interaction between the recipient and the transmitter. Thus, the critical test for the successful transmission of behavior becomes the maintenance of the transmitted behavior in the recipient following the termination of interaction with the transmitter. This criterion is intended to exclude a variety of cases, such as mobbing of potential predators (Hinde, 1954) or simple following of one animal by another, in which the behavior of one individual releases similar behavior in others. The statement of the preceding criteria is not to imply that behavioral phenomena that fail to meet them are of lesser importance than those that do. Rather, their purpose is t o differentiate interactions functioning to disseminate patterns of acquired behavior through a population from those incapable of doing so (a similar approach with respect to the definition of “culture” is to be found in Menzel et al., 1972). The following sections review a variety of field and laboratory findings which have been or can be interpreted as demonstrating the social transmission of acquired behavior. The term acquired behavior is employed here broadly, to refer both to cases in which a novel motor pattern is acquired by an organism and to cases in which a typical response comes to be elicited by a novel stimulus. Thus, for example, the incorporation of a novel item into an organism’s feeding repertoire will be treated as acquired behavior and, consequently, the spread among conspecifics of feeding on that item would be considered as a possible case of transmission of acquired behavior. The term transmission of behavior will be used to refer to any of a variety of processes by means of which the behavior of conspecifics is modified in the direction of homogeneity as a result of intraspecific interaction. In the following discussion of these processes, no implication of deliberate tuition of one organism by another is intended. No attempt

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

81

has been made to survey the literature exhaustively; rather, cases have been chosen to exemplify a number of mechanisms that are discussed in succeeding sections as possible bases for the transmission of behavior from one individual to another. 11.

FIELD AND ASSOCIATED LABORATORY STUDIES

Criteria for the identification of socially transmitted behavior are difficult to specify in most field situations. Although the careful observer can often describe differences in the behavior of subpopulations of a species, simple observation is seldom sufficient for the identification of the processes leading to their establishment. The requisite analyses would often require laboratory study of events observed in the field, but in many cases both species and phenomena suitable for field observation are not particularly practical choices for laboratory research. Conversely, species chosen for laboratory investigation are often difficult subjects for field study. As a result, the controlled analyses necessary to interpret field data fully are often not available and the importance of phenomena studied in the laboratory for the life of organisms in their natural environment frequently remains undetermined. The existing literature on the transmission of acquired behavior clearly reflects these differences between the phenomena of laboratory and field investigations, and the synthesis attempted here has necessitated a certain amount of extrapolation from the available data. The discussion has been organized around available field studies for two reasons. First, it is necessary to consider the frequency and importance of phenomena suggestive of behavioral transmission in natural settings. In terms of the approach adopted here, if possible instances of social transmission are infrequent, if social transmission does not play an important role under natural circumstances, it would be an entirely academic exercise to discuss it at any length. I do not personally view this as a serious problem. It is, for example, difficult to find an extended study of the life history of any mammalian or avian population which does not include the description of one or more behavioral phenomena amenable to consideration within the framework under discussion. Second, instances of transmission of acquired behavior in natural settings provide a necessary basis for evaluation of the importance of theoretical statements and empirical findings derived from laboratory investigations of social learning phenomena. For the purpose of organization, the data have been categorized in terms of the roles of behavior patterns discussed in the life of the organism. Where laboratory investigations relevant to the field data are available, they have been referred to in the appropriate context.

82

BENNETT G. GALEF, JR.

A. SPATIAL UTILIZATION

In many vertebrate species the territories or home ranges of subpopulations or the specialized use of certain areas within subpopulation ranges remain relatively fixed over many generations, resulting in what might be described as “traditional” boundaries psychologically restricting the movement and activity of individuals. Although the factors responsible for selection of habitation sites have been explored in only a few species, there are several investigations indicating that the preferences of some vertebrates for particular habitation sites are modifiable by experience (Klopfer, 1969; Klopfer and Hailman, 1965; Wecker, 1963). The existence in many species of consistent subpopulation differences in habitation site selection suggests that interaction with conspecifics may be an important determinant of the selection of an area for occupation. As will become apparent in the following literature review, the transmission of preferences for locations in which to carry out life’s activities may occur in any one of a variety of ways. Although the processes involved in the social transmission of patterns of spatial utilization have not in most cases been analyzed in sufficient detail to permit their precise description, it might prove useful t o cateogrize them in a general way to facilitate organization of the material presented below, even though it is not always possible on the basis of present knowledge to specify into which category a given example may fall. In the simplest instances, a parturient female can affect the choice by her young of a home range or habitat simply as a result of depositing or rearing them at one site rather than another. The social interaction responsible for selection of a specific site for habitation in the young is, in these cases, very limited, and the long-term consequences for the young of parental reproductive site selection depend on the young developing some attachment to the area in which they find themselves early in life. For example, numerous studies indicate that each of the many subpopulations of Pacific salmon return generation after generation to different streams to reproduce. The data available are consistent with the view that the young salmon become imprinted on chemical cues unique to the particular stream in which their mother spawned and in which they spend their first year (Hasler, 1966). Thorpe (1945) has proposed the term habitat imprinting to describe the well-documented tendency of some species of migratory bird with widely distributed nesting grounds to return to the area in which they were reared to engage in their own reproductive activity (Snyder, 1948). Similarly, there is evidence that sea turtles (Ehrenfeld, 1974) and many species of bat and frog show a strong tendency to migrate back to their place of birth for purposes of reproduction (Wynne-Edwards, 1962, p. 453). In slightly more complex situations, one organism may alter the environment in such a way as to channel the behavior of others with respect to it. This might be considered a more complex type of transmission in that the environmental change t o which the recipient responds is often a more active or specialized

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

83

product of the behavior of the transmitter than simple site selection and, in fact, subsumes the simpler case. For example, Atlantic salmon, which show consistent differences in spawning stream selection, like those of their Pacific relatives, are believed to respond to chemical cues deposited in breeding streams by fellow subpopulation members (Nordeng, 1971 ;Solomon, 1973). The size and position of prairie dog coterie territories remain essentially unchanged through complete population turnovers as a result of both the social organization of coteries and the effects of relatively stable burrow systems on territorial organization (King, 1955, p. 60). Similarly, Calhoun (1962, p. 142) has presented evidence that wild rats (Rams nowegicus) born to low-status clans, living in suboptimal portions of the environment, remain in the area of their birth, and become low-status adults themselves. The scented runs created by adult rat clan members define clan territorial boundaries and are rapidly learned by new recruits t o a clan (Telle, 1966, pp. 35-36). Traditional usage of restricted areas within subpopulation home ranges for specific purposes over many years have also been described and appear to result from alterations made in the environment by one individual that modify the behavior of others. Red deer, for example, use the same trails and wallows (Darling, 1937), whereas cliff swallows (Hochbaum, 1955) nest in the same locations for many generations. Yet more complex cases, in which some form of direct interaction between the transmitter and receiver are essential to the transmission process, can be subdivided into two types. In the first, general orienting or following responses on the part of receivers to conspecific transmitters introduce the receivers to selected aspects of the environment to which they then respond directly. In the second, receivers respond directly to transmitter responses to environmental features and only later come to attach those responses to the environmental features to which the transmitter responded initially. Both these types of transmission seem more complex than those previously discussed in that they require direct interaction between transmitter and receiver; because of the richness of such interactions, they are particularly difficult to analyze satisfactorily. Possible examples of the first type of direct interaction resulting in the social transmission of patterns of spatial utilization are not uncommon. For example, Geist (1971, pp. 88, 176) reports that the widely scattered home ranges of individual mountain sheep are socially transmitted, the traditions passing from lead adults to the juveniles that follow them throughout maturation. Similarly, female red deer pass on their home range traditions to their female offspring (Schloeth and Burckhardt, 1961). Further, the inherited directional tendencies of young birds during migration are readily modifiable by the example of older birds of their species; whether this intergenerational influence during migration has long-term effects on breeding and wintering ground selection is not yet known (Matthews, 1968, p. 12). Emlen (1938) has reported that the location and boundaries of wintering grounds of crows may remain unchanged for as long as 50

84

BENNETT G . GALEF, JR.

years, although the precise causes of this stability are not apparent. I have found no instance in which it seems reasonable to assume that the second type of direct interaction described above plays a role in the social transmission of acquired patterns of spatial utilization. However, instances of the occurrence of this type of social transmission are to be found in the discussions of feeding and predatory behavior and of predator avoidance, presented below. The preceding examples in which some aspect of spatial utilization appears to be determined at least in part by interaction among conspecific individuals indicate that long-term subpopulation differences in behavior may result from a variety of different mechanisms. Deposition of offspring into an environment, alteration of an environment, and the tendency of young actively to follow adults, may each serve in different species as the basis for the transmission and perseverance of subpopulation differences in spatial utilization. This multiplicity of processes responsible for the transmission of acquired behavior, although not discussed explicitly below, is common to all the examples to be considered. B.

FEEDING AND PREDATORY BEHAVIOR

The use of social transmission processes for the propagation of feeding and related behaviors appears to be quite common in vertebrates. A particularly well-documented case concerns a variety of novel eating and drinking patterns acquired by troops of Japanese macaques and apparently transmitted from individual t o individual as a result of social interaction among troop members. Examples of feedingassociated behaviors transmitted in this way range from sweet potato washing and wheat “placer-mining” (Kawai, 1965) to troop utilization of novel food resources. Descriptions of the spread of washing behavior within a troop suggest that it is transmitted as the result of one individual observing the behavior of another, as are the learned feeding patterns of juveniles to adults. The acquisition of a troop’s patterns of food utilization by juveniles seems to result from the young’s habit of ingesting scraps dropped by their mothers (Kawamura, 1959). Similar observations by Carpenter (1934, p. 74) of the feeding interaction of Howler monkey mothers and their young and by Hall (1962) of the feeding of young chacma baboons support the suggestion that adult primates can readily introduce their young to the foods they are eating as a result of the tendency of the young to ingest scraps, although corroborating studies under controlled conditions to determine the effects of ingestion of food samples in infancy on later food preferences are lacking (Hall, 1963).’ However, observations by Kuo (1967, p. 66) indicate that early feeding experience can have a profound effect on later food preferences in a variety of nonprimate vertebrates (cats, dogs, and myna birds) and support the contention of Kawamura ‘For a thorough discussion of social transmission in primates see Menzel, E. W., Jr. 1973. “Precultural Primate Behavior.” Karger, Basel.

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

85

that ingestion by infants of food samples obtained from feeding adults may affect later dietary preferences (see also Rabinowitch, 1969). Similar reports of young nonprimate organisms ingesting food samples acquired from their parents are common, and such parent-offspring interactions sometimes appear to introduce juveniles to substances they might otherwise not ingest. For example, young meerkats failed to recognize a novel food, bananas, as edible and only began to eat it when their mother, who was already familiar with bananas, did so (Ewer, 1963). Similarly, Burmese jungle fowl chicks are reluctant to ingest mealworms when they first encounter them unless the mealworms are presented by a mother hen making the “foodcall” (Hogan, 1966, p. 275, and personal communication). Information concerning edible foods available in the environment could easily be transferred from mother to young as a result of such parent4ffspring interactions (see also Wortis, 1969). Cases of the social determination of feeding patterns resulting from somewhat different sorts of interaction have also been reported in rodents. Von Steiniger (1950), in discussing the “local traditions’’ of colonies of wild rats, observed that if zinc phosphide is used in rat control in one area over an extended period of time, despite initial success, later acceptance of the poison remains low; the offspring of the survivors continue to refuse to accept the poison bait. In a series of laboratory investigations of this apparent traditional poison-avoidance behavior (Galef and Clark, 1971,1972; Galef and Henderson, 1972; Galef and Sherry, 1973), two complementary mechanisms have been described, either of which can result in rat pups preferentially ingesting the diet that the adults of their colony are eating and rejecting diets that these adults have learned to avoid. First, gustatory cues reflecting the flavor of a lactating female’s diet are incorporated into her milk, and ingestion of the female’s milk is sufficient to allow pups to recognize their mother’s diet and to cause them to ingest that diet preferentially during weaning. Second, rat pups, when seeking their first meals of solid food, have a strong tendency to approach adult rats at a distance from the nest site and to take their first meal of solid food in the immediate vicinity of a feeding adult. In situations in which food sources are spatially separate from one another, this tendency to eat in the vicinity of adults results in pups ingesting the same diet as the adults of their colony are eating. Pups soon become familiar with the flavor of the diet that they and the adults are eating and thereafter show great hesitancy in ingesting unfamiliar foods. A more complex feeding habit believed to be socially transmitted by Norway rats has been described by Gandolfi and Parisi (1972, 1973) who have found marked differences in the exploitation of bivalve mollusks as a food source by rat clans living on the banks of the Po River. Some clans feed extensively on bivalves, w h c h they collect by diving to the river bottom, while other clans d o not prey on the mollusks despite their ready availability in the river adjacent t o clan territories. There is, in addition, considerable evidence that the specific mode of opening the shells of these prey differ from colony to colony and is also

86

BENNETT G. GALEF, JR.

socially transmitted. Techniques of opening mussel shells are also believed to be transmitted by parent oyster catchers to their young (Norton-Griffiths, 1967, p. 423). The observation that members of various species of tit in restricted areas of England, Scotland, Wales, and Ireland have developed the habit of opening the tops of milk bottles to secure milk as food is suggestive of behavioral transmission of some sort. Available data on the spread of the milk bottle-opening behavior is sufficient to support the conclusion that the behavior was initiated by a number of individual birds independently learning about this food source but that the majority of birds engaging in it had “learned it in some way from others” (Fisher and Hinde, 1949; Hinde and Fisher, 1951, 1972). Turner’s (1964) observations on the tendency of chicks to peck at objects pecked at by a mechanical “hen” suggest a possible means by which such behavior could’ be transmitted from one individual t o another. The preceding examples have involved the ingestion of relatively passive food objects. Predatory species have an additional problem in that food acquisition requires the capture of the intended food object prior to its ingestion. A number of investigations suggest the possibility that patterns of predation as well as ingestive behaviors are socially transmitted. Von Steiniger (1950), for example, has reported that wild rat populations on the island of Norderoog regularly stalk, kill, and eat sparrows, whereas those in other areas of Germany are not observed to d o so. Kruuk (1972, p. 119) has collected data indicating that different hyena clans living in the Ngorongoro crater have different prey preferences which are not explicable in terms of the relative abundance of the prey in question (wildebeest and zebra) in their territories. The mechanisms responsible for these differences in prey selection have not been determined. In reviewing related evidence concerning the prey selection patterns of raptorial birds, Cushing (1944) favors the contention that differences between the prey preferences of raptor species are maintained more through the interactions of parents with offspring than through heritable factors. In the absence of parental or human guidance, young raptors are very slow t o take live prey. For purposes of falconry, even a wild-caught adult must be taught to take the particular types of live prey for which it will be used in hunting if these are not already in the animal’s diet, and it must be retrained t o any new type of prey one wishes to add to its hunting repertoire. Although the evidence hardly justifies so strong a conclusion as Cushing reaches, it does suggest the probable importance of parental influence in the prey selection of raptor young. Observations by Ewer (1963, p. 592), Schaller (1967, p. 272), Kruuk and Turner (1967), Liers (1 95 I ) , and Leyhausen (1956) on the ways in which mammalian predators (meerkat, tiger, cheetah, otter, and domestic cat) introduce their young to the killing and eating of prey species suggest that parent-young

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

87

interactions may be important in the establishment of species typical predatory behavior. However, neither the extent to which the prey selection of young is determined by the selection of prey species introduced to them by adults nor the long term effects of parent-offspring interaction on the development of species typical predatory patterns have been determined. Kuo (1930, 1938)has demonstrated that kittens reared with females that killed rodents in the presence of their young, began killing rodents at a significantly earlier age and more frequently than kittens reared alone or with a small rodent. It is unfortunate that no comparison was made between kittens reared with killing and nonkilling mothers, but the data are suggestive. Van Lawick-Goodall (1968, 1970) has indicated that the use by chimpanzees of twigs and sticks in capturing termites is transmitted between generations by observational learning, although the observations she has reported are not sufficient to support this contention. C. PREDATOR AVOIDANCE

The tendency of organisms to avoid potential predators while remaining undisturbed by the approach of harmless individuals is well documented. In most cases these differences in response are presumed to result either from instinctive responses to stimulus aspects of potential predators or from responses acquired by the individual as a result of its previous experience with similar stimulus configurations. There are, however, a few scattered reports of instances in which responses to novel stimuli appear to be learned as a result of interaction with conspecifics in the presence of those stimuli. Jackdaw fledglings, for example, learn to recognize enemies from the adults of their flock (Lorenz, 1952, p. 145). Upon the appearance of a predator, experienced individuals emit a “rattle” call that the young associate with the stimulus configuration eliciting the call in adults and which they thereafter avoid. Young gazelle, zebra, and wildebeest are believed to transmit to their young information concerning the flight distance to be maintained with respect to various predators (Hediger, 1964; Schaller, 1972, p. 389). Hochbaum (1955) reports, similarly, that loss of flight behavior from man in wild ducks arriving in a wild fowl refuge is transmitted from one bird to another and from flock to flock. All four of these reports are anecdotal and lack corroborating data, but investigation of these and similar phenomena under controlled conditions could prove interesting. In a laboratory study of the dissemination among captive chimpanzees of the habit of playing with novel objects; Menzel et af. (1972) have provided compelling evidence of the social transmission of two patterns of play behavior involving the approach to and manipulation of normally avoided novel objects. Menzel (1966, p. 134) has also described a particularly intriguing observation of apparent intentional transmission of avoidance behavior in free-living Japanese macaques (see also Menzel, 1973, p. 200). On “more than six occasions,”

88

BENNETT G . GALEF, JR.

adult females were observed to pull their offspring away from a novel object that the females themselves were avoiding. No mention is made of the long-term effects of this interaction on the behavior of the young. However, in a laboratory analog, Stephenson (1967) trained adult male and female rhesus monkeys to avoid manipulating an object and then placed individual naive animals in a cage with a trained individual of the same age and sex and the object in question. In one case, a trained male actually pulled his naive partner away from the previously punished manipulandum during their period of interaction, whereas the other two trained males exhibited what were described as “threat facial expressions while in a fear posture” when a naive animal approached the manipulandum. When placed alone in the cage with the novel object, naive males that had been paired with trained males showed greatly reduced manipulation of the training object in comparison with controls. Unfortunately, training and testing were not carried out using a discrimination procedure so the nature of the transmitted information cannot be determined, but the data are of considerable interest. D.

BIRD VOCALIZATIONS

As mentioned in the Introduction, cases in which the occurrence of species typical song in adulthood require exposure to conspecific song during the fledgling period lie outside the range of phenomena to be considered here, because the transmitter is incapable of acquiring the relevant pattern of behavior in the absence of interaction with conspecifics. However, one aspect of the ontogeny of bird ,song may exemplify the social transmission of an acquired behavior in the sense in which the term has been employed here. A number of species of song bird show regular differences in the song pattern produced by members of geographically distinct breeding populations. The detailed structure of the song varies little among animals resident in one area but is consistently different between geographic populations (Armstrong, 1965). In one species, the whitecrowned sparrow, laboratory analysis of the ontogeny of these dialects indicates that they, like the typical song, are acquired by juveniles during the first 100 days of life as a result of experiencing the song of older males that sing in the same dialect (Marler and Tamura, 1964). It is possible that the specific dialect within an area is a modification in song pattern introduced by an individual who acquired it in some way and transmitted it to his progeny. If this admittedly speculative account of dialect origins is correct, then song dialect traditions would be transmitted acquired patterns of behavior. 111. LEARNING AND CONDITIONING PARADIGMS As the preceding discussion indicates, observers of animals in their natural

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

89

habitats have reported a wide variety of behaviors in a number of vertebrate species which may be interpreted as traditional in nature, that is, as resulting from the transmission of acquired behavior from individual to individual. In most instances this interpretation has not been established by adequate experimentation in either laboratory or field settings. It has been proposed (Lehrman, 1970) that one task of the student of animal behavior is to seek an understanding of the sources of the behavior of organisms in their ontogeny and phylogeny. In pursuit of this goal many studies have been performed to determine the role of hereditary factors and individual experience in the development of apult behavioral phenotypes. There are, however, relatively few laboratory studies concerned with the role of behavioral transmission in the development of behavior. Ethologists have often implicitly assumed that such transmission is possible, and the frequently employed Kasper-Hauser or isolation-rearing design has served in part to control for behavior acquired through conspecific interaction. However, with some exceptions, little direct laboratory investigation of behavior acquisition through social interaction has been undertaken by ethologically oriented researchers. Most laboratory studies of transmission of behavior has been carried out within the ixperimental psychological framework. The approach of experimental psychologists to the problem of behavior transmission has generally been to seek to extend the Skinnerian and Pavlovian paradigms to incorporate cases in which conspecific behavior serves as a discriminative stimulus for some learned response or as an unconditioned stimulus for some reflexive behavior and, thereby, t o explain apparent “imitative” behavior in laboratory settings. The studies of learning by “imitation” undertaken by Thorndike (191 1) played a fundamental role in the development of North American psychology and determined the approach t o the study of social learning phenomena subsequently pursued. It is, therefore, worth considering the conclusions he reached from investigations of what would now be labeled observational learning. During the latter part of the nineteenth centuv, students of animal behavior, Romanes (1882), in particular, believed that animals could readily learn to perform complex tasks by imitating the observed behavior of others. Supporting data were entirely anecdotal. Thorndike undertook a careful examination of the possibility that animals (cats, chickens, dogs, monkeys) could learn by “the formation of associations by imitation” (Thorndike, 1911, p. 81). As is well known, the general results were entirely negative; neither cats, dogs, monkeys, nor chicks proved capable of learning arbitrary tasks as a result of observing trained conspecifics perform these tasks. Thorndike reached the conclusion that “learning to do an act from seeing it done” did not play an important role in the development of behavior. Apparently imitative behavior was seen as identical in its process of acquisition to other types of learned performance, as depending

90

BENNETT G. GALEF, JR.

on the interaction of instinct, the law of effect, and the law of exercise. Experimental psychologists have generally adopted Thorndike’s approach to behavioral transmission, treating it as a special case of trial and error learning, although more recent studies of observational learning in cats (Adler, 1955; Cheder, 1969; John et al., 1968; Herbert and Harsh,,1944) and monkeys (Darby and Riopelle, 1959; Warden et al., 1940; Warden and Jackson, 1935; but see Hall, 1963) suggest that there are situations in which observational learning may occur. For the purpose of this discussion we will briefly describe only the work of Miller and Dollard, of Skinner, and of Church t o indicate the approach of experimental psychologists working on problems of social learning within the Thorndikian tradition.* In their classic text, “Social Learning and Imitation,” Miller and Dollard (1941) restrict their discussion of animal social learning to what they call “matched dependent behavior.” In matched dependent behavior the behavior of one organism (the leader) serves as a cue or discriminative stimulus for a second organism (the imitator), indicating the behavior in which the imitator must engage in order to receive reinforcement. In their basic experiment, Miller and Dollard trained groups of rats either to make the same choice as their leader at the junction of a T-maze or to make the opposite choice from him in order t o receive food reinforcement. It was found that, after approximately 40 reinforced trials, animals in the appropriate groups learned either to follow or not t o follow. In successive experiments it was shown that animals trained with an albino leader in a T-maze for food reinforcement continued to behave appropriately without further training when the leader was changed from albino to black or the motivational state from hunger to thirst. Thus, a learned following response could generalize from one situation to another. Skinner (1953) in his analysis of imitative behavior, similarly indicates that one pigeon can be trained to imitate the behavior of another, but only if specific discriminative reinforcement has occurred. Thus, if one reinforces a pigeon if, and only if, it engages in the same behavior as another pigeon, the behavior of the first pigeon will come to resemble the behavior of the second. The mechanism proposed by Miller and Dollard and by Skinner is certainly sufficient to produce a certain uniformity in the behavior of contemporaneous members of a group of animals, and it is possible that some behavioral phenomena observed in field situations reflect differential schedules of reinforcement experienced by individuals when they behaved similarly to or differently from conspecific individuals. The fundamental problem with the “matched dependent” model in terms of the definition of social transmission processes proposed here is that it will not suffice as a mechanism for the maintenance of transmitted behavior beyond the period of interaction. Because the behavior of the leader 2For a more complete review see Davis, J. M. 1973. Imitation: A review and critique. In “Perspectives in Ethology” (P. P. G . Bateson and P. H. Klopfer, eds.), pp. 43-72. Plenum, New York.

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

91

is the discriminative stimulus for the occurrence of similar behavior in followers, once the leader departs those aspects of the behavior of the followers dependent on the presence of the leader are lost. For the behavior pattern initiated by the leader to become part of the behavioral repertoire of the follower in the absence of the leader, it is necessary for it to be controlled by stimuli that are not dependent on the behavior of the leader. The most direct examination of the possibility of such a transfer of stimulus control of behavior from a leader organism to some other stimuli in the environment is that of Church (1957). In one experiment, Church trained rats to follow leaders to the left and right arms of a T-maze. After 150 such trials an incidental cue was added such that the leader always entered the arm of the T-maze marked by a light. After 100 such following trials with the incidental cue present, the experimental subjects were tested for a series of 8 trials in the absence of a leader but with the incidental cue available. They showed a marked preference’for the lighted arm. As Church (1968, p. 143) has indicated, the principles of incidental learning provide a viable mechanism by which certain behavior patterns may be transmitted among conspecifics and maintained after the departure of the original instigator. For example, the observation by Galef and Clark (1 97 1) that young rats initially approach adults at a food site, eat in their vicinity, become familiar with the flavor of the diet eaten by adults, and develop a long-lasting preference for it, can be understood as an incidental learning process. Whereas Miller and Dollard, Skinner, and Church considered imitative learning in animals as a special case of discriminative operant conditioning, Humphrey (1921) discussed imitative behavior as a type of Pavlovian conditioning and cited a number of observations in support of this position. For example, Breed had observed that pigeons, placed in a cage where they could observe others pecking food, pecked the floor of their cages although no food was available t o them. According to a Pavlovian conditioning interpretation, the pigeons had in the past pecked the substrate (the unconditioned response) in the presence of food (the unconditioned stimulus). Ground pecking had frequently occurred while other pigeons were pecking the ground (the conditioned stimulus) and, as a result of these repeated pairings of the conditioned and unconditioned stimuli, the sight of other pigeons ground pecking was now sufficient to elicit ground pecking in the subjects. This observation may, however, be more parsimoniously explained within the ethological model by assuming that, in the pigeon, ground pecking by one individual serves to release ground pecking in conspecifics. Little experimental investigation has been undertaken to determine the role of conditioning processes in such situations. An alternative Pavlovian model for the transmission of behavior between conspecifics has been developed by personality theorists for the study of selected aspects of interpersonal behavior (Berger, 1962). It would seem to have consider-

92

BENNETT G. GALEF, JR.

able potential as a basis for the transmission of acquired behavioral responses in animals though it has little empirical support to date. In this model the behavior of one organism is treated as an unconditioned stimulus, the unconditioned response being a similar behavior elicited in the observer, and the conditioned stimulus the stimulus to which the original organism has learned to respond. The possibility of such conditioning depends on the existence in any given instance of an appropriate “contagious” or “infectious” behavior (Armstrong, 1965) in which the performance of a more-or-less instinctive or reflexive pattern of behavior by one individual acts as a releaser for the same behavior in a conspecific (Thorpe, 1956, p. 133). Repeated observations by one organism (the observer) of the response of others (the models) to some stimulus, those responses eliciting similar behavior in the observer, could lead t o a conditioned response on the part of the observer t o the stimulus eliciting the response in the models. The possibility exists that not only overt behavior but also emotional states may be transmitted in this way (Berger, 1962; Bandura and Rosenthal, 1966). For example, a restrained rat that has observed a conspecific receiving shocks in association with presentation of a red light will subsequently accelerate its own shock avoidance responding in the presence of a red light (Riess, 1972; see also Stephenson, 1967; Menzel, 1973, p. 209). It is possible that suchPavlovian conditioning of contagious behavior is responsible for socially transmitted avoidance behavior or learned approaches to frightening stimuli. As this brief review indicates, behavior may be transmitted from individual to individual as a result of processes formally similar to those at work in the usual cases of discriminative operant and of Pavlovian conditioning. Unfortunately, the extent to which such modes of behavioral transmission play a role in the development of the behavior of animals in their natural habitats remains undetermined. The relatively large number of trials required in the laboratory to establish the phenomena described above might seem to reduce the probability of their playing a role in field settings. It must be remembered, however, that the freedom of organisms to interact continuously in the wild may result in large numbers of interactions in a relatively brief period of time. Thus, the fixed trial procedures used to control interaction in the laboratory might disguise the rapidity with which social learning could occur under less controlled circumstances. IV. PROBLEMS OF TERMINOLOGY

A third body of literature relevant to the topic of behavioral transmission is a very broad one involving attempts to categorize the ways in which organisms may influence one another’s behavior. The psychological literature, in particular, is rich in terminology seeking to delineate various aspects of the ways in which

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

93

the behavior of one organism can influence the behavior of another. Some investigators prefer purely descriptive terms even though these can obscure differences in the mechanisms underlying surface similarities in behavioral interaction (allelomimetic behavior, social facilitation). Others utilize terminology reflecting hypothesized underlying mechanisms mediating observed behavioral interaction (coaction, local enhancement, matched dependent behavior, copying), and there are those who employ operationally defined categories (following, observational learning). Unfortunately, some terms (mimesis, contagious behavior, and social facilitation) have been used to refer to very different phenomena by different authors. As Oldfield-Box (1970) has noted, one of the major impediments to systematic investigation in this area stems from the confusion in terminology and the replacement of analysis of instances of social learning by a rather arbitrary labeling of inadequately explored phenomena. The problems with attempts at classifying the possible social processes resulting in the transmission of acquired behavior are probably not purely semantic in origin. The difficulties inherent in attempting to categorize a wide range of complex interactions within a limited conceptual framework become apparent when one begins to explore the wealth of interactions that could result in a modification of the behavior of one organism toward homogeneity with that of another. To give a partial indication of this complexity (ignoring, for the moment, alternative mechanisms mediating similar observed effects on behavior) it is sufficient to outline some of the possible alterations in behavior of an individual organism A as a result of its exposure to a conspecific organism B in some environment E. For purposes of simplicity in this discussion, it will be assumed that the observer is already familiar with both A and B’s behavior in E prior to A’s experiencing B in E, although other procedures than using A as his own control are possible and, in many situations, preferable in the study of social interactions. After experiencing B in E, (1) A may exhibit a motor pattern not previously in his repertoire while in E. (2) A may exhibit a change in the temporal or spatial distribution of his previous responses in E or in the stimuli eliciting or controlling his behavior in E. (3)Alterations in A’s behavior in E may or may not outlast the period of interaction of A and B in E. There are not only a variety of types of alterations in A’s behavior possible in response to experiencing B, but also a variety of possible interactions between A and B in E. 1. Organism A may not encounter E until B has already departed from E. Interaction in this case would depend on durable alterations in E resulting from B’s presence in E. 2 . Organism A may observe some aspect of B s activities in E without actually co-occupying E with B. 3 . Organisms A and B may be simultaneously present in E and free to interact

94

BENNETT G. GALEF, JR.

fully. This interaction between A and B in E may take many forms: A may approach or avoid B, follow B or flee from it, behave amicably, aggressively, submissively, sexually, or any of a variety of other ways toward it. In attempting to reduce the large number of possible combinations and permutations of interactions and changes in behavior to a workable number of categories, important differences in both their causes and effects become obscured. Thus, to label a change in behavior as resulting from coaction [the presence of others leading to enhancement of dominant and well-developed responses (Zajonc, 1965, 1969)] leads to the strange situation of placing in the same category the observation that each of a pair of ants will dig faster than either alone (Chen, 1937; disputed by Sudd, 1972) and the observation that 2 human cyclists ride faster than one alone, although the mechanisms mediating these effects are in all probability quite disparate. Again, wild rat pups approach adults, eat in their vicinity, learn incidental cues concerning the diet they eat, and show continued avoidance of alternative diets as a consequence of their neophobia. Juvenile mountain sheep follow adults and, thus, learn their way about their scattered home ranges. The social interaction responsible for the transmission of behavior is similar in the two cases, depending on a tendency of young to remain in the proximity of conspecifics, yet there seems to be little gain in categorizing the two behaviors as examples of socially facilitated or allelomimetic behavior. Such labeling of the interaction adds nothing to our understanding of the mechanisms or interactions responsible for the occurrence of the behavior of interest. The necessary precursors to useful classification are the precise description of the effects of social interaction, the determination of the necessary and sufficient conditions for the occurrence of observed changes in behavior, and an analysis of the mechanisms mediating those changes. At this early stage of our knowledge it seems premature t o impose arbitrary structures on the inadequately analyzed observations available. The test of a classificatory scheme lies in its heuristic value, and little seems to have been gained from the categorizations proposed to date. In fact, experimental social psychologists interested in animal interactions tend to be satisfied with an analysis of a behavioral social interaction that ends rather than begins with the discovery that the organisms in question show changes in behavior as a result of the interaction. If the result of such interaction is to produce increased homogeneity in the behavior of interactants, the use of such terms as “imitation” or “social facilitation” to describe that interaction seems to reduce the perceived need for an analysis of the mechanisms mediating the observed alteration in behavior or for a determination of the necessary and sufficient conditions for its occurrence. It is possible that the discussion of behavioral phenomena resulting from social

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

95

interaction in terms of their functions or effects (Tinbergen, 1963) rather than their poorly understood causes or underlying mechanisms would prove useful. The focusing of attention on the results of an interaction might point to the need for analyses, in terms of observable events, of the processes by which those results were achieved and might help to avoid the errors of explanation by denotation that, as Oldfield-Box (1970) has pointed out, hinder progress in the area.

V.

CONCLUSIONS

The preceding review of the literature, although in no way comprehensive, is intended to give a broad overview of the current state of knowledge of the means by which organisms are able to transmit acquired behavior. It is clear that much work is to be done before a complete picture of the processes involved is reached. Numerous instances of apparent behavioral transmission reported by field observers must be examined under controlled conditions to permit analysis of the mechanisms by which transmission proceeds, and an effort must be made to ascertain the adequacy of laboratory-derived models of behavioral transmission to explain the behavior of organisms in their natural environment. There are at least two central questions. First, what is it that is being transmitted and, second, what are the mechanisms by which transmission is achieved? Although it is probably premature to attempt t o answer these questions in the light of current knowledge, two general principles seem to emerge from consideration of the data described here. First, with respect to the nature of transmitted material, there would appear to be relatively few cases (bird song dialect, “placer-mining’’ of wheat) in which actual motor patterns are communicated from one individual to another. In almost every case, the motor patterns involved seem t o develop independently of social interaction. Acquired stimulus control of behavior rather than acquired motor patterns themselves are the usual messages passing between individuals. Jackdaws, for example, do not learn to flee nor Japanese macaques to eat as a result of social interaction. Knowledge of the appropriate context within which to engage in these activities is acquired as a result of experience with conspecifics. Second, if the preceding analysis of the message content is sound, then the mechanisms by which social transmission of behavior proceed should be ones that enable stimulus control of behavior t o pass from one individual to another. This does, in fact, seem to be the case. In all but a few instances transmission of behavior appears to result in large part from the introduction by one organism of another into a stimulus situation to which the second organism is predisposed,

96

BENNE?T G . GALEF, JR.

either as a result of previous experience or of instinctive tendencies, to respond in such a way as to acquire the behavior of the first organism (Ewer, 1969). In many cases, especially those involving transmission of behavior from parent to offspring, the social component of the interaction may be almost trivial, involving nothing more than giving birth to offspring in one locale rather than another. The continuity of reproductive behavior in one location from generation to generation, for example, depends on a tendency in the young t o return t o their place of birth to reproduce. Similarly the tendency of young to remain in close proximity to conspecific adults, may result in their acquiring adult feeding habits, home ranges, predatory patterns, or responses to potentially dangerous stimuli; but the particular response acquired depends on the reaction of the follower to the stimulus events to which the leader introduces him. This may be the case whether the “remaining close” response or the response to the stimulus situation is, for want of better terms, conceived as “instinctive” or “learned.” Even in those cases in which the social interaction between initiator and acquirer seems more directly evolved for purposes of transmission of behavior, as is the case, for example, in the feeding behavior of maternal meerkats (Ewer, 1963, p. 592), otters (Liers, 1951), or mother hens (Hogan, 1966), the acquisition of the adult feeding patterns by the young depends on the tendency of the young to respond appropriately to the stimuli presented to them by their parents. The observed behavioral transmission, thus, results from a combination of social interactions and a predisposition to respond in a particular way to the stimuli encountered as a result of these social interactions. Maintenance of the transmitted behavior may depend either on the reinforcement contingent on engaging in the pattern of behavior in question or a predisposition t o behave in certain ways toward some class of stimuli once they are experienced. The conceptual dichotomy between the inheritance and individual acquisition of behavior embodied in the nature-urture controversy of the 1950s has tended to obscure the existence of developmental processes involving the interaction of genetic and environmental effects in determining the behavior of individuals. Perhaps, as a consequence, the empirical analysis of the social transmission of behavior resulting from such interactions has been largely ignored by contemporary students of behavior. It is hoped that this intrinsically interesting aspect of the ontogeny of behavior will receive greater attention in the future. Acknowledgments 1 am grateful to the National Research Council of Canada for their support while this work was undertaken and to Lorraine Allan, Abraham Black, Mertice Clark, Herbert Jenkins, Michael Leon, John Platt, Paul Rozin, and W. John Smith for their advice, critical reading, and helpful discussion. Particular thanks are due the editors of the present volume for their most thoughtful and constructive critiques of an earlier draft. All errors both of commission and omission remain, of course, the sole responsibility of the author.

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

97

References Adler, H. E. 1955. Some factors of observational learning in cats. J. Genet. Psychol. 86, 159-1 7 7 . Armstrong, E. A. 1965. “Bird, Display and Behavior.” Dover, New York. Bandura, A., and Rosenthal, T. L. 1966. Vicarious classical conditioning as a function of arousal level. J. Pers. Social Psychol. 3, 5442. Berger, S. M. 1962. Conditioning through vicarious instigation. Psychol. Rev. 69,450-466. Calhoun, 1. B. 1962. “The Ecology and Sociology of the Norway Rat.” US. Dep. Health, Educ. Welfare, Bethesda, Maryland. Carpenter, C. R. 1934. A field study of the behavior and social relations of howling monkeys. Comp. Psychol. Monogr. 10, No. 2, Serial No. 43. Chen, S. C. 1937. Social modification of the activity of ants in nest-building. Physiol. Zool. 10,420436. Cheder, P. 1969. Maternal influence in learning by observation in kittens. Science 166, 901-903. Church, R. M. 1957. Transmission of learned behavior between rats. J. Abnorm. Social Psychol. 34, 163-165. Church, R. M. 1968. Applications of behavior theory to social psychology: Imitation and competition. In “Social Facilitation and Imitative Behavior.” (E. C. Simmel, R. A. Hoppe, and G. D. Milton, eds.), pp. 135-167. Allyn & Bacon, Boston, Massachusetts. Cushing, J. E., Jr. 1944. The relation of nonheritable food habits to evolution. Condor 46, 265-27 1. Darby, C. C., and Riopelle, A. J. 1959. Observation learning in the rhesus monkey. J. Comp. Physiol. Psychol. 5 2 , 94-98. Darling, F. F. 1937. “A Herd of Red Deer.” Oxford Univ. Press, London and New York. Ehrenfeld, D. W. 1974. Conserving the edible sea turtle: Can mariculture help? Amer. Sci. 62,23-31. Emlen, J. T., Jr. 1938. Midwinter distribution of the American Crow in New York State. E C O ~ O 19,264-215. ~V Ewer, R. F. 1963. The behaviour of the Meerkat, Suricata mricatta (Schreber). Z . Tierpsychol. 20,570407. Ewer, R. F. 1969. The “instinct to teach.” Nature (London) 222,698. Fisher, J., and Hinde, R.A. 1949. The opening of milk bottles by birds. Brit. Birds 42, 347-357. Galef, B. G., Jr., and Clark, M. M. 1971. Social factors in the poison avoidance and feeding behavior of wild and domesticated rat pups. J. Comp. Physiol. Psychol. 75, 341-357. Galef, B. G., Jr., and Clark, M. M. 1972. Mother’s milk and adult presence: Two factors determining initial dietary selection by weanling rats. J. Comp. Physiol. Psychol. 38, 220-225. Galef, B. G., Jr., and Henderson, P. W. 1972. Mother’s milk: A determinant of the feeding preferences of weaning rat pups. J. Comp. Physiol. Psychol. 78,213-219. Galef, B. G., Jr., and Sherry, D. F. 1973. Mother’s milk: A medium for the transmission of cues reflecting the flavor of mother’s diet. J. Comp. Physiol. Psychol. 83, 374-378. Gandolfi, G., and Parisi, V. 1972. Predazione su Unio pictorum L. da parte del ratto, Rattus norvegicus (Berkenhout). Acta Natur. 8 , 1-27. Gandolfi, G., and Parisi, V. 1973. Ethological aspects of predation by rats, Rattus norvegicus (Berkenhout), on bivalves Unio pictorum, L. and Cerastoderma lamarcki (Reeve). Boll. ZOO^. 40,69-74. Geist, V. 1971. “Mountain Sheep.” Univ. of Chicago Press, Chicago, Illinois I

I

98

BENNETT G. GALEF, JR.

Hall, K. R. L. 1962. Numerical data, maintenance activities and locomotion of the wild chacma baboon, Papio ursinus. Proc. Zool. Soc. London 139,181-220. Hall, K. R. L. 1963. Observational learning in monkeys and apes. Brit. J. Psychol. 54, 201-226. Harlow, H. F., and Harlow, M. K. 1965. The affectional systems. In “Behavior of Nonhuman Primates” (A. M. Schrier, H. F. Harlow, and F. Stollnitz, eds.), Vol. 2, pp. 287-334. Academic Press, New York. Hasler, A. D. 1966. “Underwater Guideposts.” Univ. of Wisconsin Press, Madison. Hediger, H. 1964. “Wild Animals in Captivity.” Dover, New York. Herbert, M. J., and Harsh, C. M. 1944. Observational learning by cats. J. Comp. Psychol. 37, 8145. Hinde, R. A. 1954. Factors governing the changes in strength of a partially inborn response, as shown by the mobbing behaviour of the chaffinch (Fringilla coelebs). 1. The nature of the response, and an examination of its course. Proc. Roy. SOC.,Ser. B, 142, 306-33 1. Hinde, R. A., and Fisher, J. 1951. Further observations on the opening of milk bottles by birds. Brit. Birds 44,392-396. Hinde, R. A,, and Fisher, J. 1972. Some comments on the republication of two papers on the opening of milk bottles by birds. In “Function and Evolution of Behavi.or” (P. H. Klopfer and J. P. Hailman, eds.), pp. 377-378. Addison-Wesley, Reading, Massachusetts. Hochbaum, H. A. 1955. “Travels and Traditions of Waterfowl.” Univ. of Minnesota Press, Minneapolis. Hogan, J. A. 1966. An experimental study of conflict and fear: An analysis of behavior of young chicks toward a mealworm. Part 11. The behavior of chicks which eat the meal worm. Behaviour 27,273-289. Humphrey, G. 1921. Imitation and the conditioned reflex. Pedagog. Seminary J. Genet. Psychol. 28, 1-21. Immelman, K. 1972. Sexual and other long-term aspects of imprinting in birds and other species. In “Advances in the Study of Behavior” (D. S . Lehrman, R. A. Hinde, and E. Shaw, eds.), Vol. 4, pp. 147-169. Academic Press, New York. John, E. R., Chesler, P., Bartlett, F., and Victor, I. 1968. Observation learning in cats. Science 159, 1489-1491. Kawai, M. 1965. Newly acquired pre-cultural behavior of the natural troop of Japanese monkeys on Koshima Inlet. Primates 6, 1-30. Kawamura, S. 1959. The process of sub-culture propagation among Japanese macaques. Primates 2 , 4 3 6 0 . King, J. 1955. Social behavior, social organization and population dynamics in a black-tailed prairie dog town in the Black Hills of South Dakota. Contrib. Lab. Vertebr. Biol., Univ. Mich. No. 67. Klopfer, P. H. 1961. Observational learning in birds: The establishment of behavioral modes. Behaviour 17,71430. Klopfer, P. H. 1969. “Habitats and Territories.” Basic Books, New York. Klopfer, P. H., and Hailman, J. P. 1965. Habitat selection in birds. In “Advances in the Study of Behavior” (D. S. Lehrman, R. A. Hinde, and E. Shaw, eds.), Vol. 1, pp. 279-303. Academic Press, New York. Kruuk, H. 1972. “The Spotted Hyena.” Univ. of Chicago Press, Chicago, Illinois. Kruuk, H., and Turner, M. 1967. Comparative notes on predation by lion, leopard, cheetah and wild dog in the Serengeti area, East Africa. Mamrnalia 31, 1-27. Kuo, Z. Y.1930. The genesis of the cat’s response to the rat. J. Comp. Psychol. 11, 1-35.

SOCIAL TRANSMISSION OF ACQUIRED BEHAVIOR

99

Kuo, Z. Y. 1938. Further study on the behavior of the cat toward the rat. J. Comp. Psychol. 25,143. Kuo, Z. Y. 1967. “The Dynamics of Behavior Development.” Random House, New York. Lehrman, D. S. 1970. Semantic and conceptual issues in the nature-nurture problem. In “Development and Evolution of Behavior” (L. R. Aronson, E. Tobach, D. S. Lehrman, and J. S. Rosenblatt, eds.), pp. 17-52. Freeman, San Francisco, California. Leyhausen, P. 1956. Verhaltensstordien an Katzen. Z. Tierpsychol. No. 2. Liers, E. E. 1951. Notes on the River Otter (Lutra canadensis). J. Mammal. 32, 1-9. Lorenz, K. 1952. “King Solomon’s Ring.” Crowell, New York. Mainardi, D. 1970. La trasmissione culturale. Boll. Zool. 37,449456. Mainardi, D. 1973. Biological basis of cultural evolution. Atti Accad. Naz. Lincei, 0. Sci. Fis.,Mat. Natur., Rend. 182, 175-188. Marler, P., and Tamura, M. 1964. Culturally transmitted patterns of vocal behavior in sparrows. Science 146, 1483-1486. Matthews, G. V. T. 1968. “Bird Navigation,” 2nd Ed. Cambridge Univ. Press, London and New York. Menzel, E. W., Jr. 1966. Responses to objects in free-ranging Japanese monkeys. Behaviour 26, 130-150. Menzel, E. W . , Jr. 1973. Leadership and communication in young chimpanzees. In “Precultural Primate Behavior” (E. W. Menzel, Jr., ed.), pp. 192-225. Karger, Basel. Menzel, E. W., Jr., Davenport, R. K., and Rodgers, C. M. 1972. Protocultural aspects of chimpanzees’ responsiveness to novel objects. Folia Primatol. 17, 161-170. Miller, N. E., and Dollard, J. 1941. “Social Learning and Imitation.” Yale Univ. Press, New Haven, Connecticut. Nordeng, H. 1971. Is the local orientation of anadronous fishes determined by pheromones? Nature (London) 233,411413. NortonCriffiths, M. 1967. Some ecological aspects of the feeding behaviour of the Oystercatcher Haematopus ostralogus on the edible mussel Mytilus edulis. Ibis 109,412-424. Oldfield-Box, H. 1970. Comments on two preliminary studies of “observational” learning in the rat. J. Genet. Psychol. 116,454 1. Rabinowitch, V. 1969. The role of experience in the development and retention of seed preferences in Zebra finches. Behaviour 33, 222-236. Riess, D. 1972. Vicarious conditioned acceleration: Successful observational learning of an aversive Pavlovian stimulus contingency. J. Exp. Anal. Behav. 18, 181-186. Romanes, G. J . 1882. “Animal Intelligence.” Kegan, Paul, Trench, London. Schaller. G. B. 1967. “The Deer and the Tiger.” Univ. of Chicago Press, Chicago, Illinois. Schaller, G. B. 1972. “The Serengeti Lion.” Univ. of Chicago Press, Chicago, Illinois. Schloeth, R., and Burckhardt, D. 1961. Die Wanderungen des Rotwildes Cervuselaphus L. im Gebeit des Schweizerischen National-parkes. Rev. Suisse Zool. 68, 146-155. Skinner, B. F. 1953. “Science and Human Behavior.” Macmillan, New York. Smith, N. G. 1968. The advantages of being parasitized. Nature (London) 219,690694. Snyder, L. L. 1948. Tradition in Bird Life. a n . Field Natur. 62,75-77. Solomon, D. J. 1973. Evidence for pheromone influenced homing by migrating Atlantic salmon, Salmo salar (L.). Nature (London) 244,231-232. Stephenson, G. R. 1967. Cultural acquisition of a specific learned response among rhesus monkeys. In “Progress in Primatology” (D. Starek, R. Schneider, and H. J. Kuhn, eds.), pp. 279-288, Fischer, Stuttgart. Sudd, J. H. 1972. The absence of social enhancement of digging in pairs of ants (Formica lemani Bondroit). Anim. Behav. 20,813-819.

100

BENNETT G. CALEF, JR.

Telle, H. J. 1966. Beitrag zur Kenntnis der Verhaltensweise von Ratten, vergleichand dargestellt bei Rattus norvegicus und Rattus rattus. Z. Angew. Zool. 5 3 , 129-196. (Eng. transl. Nut. Res. Counc. Can., Tech. Transl. No. 1608.) Thorndike, E. L. 191 1. “Animal Intelligence.” Hafner, Darien, Connecticut. Thorpe, W. H. 1945. The evolutionary significance of habitat selection. Anim. Ecol. 14, 67-70.

Thorpe, W. H. 1956. “Learning and Instinct in Animals,” 2nd Ed. Methuen, London. Tinbergen, N. 1963. On aims and methods of ethology. Z. Tierpsychol. 20,410429. Turner, E. R. A. 1964. Social feeding in birds. Behaviour 2 4 , 1 4 6 . van LawickGodall, J. 1968. The behaviour of free-living Chimpanzees in the Gombe Stream Reserve. Anim. Behav. Monogr. 1(3), 161-31 1. van LawickGoodall, J. 1970. Tool-using in primates and other vertebrates. In “Advances in the Study of Behavior” (D. S. Lehrman, R. A. Hinde, and E. Shaw, eds.), Vol. 3, pp. 195-249. Academic Press, New York. von Steiniger, F. 1950. Beitrage zur soziologie und sonstigen Biologie der Wanderratte. Z. Tierpsychol. 7, 356-379. Warden, C. J., and Jackson, T. A. 1935. Imitative behavior in the Rhesus m0nkey.J. Genet. Psychol. 46,103-125 Warden, C . J., Fjeld, H. A., and Koch, A. M. 1940. Imitative behaviour in Cebus and Rhesus monkeys. J. Genet. Psychol. 56, 311-322. Wecker, S. C. 1963. The role of early experience in habitat selection by the Prairie Deer Mouse, Peromyscus maniculatus bairdi. Ecol. Monogr. 3 3 , 307-325. Wortis, R. P. 1969. The transition from dependent to independent feeding in the young ring dove. Anim. Behav. Monogr. 2 , 1 4 4 . Wynne-Edwards, V. C. 1962. “Animal Dispersion in Relation to Social Behaviour.” Hafner, New York. Zajonc, R. B. 1965. Social facilitation. Science 149,269-274. Zajonc, R. B. 1969. “Animal Social Psychology.” Wiley, New York.

Care and Exploitation of Nonhuman Primate Infants by Conspecifics Other Than the Mother SARAHBLAFFER HRDY PEABODY MUSEUM HARVARD UNIVERSITY CAMBRIDGE. MASSACHUSETTS

. .

I I1

I11. IV

.

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

Introduction Male Care vs Exploitation of Infants A Protection and Rescue B . Baby-sitting C. Adoption D . Agonistic Buffering E Infanticide F Care vs Exploitation and Degree of Relationship Nurture vs Abuse-Male and Female Roles The Pros and Cons of Aunting A Learning to Mother B. Incompetence, Kidnapping, and “Aunting to Death” C. Adoption D Other Benefits for the Mother-Infant Pair E Aunts and Infant Independence F Status Benefits for Mothers, Aunts, and Infants G Preferred and Available Aunts and Infants Selective Pressures on the Infant A Natal Coats and Other Traits of Attraction B. Phylogeny, Environment, or an Inducement to Caretakers Summary References

.

. .

.

. . . . V. . VI .

.

101 104 105 106 107 108 110 113 118 120 122 125 128 130 133 136 137 142 142 145 148 150

I . INTRODUflION Maternal care of offspring is both a widespread and relatively unsurprising phenomenon: by investing care the mother is presumably maximizing her 101

102

SARAH BLAFFER HRDY

chances of leaving surviving offspring. T o the extent that care represents an investment, care by conspecifics other than the parents is more puzzling. In the primate literature, such relations between infants and older animals have been referred to as mtnting (Rowel1 et al., 1964), paternal (Itani, 1959), or parental (Alexander, 1970) behavior. Despite this nomenclature, none of these terms necessarily designate any genetic relationship, although they do not preclude it. As used in this paper, the term “aunt” by definition excludes the mother; a male caretaker, on the other hand, may in fact be the biological father since paternity is rarely known. Recent theoretical suggestions about the role of kinship in the determination of behavior (Hamilton, 1964) and in particular current hypotheses concerning “inclusive fitness”-that is, the sum of an individual’s own fitness plus the effects that his behavior has on the fitness of his relatives and vice versa-make it increasingly important to know the extent to which such “aunts” and “uncles” really are related to their charges. On the basis of Hamilton’s theories, one would expect degree of relationship to be a rough predictor of the type of behavior that will be directed toward an infant. Lucid explanations of what Hamilton means by “degree of relationship” and its bearing on behavior are available in his own work (Hamilton, 1964, Part II), in Trivers (1974), and in Wilson (1971, Ch. 17). Very briefly, in diploid organisms such as primates, a parent and offspring, and full siblings share onehalf of their genes by common descent; half-siblings are related by one-quarter, cousins by one-sixteenth, and so forth. The likelihood of altruistic behavior will be a reciprocal function of the degree of relationship involved. In order for any given social trait t o be favored by natural selection, it should have a positive net effect for the inclusive, as well as the individual, fitness of the carrier. Where benefit differences are attached to different behavior, and where discernment of kinship is possible, as in the case of a sibling or maternal relationship, one would expect discrimination t o occur. In the case of paternity, where kinship is less easily determined by an observer, one would expect role differences between those animals in potentially progenitorial positions and those in positions peripheral to the breeding system. In this paper, instances of care for infants by individuals other than the natural mother, and also instances of abuse of infants by males (Section 11) and females (Section IV) of various primate species, are examined, and the advantages and disadvantages of such behavior for the parties concerned enumerated. As an extension of this approach, I explore natural-selection pressures on the infant and on the mother-infant pair to either attract or discourage conspecific attentions (Section V,B). Some differences between male and female treatment of infants are also discussed (Section 111). Throughout this paper, the purpose of enumerating costs and benefits as proposed above is to relate observed behavior to evolutionary theory. Needless to say, the data necessary to test the predictions generated by those theories do not

CARE AND EXPLOITATION OF PRIMATE INFANTS

103

exist in the primate literature. Long-term (5 years or more) genealogical information gathered under relatively natural conditions is available for only four species: Pan troglodytes, in the published and unpublished records of van Lawick-Goodall; M a m a fuscata, in the work of Kawai, Kawamura, and others; for Macaca mulatta in the work of Koford, Sade, Vessey, and others, and in unpublished records; and for Presbytis entellus in the unpublished notes of SM. Mohnot. In the case of the first three species, where such genealogical data have been used in behavioral analyses (van Lawick-Goodall, 1967, 1971; Itani, 1959; Kawai, 1958; Kawamura, 1958; Koford, 1963a,b; Sade, 1965, 1967; Yamada, 1963; and elsewhere), matrilineal kinship in connection with other life-history parameters has emerged as a crucial determinant of both social status and frequency of association with other animals. Whether the importance of maternal kin, as seen in chimpanzees and Japanese and rhesus macaques, will hold true for other species remains to be demonstrated as current studies yield more genealogical information and as new techniques are applied to this problem. For obvious reasons, matrilineages have been easier to determine than patrilineages. New possibilities for captive and trapped study populations include biochemical techniques for paternity exclusions and determination of probable paternity. Several analyses of blood proteins carried out for Macaca nemestrina (Simons and Crawford, 1969) and Macaca mulatta (C. Alper, 1973 personal communication) have already led to paternity exclusions. For most primate studies, there are no firm data on kinship; the researcher‘s impression that “there is no particular relationship,” or that one monkey is “probably an older sibling” may or may not be reliable. However, the following assumptions can be made with some degree of assurance. 1 . In multimale troops, dominant males are most likely to copulate with females at the height of estrus, and females are most likely t o be impregnated at this time, e.g., baboons and macaques (DeVore, 1965; Rowell, 1967; Michael and Zumpe, 1970); younger and more subordinate males are less likely either t o have consort relationships or to impregnate females. 2. In harems, the length of the leader’s reign, and his success in maintaining the breeding integrity of his troop, must be taken into account, e.g., patas monkeys and one-male troops of langurs (Hall, 1968; Yoshiba, 1968), but in general, this male will be the progenitor of that troop’s recent offspring. 3. In matrifocal societies in which contact with the mother may continue after birth of the next infant, e.g., Japanese macaques, chimpanzees, Nilgiri langurs (Yamada, 1963, p. 50; van Lawick-Goodall, 1971; Poirier, 1968, p. 49), juveniles or adults that seek recurrent contact with an older multiparous female may be assumed to be her offspring and, thus, her new children their half-siblings. Using behavioral indices to determine probable degree of relationship becomes

104

SARAH BLAFFER HRDY

dangerously circular when kinship derived this way is then used as a part of the explanation for observed behavior. Nevertheless, pending the availability of precise genealogical information, these assumptions will allow us t o formulate some predictions and to test these tentatively against the data that we do have. It is to be hoped that such an attempt $11 stimulate research that will allow for a more rigorous consideration of evolutionary theory among the primates. In the discussion that follows, 1 hope that it is clear that motivations ascribed to nonhuman primates refer to theoretical interpretations of observed behavior. It is assumed that on average genes of those animals that respond to certain situations in a manner which is reproductively advantageous to them will be disproportionately represented in subsequent generations. On this survival and reproduction level of causation (discussed by Tinbergen, 1963), increased reproductive success is a sufficient explanation for an animal’s behavior (Williams, 1972). Proximate mechanisms leading to specific behavior (e.g., endocrinological bases and behavioral conditioning) are not considered in this paper.

11. MALE CARE VS. EXPLOITATION OF INFANTS

From the assumptions listed in Section I, one would expect dominant males (which are probable progenitors, likely to have a greater stake in the well-being of infants born in the troop) and young males closely associated with an infant from its birth (whch are likely to be siblings) to engage in behavior that benefits an infant, even at some cost to themselves. Such altruistic behavior is described under Protection and Rescue (Section HA),Baby-sitting (Section II,B), and Adoption (Section I1,C). Subordinate males, which are unlikely to be progenitors and which have much to gain in terms of “fitness,” would be more likely to engage in behavior that benefits them even at the expense of the infant. Whether or not such males discriminate in favor of some infants (such as siblings), should depend on both their precise degree of relationship and how much they stand to gain. Behavior that primarily or exclusively benefits the male is discussed under Agonistic Buffering (Section II,D) and Infanticide (Section 11,E). This chapter is concerned with the potential advantages and disadvantages for the parties involved of each of the five categories of male-infant interactions just listed. Relevant instances are cited from various primate species. No attempt is made to be all-inclusive since patterns of male-infant interactions for all species for which information exists have been recently reviewed (Mitchell and Brandt, 1972). Detailed accounts of what has been termed paternal or parental behavior are available for Japanese macaques and Barbary apes (Itani, 1959; Lahiri and Southwick, 1966; Alexander, 1970). Recent studies that were not included in the review by Mitchell and Brandt will be emphasized here, especially a paper by

CARE AND EXPLOITATION OF PRIMATE INFANTS

105

Deag and Crook (197 1) and one by Ransom and Ransom (197 l), which provide the most detailed descriptions to date of males using infants to further their own purposes. Deag and Crook suggest two major groupings for the behavior exhibited in male-infant interactions: mule cure and ugonistic buffering. By care they mean maternal-like behavior including holding, grooming, and carrying the infant as well as protecting the infant from other individuals and dangerous situations. By agonistic buffering Deag and Crook refer to situations in which an infant is used by a male as a “passport” (Itani, 1959) or as a buffer to inhibit aggression in some social situation, usually one that involves other males. In other words, two types of behavior are being described: behavior that benefits the infant and behavior that benefits the male but, if at all, only indirectly benefits the infant. Just how this distinction relates to the likelihood that a male and a given infant will be related is discussed in Section IIP. At the outset, I need to make clear that I focus here on those cases in whick the male approaches the infant. In some primate species, older infants and juveniles do actively solicit male attention. For example, among Hanuman langurs, males generally ignore infants and it is the infants that must initiate contact. Among vervets, juveniles sometimes solicit the aid of one adult male against a third animal (Struhsaker, 1967b). The possibility that younger animals might be using adult males is a subject in itself. This topic will not be discussed here and is dismissed with the following two generalizations: only older infants could be expected to take the initiative in this fashion, and, although male exploitation of infants may have serious repercussions for the infant, the converse would rarely be true.

A.

I I I

I

r

j

PROTECTION AND RESCUE

In those species in which the male remains in the vicinity of the motherinfant pair, protection, which may include threats and actual fighting, is the male’s most important contribution t o infant survival. The male may protect the infant from external, usually interspecific danger, and defend it in intragroup encounters. It is important to distinguish between generalized troop defense, which indirectly affects the infant, and male reactions aimed specifically at defense of the infant. This distinction is illustrated by the difference between those species in which males show little interest in newborns (Presbytis entellus, Erthyrocebus putus) and in which females with infants may even avoid adult males (Presbytis johnii), and those species in which males exhibit such solicitude toward infants that mothers of newborns may avoid contact with other group members while staying in close contact with adult males (e.g., Pupio unubis, Pupio cynocephulus). When in trouble a young juvenile baboon may be more likely to seek out an adult male than its own mother (Hall and DeVore, 1965, p. 84).

106

SARAH BLAFFER HRDY

In those species in which males are least likely to be in close, attentive association with mother-infant pairs (e.g., langurs), the male plays a relatively small role in group defense except to prevent intrusion by alien males. However, the frequently cited indifference of such males toward infants (see Mitchell and Brandt, 1972, p. 175) may be overemphasized. Males of both Presbytis entellus (McCann, 1934, p. 620) and Presbytis cristatus (Bernstein, 1968, p. 13) have been reported to respond to individual infants in distress. From observations of squirrel monkeys (Saimin sciureus) in a seminatural Florida environment, Dumond (1968) reports that “on one occasion a subadult male came from ten feet away t o retrieve a baby that was alone and which (Dumond) was menacing wildly”; shortly after, the male pushed the baby off but remained nearby. As Dumond continued to stare at the infant the male returned and took the baby onto his back. Such episodes involving male rescue of an externally threatened infant have been reported both for species with multimale defense-oriented troops, e.g., Japanese macaques (Itani, 1959, pp. 66, 84) as well as for those living in one-male groups or groups not normally considered defense oriented, e.g., black and white Colobus (Booth, 1962, p. 484; Haddow, 1952), Hanuman langurs (McCann, 1934; and possibly Jay, 1965), perhaps chimpanzees (Rahm, 1967, p. 206), lutongs and squirrel monkeys. B.

BABY-SITTING

Individualized male care of infants may also occur in the absence of any immediate danger. As defined by Ransom and Ransom (1971, p. 183), such baby-sitting refers t o any association between an infant and an older male, in the temporary absence of its mother, in which the male fosters the infant’s wellbeing. This might involve grooming, reassuring contact, or removal of the infant from harm’s way. One anubis baboon mother would leave her son “confidently” with her consort for periods up to 30 minutes, several times a day. Such care may mean important advantages t o the infant. In addition to protection from nearby chimpanzees and other predators, benefits may include access to food and increased influence over other animals, especially peers. This influence may mean an improved dominance status, even in the subsequent absence of the male protector. These advantages may or may not extend into adulthood. There is apparently great variation in the occurrence of care and in its quality. Because so little is known, behavior that may not, in fact, be comparable is lumped into this category. In the case of the macaques, the group about which most is known, the extent of male care vanes within the genus (Lahiri and Southwick, 1966; Brandt et al., 1970), and in the case of Japanese and Barbary macaques, between troops of the same species (Itani, 1959; Burton, 1972). Furthermore, the presence or absence of male care may vary according to the situation or the season. Even in the case of rhesus macaques, where male care is

CARE AND EXPLOITATION OF PRIMATE INFANTS

107

relatively rare, males may be attracted to newborns or to distressed infants. Koford (1963b) reports that 1-year-old rhesus males may be especially attracted to their newborn siblings (though not so much as their sisters are). SpencerBooth (1968a, p. 546) observed caged rhesus males who were 3 years older cuddling infants whose mother was absent. In some troops of Japanese macaques, males care for yearlings and neonates only during the birth season (Itani, 1959; Alexander, 1970). In a few species infants at some ages may be more frequently with males than any other animals except mothers. For example, during their fourth and fifth months, young mangabeys (Cercocebus albigena) spent nearly 70% of their time with an adult male, the remaining 30% with their mothers. According to the observer, these males displayed a generally helpful attitude toward infants, although in the sample of 2 infants the first contact with males did not occur until the tenth week (Chalmers, 1968, p. 268). In other cases, males as well as females are allowed to hold infants soon after birth, e.g., among caged Colobus guereza (Wooldridge, 1969). In the case of wild Macaca sylvana, adult females do not normally carry infants other than their own, whereas juvenile, subadult, and adult males (as well as subadult females) carry and care for infants as young as 1 week old (Deag and Crook, 1971; Burton, 1972). In one group of captive Barbary macaques studied by Lahiri and Southwick, dominant males played a particularly active role in infant care; during the first 12 weeks after birth, infants spent an average of 8%of their time being groomed and carried about by them (Lahiri and Southwick, 1966, p. 263). Similar involvement by head males in two troops of Barbary macaques was observed by Burton. In one case the leader held the neonate on four different occasions during its first day of life (Burton, 1972, p. 33). Individual variation will obviously play a role in the quality of male “sitting,” but this is a difficult topic on which to gather information. Van LawickGoodall’s work with chimpanzees and the Ransom and Ransom study of anubis baboons are of particular interest in this respect, since individual case histories illustrative of different types of male-infant relations are presented; some of these are discussed in Section II,F. These authors raise interesting questions about the effects that adult male-infant encounters could have on the subsequent emotional development of the infants. C.

ADOPTION

Of even greater importance for the infant than such temporary fostering are permanent adoptions of orphans by males. Male adoptions have been reported for the three most studied nonhuman primate groups: baboons, chimpanzees, and macaques. DeVore (1963) reported the adoption of a sick and orphaned baboon infant by a beta male. Itani (1959, p. 66) reports a semiadoption of a

108

SARAH BLAFFER HRDY

6-month-old Japanese macaque infant by a male of subleader status: although the foster parent never hugged or carried the infant, he groomed it and stayed near it. This relationship lasted for 6 months. In Pupio humudtyus,a young male on the make may depend on his ability to adopt weaned females and “mother” them to maturity. Motherless infants are invariably adopted by young adult males (Kummer, 1967, p. 70). In several cases of adoption, the immediate degree of relationship was known; invariably, these cases involved older infants and the foster parent was either the biological father or a brother. Sade (1967) reported that an adolescent male rhesus had his 6-year-old brother as his most frequent companion after he was orphaned at age 4. Van Lawick-Goodall (1968) reported a similar adoption by an older male sibling chimpanzee. Even where a 2-year-old orphan was adopted byaan older sister, the adolescent brother “moved around with him and protected him on occasion” (report of Edna Koning in van LawickGoodall, 1967, p. 30811). I know of only one instance of a male adopting a very young infant. This occurred under extremely abnormal conditions, in a caged group of rhesus macaques. With the exception of an adult male (the only one in the group) and a 4-month-old infant, each monkey was removed from the cage, operated on, and returned. The male adopted this infant (probably his offspring) subsequent to the mother’s operation (Barbara Smuts, personal communication). The benefits of adoption are obvious: a young primate without a caretaker would be unlikely t o survive. The above-mentioned point about weaned infants, however, brings u p the great risk involved when adoption means taking an infant away from a lactating female. Furuya (cited in Itani, 1959, p. 66) reports a Mucucu fusiculuris male that took an infant away from its mother by force and retained it until the infant starved to death. A Gzllicebus moloch infant from a caged group died when the male, which in this species normally carries the infant at most times, refused to return the infant to its mother, even for nursing (Lorenz, 1970, p. 79). D.

AGONISTIC BUFFERING

Male-infant interactions d o not necessarily benefit the infant. Exploitation of infants by males has been reported for species as different as anubis baboons, Nilgiri langurs, vervets, Barbary macaques, Japanese macaques, and Hamadryas baboons. By exploitation I mean behavior from which the male stands to gain but which may or may not benefit the infant, and may actually harm it. The most typical instances involve some variation of the behavior Deag and Crook label “agonistic buffering.” There is good evidence that the presence of an infant, especially a young infant still in its natal coat (Ransom and Ransom, 1971 , p. 190) acts as a signal to inhibit aggression in the adults of most species. Van LawickGoodall has reported for chimpanzees that “Only on one occasion

CARE AND EXPLOITATION OF PRIMATE INFANTS

109

was a male seen to attack, very mildly, a female with an infant on her back, whereas twenty-five attacks were recorded on females with infants in the (less visible) ventral position” (van Lawick-Goodall, 1967, p. 31 1). Whether or not an age difference was involved was not reported. Wooldridge (1969 p. 32) also notes that Colobus monkeys carrying infants were less likely to be the object of another monkey’s aggressive impulses. In several baboon and macaque species, this inhibition is used by males t o approach other males, usually dominant animals to which they would not normally have access. The following excerpt from a study of Macucu sylvunu is typical of this procedure: “It was not unusual to see a male running on three legs holding a baby under him with one hand for as much as 30 or 40 m, and taking it straight to another male to which it was then ‘presented’ ” ( h a g and Crook, 1971, p. 191). Commonly the baby would be pulled off by one of the other males and placed between them, o r else the presenting male might be mounted by the more dominant animal; during the mount, the baby might be mouthed or else simply pulled off by the mounter. Virtually the same pattern of behavior is exhibited by anubis baboon males. According t o Ransom and Ransom (1971, p. 187): Some of the males tended to establish close proximity to an infant under conditions of stress, proximity which in its most intense form consisted of carrying the infant on belly or back. . ..This kind of relationship appeared to be based on the adult male’s ability to increase his effectiveness in interactions with other males, insofar as close contact with an infant seemed to inhibit aggressive behavior from them.

Among Japanese macaques, the center of the troop, with its concentration of troop leaders and dominant females, offers a young male opportunities to enhance his status or to share in resources monopolized by those at the center; one common ploy utilized by males t o gain access is close association with infants. One male described by Itani (1959, p. 85) rarely entered the center alone; almost always he took an infant along as a “passport.” Subadult Hamadryas baboon males have likewise been reported to use infants to inhibit attacks against them from more dominant animals (Kummer, 1967). Poirier reports a related phenomenon for Presbytis johnii where the key stratagem in an alien male’s campaign to join a troop may be associating with infants and juveniles. On one occasion, 3 males approached a troop. During the first 2 weeks of merging, play accounted for 31%of all interactions between this trio and the troop; contact was almost entirely with 1 older infant from the troop. Although the dominant male of the 3 frequently played with this infant during the initial period, once acceptance by the troop was gained, he totally ignored the infant (Poirier, 1969, p. 32). Similarly, peripheral male juvenile vervets may facilitate the entrance of a strange adult male into a troop (Struhsaker, cited in Mitchell, 1969, p. 410).

110

SARAH BLAFFER HRDY

Other advantages to be gained from contact with an infant may involve the services of a third animal. Ransom and Ransom (1971, p. 189) describe an adult male that repeatedly stole an infant and used i t to force its mother to groom him. Besides these advantages for the male, a number of benefits may accrue to the infant from male attentions. The infant widens his experience of the social, and especially the male, world, and makes influential “contacts.” In those species where predators present a frequent danger (i.e., dogs in the case of Macaca sylvana; chimps in the case of anubis baboons), nearby males-whether exploiters or caretakers-could carry the infant to safety. Nevertheless, these positive aspects of male care have perhaps been overemphasized in the literature, leaving out the potential dangers for the infants involved. Attention to the details of these interactions from the infant’s point of view suggests some of the drawbacks. In making the point that the use of infants as agonistic buffers “may .. . keep antagonism between males in the group down to a minimum,” Deag and Crook (1971, p. 198) mention “a few observations showing that when actually involved in agonistic encounters males may grab babies and carry them.” There is n o information concerning occasions when the antagonist failed to notice the infant (as has been reported for chimpanzee females carrying the infant ventrally), but it surely cannot do the infant any good to be caught u p in these skirmishes. Several photographs, a series from Deag and Crook (1971, Figs. 5,a-j) and Fig. 5 from Ransom and Ransom (1971), illustrate t o what extent the infant’s keepers are pursuing their own ends. The Macaca sylvana series shows that when not “in use” an infant, which may have been toted some distance from its mother, is simply left sitting alone. Figure 5 in Ransom and Ransom shows an adult male baboon carrying a 3-week-old infant by one leg and upside down! In the case of a very young infant, a mother may prevent males from taking or even approaching it (van Lawick-Goodall, 1971, pp. 146-147). Hopf (1967, p. 258), describing the attractiveness of a Saimiri newborn for its cagemates, writes: “Females sniff, nuzzle and touch it; juveniles tug at its tail or limbs. These manipulations can be dangerous for the newborn. . . . Depending on her rank in the group the mother may prevent large males from touching the infant by threatening or avoiding them.” E.

INFANTICIDE

Carried to an extreme, male exploitation could conceivably lead to injury of the infant. In fact, instances of adult males killing infants have been reported for a number of primate species, including several prosimians (Mitchell and Brandt, 1972); free-ranging Macaca mulatta (Carpenter, 1942); caged Macaca fasicularis (Thompson, 1967); free-ranging Macaca sylvana (Burton, 1972); wild Papio ursinus (Saayman, 197 1); caged Papio hamadryas (Zuckerman, 1932); wild Pan

CARE AND EXPLOITATION OF PRIMATE INFANTS

111

troglodytes (Bygott, 1972); and Presbytis entellus (Sugiyama, 1967; Mohnot, 1971). In addition, it is suspected that adult males may have killed infants among wild Alouuttu (Collias and Southwick, 1952), among caged Suimin (Bowden e l ul., 1967), and among wildPresbytis senex (Rudran, 1973). Both chacma baboon and chimpanzee accounts involved cannibalism. Where the infant was not eaten, however, the suggestion that these incidents represent “male exploitation” of the infants must be accompanied by some demonstration of how infanticide would benefit the male. The circumstances surrounding infanticide are known in only a few instances and are discussed below. In each case where details are known, the male attacked an infant that was almost surely sired by some other male. One possibility is that infanticide here represents a strategy whereby a male increases his own reproductive success while proportionally decreasing that of his competitors (Trivers, 1972). The most detailed evidence in support of this hypothesis comes from studies of hanuman langurs, among which the killing of an infant quickly brings the mother back into estrus. Infanticide has been frequently reported among langurs (hesbytis entellus) under conditions that are both widespread and of long duration (Hughes, 1884). In recent years, infanticide has been reported at Dharwar, in Mysore state, South India (Sugiyama, 1967); at Jodhpur in northwestern Rajasthan (Mohnot, 1971); and at Abu, a hill station in southernmost Rajasthan (Hrdy, 1974). Circumstantial evidence also suggests that infanticide occurs among langurs at Polonnaruwa, Sri Lanka (S. Ripley, 1973 personal communication). The type case of langur infanticide was reported by Sugiyama (1965b) at Dharwar when a band of 7 males invaded a bisexual troop. The single resident male was wounded while defending his troop and eventually driven out. Subsequently, 1 male from among the invaders usurped troop leadership and drove out his former accomplices. Soon after the takeover, 5 infants in the troop were bitten to death by the new leader. Of ten takeovers by males from outside the troop, which have been reported at Dharwar, Jodhpur, and Abu, seven were accompanied by infanticide and resulted in the deaths of some 30 infants (Hrdy, 1974, Table VI). To date, assaults by langur males upon infants have only been reported when a male entered the troop from outside it. All females that were under observation after their infants had been killed exhibited estrous behavior within days after the death of the baby and copulated with the new male. In one troop at Abu, infant mortality over a 3-year period was as high as 80%: 9 of 11 infants present in this troop between July of 1971 and February of 1973 disappeared when males entered the troop from outside it. Local people witnessed the murder of 3 of these infants by an adult male langur; on fourteen occasions, adult males were seen by the observer to attack 3 other infants that subsequently disappeared. The complex events surrounding these attacks and the problems of interpreting them are discussed elsewhere (Hrdy, 1974). High langur population densities are found at both Dharwar (220-3491square

112

SARAH BLAFFER HRDY

mile) (Yoshiba, 1968) and Abu (more than 130/square mile). Large numbers of nomadic nontroop males circulate about the bisexual troops, and there is intense competition between males for access to troop females leading t o great social instability. Sugiyama (1967) estimates that new males take over troops on average once every 3-5 years. Given these circumstances, a usurping male might make the best of a short reign by eliminating unweaned infants and, hence, short-cutting a 2-3 year birth interval. Although positive assignments of paternity and, hence, precise measures of reproductive success are nonexistent, it does appear that males are enhancing their reproductive success by killing infants: in three troops for which information on subsequent births is available (Sugiyama, 1965b, 1966; Hrdy, 1974), 70% of the 15 females in these troops whose infants were killed gave birth within 8 months, or just over one langur gestation period later.' In the desert area of Jodhpur, however, as many as 27 months elapsed before one infantdeprived female gave birth. The average time between death of their infants and birth of the next live one for 4 Jodhpur females was 17 months (S. M. Mohnot, 1973 personal communication). In almost every instance in which infanticide may not have been advantageous to the male that killed the infants, his failure to benefit could be attributed either t o interference from another male or to noncooperation from females. Confronted with a population of males competing among themselves, often with adverse consequences for females and their offspring, one would expect natural selection to favor those females best able to defend their interests and the interests of their close relatives. At Abu, females formed temporary alliances against attacking males. On at least nine occasions when a male attacked an infant, 2 older females in the troop that did not at that time have infants of their own intervened. These 2 females would engage the male in fierce slapping encounters and would chase him away from the mother-infant pair. The probable relationshp between these females and the infant they defended is discussed in Section IV,D.One mother with an infant actually left the troop to travel on her own. A third factor that potentially detracts from the reproductive success of the invading male is that females may thwart his attempts to retain exclusive sexual access to them. At all three locations where infanticide occurred, a few females were also observed sexually soliciting males other than those that had killed their infants. Even though competing males and troop females may limit the advantages of infanticide for the invading male, on average infanticide appears to benefit males that practice it by rendering females reproductively available. A similar sexual selection interpretation has been offered by Thompson (1967) to explain the 'Langur gestation periods of 2 O O a O days have been reported at the National Center for Primate Biology at Davis (L. J. Neurater, 1971 personal communication) and are close to those of 6-7 months reported by Sugiyama in the wild.

CARE AND EXPLOITATION OF PRIMATE INFANTS

113

incident he observed among crab-eating macaques. This infanticide was the unexpected outcome of an experimental study of the effects of familiarity or the lack of it on opposite sex pairs. When paired with his usual companion and her infant, the adult male displayed typical behavior: he mounted the female briefly and then set about exploring his surroundings; he entirely ignored the infant. Paired with an unfamiliar mother-infant pair, the male responded quite differently. After a brief attempt at mounting, he attacked the infant as it lay clutched to its mother’s ventral surface. When the mother tried to escape, he pinned her to the ground and gnawed the infant, making three different punctures in its brain case with his canines. As Thompson points out, the intensity of the male’s attack, and his selectivity were remarkable; only the infant and only a strange infant was harmed. If, indeed, males profit from killing infants sired by their competitors, this interpretation might explain events that occurred during the first year at the Cay0 Santiago colony before the groups had stabilized: “more infants were killed usually by adult males but also sometimes by females . . . than died of all other causes” (Carpenter, 1942). Similarly, it may be significant that in the case of chimpanzee infanticide and cannibalism (Bygott, 1972) the infant eaten was the offspring of a strange female who had not been seen before in the area. Obviously though, acceptance of this male-male competition hypothesis must await more precise information on the efficiency of infanticide in increasing the reproductive success of those males that practice it. ’

F. CARE VS. EXPLOITATION AND DEGREE OF RELATIONSHIP

For several species the examples cited appear to fit predictions generated by kin-selection theory. Males most likely to be fathers are apparently those males that also protect and care for infants. Males caring for infants would be expected to discriminate in choices of charges: familiarity with the mother may play a crucial role in such discrimination. On the other hand, males peripheral to the breeding system may be those most likely t o exploit infants indiscriminately. Unfortunately, no conclusions can be drawn from this apparent fit, as the data relevant to the problem are too skimpy. Furthermore, multiple biases were implicit in their collection and a new bias has been added here, that is, examples were selected for their “pertinence” to the theory. The presentation of these examples here was based on three assumptions (see Section I) which, although reasonable, are far from proven. With these qualifications understood, the following is a synopsis of present knowledge of male care vs. exploitation of infants and how such behavior might be a function of genetic relationship. For those rare occasions when a male was observed defending a particular infant at some risk to himself, the expectation raised by Hamilton’s theory is that these two individuals would be closely related. In fact, for three of the

114

SARAH BLAFFER HRDY

rescue instances encountered (for Colobus, langurs, and lutongs; see Section II,A), the male involved was a dominant male of the harem and very likely the father. In the case of the Suimin rescue by a subadult male, there was no clue to probable relationship. Multiple copulations which may include younger animals are common in this species; the subadult could have been a father, a sibling, an unrelated individual, or anything in between. For those cases of male adoption in which information on relationship was available (for free-ranging rhesus, caged rhesus, and for chimps), the foster parent was either the probable father or else an older brother. In the case of macaques, young and relatively subordinate males are rarely involved in infant care; brothers, however, are reported sometimes to groom and protect younger siblings (Kaufmann, 1967; Sade, 1965,1967). Although Hamadryas males are known to adopt (or kidnap) unrelated or distantly related females, this fostering can best be considered as an installment toward a future harem. A future consort relationship may also be an issue in some cases of male care reported for anubis baboons (Ransom and Ransom, 1971). The Japanese macaque records provide suggestive data on this point: whereas there was little sex difference (28 males versus 34 females) among 62 yearlings cared for by males, there were 20 females in a group of 25 2-year olds cared for by males (Itani, 1959). Are these females more likely t o breed with their former caretakers when they mature? It is possible that the information to answer this question already exists in the records of the Japanese Monkey Center, although it has not yet appeared in English. Most interesting in terms of kin-selection theory is the possibility of differential treatment of closely related vs. more distantly related infants. Since so few data have been collected with this problem in mind, a statistical analysis of which infants are cared for and exploited most frequently by which males is not possible. To phrase this as a question deserving further research: Are there detectable trends in the age and status of the males involved in infant care, and what difference, if any, does previous association with the mother make? In a recent study, Ransom and Ransom (1971) were the first to collect relatively long-term data (over an 18-month period) relevant to this problem. Their findings suggest that in the case of at least one baboon species, a male’s status dictates the type of relationship that he has with fertile females and that this relationship affects his behavior toward her offspring. Among anubis baboons, fully mature males are more likely than younger animals to be engaged in consort relationships, and these males participate more frequently in “paternalistic” care including baby-sitting and active protection. The attentions of males that were involved in sustained consort or pairing relationships or that were potentially involved in such relationships were aimed at a specific infant or at the offspring of a specific female over either a sustained or a temporary timespan. Six such relationships were observed by the authors; five of these involved

CARE AND EXPLOITATION OF PRIMATE INFANTS

115

mature males, whereas the sixth involved a male that was “barely mature” and still somewhat peripheral (T. Ransom, personal communication). The type example of such “paternalistic” care might be the old male Harry (actually past his prime at the time o f the study) that was bonded with the multiparous female Myrna. This bond was expanded to include the infant Moley and its juvenile sister Loy; both associated.with him freely and were extended contact, comfort, and protection preferentially; that is, Harry did not extend these privileges to other youngsters (Ransom and Ransom, 1971. PP. 184-185). Although usually high-ranking anubis males d o not form pair bonds with females that have not had more than one offspring (Ransom and Ransom, 1971, p. 193), such a male may occasionally focus his attentions on a first infant in response to certain special conditions. For example, in the case of one primiparous and casual mother which was slow to react to her infant’s distress signals and which was not sufficiently heedful of her infant’s proximity to potential predators such as chmpanzees, a high-ranking male took over the role of protector; he stayed close to the infant and carried it for extended periods. His attentions were confined to that infant, and he was never seen to generalize such behavior to include another infant; when the infant died of unknown causes his relationship with the mother ended (Ransom and Ransom, 1971, p. 185). In contrast to males that appear to be “choosing” infants, a number of anubis males that had no previous consort relationships were less discriminating and more opportunistic in their relations with infants: other criteria such as availability and usefulness proved more important than familiarity with the mother. Young males (approximately 4-10 years old) often took an interest in the infants of young low-ranking females. Due either to inexperience or to lack of other bonds, these young mothers were more willing than higher-ranking females to allow males to take their infants (Ransom and Ransom, 1971, p. 186). One male using infants to enhance his social effectiveness was observed to switch from an older to younger babies (the most effective agonistic buffers) as they were born into the group (Ransom and Ransom, 1971, p. 190). If agonistic buffering is a maneuver allowing a subordinate animal to approach a dominant male, one would expect younger males to depend o n infant contact more than older males do for social effectiveness. No quantitative data are known to support this point, and a number of apparent exceptions are known. Deag and Crook (1971, p. 191) write that Macucu syhunu “of all ages” are involved in agonistic buffering. Furthermore, Crook reports that “the wild Barbary macaque does not seem, on present evidence, to limit his interest to a particular infant” (Crook, 1971, p. 244), suggesting that no discrimination is going on, regardless of age or the probability of being in a progenitorial position. In the case of anubis baboons, where mature male consorts seem to discriminate in favor of their probable offspring, high-ranking males are reported to use infants as buffers when confronted with the presence of a potentially powerful

116

SARAH BLAFFER HRDY

and relatively untested animal (a young male or a newcomer to the group). Of 8 adult and subadult males that sought contact with 9 infants for this purpose, the 3 engaging most frequently in agonistic buffering were high-ranking animals (Ransom and Ransom, 1971, pp. 187-188). However, this particular set of examples does not necessarily contradict the above prediction, for the importance of predominating over a new male (potentially a threat to all future reproductive success) may overrule any risk entailed in using an offspring. There is no information on discrimination in other species comparable to that for anubis baboons. The two reports available for Mucucu fiscutu suggest that, within the leader and subleader class, interactions with infants are not dependent on individual ranking. Itani (1959, p. 62) writes that there exists “no great difference between (the males’) behavior towards their infants and the behavior of a mother towards her infant.” Alexander (1970) classifies all contacts between males and infants as “affdiative” (defined as gross body contact, cofeeding, or grooming); no distinction is made between “care” and behavior that might not benefit the infant. Nevertheless, several features of Itani’s description suggest that (1) males likely to be fathers are behaving differently from those that are not, and (2) male care is in some instances inferior to maternal care and that agonistic buffering is going on. Care of infants during the birth season has been reported in four separate Japanese macaque troops: at Takasakiyama and Takasakiyama B (Itani, 1959), at Takahasi (Furuya, cited in Itani, 1959), and in the enclosed troop at the Oregon Regional Primate Center (Alexander, 1970). Among the free-ranging troops, only males of the leader (ca. 20 years or more) and subleader (1 5 years or more) class were commonly involved. In the Takasakiyama troop most intensively studied by Itani, thirty-five instances of paternal care were observed for 6 males of leader status; sixty instances for 10 males of subleader status; and four instances for 10 young adult peripheral males. There were virtually no occasions involving the 2-3 year old males that live on the periphery of the troop and exhibit little interest in babies. It is not known to what extent these interactions with infants reflect opportunities of access. Itani determined that interest in infants was most characteristic of males in the middle rank of each of the two top classes and of animals that exhibited an interest in the central part of the troop. In the enclosed Oregon troop, subordinate males were seen to interact with older infants, but only dominant males participated in “nursery groups” containing the very young; of thirty-two “play” and “affiliative” interactions between adult males and neonates, 88% involved dominant animals (Alexander, 1970, p. 281). It is important to note that the motivation for subordinate animals to use infants as “passports” may have been reduced in this troop; among the enclosed animals the central-peripheral troop structure with young males on the outside had largely disappeared (Alexander, 1970, p. 277).

CARE AND EXPLOITATION OF PRIMATE INFANTS

117

If infants are being used as passports, it is not surprising that the males that engage most frequently in infant care are those ranking in the middle of their class and those termed by Itani as most “interested” in the center of the troop; that is, these are the less well-established, middle-ranking animals that have an ambitious interest in being near dominant animals and that could profitably use infants to achieve this end. Itani (1959, p. 72) mentions that closely ranked subleader males sometimes vie with one another to care for the first infants born each season. Incidents in which males drop the infant that they are carrying or else pull them about by force were reported (Itani, 1959, p. 62). This apparent nonchalance and self-absorption in the young macaque males in caring for infants is reminiscent of the agonistic buffering reported for anubis baboons (see Section 11,D). A further comparison of macaques and baboons must await quantitative data on the breeding success of leader and subleader males, and information on the quality of care dispensed by members of each class. If the kin-selection interpretation offered for the anubis situation is valid and if it applies to the Japanese macaques as well, one would expect that a male would direct solicitude toward the infants of females that were familiar to him either by virtue of common sibship or because the females were former consorts and that the ambitious middle-ranking leaders that appeared to be using infants would not be in progenitorial roles. This possibility is supported by the work of Imanishi (1957a,b) and Nishida (1966) indicating that among Japanese macaques increased frequency of consortship with estrous females is correlated . with higher status, just as it probably is for baboons. However, progenitors or not, such middle-ranking males could be siblings or uncles. Yamada (1963, pp. 46-47) points out that the frequency of Macaca fuscatu infants cofeeding with their brothers and sisters was second only to their frequency of doing so with their mothers. The basis for a familiarity that could potentially influence choice of infants to care for is there, although, for a number of Japanese macaques, such sibling-nephew-niece preferences would be ruled out by the departure of young males from their natal troops (Smuts, 1972, p. 72; Koyama, 1970). In this section, degree of relationship and how it affects interindividual behavior has been discussed only in terms of a limited span of genealogical time. A totally unexplored level of inquiry involves the interplay between population genetics and behavior. Breeding integrity of the troop and stability of troop composition over a period of time will greatly influence the inbreeding coefficient and the degrees of relationship between individuals in the group. For example, in geographically isolated troops or troops with little immigration and social turnover, individuals will be more closely related than will those individuals living in population belts (that is a number of interbreeding populations over a large area) or where there is social change resulting in new leaders (i.e., the langurs of Dharwar and Mt. Abu). One would expect behavioral differences

118

SARAH BLAFFER HRDY

between those groups in which most animals share a number of genes and have done so for some time and those groups in which animals are closely related to only a few individuals. 111.

NURTURE VS. ABUSE-MALE AND FEMALE ROLES

The preceding section reviewed instances of male care and male exploitation of infants. From the published examples, it appears that infants are more frequently injured by males than they are by females. Without exception those incidents of maternal abuse in which the infant was killed (reported for Macaca mulatta, Harlow et al., 1966; for Sairniri sciureus, Bowden el al., 1967; Gorilla gorilla, Schaller, 1963) occurred among captive animals and could be attributed to conditions of stress and severe social deprivation. Excluding a special phenomenon termed “aunting to death” (discussed in Section IV,B), only 1 case of serious injury has been reported for females living under natural conditions. In this instance, an infant langur was mortally wounded by a female from another troop ( S . Ripley , personal communication). Minor mistreatments due to incompetence are discussed in Section IV,A. If valid, this observation regarding male infanticide would hardly be surprising. Whereas males of most species may greatly improve their reproductive success by aggressive behavior, females usually cannot. With a physiological ceiling on her fertility, a female’s best strategy will be adequate care of the infants she does produce; the fitness of any female insensitive to an infant’s needs would be drastically reduced. [Without any direct comparison intended, it is perhaps of interest that American males are far more frequently involved in damaging abuse of children than females. Although children were abused by their mothers or a mother substitute in 47.6% of a recent sample (N = 1380), 29.5% of these instances occurred in fatherless homes. Where males were present in the home, fathers or substitutes were involved in two-thirds of the incidents. One-third of the males involved were stepfathers rather than biological fathers (Cil, 1970,pp. 116-117).] These differences between male and female roles are not the same in all primate species. In tamarins (Saguinus species) for example, the female is reportedly more aggressive than the male (Hampton, 1964; Hampton el al., 1966). A theoretical paper by Trivers (1972) provides a neat explanation for such phalarope-like role reversals as are found in tamarins. In these monogamous New World monkeys, the male investment in offspring may be almost as great as that of the female with the result that females are no longer the resource limiting male reproductive success. According to Trivers’s model, males in such a species would not be involved in intraspecies competition for females and, hence, would not be subject to sexual selection for increased aggressiveness.

CARE AND EXPLOITATION OF PRIMATE INFANTS

119

A basic role distinction can only be part of the answer for the greater part played by males in infant abuse. Although quantitative cross-species information on this point is not available, one would expect that the likelihood of male abuse would vary from species to species and would be correlated with the optimum strategy for maximizing reproductive success in that species; this strategy, however, might vary with conditions such as population density. Furthermore, since females can and do discriminate .between their own and other infants, and in some cases between infants of close relatives, one would expect a high degree of selectivity in genetic relationships connected with nurturing activities. As yet, there is no conclusive evidence for this assumption. In most primate species for which there is information, individual mothers vary greatly in permissiveness and nurturing activities. For example, Jay observed differences among langur females as to whether they would allow infants other than their own to nurse once the alien infant had found the nipple. Of all those females observed holding an alien infant, however, less than one-quarter deliberately helped the infant find the nipples. Childless, nonlactating females were less discriminating. Similarly, Hinde (1965, p. 71) noticed that females with an infant of their own were more aggressive toward alien infants than were childless females. From their work with pigtail and other macaques, Jensen and Bobbitt (1968, p. 43) write that “most monkey mothers are quite punitive towards a strange infant.” Rosenblum’s (1968, p. 228) work with caged Mucacu nemestnnu confirms this impression; pigtail infants separated from their mothers were generally ignored or actually rejected by other group members. Rosenblum found bonnet macaques more solicitous toward separated infants; however, as with langurs there was great variation. One Macuca rudiuta “supermother” named Brunie nursed 2 infants in addition to her own (3 in all), 2 at a time. Almost certainly, however, Brunie’s generosity was influenced by her experimenter’s methods; the first alien infant was introduced to Brunie after her own had been removed, and the third infant was introduced in the absence of the first 2. The bonnet mothers observed in the field were more discriminating: “The female resents another’s baby trying to cling to her and drastically removes it” (Rahaman and Parthasarathy, 1962, p. 157). In general, nonhuman primate mothers nurse only their own infants, although individuals may vary in their tolerance toward other infants. Two exceptions t o this rule have been reported among the Colobinae. Wooldridge (1969) reports that an infant Colobus guereza born at the National Primate Research Center suckles regularly from another lactating female in addition to its mother. In his study of wild Nilgiri langurs, Poirier (1968) noted that “When a female had two infants at her chest, there was often a struggle as to which infant would nurse. Even if one of the infants was her own offspring, a mother did not help it obtain the nipple. It seems possible that any lactating female might nurse another’s

120

SARAH BLAFFER HRDY

offspring.” It would be of interest here to have further information on the genetic relationship, if any, between nurse and nurslings. Experimental studies with rhesus macaques reveal temperamental differences between the sexes which are apparent early in development (Harlow and Zimmerman, 1959; Jensen et al., 1968; Mitchell, 1968). Male infants, for example, were generally rougher in their play (Hansen, 1966), whereas preadolescent females directed significantly more positive social behavior and less hostility toward an infant than did young males (Chamove et al., 1967; Spencer-Booth, 1968a). Captivity studies with chacma baboons and field studies with other savannah baboons (Bolwig, 1959; Ransom and Rowell, 1972, p. 130; DeVore, 1965; also cited in Hamburg, 1969) revealed a greater interest in newborns by immature and postpubertal females than by young males. The sexes differed in similar respects among free-ranging vervets (Lancaster, 1971, p. 174). These studies, undertaken for only a limited number of primate species, do not mean that differences between the sexes will exist t o the same extent or even in the same direction for all primates. Not counting motherhood, rhesus macaque females still have more intimate contact with infants than males d o at all stages of their lives, and the same is true to a lesser degree for chacma baboons (Bolwig study). This is not, however, universally the case in macaque and baboon species, namely in Macaca fuscata, Macaca sylvana, Papio anubis, and Papio hamadryas (see Section 11,B). There is great variation both within (see Itani, 1959) and between species in the amount of time males spend with infants. In some species (e.g., marmosets) males possess maternal qualities commonly associated with females. Although females rarely nurse another female’s infant, other forms of nurturing-cuddling, grooming, protection, and reassuring contact-are common. Such aunting behavior is discussed in the next section.

N. THE PROS AND CONS OF AUNTING The relationship between infants and other group members has been a topic of particular interest in some, and of at least peripheral interest in most, primate field studies. Universally, primate neonates are objects of attention, and females may be especially attracted. Within and between species, however, individual mothers vary as to the freedom that they will allow such females with their infants. From current information, four species are remarkable for their permissiveness: among Presbytis entellus (Jay, 1962), Colohs guereza (Wooldridge, 1969), Presbytis obscurus (Badham, 1967), and Pygathrix nemeus (Hill, 1972), infants may be held by other group members and carried to some distance from the mother within hours after birth. For other species in which aunting is common,

CARE AND EXPLOITATION OF PRIMATE INFANTS

121

first transfer is substantially later, that is, around 8 days for Cercopithecus uethiops (Lancaster, 1971), although exceptional transfers have been recorded at 14 hours (Struhsaker, 1967a, p. 37) and at 2 days (Gartlan, 1969); around 10 days for Presbytis johnii (although the first week of life was not observedPoirier, 1968, p. 52) and Cercopithecus cumbelli lowei (Bourliire etal., 1970); and as late as 2-3 weeks in Suimiri sciureus (Dumond, 1968, p. 125) and in caged Mucaca mulatta (Rowell et al., 1964). In some species, mother-infant contact is more intense and relatively uninterrupted throughout the early weeks, lasting until mother and infant of their own accord begin to spend time apart. Macucu radiutu.’ (Simonds, 1965, p. 192; Rahaman and Parthasarathy, 1962, p. 157), Mucaca nemestrinu (Rosenblum, 1968, p. 227), and Macuca fuscuta (Sugiyama, 1965a) appear to fit this description, although with great individual and contextual variation. Itani describes such variation among Japanese macaques: “There are fond mothers who hate to let their infants go for a long time after birth, while there are also such cold mothers as Elk . . . who left her infant two days old on the ground and busied herself in feeding” (1959, p. 68). There are conflicting reports as to whether caged monkeys are more or less possessive (obviously conditions will vary). I t may be that free-ranging macaques (Southwick etal., 1965; Jay, 1965, p. 577; Itoigawa, cited in Wolfheim etul., 1970) are less permissive than the caged animals studied by Rowell etul. (1964). Similarly, in one group of captive Eiythrocebus putm, a mother allowed her 14-day-old infant t o be taken from her by another female even though such permissiveness has never been observed under natural conditions (Hall and Mayer, 1967, p. 232). With the exception of the “greeting” behavior allowed by baboon mothers, the savannah baboons (DeVore, 1963) and Erythrocebus putus (Hall, 1968, pp. 105-107) epitomize possessive mothers. As described by Hall (1963), greeters are animals permitted to pick up the infant, usually by the hind legs, touch its rump with their mouth, embrace it, and so forth; these greeters are most often adult females, but may also be males and younger animals. Aunt-infant relationships were first observed in caged rhesus at Madingley, Cambridge (Rowell et al., 1964; Hinde, 1965; Spencer-Booth, 1968a). More recently, mother-infant relationships have been studied in caged squirrel monkeys (Rosenblum, 1968, 1971). To date, the most detailed field report and functional analysis of aunting is based on a study of vervets (Lancaster, 1971). Except for a general review of the relationships between infants and conspecifics other than mother or peers for all mammals (Spencer-Booth, 1970), the primate literature on aunting behavior is scattered and as yet unreviewed. Only special 2Rosenblum (1968, p. 221) has characterized captive bonnet macaque mothers as “permissive,” but he means permissive relative to pigtail macaques, the animals with which they were being compared.

122

SARAH BLAFFER HRDY

aspects of aunt-infant interactions, related to the potential advantages and disadvantages of aunting behavior for the parties involved, is discussed here.

A.

LEARNING TO MOTHER

In her article on “play-mothering” among juvenile vervets, Lancaster emphasizes the relatively small number of offspring that monkey and ape mothers have during their lifetimes. “Most do not mate until their third year of life or even later and the long gestation combined with the annual breeding patterns and single births, make the loss of an infant through neglect or inexperience very costly” (Lancaster, 197 1, p. 162). The position taken by Lancaster, Gartlan, and others (Jay, 1962; Struhsaker, 1967a; van LawickGoodall, 1967, p. 293) is that “maternal behaviour is a highly skilled performance, and there is ample evidence, that although the basic patterns may be innate, the behaviour is subject to the normal rules of learning. It is clearly more efficient for an adult female to be capable of dealing with an infant by the time her infant is born than to lose it through clumsiness” (Gartlan, 1969, p. 148). In the opinion of Gartlan and Lancaster, aunting behavior is practice for motherhood. The learning to mother argument rests on three points: ( I ) the existence of a disparity in maternal competence between primiparous and multiparous mothers which may be lessened by aunting experiences prior to motherhood; (2) predominance of nulliparous females participating in aunting behavior; and (3) some demonstration that maternal competence is correlated with reproductive success. 1. Primiparous vs. Multiparous Mothers

The literature on primiparous chimpanzee and monkey mothers has been reviewed by Lehrman (1961) who concluded that the primiparous mothers tended t o provide their offspring less adequate care than multiparous ones. In a reconsideration of the same observations, however, Seay (1966, p. 163ff.) finds them “inconclusive.” Seay’s (1966, p. 162) results from an experimental comparison of primiparous and multiparous wild-raised rhesus mothers demonstrated striking similarities in maternal categories such as cradling, restraining, retrieving, embracing, and nipple contact. The only significant difference involved maternal confidence as reflected by the higher anxiety of the primipara, and the higher percentage of physical rejections as well as the increased firmness with which rejection was accomplished among multiparous mothers. Seay (1966, p. 163) concluded that “primiparous rhesus mothers normally give adequate care t o their infants.” Field observations of vervets (Gartlan, 1969) and rhesus macaques (Kaufmann, 1966) lead to a similar conclusion. Within species, individual variation and life history appear to be far more important than panty. Whereas some primi-

CARE AND EXPLOITATION OF PRIMATE INFANTS

123

parous mothers are extremely nonchalant about such things as separation from their infants (e.g., Gartlan, 1969, p. 147; Itani, 1959, p. 68; van LawickGoodall, 1971) or are otherwise incompetent (e.g., a study of caged rhesus reported in Rowell, 1963a, pp. 48-49), other primiparous mothers in these same studies were perfectly adequate. In the wild, the case for incompetent female care rests almost entirely on observations of juvenile or subadult nulliparae. By the time of motherhood, most females are practiced. Of the seven occasions when Jay (1962, 1963) observed langur females carrying infants so awkwardly that they dropped them, all were very young females and 4 were known to be subadults or nulliparous. Similarly, vervet females seen carrying infants upside down or otherwise awkwardly were subadults (Gartlan, 1969). The important point for the learning to mother argument is that those animals (including Seay’s subjects) for which parity was relatively unimportant had all been raised in the wild. The strongest case for multiparity making a difference derives from caged and socially deprived animals. Harlow et al. (1966) found that mothers that were themselves “motherless” made abusive and even murderous mothers themselves. These same mothers that were abusive with their first infant, might care for their second and third offspring: of 6 rhesus mothers that were indifferent or abusive toward their first offspring, 5 had second infants that received “adequate” treatment. This familiarization process may also apply to apes. A caged female gorilla that had killed her first infant, cared for a second 2 years later (Schaller, 1963, p. 287). Inexperienced captive chimpanzee mothers likewise are often afraid of their firstborns, refusing to touch them or t o allow them to cling (van Lawick-Goodall, 1967, p. 292). These reports do not distinguish primiparous from multiparous mothers, but rather mothers that have had prior experience with infants, whether with their own or with another female’s, from those that have not. 2. Primiparous Participation in Aunting

As both Hamburg (1969) and Lancaster have pointed out, in virtually all species, females raised in the wild will have had some contact with infants prior to motherhood. Differences exist, however, in the extent of this contact and the age of the infant at first access. In some species, other females are not allowed access to very young infants (see beginning of Section N);in others only older females are allowed to hold infants. Poirier (1968, pp. 54-55) reports for Nilgiri langurs that juveniles and subadults never tried t o take an infant from its mother and that transfer of infants occurred only among adult females. Among Lemur catta only other mothers are allowed access to very young infants (Jolly, 1966, p. 115). Given that the mother often determines who holds her infant (see Section IV,G), one would expect that in those cases in which the mother allows only other mothers (i.e., experienced females) to approach, the benefits of having a

124

SARAH BLAFFER HRDY

young aunt d o not outweigh the potential disadvantages, such as harm to the neonate due to inexperience. Such discrimination might occur in species in which infants are relatively undeveloped at birth o r in which the mothering of a newborn infant entails delicate skills, e.g., Colobus vetus (see Section lV,B). Because Poirier also mentions the “strong desire” of nulliparous females t o participate in transfer sequences, the preference for older nulliparous Nilgiri langurs may be a compromise between (u)young females that are dangerously inexperienced and (b) no aunts at all; this kind of trade-off is discussed in later sections. In species in which only older females are involved in infant transfer, learning to mother does not appear to be a sufficient explanation for the existence of aunting behavior. However, for most species in which aunting frequently occurs, and for some species in which it rarely occurs, juvenile and subadult females play the prominent role. In squirrel monkeys, aunts are often nulliparous females that are either pregnant for the first time or that were too young to become pregnant during the previous mating season (Dumond, 1968, p. 123). Similarly, Itani (1959, p. 69) reports that among Japanese macaques, nulliparous females are strongly interested in infants and make them their “playthings.” Quantitative information regarding which females exhibit the greatest interest in infants is available for three species: caged rhesus macaques (Spencer-Booth, 1968a), vervets (Lancaster, 1971), and Hanuman langurs (unpublished data from February and March of 1973). In each case, a disproportionate number of nulliparous females participated in aunting behavior. Of 347 “affectionate contacts” between vervet infants and females other than their mothers, 295 involved females between 1 and 3 years old that had never had an infant. Nulliparous females composed 38% of the females, yet were responsible for 85% of the aunting. Furthermore, contacts between infants and juveniles tended t o be more sustained. Gartlan (1969, p. 149) reported that even vervet females that were too small to carry the infant for long distances would attempt to carry one and would play with infants. Similarly, in a study of langurs at Abu, Rajasthan, nulliparous females constituted 15% of the available caretakers (including juvenile males that also occasionally held infants), yet were responsible for 140 of 196 observed episodes in which a troop member other than the infant’s mother held or carried an infant. Among species such as bonnet macaques (Rahaman and Parthasarathy, 1962, p. 157) in which first transfer occurs relatively late, juvenile and subadult females are among the first aunts. Spencer-Booth (1968a, pp. 556-557) reported that female rhesus macaques around 2 years old are the most likely to participate in aunting b e h a ~ i o r .She ~ also noticed that whereas nulliparous females were more hesitant in approaching, they exhibited a greater proportion of 31n their early report, Rowell et ul. (1964)defined an aunt as a female around 2 years old.

CARE AND EXPLOITATION OF PRIMATE INFANTS

125

touching and cuddling behavior than did multiparous females (Spencer-Booth, 1968a, p. 546). DeVore also has emphasized that older juvenile and subadult baboon females appear highly motivated toward a new infant, much more so than young males (DeVore, 1965; also cited in Hamburg, 1969, p. 10).

3. Reproductive Success and Previous Experience The observations of Seay and others suggest that primiparous mothers are more anxious, more affectionate, and more hesitant at weaning than multiparous females. From an evolutionary point of view, however, maternal “inadequacy” is measured only in terms of surviving, reproducing offspring, and evidence from field studies on child-rearing casualties is too slim to derive measures of primiparous versus multiparous efficiency. Although there is virtually no primate species in which wild females have not been exposed to infants prior to childbirth, great variation exists in the age at first access to these “practice” infants. Whether or not differences in the suivivorship of first infants exist between species such as langurs and vervets, in which aunts have early access, and species such as patas and baboons, in which first access is relatively late, remains to be determined. The possibility of disparate developmental rates must also be considered; that is, neonates of some species may be more or less vulnerable to maternal inexperience. No conclusions are possible without data on the reproductive success of large samples of mothers. Very recently, Drickamer (1974) published “A ten-year summary of reproductive data for free-ranging Macucu mulattd’ showing that in this La Parquera population between 40-50% of infants born first or second to a female did not survive. Drickamer also found that infants born to high-ranking females had a higher rate of survival and that daughters of such females themselves gave birth at an earlier age. (This well-demonstrated correlation between female rank and reproductive success is highly relevant to comments in Section lV,F). Thus, the available evidence does support rather than contradict the importance of learning to be a competent mother. Lancaster’s (1971) hypothesis that juvenile aunting or “play-mothering” is practice for motherhood almost surely is correct as it applies to vervets and langurs, and perhaps as it applies to all species where aunting is common. B.

INCOMPETENCE, KIDNAPPING, AND “AUNTING TO DEATH”

Assuming that aunting is practice for the aunt and assuming that it increases the aunt’s reproductive success by making her a better mother, the question remains: Does aunting benefit the mother-infant pair? If the mother controls access to her infant, one would not expect her to allow another female to take her infant, running the risk of losing it unless ( I ) the aunt was a close relative of hers, or (2) certain benefits for the mother-infant pair accrue t o aunting which

126

SARAH BLAFFER HRDY

offset the potential disadvantages. Since instances are known where aunting harmed, or could have harmed the infant, and since in some species mothers do permit unrelated females to take their newborn or slightly older infants, such benefits must exist. These potential advantages will be discussed in Sections IV,C and D; here, only the potentially disadvantageous consequences of aunting for the mother-infant pair will be considered. A number of differences observed between young females and multiparous animals entail techniques of holding and carrying the infant and related motor skills. Incompetence could result in dropping an infant, holding i t in an awkward position (i.e., upside down, where it cannot reach the nipple, where the infant cannot orient itself, etc.), holding the infant too tightly, and so forth. The female holding an infant in such an awkward position could belong to the small percentage of “clumsy” mothers. More frequently, however, she is a young aunt at practice. Except for distress vocalizations, no signs of damage t o the infant from such treatment have been reported, but it seems inevitable that occasionally injuries do occur. Another potential source of damage is altercations between aunts or between the aunt and the mother as to which should hold the infant. Gartlan, for example, mentions how vervet infants are squeezed and pulled about in such disputes (Gartlan, 1969, pp. 148-149). Other potential drawbacks to aunting involve the naitvet6 of aunts concerning environmental hazards (e.g., Dumond, 1968, pp. 125-126). In most species where aunting has been reported, so has maternal supervision of the aunts (e.g., Dumond, 1968, pp. 125-126; BourliCre etal., 1970, p. 316; Lancaster, 1971). At the first symptoms of distress, usually the vocalizing of her infant, the mother retrieves it. Lancaster has suggested that such watchfulness on the part of the mother enhances the process of learning t o mother by conditioning the aunt t o keep the infant contented and quiet: “Instances of carelessness, clumsiness, or real abuse will, in effect, be punished. . . . Normally, if anything should make an infant cry out, its mother will come and retrieve it. If the infant is being abused, she may even bite the juvenile female (Lancaster, 1 9 7 1 , ~ 175-176). ~. The vulnerability of an infant monkey decreases rapidly with age as its grip strengthens and it grows more robust. This necessary period of development provides a reasonable explanation for the postponement of aunting behavior in most species, although it certainly does not explain all of the time differences (see beginning of Section IV). The possibility that rates of development could be speeded up through selection in proportion to the advantageousness of aunting must be kept in mind. Other species-specific traits could also be involved. For example, it has been reported that newborn olive Colobus monkeys are carried in their mothers’ mouth, perhaps as an adaptation to the extremely thick forest through which these arboreal monkeys move or perhaps because the adult pelage is too short for a four-fingered infant to cling to (Booth, 1957, p. 427; Wool-

CARE AND EXPLOITATION OF PRIMATE INFANTS

127

dridge, 1971, p. 483). Whatever the reason, the risks of early aunting in such a species might be considerably greater than for monkeys that cling from birth. The great attractiveness of the newborn infant for other members of the group has been mentioned; this appeal may be at the root of both inter- and intraspecific kidnappings. Chimpanzees are notorious for stealing baby monkeys (Kortlandt, 1967), although this behavior might also be classified as predation (Ransom and Ransom, 1971; van Lawick-Goodall, 1971; Teleki, 1973). Other interspecific examples include a female spider monkey that carried a howler infant until it died of starvation and interspecific adoptions among caged animals. Intraspecific kidnappings also occur between troops. At Dharwar (Sugiyama, 1966; Yoshiba, 1968) Jodhpur (S. M. Mohnot, personal communication), and Abu (personal observations) langur females occasionally steal infants from a neighboring troop. Other females from the kidnapping troop then prevent the mother from retrieving her offspring. A number of intraspecific kidnappings appear to be direct outgrowths from aunting behavior. Gartlan (1969, p. 149) describes an extremely tenacious vervet aunt that took a 3-weeks-old infant, retaining it for over an hour. Whenever the mother approached, she ran away. Temporary stealing has also been reported for captive Colobus guerezu (Wooldridge, 1969, p. 81) and macaques (Schultz, 1969, p. 331; Hinde and Spencer-Booth, 1967a, p. 268). In relating how “the importance of the aunts was first brought to our attention,” Hinde and Spencer-Booth (1 967a, pp. 344-345) suggest a surprising side effect of aunting-kidnap behavior. They describe an adolescent female whose attempts t o take an infant were so persistent that she made the mother ill: “During the period of illness such particularly acute deteriorations in the mother‘s condition were noted nine times and in at least seven of them, the baby was known to have been stolen within the previous twenty-four hours. It seems clear, therefore, that these were effects of the aunt’s behavior.” (Hinde and Spencer-Booth, 1967a, p. 345). More serious results of kidnapping have occurred when’nonlactating aunts took an unweaned infant and did not return it and, subsequently, it starved to death. Such occasions of “aunting to death” have been reported for wild Cercopithecus cambelli lowei (Bourlitre et al., 1970, p. 317) and caged Saimin sciureus (Rosenblum, 1971, p. 105). The kidnapping of the Lowe’s guenon is of particular interest. Soon after parturition, the mother became ill, and her infant was taken by another female. On the second day, other aunts (including a 23-monthold female sibling of the infant) that had been following the real mother transferred their attentions to the “new” mother. The sick female was ignored except when she attempted to approach; on these occasions she was threatened away by the two oldest females involved. The infant died at 4 days old, apparently of starvation, and the mother subsequently recovered. This incident illustrates two extremes, the worst possible and most positive

128

SARAH BLAFFER HRDY

consequences of aunting. As it turned out, aunting in this case was lethally inopportune. However, had one of the females been lactating, or had the mother recovered sooner and retrieved her infant, aunting could have meant survival for both mother and infant. In Japanese macaques, and other species, kidnapping may occur when a mother has lost her own infant and she attempts t o steal another, often from a female of lower rank (Itani, 1959, p. 64). Not all infantstealing females, however, are childless themselves. One curious outcome of a neonate’s attractiveness is that monkeys have been known to neglect their own, slightly older, infants in their eagerness t o hold a newborn belonging to another female; so far as I know, this phenomenon has been reported only among caged animals, for example, Mucucu mdiafa (Bullerman, 1950) and Cerocebus ulbigenu (personal observation). C.

ADOPTION

If a kidnapper is lactating, the consequences for the mother-infant pair are less severe, and in terms of reproductive success may even be advantageous. The mother is free to resume cycling while the foster-mother bears the cost of raising her offspring. If, as in Mucucu fuscuta and Mucuca nemestrinu examples, the foster-mother ranks higher than the real mother, the infant as well stands to gain in fitness, to the extent that its foster-mother’s rank entitles it to differential access to food and protection, and to higher status in dealings with other group members (see especially Bernstein, 1969b, p. 456). Such kidnappings by lactating females may occur when a mother has lost her own infant and aggressively sets about obtaining a substitute (Itani, 1959, p. 64; Rowell, 1963a, p. 43). Although clearly starvation does sometimes occur, several factors operate in favor of the infant. In those species in which infants are born in the space of a limited birth season (e.g., Japanese macaques, some baboons, and some langurs), the likelihood that an orphan will be adopted by another lactating female is increased. Also,the odds are in the orphan’s favor in that a mother loses her infant more commonly than an infant loses its mother. Even in cases where the foster-mother was not lactating, adoption has been known to induce lactation. Production of apparently normal milk by nonpregnant, initially nonlactating females has been observed in caged rhesus macaque foster-mothers (Harlow e t d . , 1963; Hansen, 1966; also cited in Spencer-Booth, 1970, p. 45). A further possibility is that a female that was already lactating might be hormonally “geared” for motherhood and, hence, more motivated to adopt an orphan. The Mucaca rudiufu “supermother” Brunie (see Section 111) may be such an exmaple. Orphans are uncommon in the wild, but when observed they have almost invariably been adopted by another female in the group. Usually, the foundations for adoption have been laid before the actual transfer became necessary,

CARE AND EXPLOITATION OF PRIMATE INFANTS

129

through aunting behavior. This priming has been best documented for caged rhesus (Rowell et al., 1964) and for Cercopithecus sabaeus (Marsden and Vessey, 1968). This last, caged green monkey example, is abnormal in that ( I ) the infant adopted was a hybrid (the son of a Cercopithecus sabaeus female and a Cercopithecus aethiops male) and ( 2 ) the true mother continued to live in the same cage after the adoption took place without making any effort to regain it. When the hybrid infant was 2 weeks old, the second female lost her own infant; transfer occurred soon after. Prior to this adoption, however, during the second week of the hybrid’s life, it spent 65% of its time (down from 100% during the first week) on the nipple of its own mother, 18% on the nipple of its future fostermother, and 20% on nipples at large. Prior to adoption the infant was actually groomed more frequently by his future foster-mother than by his own mother (Marsden and Vessey, 1968, Table 2.). It is true for several species that mothers may not groom their infants as frequently as less closely related females. For example, Presbytis johnii mothers are seldom observed to groom their own infants, and even under normal conditions do so only about 7% of the total time that the infant is groomed (Poirier, 1968, p. 5 9 , about the same as this green monkey mother. Among Lemur catta as well, aunts groom the infant more than its own mother does (Jolly, 1966, pp. 115-1 16). When the adoptive animal is a close relative, such previous familiarity can be assumed, especially in species such as macaques and chimpanzees where matrilineal relatives have preferential access to the infant. Van Lawick-Goodall (1967, 308n, 1968) has reported three instances of adoptions among chimpanzees, in two cases by older juvenile sisters, and in a third by an older brother; Sade (1965) reports similar adoptions by older sisters for rhesus macaques. Adoption (between generations) within matrilines may also occur. When a female Japanese macaque of the Takasakiyama troop gave birth to a pair of twins, one of these was cared for by the mother’s presumed mother. However,the grandmother had not bred that year and was unable to nurse her twin; i t died about a week later, apparently of starvation (Itani, 1959). Twins should be somewhat more common than orphans. Schultz suggests that twinning probably occurs at roughly the same rate in most primates, including man, i.e., at around 1 pairl100 births. Nonhuman primate twinning rates have rarely been calculated using large samples but from numbers of twins known for chimpanzees and langurs, there is no reason to question Schultz’s approximation. [His estimate does seem high, however, for rhesus macaques; Koford et al. (1966; also cited in Spencer-Booth, 1968b) found 4 pairs of twins in 1748 births.] One would expect aunting behavior, whether from juvenile females, siblings, or older relatives to be particularly advantageous in the case of twins. According to Schultz (1969, p. 184), marmosets and tamarins are the only members of the entire suborder of Anthropoidea that regularly produce more than one offspring at a time (single births are the exception). It is of some interest

130

SARAH BLAFFER HRDY

that it is in this species group that paternal care is most pronounced. The possibility that twinning and paternal care are related raises a question about maternal capacity to care for twins unassisted. Three cases of twins being reared successfully in the wild are known for Hanuman langurs: one pair to 2 years, one pair to 1 year, and the third to 4 months when the twins died from external causes (personal observations of Mohnot and Blaffer). Such mother-twin trios have never been studied intensively in the wild, and it is not known whether survivorship was influenced by aunting. The successful nursing of 2 infants younger than 2‘ months by 1 female (1 her own and 1 a presumed orphan) has been reported for bonnet macaques (Jay, 1965, p. 577), a species in which aunting may be much less common than in langun. Both infants appeared in “excellent physical condition.” Cage studies of Macaca radiata (Rosenblum, 1968) and Macaca mulatta (Spencer-Booth, 1968b) corroborate the finding that some females, under some conditions, are capable of rearing multiple young. In terms of aunting, it is important to note that nursing might be the most important limiting factor on twin survivorship (probably dependent on the individual mother and on environmental circumstances) and that aunts do not normally nurse their charges. In other words, the main advantage of aunts for the mother-twins would be in case of danger when the aunt could carry I twin. Under normal conditions, monkeys can and do carry 2-3 offspring at a time, for instance, the bonnet macaque and Hanuman langur examples and also Nilgiri langurs (Pokier, 1966, cited in Bernstein, 1967, p. 12). Poirier (1968, p. 49) has reported, however, that overall movement in Nilgiri langurs decreases as soon as any female in the group gives birth and that the group may be slowed down for as long as there is an infant under 3 months. Terrestrial primates that need to cover long distances during the day might find an extra infant an even greater burden than it is for more sedentary arboreal monkeys. D.

OTHER BENEFITS FOR THE MOTHER-INFANT PAIR

Aunting to death and successful adoption of orphans represent extreme and relatively rare outcomes of other than maternal care of infants. The effects of day to day aunt-infant interactions, which might include grooming, play, infanttending, or minor rescues, are cumulative and inconspicuous. Some of these benefits from aunting include ( I ) foraging freedom for the mother, (2) socialization of the infant, and (3)potential help for the infant in case of contingencies. The benefits of such routine aunting may be quite subtle, as in the case of foraging freedom. One of the common patterns of aunting among Nilgiri langurs (Poirier, 1968), vervets (Lancaster, 1971), caged patas (Hall and Mayer, 1967, and personal observation), as well as among caged rhesus (Rowell, 1963a), is for a mother to deposit her infant near another female and proceed to feed some distance away. Among Nilgiri langun, for example, approximately one-half of observed infant transfers were followed by the mother going off to feed, al-

CARE AND EXPLOITATION OF PRIMATE INFANTS

131

though the frequency of this sequence varied somewhat with the age of the infant (see Poirier, 1968, Chart V). In such cases the mother gains unencumbered moments for foraging in the shrub level where she is relatively more vulnerable t o predators (i.e., dogs). She can afford these excursions because her infant is safe above her. On these occasions when the mother seeks freedom to forage, the baby sitter does not necessarily take the initiative. This pattern has been described as follows for Nilgiri langurs: A female need not have indicated a desire to mind the infants left in her care; rather she was often the last individual remaining in a rest or sleeping area. The “baby-sitter” role alternated frequently as the original “sitter” left and another female took its place with as many as three females assuming the role in a short period of time [Poirier, 1968, p. 55 1

Such sequential baby-sitting has also been noticed among patas monkeys living in partitioned but connected cages (see Section IV,E). It often appeared that the mother’s “decision” to move into the next cage to feed was correlated with another female’s proximity to her infant. Once the mother had moved away (although never out of sight since she could see through the partitions) the first sitter might herself leave if there was a second sitter nearby. Only infrequently was the infant left alone in a cage. One benefit of baby-sitting for the mother-infant pair is that, without much risk to her infant, the mother is better fed, and hence more “fit” to be a mother. Why the sitter should cooperate is more complex. If she is a subadult or nulliparous female, the experience may of course be mutually beneficial. If, however, the aunt is not related, nor learning to mother, and if her status does not improve from holding an infant, she has little to gain, and I believe that this is reflected in the apparent nonchalance reported for some sitters. For example, Poirier writes that “The ‘baby-sitter’ did not protect a youngster(s) left in her care and the youngster was frequently unattended when she left” (Poirier, 1968, p. 55). Even such a “neglected” Nilgiri may be better off than an infant in a similar situation which is not left at all. For example, in wild bonnet macaques, the mother may temporarily abandon her infant, leaving it alone in the trees or bushes while she goes into the fields to feed (Simonds, 1965, p. 191). Other benefits of aunting behavior affect the mother only indirectly by enhancing the fitness of her offspring. The infant, however, may be directly affected insofar as aunting contributes t o its development of skills, socialization, and survival. The “general helpfulness” of aunts has been widely documented. This solicitude is perhaps best described in the following excerpt from Rowell et al.: As the infants grew, aunts sometimes watched them when they tried new physical feats and hovered anxiously nearby, going to the rescue if necessary. They seemed to be aware of dangers to young infants-for instance showing care when using the heavy swing door connecting the two parts of the pen if babies were near, and occasionally holding it open for an infant to scramble through.

132

SARAH BLAFFER HRDY When a baby approached the observer an aunt would sometimes threaten, with the result that the baby went away, and on a few occasions an aunt punished another female who had been aggressive to a baby. Occasionally a baby rejected by its mother would go to an aunt and be cuddled. [Rowell eta/., 1964, pp. 22 1-2221

Bourlihe etal. (1970) report that a Lowe’s guenon aunt may carry an infant in difficulty, for example, when the infant is climbing on wet tree trunks after rain. A langur aunt has been observed t o push a timorous infant off of a limb into the waiting arms of its mother in the next tree (personal observation). Less subtle and also less common than this general solicitude are the benefits that infants derive when the aunt protects them in sudden danger when the mother is out of reach, or in case of orphanage. A curious practice reported for black and white Colobus monkeys underscores the rescue potential provided by caretakers. A mortally wounded mother pushed her infant away from her before she fell (Booth, 1962, p.484). If another animal then takes the infant, this practice would be adaptive. As in the case of common langurs, Colobus guereza infants are passed around soon after birth. Moreover, the snow-white newborn is a striking object eliciting group-wide attention. These two characteristics, infantsharing and dimorphic natal coats, may be instrumental in the success of the Colobus guereza mother’s strategy. It is interesting that her behavior is exactly opposite to that of the related Colobus vems mother, which when wounded does not release her infant (which is carried in her mouth) and, if anything, grips it more tightly. Although it is not known whether aunting is as frequent among olive Colobus as among black and white Colobus, care of the very young by other-than-maternal females seems unlikely (see Section IV,B). A number of cases illustrate that prior contact with an aunt increases the likelihood that an infant will be rescued by that female. Dumond, for instance, reports: A (Sairniri) mother and an aunt that was carring the infant were travelling as a pair.. As the pair approached a grey squirrel in their route, the mother violently shook the branch causing the squirrel to move away. A few moments later the baby was off the aunt’s back alone, and both the mother and aunt had gone about fifteen feet ahead. As the grey squirrel was returning to where the infant had been left, the aunt ran to the baby and presented her shoulder to it, making a purr call as the infant climbed on. [Dumond, 1968, pp. 126-1271

..

Such examples, however, do not answer the question of why the mother-aunt bond formed in the first place. Even when no special relationship between the aunt and the mother-infant pair was previously apparent, the aunt may defend the infant. The langur case where 2 childless females penistently and audaciously interposed themselves between an infant and the adult male attacking it was mentioned in the discussion of infanticide. Several features of langur life may contribute to the occurrence of this protective behavior. Whereas males frequently leave their natal

CARE AND EXPLOITATION OF PRIMATE INFANTS

133

group to join male bands, the composition of females in a langur troop remains more or less stible over time, increasing the likelihood that any 2 females will be related. To the degree that they are related, altruistic behavior will be adaptive (Trivers, 1971). In addition, infant-sharing soon after birth may serve t o familiarize a number of older females with the infant. Outside of “general helpfulness” and care in case of contingencies, little is known about how experiences with other-than-maternal females of various ages influence infant development. It is possible, however that by offering an alternative source of solicitude, the presence of one or more aunts increases the infant’s confidence in his surroundings which may promote separation from the mother and lead to earlier independence. Conflicting information on this matter will be presented in the next section. E.

AUNTS AND INFANT INDEPENDENCE

Universally, mammalian infants spend more time away from their mothers as they develop. In rhesus macaques, baboons, and probably most primate species, the responsibility for this independence4 lies with both the infant that wanders more and the mother that rejects it more frequently (Hinde and Spencer-Booth, 1967a) and otherwise encourages its departure (e.g., Ransom and Rowell, 1972, p. 119). Some evidence suggests that insecure mothers are less likely to facilitate the departure of their infants. For example, caged primiparous macaque females, which are presumably less experienced and less confident, hesitate more in initiating the separation process that normally begins around 3 months. Similarly, Harlow’s “motherless mothers” exhibit a much lower rate of rejection after the %month period than normal mothers do (Harlow etul., 1965). (Before this period, however, motherless mothers are much more rejecting than normal mothers.) Chalmers (1972) has shown that caged Cercopithenrs mitis mothers stayed closer to their infants and restrained them more when the adult male had been temporarily removed, presumably because they felt less secure in his absence. As this example suggests, external factors may greatly influence the amount of mother-infant separation. A totally safe but stimuli-poor environment in which mother-infant pairs are isolated from other monkeys (Jensen et ul., 1967, p. 49; Hinde and Spencer-Booth, 1967a, p. 363) may be as unconducive to infant independence as an overly stimulating one which is perceived by the mother as dangerous. Hinde and Spencer-Booth (1967b) found that isolated mother-infant pairs initially spent more time apart than group-living monkeys did, presumably because the mother was less restrictive. Later, however, in the second 6 months of development, these infants spent more time with their 41ndependence means physical separation from the mother; other implications of the term are not considered here.

134

SARAH BLAFFER HRDY

mothers. In both types of situation, isolated and overstimulating, the mother and infant may maintain intense contact long after the normal onset of separation. In two studies, the presence of aunts worked counter to infant independence (Hinde and Spencer-Booth, 1967b; Wolfheim et al., 1970). For example, rhesus mothers were more permissive in the absence of aunts; in the presence of aunts, infants spent a smaller proportion of their time away from their mothers. Hinde and Spencer-Booth suggest that the presence of aunts that might take infants irretrievably was perceived as a threat by their mothers. Wolfheim et al. compare their results with those of a Japanese observer who noticed that Macuca fuscata mothers restrained their infants more frequently in the wild than in the laboratory. According to them, this parallel illustrates an adaptive mechanism whereby the mother becomes more protective in potentially dangerous situations. According to Rowell etal. (1964), just how “threatening” an aunt is t o the mother may depend on the rank of the females involved. Females that were allowed to cuddle and carry the first 5 of 7 infants born into the group were all subordinate to the mother. Whereas high-ranking females were able to control other females’ interactions with their infants, low-ranking females were unable to d o so. To avoid giving up their infants, these low-ranking mothers would have to pick them up and move away. If the mother’s status in relation to the aunt affects her chances of retrieving the infant, this could provide an explanation as to why rhesus mothers should show this preference for subordinate females. To me, this preference is curious. From one point of view it would make sense for the mother t o prefer the most prestigious aunt available since among rhesus macaques, as in other macaques, the status of the mother or caretaker affects the status and privileges of the infant. Also, it may be that such discrimination is shaped by the circumstances of captivity. The finding that rhesus infants old enough to spend time away from their mothers d o so less in the presence of aunts is somewhat surprising. Assuming that the amount of time spent away from the mother is determined by ( 1 ) differences between mothers, ( 2 ) the mother’s confidence in her environment and especially her ability to retrieve her infant, and (3)the infant’s “motivations” to wander deriving from both physical maturation and the availability of attractive alternatives, one would expect aunts to increase infant independence insofar as they increased mother-infant confidence in their surroundings (e.g., the depositand-forage pattern in Nilgiri langurs and patas; Section IV,D)and insofar as they presented alternative sources of solicitude. Part of the problem here is my definition of aunt-any female, older than the infant, that associates with it; as with Lancaster’s (1971) definition, this one overlaps with “playmates.” In their 1964 work, Rowell etal. were referring to females over 2 years old, Even among caged rhesus, one would expect that the presence of very young aunts (especially juveniles whose mothers were either less

CARE AND EXPLOITATION OF PRIMATE INFANTS

135

dominant or absent) might encourage mother-infant separation. A great deal of the problem must also have to d o with peculiar features of rhesus macaque female dominance hierarchies. At any rate, it is not clear that the presence of aunts decreases mother-infant separation in other species. Unfortunately, the evidence to support this point among other species is less adequate than that from the rhesus studies; it is either qualitative or else based on samples that are too small to indicate anything except that further research is needed. In addition, age differences and differences in maturation rates, which cannot at this time be controlled, raise questions as to the comparability of cross-species informati on. Baldwin and Baldwin (1971) have suggested that the availability of infant and juvenile peer play experience is important in determining the degree to which squirrel monkeys accommodate to and engage in social and nonsocial activities. They point out that in small Suimiri troops, infant and peer socialization groups are small; because a youngster in such a group had fewer animals in the same sex and age class to play with, it therefore might play less. Another possible effect would be that infants played with whatever other young animals were available, regardless of age and sex. This was the case in a caged group of 6 patas monkeys at the Tigoni Primate Center in Kenya: the single infant (6 weeks old) in the group spent most of the time that i t was away from its mother with an older juvenile female named Anxious as well as time with an undersized subadult female, Huiha. In an experimental study of the effects of other group members on motherinfant contact (see Chalmers, 1972), various animals were removed and replaced at 2-day intervals over a period of 14 days. This particular experiment with a patas group was part of a series of experiments under the direction of Dr. Neil Chalmers; the procedure and results are described in an unpublished manuscript (Blaffer, 1970). Briefly, in 35 hours of monitoring, the infant spent 20% fewer 0.5-second intervals away from its mother during the 2 days when his “favorite” aunt Anxious was removed than it did during either of the adjacent controls; significant differences were not observed during the absence of other animals. The presence of the juvenile femaIe Anxious appeared to influence ( I ) the likelihood that the infant would leave its mother, (2) the distance that would separate them, and (3)the length of time the separation would last. If the distance separating mother and infant is taken as a measure of their confidence in their surroundings, the proximity of the aunt apparently increased this confidence. An analysis of the proportion of times that the infant approached the mother minus the proportion that he left her (% Ap, %Li) during the interval when Anxious was absent, indicates that the increased proximity to the mother during this time was largely due to the infant (Chalmers, 1972). Until more information is available, it is not possible to say conclusively that aunts contribute to mother-infant separation, although it is likely that this will

-

136

SARAH BLAFFER HRDY

turn out to be the case for some groups. Because of great variaton in habitat use and social organization both between and within species, and because of differences in maternal responses to aunts, the effects of aunting on infant independence will differ from case to case. Nor is it clear that early independence is necessarily advantageous. Whereas Hinde (1965, p. 71) reports that maternal restrictiveness among rhesus mothers in the presence of aunts retarded motor achievements of the restrained infants, the relatively late independence of some arboreal monkeys (Chalmers, 1972) may be important for infant survival. More subtle effects could be reflected in the infant’s dealings with other animals. F.

STATUS BENEFlTS FOR MOTHERS, AUNTS, AND INFANTS

In a number of species, mothers with infants are treated differentially and their role in group life may be changed after parturition, and in some cases even in pregnancy (e.g., baboons, Japanese macaques, black and white Colobus, chimpanzees, and langurs). As has been mentioned, animals carrying an infant are less subject to attack from conspecifics (e.g., chimpanzees and Colobus monkeys; see Section I1,D). After the birth of their infants, baboon mothers stay closer to the center of the troop, protected by the dominant males (Hall and DeVore, 1965). Although Hanuman langur females d o participate along with males in intertroop encounters, pregnant females and mothers carrying infants are rarely involved (Ripley, 1967, p. 247). Assuming that the special status accorded to mothers is advantageous, the question arises: Does an aunt holding an infant share in maternal prerogatives? If so, how equivalent is aunting in these instances to “agonistic buffering”? Wooldridge has reported for Colobus guereza that whichever female was holding an infant, whether she was the mother or not, was immune to attack from the adult male. Once she had given up the infant, however, she was again vulnerable (Wooldridge, 1969, p. 32). It is unlikely that such aunting is ever as exploitive as its masculine counterpart, but, in fact, this possibility has never been investigated. Another possible status benefit to an aunt from aunting might be the contact she makes with other females. Ploog (1967), for example, reports that among squirrel monkeys a relationship was occasionally formed de ~ O Y Obetween 2 females several weeks after 1 of them had given birth-apparently due to the aunt’s interest in the infant. Rosenblum (1972, personal communication) has suggested that in squirrel monkeys aunting may be reciprocal; that is, the mother whose infant is aunted may repay the compliment when the aunt herself gives birth. Obviously though, such reciprocity could only apply in those species in which multiparous females participate in aunting. An infant too might be deriving status benefits from association with a highranking aunt or foster-mother. If this is so, one would expect that, in those

CARE AND EXPLOITATION OF PRIMATE INFANTS

137

species with a pronounced female dominance hierarchy, mothers would invite aunting and in doing so prefer dominant patrons. However, there is no evidence that a mother ever discriminates in this way, and in the case of caged rhesus macaques, the opposite appears to be true. Apparently, the drawbacks to permissiveness among rhesus macaques outweigh the advantages; for example, a subordinate mother could risk losing her infant, whereas a dominant one might be wasting the privileges of her position if she were to loan her infant to another animal. It may be that only in species with less pronounced hierarchies could the benefits of early aunting outweigh the disadvantages. Observations of postnatal infant-sharing in the wild are limited to Presbytis entellus of India (Jay, 1963; Sugiyama, 1965a) and to Colobus guereza of East Africa (P. Marler, personal communication, cited in Wooldridge, 1969). Bernstein (1968) has also reported infant-sharing in Presbytis crisfutus of Malaysia, but the timing of the first transfer was not mentioned. Instances of transfer within the first 24 hours after birth have also been reported for caged colobids, including two Southeast Asian langurs, Presbytis obscunrs (Badham, 1967) and Pygathrix nemaeus (Hill, 1972). The occurrence of infant-sharing in geographically disparate species belonging to the same subfamily (Colobinae), strongly suggests phylogenetic determination of the trait. Such a phylogenetic interpretation, however, does little to explain why postnatal sharing should have been adaptive in the first place. Any explanation for such a complex behavioral trait must take into account the social context in which it evolved. For example, if it turns out that female dominance hierarchies are as "relatively unstable and poorly defined" among other Colobinae as Jay (1965, p. 233) found them to be among the langurs she studied, then several of the disadvantages of early sharing suggested in the case of rhesus macaques cease to apply, possibly predisposing members of this subfamily to the evolution of early aunting. Needless to say, this suggestion, if true, would lead to a host of questions. G. PREFERRED AND AVAILABLE AUNTS AND INFANTS

Observations from a number of species indicate that to a large extent the mother controls access to her infant. Even in cases of relatively low-ranking females, a mother may either fight off or avoid more dominant animals attempting to take her infant (for baboon and hesus macaque examples, see Ransom and Ransom, 1971, p. 191; Rowel1 etal., 1964). Mothers have been observed to push away, threaten, bite, or otherwise thwart any animals, including adult males, on behalf of their infants, for instance, sifakas (Jolly, 1966, pp. 67-68), Hanuman langurs (Jay, 1965), and vervets (Lancaster, 1971). Assuming that the mother controls access to her offspring, does a mother discriminate in the matter of aunts? Clearly, in some species she does. As men-

138

SARAH BLAFFER HRDY

tioned, Nilgiri langurs only permit adult females (Poirier, 1968), and Lemur catta (Jolly, 1966) only other mothers, to approach their infants; caged rhesus favor subordinate females and “best friends.” Although among wild patas other females have never been seen to take very young infants away from their mothers, on one occasion Hall (1963; also cited in Hall, 1968) saw an adult female briefly hold an infant while remaining next to the mother; in order t o do so, this female had glanced up at the mother in the manner typical of a subordinate animal anticipating attack. These cases suggest that, in animals with pronounced female hierarchies (i.e., rhesus and patas), subordination may be a prerequisite for infant access. In other species maternal preferences are not apparent; the eagerness of the other female t o take the infant may determine aunthood. Occasionally aunts may resort to subterfuge. Instances of a prospective aunt grooming the mother in order to gain access to the infant have been reported for chimpanzees (van Lawick-Goodall, 1971), for vervets (Lancaster, 1971, p. 173; Gartlan, 1969, pp. 148-149), and for caged patas (personal observation). However, it is unlikely that such stratagems would succeed if the mother were determined to hold her infant. Since the “cost” of permitting aunting may vary according to the age and status of the aunt involved, in those cases in which the mother discriminates, one would expect her to do so on the basis of which female provides the advantages of aunting (in terms of foraging freedom for the mother, socialization and protection for the infant, adoption when the mother is sick or if she dies, etc.) with the minimum of its disadvantages (i.e., incompetence, kidnapping, etc.). Thus, among caged rhesus the mother prefers subordinate females that are least likely to succeed in kidnapping her infant (see Section IV,E); among ring-tailed lemurs or Nilgiri langurs, only older animals, which are least likely to damage the infant through inexperience, are permitted access. When such an “optimum” aunt is not available, the possibility of a “dangerous” aunt has to be weighed against the potential disadvantages of no aunt at all. If aunting behavior does not occur to the same extent in all primates, it must be because this trade-off varies both between and within species. Although many patterns of primate social behavior are phylogenetically determined, variations may also be induced by historical and environmental factors. Just as species differences in the maturity of infants at birth will affect the amount of aunting behavior, so will predation pressure, a particularly vulnerable habitat, troop composition as it affects numbers of available aunts, individual differences such as maternal status (i.e., a dominant rhesus female will have more aunts to choose from than a low-ranking mother), and so forth. Where aunting, even from an inexperienced animal, is more advantageous than no aunt at all, mothers may allow any female, including juveniles and subadults, to take their infants. The willingness of the aunt to take the infant or else her

CARE AND EXPLOITATION OF PRIMATE INFANTS

139

availability (i.e., “the last female in the sleeping tree,” see Poirier, 1968, p. 55, discussed in Section IV,D) will be the deciding factor. In some cases, for example, in caged squirrel monkeys (Rosenblum, 1968, p. 227), no previous association between aunts and the mother is apparent. Where the mother is related to the aunt, the odds as t o when aunting is favored and when it is not are complicated by the mother’s double stake in the acquisition of maternal competence by her older daughters and her nieces, as well as in the well-being of her own infant. Among chimpanzees, rhesus macaques, Japanese macaques, perhaps Nilgiri langurs, possibly squirrel monkeys, and undoubtedly others, the strongest and most persistent bond is between mother and infant (see Southwicket al., 1965, p. 155;Yamada, 1963). For many species the reports on the duration of motherinfant contact are ambiguous. For example, Jay (1963, 1965) reported for Northern langurs that mother-child relations were totally severed at weaning prior to the birth of the next offspring. Yoshiba (1968) reported, however, that weaning among Southern langurs could take place as long as a year later. In contrast to DeVore (1963), Ransom and Ransom (1971, p. 81) reported that at Gombe the bond between mother and infant is not severed at the birth of the next infant and may even be intensified, inducing renewed proximity and grooming and nursing. In most cases, female-juvenile relationships have not been traced because of the short duration of study, although they have sometimes been inferred (e.g., Poirier, 1968, p. 49). In the absence of concrete information, however, it has been tacitly assumed that female infants in some species maintain contact with their mothers throughout life whereas in others they do not. This presumed distinction makes a term such as matrifocal worthwhile temporarily. The fact that the only species for which long-term information is available are all matrifocal (or at least females maintain contact with their mothers) suggests that more research is needed to validate the distinction if it is to be really useful. In such matrifocal species, infants and juveniles maintain close contact with their mother after weaning, and often, after the birth of a new offspring, these siblings may have preferential access to the infant. For example, chimpanzee babies less than 5 months old are usually protected from contact with other animals except their own siblings (van Lawick-Goodall, 1967, p. 148). Such access affiliates older siblings into the new mother-infant bond, and may be extended to include more distant matrilineal relatives (e.g., maternal grandmothers; see Section IV, C). The effects of this early association may be longlasting. Yamada (1963, p. 50) reported that among Japanese macaques the frequency with which an infant cofeeds with its siblings is second only to the amount of time spent feeding in the company of its mother; by the time the infant is a juvenile, however, it may feed more often with siblings than with its mother.

140

SARAH BLAFFER HRDY

A corollary of this close association and preferential access is the special attitude of siblings toward the new infant. The Madingley caged rhesus work has demonstrated that siblings show more attention toward an infant than do less closely related animals of the same age and sex in the group (Spencer-Booth, 1968a, p. 557). Field reports have confirmed this among wild chimpanzees (van Lawick-Goodall, 1967). In these species in which there is early sibling-infant association, the foundation for care and potential adoption by older siblings is laid almost from the infant’s birth. At the same time the mother’s female offspring have priority in learning to mother with her new infant. One interesting question here is, Which animals if any are preferred by primiparous mothers? Macaques and chimpanzees may be at the extreme end of the matrifocal continuum; this remains to be determined. From current evidence, it appears that among other species allocation of training is less nepotic. Although a black vervet infant spends much of its first few months in the company of its mother and siblings, it may also be in contact with adolescent females from other genealogies (Lancaster, 1971, p. 166). From Lancaster’s impression and from what data there are, it appears that availability of the infant (in this case due to maternal permissiveness) was more important than genetic relationship in choice by the aunt of an infant, and that maternal permissiveness was not influenced by degree of relationship (Lancaster, 1971, p. 172).5 Eagerness to take the infant seems to vary with the age and status of the female, and one would suspect that this variation reflects the differential benefits derived from being an aunt. A female nursing an infant of her own may be more punitive toward alien infants than a childless one, presumably because nursing another infant could detract from her own reproductive success. In a group of wild sifakas, 2 mothers with infants of their own were the only group members not to show interest in other newborns (Jolly, 1966, p.66). In a number of species (e.g., vervets, savannah baboons, squirrel monkeys, Lowe’s guenons, langurs), nulliparous females show the greatest interest in holding infants. Among caged squirrel monkeys, pregnant females are the most likely to retrieve an infant separated from its mother (Rosenblum, 1972, Fig. 2); generally, such aunting is nonexclusive. According to Rosenblum (1968, p. 227), females may act as aunts to several infants. This undiscriminating eagerness to hold infants on the part of pregnant or nulliparous females implies that they have something to gain; almost surely they are “learning to mother” (see Section IV,A). Unfortunately (for mothers and ’In Table 111, Lancaster (1971) presents frequency of contacts between infants and juvenile or adolescent females; also where known, individuals belonging to the same genealogy are designated. A Mann-Whitney nonparametnc ranking test for contacts of infants on the basis of kinship and nonkinship showed that there was no significant relationship between contact and genetic relationship. The obvious limitation of the data here, however, is that the fact that no genetic relationship was known to exist does not mean that one could not have existed.

CARE AND EXPLOITATION OF PRIMATE INFANTS

141

infants), the willingness of aunts t o aunt may be inversely correlated with their competence in handling infants (see Section IV, B). More experienced multiparous females have less to gain. Unless they either have a preexisting bond with the mother (including genetic relationship), or they are forming such a bond, these females would be less likely to initiate aunting behavior. When females who have not initiated aunting are forced into the role by “infant deposit,” such aunts would be expected to make nonchalant caretakers (see Section IV,D). The infant, too, may exercise choice in its response to an aunt or uncle, especially as it matures. Rosenblum (1968, p. 214) reports that an older infant squirrel monkey may temporarily prefer an aunt to its mother. This was true of the young patas infant described in Section IV,E, which would actively seek his “favorite” aunt. One of the infant baboons at Gombe would avoid all contact with a particular adult male that treated it roughly, while seeking out the more solicitous male Harry (Ransom and Ransom, 1971, pp. 189-190). Very young infants have less choice and cling t o the female currently holding them. In 19 of 49 infant transfers witnessed among hesbytis cnstatus, the infant vigorously resisted (Bemstein, cited in Poirier, 1968). Similarly among Presbytis entellus, females wishing to hold an infant often had to obtain it by force (Sugiyama, 1965a, p. 228). Although an infant may recognize its mother within days of birth (Jay, 1963, p. 443), before this point infants occasionally resist returning to their own mothers; after an infant learns to recognize its mother, he may resist being taken by another female (Wooldridge, 1969, p. 81, 1971, p. 483). This tendency to cling has an obvious adaptive value: presumably the infant is safest with its mother, but once any other female has taken it, survival depends on not becoming se arated. The apparent tit getween the evidence on aunting behavior in this section and kin-selection theory is subject to the same qualifications as were mentioned for male care (see Section 11,D); in particular, the data were selected for relevance to the theory. By way of a summary, some predictions concerning which females should attempt to aunt and which should be preferred as aunts will be presented. Until these are tested, a proper conclusion is pointless. Maternal permissiveness should depend on a tally of the pros and cons of aunting for the mother and her infant under the circumstances in which they find themselves. If the available aunts are related to her, the mother’s behavior should reflect both this tally, and her stake in the acquisition of maternal competence by her female relatives. In those species in which the period of contact between a mother and her female offspring overlap with the birth of subsequent offspring, these daughters will be the preferred aunts. Daughters as aunts means that the “cost” of aunting is deductible in that a close relative profits; the cost may even be reduced. For example, the status of the mother in relation to her offspring should be clearly defined; disputes over who holds the infant and kidnapping will be less likely. The cost from incompetence remains the same for related as for unrelated females. Due to the proximity of siblings to the mother

142

SARAH BLAFFER HRDY

and newborn, and to preferential access, related infants might also be the most available candidates for an older sibling’s aunting attentions. One would expect that such females would be discriminating and prefer infant siblings to unrelated infants (whether aunts are choosing siblings because they are more attracted t o them or because they are more available due to familiarity with the mother needs to be investigated). Aunts unrelated or distantly related to the infants tended may include: (I) nulliparous females eager to hold infants; (2) adult females in the process of establishing a relationship with the mother; and (3) more or less uninterested females that have been conned into aunting. Whereas the first two will be solicitous in order to prevent retrieval of the infant or even termination of the relationship by the mother, the third need not be. Only the female interested in a bond with a particular mother could be expected to discriminate; availability of the infant would be the most important single factor, and this availability will depend on the mother’s assessment of the situation. These predictions represent a combination of what one would expect t o be true if kin-selection theory applies to aunting and of what does seem to be true. In other words, current evidence does not contradict these predictions, but more research is needed to confirm them.

V. SELECTIVE PRESSURES ON THE INFANT A.

NATAL COATS AND OTHER TRAITS OF ATTRACTION

Generally, primate neonates are attractive to some and occasionally, as in the case of Colobus guereza, to all, nearby conspecifics. There is great variation in the strength of this attraction; its onset and duration;. the age, sex, and status of the animals attracted; and the likelihood that perception of the infant will elicit solicitude. Features that may contribute to the infant’s attractiveness include: size at birth, peculiar sounds (e.g., the “purring” noises made by howler and rhesus babies), infantile facial expressions and motor patterns, skin color (often white or pink), relative hairlessness, distinguishing morphological features such as big ears or tail tufts, and distinctive coat color. Of the natal features, coat color is often the most variable and most striking. A number of observers have noted the apparent correlation between the natal coat stage and the concern for the infant exhibited by adult females and other conspecifics among Colobinae and African cercopithecines; as the natal coat changes to a color characteristic of older animals, interest in the infant declines (Booth, 1962, p.485; Gartlan, 1969; Jay, 1965; Lancaster, 1971, p. 177; Poirier, 1968, p. 50). Other observers have speculated on the effectiveness of natal coats and other distinguishing features (e.g., the chimpanzee white tail

CARE AND EXPLOITATION OF PRIMATE INFANTS

143

tuft) in inhibiting aggression (Ransom and Ransom, 1971) and in eliciting protection and rescue. According to Booth (1962, pp. 484-485), among vervets and black and white Colobus, “the sight of an infant in natal coat in human possession resulted in marked agitation on the part of adult wild monkeys of both sexes.” A similar episode, when wild adults approached a human observer holding an infant, has been reported for Presbytis cristafus (Bernstein, 1968, pp. 12-13). Yoshiba (1968, p. 242) reported that the leader of a Presbytis entellus troop attacked an observer who “showed him a newborn infant from another troop.” Yoshiba suggests that the male attacked because the man held a strange infant; another possibility could be that the male was attacking the human who had captured a dark infant. The only experimental work on this subject seems to be that mentioned in Booth (1962, p. 485). According to her, a stuffed natal coat skin will agitate adult Cercopithecus monkeys if it is being moved. Their agitation dies down if the skin lies still. Booth (1962, pp. 483-484) also states that Cercopithecus mothers do not show much interest in dead babies. This observation is in marked contrast to reports for other species. Among savannah baboons (DeVore, 1963), bonnet macaques (Rahaman and Parthasarathy, 1962, p. 157), Hanuman langurs (personal observation), and squirrel monkeys (Clewe, 1969), mothers carry and protect dead infants for days after their death, suggesting that factors other than movement are involved. Clewe has suggested that the presence of hair may be the crucial stimulus, since squirrel monkeys born without hair are dropped to the cage floor, whereas those born with it are held (Clewe, 1969, p. 154). However, without controlling for length of pregnancy and the mother’s hormonal state, it would be impossible to attribute confidently the mother’s response to the state of the vellus. The remainder of this section and the next one focus on natal coat colors; for the purpose of this discussion, species will be divided into three classes: species born with “flamboyant” natal coats, species with coats that are distinctive but discreet, and those with coats that are scarcely distinguishable from the adult pelage. 1.

“Hamboyant” Natal Coats

Here flamboyant refers to striking differences from adult coloration perceptible at a distance to members of other species (including predators) as well as to conspecifics. In this category are included at least five species in the Colobinae subfamily. Newborn infants in Presbytis nrbicundus and Colobus guereza are pure white at birth. Presbytis geei newborns are almost white, but this coloring does not differ greatly from the golden pelage of adults. Among Presbytis aygula and Presbytis melalophus the newborn is white with a dark stripe from head to tailtip which is crossed by a stripe between the shoulders in

144

SARAH BLAFFER HRDY

what has been called a “cruciger” pattern. Presbytis cristatus are born with white skin, white faces, hands and feet, and bright orange body hair; skin and coat color begin to darken within days of birth (Bernstein, 1968, p. 3). Presbyfis obscurus newborns are whitish yellow all over (Furuya, 1961-1962, p. 42). Presbytis johnii have little pigmentation and are sparsely covered with reddish brown hdr; skin and fur begin to turn black like adults at around 10 weeks (Poirier, 1968, p. 49). Outside of the Colobinae, striking natal coats are less common. The young of Macaca arctoides are much paler than the adults. The same is true for Hylobates lar and Hylobates hoolock, although in the case of the hoolock gibbon the situation is complicated by the occurrence of a similar color dimorphism between adult males and females. Both males and females are a pale grayish white at birth, turning dark with age. At puberty, females turn a pale yellowish brown whereas males remain black (McCann, 1933). The color similarity between infants and females of child-bearing age suggests that camouflage (i.e., the infant would not be visible on its mother) as well as distinctiveness may be involved. Whereas the flamboyant neonates mentioned above would be hard t o camouflage unless they were covered by the body of another animal, in several species flamboyant natal features are localized and more discreet. If natal features are indeed a message, the broadcast in these cases could be limited to conspecifics. For example, among Nasalis larvatus, newborns have a small up-tilted nose and vivid blue facial skin which is quite distinct from the flesh-colored faces of adults (Pournelle, 1966, p.4). Pan troglodytes infants have coats that are approximately the same color as those of adults, but they have white tail tufts (van Lawick-Goodall, 1965). Similarly, newborn orangutans are distinguished by white circles around their eyes. 2.

“Discrete but Discreet” Coats

Newborns in this category are characterized by distinctive coats that are not strikingly different from those of adults; usually, these are a darker or else a paler version of the adult pelage. Dark natal coats, pink faces, and large ears are typical of the savannah cercopithecines, i.e., the baboon species, Cercopifhecus aethiops and Erythrocebus patas. Some forest-dwelling New World monkeys also have black newborns (e.g., Ateles). By contrast, the majority of the forestdwelling Cercopithecus species in West Africa have natal coats that are “not significantly different” from those of adults (Gartlan, 1969, p. 149). Although the majority of the Presbytis and Colobus genera for which information is available have flamboyant natal coats, some species in the Colobinae subfamily, such as Presbytis entellus, have dark natal coats. Newborns in the Presbytis senex group are gray with white cheeks (adults are gray or black), although a tendency for “partial albinism” has been reported (Napier and Napier, 1967). In Presbytis entellus, as in some other species, there is an inter-

CARE AND EXPLOITATION OF PRIMATE INFANTS

145

mediate, juvenile coat color. At about 5 months of age, the black coat is replaced by creamcolored fur which persists until the young langur turns gray around 2 years of age. Several of the Colobinae, Procolobus v e m and Colobus badius, have natal coats resembling those of adults (Booth, 1957, p. 422; Dekeyser, 1955, cited in Booth, 1957), and these probably belong in the third category.

3. Adultlike Pelage

This represents a somewhat arbitrary category simply because all newborns are distinguishable from older animals. Regardless of coat color, newborns are invariably small, relatively hairless, etc. Nevertheless, in some species, such as marmosets, Saimin, possibly orangutans and rhesus macaques, newborns appear to resemble adults more than newborns do in other species.

B. PHYLOGENY, ENVIRONMENT, OR AN INDUCEMENT TO CARETAKERS

In this section various explanations for the presence or absence of striking natal costs will be considered. The main argument here is based on two assumptions which, although they seem reasonable, remain to be proven. First, it is taken for granted that color dimorphisms are not accidental and that they serve, or once served, some purpose-in this case, to single out neonates as objects of special attention. Second, it is assumed that flamboyant natal coats increase vulnerability to predation. To phrase this as a testable query, one might ask: Does a raptorial bird or other predator respond more readily to a white or golden colored infant, and will predators choose such an infant more often than a discreetly colored one when presented with both choices? Actually, except for predation by other primates, predation upon primates has rarely been witnessed, possibly because the human observer was a deterrent. Only a few incidents, such as Cynthia Booth’s account of a monkey eagle carrying off a Colobus infant, are known (cited in Jolly, 1972). The predominance of Colobinae among the species with flamboyant natal coats suggests the importance of phylogeny for this trait. However, flamboyant natal coats are not universally found in this subfamily (exceptions include Presbytis entellus, Procolobus verus, and Colobus badius), and several nonColobinae exhibit the trait to some degree (e.g., Macma arctoides and Hylobates lar). Furthermore, such a phylogenetic explanation only leads t o further questions. The first of these might be: Why, when in a number of species neonates manage without flamboyance, should this potentially disadvantageous trait evolve at all? Any answer to the question of why flamboyance occurs must take into account the selective pressures on both the infant and on the infant’s potential

146

SARAH BLAFFER HRDY

caretakers. To the extent that infants benefit from caretakers, and to the extent that aunts and uncles benefit from temporary possession of the infant, their interests coincide; but they do not necessarily do so. Selective pressures may work on either party to behave in a way counter to the interests of the other, as shown by male exploitation of infants (see Section II,E) and cases of infants resisting transfer (Sugiyama, 1965a; Bernstein, 1968). When it is to the advantage of an uncle to take an infant, he attempts to do so regardless of whether the infant possesses a natal coat or not-although very young infants may be preferred in such cases simply because their natal coats are effective in forestalling aggression. Typically, the natal coat lasts for the first 3-5 months, but, after this period, in several species the most intense interest in holding the infant is displayed by nulliparous females, e.g., in vervets (Lancaster, 1971) and by subadult to adult males, e.g., among baboons (Ransom and Ransom, 1971) and Japanese macaques (Itani, 1959). In these species, the dark coat color does not coincide with the period when caretakers are apparently benefiting the most from caring, but rather it is the time when the infant is most helpless and in need of benefits, such as rescue and adoption, from conspecific attention. One would expect that the presence or absence of an extravagant natal coat reflects the needs of an infant within a given socioecological setting; degree of flamboyance should be related to its advantages and disadvantages in each species. Factors that affect the balance might include: degree of terrestriality, or, in the case of completely arboreal species, canopy preference; mode of infant transport (i.e,, ventral or oral carriage among arboreal species as opposed to the jockey style adopted by terrestrial species); other factors affecting the visibility of the infant; maturation rates and the period of infant dependency (very few comparative data are available on maturation rates; t o date, the most informative study is that of Chalmers, 1972); and especially the competence and availability (i.e., the motivation and proximity) of caretakers. One would expect that in those species with bright coats at birth, either special advantages accrued to attracting conspecific attention or else that the risk of attracting predators was diminished. The fact that no terrestrial nor partially terrestrial species are reported to have flamboyant natal coats suggests that bright coloring may be related to being found among the leaves or else to not being found on the ground, or both. For example, in a species such as Erythrocebus putas, which depends for survival on concealment from predators, the disadvantages of a striking natal coat would outweigh all possible advantages. The vivid blue faces of newborn Nasalis Zur vatus (a colobid species with a markedly “terrestrial tendency,” Kawabe and Mano, 1972), on the other hand, represents one possible solution to the problem of how t o attract caretakers without inviting predators as well. Not all arboreal monkeys have flamboyant natal coats, but many arboreal monkeys with bright natal coats have predators. Living in the trees may make bright colors more

CARE AND EXPLOITATION OF PRIMATE INFANTS

147

feasible, but habitat alone does not provide an explanation as t o why natal coats are advantageous. Gartlan (1969, p. 149) and Lancaster (1971, p. 179) have suggested that natal coats in vervets are related to the greater vulnerability of savannah dwellers to predation. According to them, the evolution of distinctive natal coats in conjunction with the intense interest in newborns would ensure that infants were always watched out for. However, it is just as possible that in the context of all primates, the dark natal coats of ground-dwelling species such as baboons and vervets represents a compromise between flamboyance and no distinction at all. Lancaster (1971, p. 177) has also suggested that among species with n o contrasting natal coats, infants may be kept in close contact with their mothers for a longer period of time; in this case, attracting the attention of other group members as a means of protection would be less important. Examples of such undistinguished neonates in close contact with their mothers might include chimpanzees, rhesus macaques, and orangutans. The fact that in these species infants do not need to attract attention may also be a function of “automatic” aunts in a matrifocal system where siblings or true aunts are at hand. Several of the species that practice infant-sharing soon after birth (Presbytis obscurus and Colobus guereza) have striking natal coats; the fact that a third species, Presbytis entellus has dark rather than flamboyant newborns, is almost surely related t o habitat use; common langurs are the most terrestrial members of the Colobinae subfamily. In some areas these animals may spend over 50% of their day on the ground. Infant-sharing also occurs in other Colobinae with flamboyant natal coats (e.g., Presbytis johnii and Presbytis cristatus), but first transfer of the infant may be substantially later than in the foregoing examples. If maternal permissiveness is equated here with an invitation to aunts, the apparent correlation between flamboyant neonates and infant-sharing lends back-handed support to the following hypothesis: in species without flamboyant natal coats, individuals do not benefit (overall) from encouraging other than maternal caretakers in the first months of life, and in species with extravagant natal coats, infants may benefit by attracting other group members. For instance, Procolobus verus neonates resemble adults at birth. If the suggestion offered in this paper (Section W,B) is valid, namely that early aunting would be dangerous for an infant in this species, then that finding would support the hypothesis. In summary, three distinct strategies have been suggested here. 1. As in Colobus guereza and Presbytis obscurus, infants may be strikingly colored and passed around soon after birth; Presbytis entellus represents a ground-adapted modification of this system: there is postpartum infant-sharing, but a more discreet natal coat. 2. As in vervets and baboons, infant coloration may be discrete but discreet, and mothers may be more possessive. Handling of infants by aunts and uncles

148

SARAH BLAFFER HRDY

occurs much later and care in these instances may be initiated by these animals for purposes of their own. 3. As in rhesus macaques, infants may be virtually undistinguished from adults in their coloration. In such matrifocal systems, matrilineal relatives will be at hand, and mothers may be more possessive of their infants in the presence of less familiar animals. It is suggested here that elements of the first strategy would not work or would not be advantageous in the social context of species such as the rhesus macaque. The rhesus social system with its pronounced female dominance hierarchy (see Section IV,E)would preclude widespread infant-sharing as a solution to the need for caretakers; mothers could not afford to lend their infants to dominant females, and infants might lose in terms of status benefits by being under the charge of subordinate females. Just as the rhesus social system precludes sharing, their terrestriality precludes the possibility of striking natal coats; instead, infant-care is assured in other ways, by the availability of siblings, true aunts, and grandmothers.

VI.

SUMMARY

Field and laboratory instances of infant care and abuse by conspecifics other than the infant’s mother have been reviewed and an attempt made to analyze these in terms of the individual and “inclusive” fitnesses of the participants. Partial summaries of this synthesis are provided at the end of Sections I1 and IV. The possibility that flamboyant natal coats and postnatal infant-sharing reflect past selective pressures on the mother-infant pair to invite conspecific care was also considered; this argument is summarized at the end of Section V. It was stressed in the first section and throughout the paper that the data necessaty to test Hamilton’s (1964) theories among the primates simply d o not exist and that all statements made can only be regarded as hypotheses and predictions in need of testing. In particular, almost n o quantitative information is available on the reproductive success of animals involved in various infantcare and exploitation strategies. For this reason it is not yet possible to assign realistic weights to the costs and benefits that such behavior has for those involved. Nevertheless, it is assumed that the animal‘s behavior does reflect a preponderance of advantages over disadvantages and that this balance sheet of effects must be calculated within the individual’s ecological and social context. This tally would be expected to change as the individual adopted different roles in the course of a lifetime. In the case of male care, several expectations are raised by evolutionaIy t h e o y . To the extent that dominant males sire a disproportionate number of offspring, one would expect high-ranking males to exhibit greater solicitude

CARE AND EXPLOITATION OF PRIMATE INFANTS

149

toward infants than subordinate males do. Instances where troop leaders take risks t o rescue infants were cited, and evidence presented from enclosed troops of Japanese macaques and from Barbary macaques in which dominant males participate actively in infant care. An adequate testing of the prediction awaits fine-grained studies of male-infant interactions which take into account the effects that contact with males of different ranks have upon the infant. Subordinate males would be expected to care for infants to whom they were closely related as siblings or true uncles. Cases of sibling adoptions among rhesus and chimps were cited; the possibility that older male siblings would not be available due t o migration was also discussed in the case of macaques. Whereas males caring for infants would be expected to discriminate in regards to the infants that they adopt, protect, or otherwise benefit, males abusing infants would be less likely to do so. Case studies of anubis baboons (Ransom and Ransom, 197 1) suggest that males either actually or potentially involved in consort relationships are more likely to baby-sit and to protect infants, and that these males often direct their attentions toward a specific infant. Researchers on other species have not focused on this problem, and only slim evidence is available; what data there are for Barbary apes d o not support the above prediction. There is some evidence for baboons, macaques, and langurs that the males most likely t o exploit infants are those in positions peripheral to the breeding system or that are just entering it; these are subordinate and “outsider” males that would have the most to gain and the least to lose from behavior (such as agonistic buffering) which benefits them at potential risk to the infant involved. It has been suggested that cases of infanticide reported for langurs and crab-eating macaques represent an extreme example of such exploitation; the case for this cannot be settled, however, until some quantitative measures become available for infanticide’s efficiency in increasing reproductive success. In the case of aunting, i t was suggested that maternal permissiveness and the willingness of aunts to aunt reflect a balance between potential benefits and risks. From the mother’s point of view, the possible advantages of aunting for her infant (e.g., rescue, adoption, status, and socialization benefits) and for herself (e.g., foraging freedom) must be weighed against the likelihood that her infant will be adopted or kidnapped by a nonlactating or incompetent female who either injures it or exposes it t o danger. Factors involved here include not only age, experience, and parity of the aunt, but the vulnerability of the infant at birth, its rate of maturation, and the availability of desirable caretakers. With the exceptions of studies of aunting by Rowell, Hinde and Spencer-Booth and research by Chalmers on comparative maturation of Old World monkeys, evidence on these points is derived largely from chance observations and from peripheral data included in general reports of social behavior. The eagerness of the aunt to take the infant may reflect quite different interests from those of the mother and conceivably could conflict with those of

150

SARAH BLAFFER HRDY

the infant. For example, the advantages to the aunt of holding the infant may be inversely proportional to her competence in infant care, as an inexperienced female has the most to gain from learning to mother. Similarly, the younger and more vulnerable the infant, the more potentially relevant holding it might be for the unpracticed nulliparae. The infant's mother and the aunt then would not necessarily agree on the optimum time for first transfer. Among squirrel monkeys, baboons, and bonnet, Japanese and rhesus macaques, young females appear highly motivated toward infants. A predominance of nulliparous females participating in aunting has been quantitatively demonstrated for vervets, langurs and caged rhesus macaques. An aunt may also be influenced by the desirability of an alliance with the mother or status benefits attached to holding an infant. There are no data on this point. To the extent that aunting would detract from care of her own infant, mothers are not expected to care for or nurse unrelated infants unless the probability of reciprocation is high. Where an aunt is closely related t o the infant, the balance will be complicated both by the mother's stake in the aunt's competence and the aunt's stake in the infant's well-being, No research to date has addressed itself specifically to these problems. Acknowledgments Without the advice and encouragement of Professor E. 0. Wilson, I never could have completed this paper. Without the input of Dr. R. L. Trivers, it would not have been worth writing; in his lectures and private discussions he has exposed me to a theoretical construct that I believe begins t o make sense of many of the problems with which anthropologists must deal. I am grateful to have had such generous teachers. I am also indebted to my advisor, Professor Irven DeVore, and to Dr. Neil Chalmers who first introduced me to the subject of aunting. References

'

Alexander, B. K. 1970. Parental behavior of adult male Japanese monkeys. Behuviour 36, 270-285. Altmann, S.A. 1962. A field study of the sociobiology of rhesus monkeys Wucuca mulatta). Ann. N.Y. Acad. Sci. 102,338-435. Altmann, S. A. 1969. Sociobiology of rhesus monkeys. IV. Testing Mason's hypothesis of sex differences in affective behavior. Behauiour 3 2 , 4 9 4 8 . Badham, M. 1967. A note on the breeding of the spectacled leaf monkey. Int. Zoo Yearb. 7, 89. Baldwin, J. 1969. The ontogeny of social behavior of squirrel monkeys (Saimin sciureus) in a semi-natural environment. Foliu PrimatoL 11, 35-79. Baldwin, J. D., and Baldwin, J. I. 1971. Squirrel monkeys (Suimin') in natural habitats in Panama, Colombia, Brazil and Peru. Primutes 12(1), 45-61. Benedict, B. 1969. Role analysis in animals and men. Man 4(2), 203-214. Bernstein, L 1966. An investigation into the organization of pigtail monkey groups through the use of challenges.Pn'mutes 7(4), 471-480. Bemstein, 1. 1967. A field study of the pigtail monkey (Mucucu nemestrinu). Primates 8(3), 21 7-228.

CARE AND EXPLOITATION OF PRIMATE INFANTS

151

Bernstein, I. 1968. The Lutong of Kuala Selangor. Behaviour 32(1/3), 1-16. Bemstein, I. 1969a. On the morphology, behavior and systematic status of the Assam macaque (Macaca assamensis McClelland 1839). Primates 10, 1-17. Bernstein, I. 1969b. Stability of the status hierarchy in a pigtail monkey group (Macaca nemestrina). Anim. Behav. 17,452458. Bernstein, I. 1970. Some behavioral elements of the Cercopithecoidea. In “Old World Monkeys” (J. R. Napier and P. H. Napier, eds.), pp. 263-296. Academic Press, New York. Blaffer, S. 1970. Infant independence and the presence of an aunt in one group of Erythrocebus patas. Unpublished manuscript. Bolwig, N. A. 1959. A study of the behavior of the chacma baboon, Papio m i n u s . Behaviour 14, 136-163. Booth, AH. 1957. Observations on the natural history of the Olive Colobus monkey, Procolobus verus (van Beneden).Proc. Zool. Soc. London 129,421-431. Booth, C. 1962. Some observations on behavior of Cercopithecus monkeys. I n “The Relatives of Man.” Ann. N. Y. Acad. Sci. 102, Art. 2,477487. Bourlitre, F. Hunkeler, C., and Bertrand, M. 1970. Ecology and behavior of Lowe’s guenon (Cercopithecus campbelli Lowei) in the Ivory Coast. In “Old World Monkeys” (J. R. Napier and P. H. Napier, eds.), pp. 297-350. Academic Press, New York. Bowden, D., Winter, P., and Ploog, D. 1967. Pregnancy and delivery behavior in the squirrel monkey (Saim’risciureus) and other primates. Folia Primatol. 5, 1-42. Brandt, E. M., Irons, R., and Mitchell, G. 1970. Paternalistic behavior in four species of Macaques. Brain Behav. Evol. 3,415420. Bullerman, R 1950. Notes on kidnapping by bonnet monkeys. J. Mammal. 31,93-94. Burton, F. D. 1972. The integration of biology and behavior in the socialization ofMacaca sylvana of Gibraltar. In “Primate Socialization” (F. Pokier, ed.), pp. 29-62. Random House, New York. Butler, H. 1967. Seasonal breeding of the Senegal galago (Galago senegalins&) in the Nuba Mountain Republic of Sudan. Folia Primatol. 5 , 165-175. Bygott, J. D. 1972. Cannibalism among wild chimpanzees. Nature (London) 238,410-411. Carpenter, C. R 1934. A fiild study of the behavior and social relations of howling monkeys (Alouatta palliata). Comp. Psychol. Monogr. 10(48), 1-168. Carpenter, C. R 1942. Societies of monkeys and apes. BioL Symp. 8,177-204. Chalmers, N. R 1968. The social behavior of free-living mangabeys in Uganda. Folia Primatol. 8,263-281. Chalmers, N. R. 1972. Comparative aspects of early infant development in some captive Cercopithecines. In “Primate Socialization” (F. Poirier, ed.), pp. 63-82. Random House, New York. Chamove, A, Harlow, H. F., and Mitchell, G. D. 1967. Sex differences in the infant-directed behavior of preadolescent rhesus monkeys. Child Develop. 38,329-335. Chance, M. 1971. Mother monkeys and their infants. Science 7(1), 3 5 4 0 . Clewe, T. H. 1969. Observations on reproduction of squirrel monkeys in captivity. J. Reprod. Fert., Suppl. 6, 151-156. Collias, N., and Southwick, C. H. 1952. A field study of population density and social organization in howling monkeys. Proc. Amer. Phil. soc. 96, 143-156. Crook, J. H. 1970. The socioecology of primates. In “Social Behavior in Birds and Mammals’’ ( J . H. Crook, ed.), pp. 103-166. Academic Press, New York. Crook, J. H. 1971. Sources of cooperation in animals and man. In “Man and Beast” (J. F. Eisenberg and W. S. Dillon, eds.), pp. 235-262. Smithsonian Inst. Press, Washington, D.C.

152

SARAH BLAFFER HRDY

Crook, J. H., and Gartlan, 1. S. 1966. On the evolution of primate societies. Nature {London) 210,1200-1203. Deag, 1. M., and Crook, J. H. 1971. Social behavior and ‘agonistic buffering’ in the wild barbary macaque Macaca sylvanus. Folia Primatol. 15, 183-200. DeVore, I. 1963. Mother-infant relations in free-ranging baboons. In “Maternal Behavior in Mammals” (H. L. Rheingold, ed.), pp. 305-335. Wiley, New York. DeVore, I. 1965. Male dominance and mating behavior in baboons. In “Sex and Behavior” (F. Beach, ed.), pp. 266-289. Wiley, New York. Ditmars, R L. 1933. Development of the silky marmoset. Bull. N.Y. Zoo Int. Zoo Yearb. 8 , 139-143.

Drickamer, L. 1974. A ten-year summary of reproductive data for free-ranging Mucuca mulatta. Fotia Primatol. 21,61-80. Dumond, F. V. 1968. The squirrel monkey in a semi-natural environment. In “The Squirrel Monkey” (L. Rosenblum and R. W. Cooper, eds.), pp. 88-146. Academic Press, New York. Furuya, Y. 1961. The social life of silver leaf monkeys (Trachypithecus cristatus).Rimtes 3,41-60.

Gartlan, J. S. 1969. Sexual and maternal behavior of the vervet monkey, Cercopithecus aethiops. J. Reprod. Fert., SuppL 6, 137-150. Gil, D. G. 1970. “Violence Against Children.” Harvard Univ. Press, Cambridge, Massachusetts. Goswell, J. J., and Gartlan, J. S. 1965. Pregnancy, birth and early infant behavior in captive patas monkeys, Erythrocebus patas. Folia Primatol. 3, 189-200. Haddow, A. J. 1952. Field and laboratory studies on an African monkey, Cercopithecus ascanius schmidti Matschie. R o c . ZooL SOC.London 122,297-394. Hall, K. R. L. 1963. Some problems in the analysis and comparison of monkey and ape behavior. In “Classification and Human Evolution” (S. L. Washburn, ed.), pp. 263-300. Viking Fund Publ., WennerGren Found., New York. Hall, K. R. L. 1968. Behavior and ecology of the wild patas monkey. In “Primates” P. Jay, ed.) pp. 32-119. Holt, New York. Hall, K. R. L., and DeVore, I. 1965. Baboon social behavior. In “Primate Behavior” (I. DeVore, ed.), pp. 53-110. Holt, New York. Hall, K. R L., and Goswell, M. J. 1964. Aspects of social learning in captive patas monkeys. Primates 5(3/4), 59-70. Hall, K. R. L., and Mayer, B. 1967. Social interactions in a group of captive patas monkeys (Erythrocebur patas). FoliaPnmatoL 5,213-236. Hall, K. R. L., Boelkins, R. C., and Goswell, M. J. 1965. Behaviour of patas monkeys, Erythrocebus patas, in captivity, with notes on the natural habitat. Folia R i m t o l . 3, 22-49.

Hamburg, D. A. 1969. Observations of mother-infant interactions in primate field studies. In “Determinants of Infant Behavior” (B. M. Foss, ed.), Vol. IV, pp. 3-14. Methuen, London. Hamburg, D. A. and Lunde, D. T. 1966. Sex hormones in the development of sex differences in human behavior. In “The Development of Sex Differences” (E. Maccoby, ed.), pp. 1-24. Stanford Univ. Press, Stanford, California. Hamilton, W. D. 1963. The evolution of altruistic behavior.Amer. Natur. 97, 354-356. Hamilton, W. D. 1964. The genetical evolution of social behavior. Parts 1 and 11. J. Theor. B i d . 7, 1-16, 11-52.

CARE AND EXPLOITATION OF PRIMATE INFANTS

153

Hamilton, W. D. 1971. Selection of selfish and altruistic behavior in some extreme models. In “Man and Beast” (J. F. Eisenberg and W. S. Dillon, eds.), pp. 57-92. Smithsonian Inst. Press, Washington, D.C. Hampton, J. K. 1964. Laboratory requirements and observations of Oedipomidas oedipus. Amer. J. Phys. Anthropol. 22,239-244. Hampton, J. K., Hampton, S. H., and Landwehr, B. T. 1966. Observations on a successful breeding colony of the marmoset, Oedipomidas oedipus. Folia Primatol. 4, 265-287. Hansen, E. W. 1966. The development of maternal and infant behavior in the rhesus monkey. Behaviour 27,107-149. Harlow, H. F. 1958. The nature of love. Amer. Psychol. 13(12), 673-685. Harlow, H. F. 1959. The development of affectional patterns in infant monkeys. In “Determinants of Infant Behavior” (B. M. Foss, ed.), Vol I. Methuen, London. Harlow, H. F. 1963. The maternal affectional system. In “Determinants of Infant Behavior” (B. M. Foss, ed.), Vol 11. Methuen, London. Harlow, H. F., and Harlow, M. K. 1965. The affectional systems. In “Behavior of Nonhuman Primates” (A. M. Schrier, H. F. Harlow and F. Stolinitz, eds.), Vol. 2 , pp. 287-334. Academic Press, New York. Harlow, H. F., and Harlow, M. K. 1969. Effects of various mother-infant relationships on rhesus monkey behaviors. In “Determinants of Infant Behavior” (B. M. Foss, ed.), Vol. IV, pp. 15-36. Methuen, London. Harlow, H. F., and Zimmerman, R. R. 1959. Affectional responses in the infant monkey. Science 130,421-432. Harlow, H. F., Harlow, M. K., and Hansen, E. W. 1963. The maternal affectional system of rhesus monkeys. In “Maternal Behavior in Mammals’’ (H. Rheingold, ed.). Wiley, New York. Harlow, H. F., Harlow, M. K., Dodsworth, R O., and Arling, G. L. 1966. Maternal behavior of rhesus monkeys deprived of mothering and peer associations in infancy. Proc. Amer. Phil Soc. 110,58-66. Hill, C. A. 1972. Infant-sharing in the family Colobidae emphasizing Pygathrix. primates 13(2), 195-200. Hinde, R. A. 1965. Rhesus monkey aunts. In “Determinants of Infant Behavior” (B. M. Foss, ed.), Vol. 111, pp. 67-71. Methuen, London. Hinde, R. A. 1969a. Analyzing the roles of the partners in a behavioral interaction-motherinfant relations in rhesus macaques. Ann. N.Y. Acad. Sci. 159,651-667. Hinde, R. A. 1969b. Influence of social companions and of temporary separation on mother-infant relations in rhesus monkeys. In “Determinants of Infant Behavior” B. M. Foss, ed.), Vol. IV, pp. 37-60. Methuen, London. Hinde, R. A., and Spencer-Booth, Y. 1967a. The effect of social companions on motherinfant relations in rhesus monkeys. In “Primate Ethology” (D. Morris, ed.), pp. 267-286. Aldine, Chicago, Illinois. Hinde, R. A., and Spencer-Booth, Y. 1967b. Behavior of socially living rhesus monkeys in their first two-and-a-half years. Anim. Behav. 15, 169-196. Hopf, S. 1967. Ontogeny of social behavior in the squirrel monkey. In “Neue Ergebnisse der Primatologie.” Fischer, Stuttgart. Hrdy, S. B. 1974. Male-male competition and infanticide among the langurs (Presbytis entellus) of Abu, Rajasthan Folia F’rimatol. 22, 19-58. Hughes, T. H. 1884. An incident in the habits of Semnopithecus entellus, the common Hanuman monkey. Proc. Ash. SOC.Bengal pp. 147-150.

154

SARAH BLAFFER HRDY

Imanishi, K. 1957a. Identification: a process of enculturation in the subhuman society of Macaca fuscata. Primates 1(1), 14. Imanishi, K. 1957b. Social behavior in Japanese monkeys, Macaca fuscnta. Psychologia 1, 47-54. Imanishi, K. 1960. Social organization of subhuman primates in their natural habitat. Curr. Anthropol. 1, 393407. Itani, J. 1959. Paternal care in the wild Japanese monkey, Macaca fuscata fuscata. Primates 2(1), 61-93. Jay, P. 1962. Aspects of maternal behavior among langurs. Ann. N. Y . Acad.Sci. 102, Art. 2, 468-476. Jay, P. 1963. Mother-infant relations in langurs. In “Maternal Behavior in Mammals” (H. L. Rheingold, ed.). Wiley, New York. Jay, P. 1965. Field studies. In “Behavior in Nonhuman Primates” (A.M. Schrier, H. F. Harlow, and F. Stollnitz, eds.). Vol. 2, pp. 525-591. Academic Press, New York Jensen, G . D., and Bobbitt, R. A. 1965. An observational methodology and preliminary studies of mother-infant interaction in monkeys. In “Determinants of Infant Behavior” (B. M. Foss, ed.), Vol. 111, pp. 47-63. Methuen, London. Jensen, G. D., and Bobbitt, R. A. 1968. Monkeying with the mother myth. Psychof. Today 1(12), 4143,68-69. Jensen, G. D., Bobbitt, R. A., and Gordon, B. N. 1967. The development of mutual independence in mother-infant pigtailed monkeys, Macaca nemestrina. In “Social Communication among Primates” (S. Altmann, ed.), pp. 43-53. Univ. of Chicago Press, Chicago, Illinois. Jensen, G. D., Bobbitt, R. A., and Gordon, B. N. 1968. Sex differences in development of independence of infant monkeys. Behaviour 3 0 , 1 4 . Jolly, A. 1966. “Lemur Behavior.” Univ. of Chicago Press, Chicago, Illinois. Jolly, A. 1972. “The Evolution of Primate Behavior.” Macmillan, New York. Kaufmann, J. H. 1966. Behavior of infant rhesus monkeys and their mothers in a freeranging band. Zoologica (New York) 5 1, 17-29. Kaufmann, J. H. 1967. Social relations of adult males in a free-ranging band of rhesus monkeys. In “Social Communication among Primates” (S. Altmann, ed.), pp. 73-98. Univ. of Chicago Press, Chicago, Illinois. Kaufman, I. C., and Rosenblum, L. A. 1969. The waning of the mother-infant bond in two species of macaque. In “Determinants of Infant Behavior” (B. M. Foss, ed.), Vol. IV, pp. 41-59. Methuen, London. Kawabe,M., and Mano,T. 1972. Ecology and behavior of the wild proboscis monkey, Nasalis larvatus (Wurmb), in Sabah, Malaysia. Primates 13(2), 213-228. Kawai, M. 1958. On the system of social ranks in a natural troop of Japanese monkeys: I. Basic rank and dependent rank. Primates 1/2, 111-130. [Trans. in “Japanese Monkeys” (S. Altmann, ed.) The Editor, Edmonton, Canada, 1960.1 Kawamura, S. 1968. The matriarchal social order in the Minoo-B Troop. A study on the rank system of Japanese macaques. Ptimates 1/2 148-156. Koford, C. B. 1963a. Group relations on an island colony of rhesus monkeys. In “Primate Social Behavior” (C. H. Southwick, ed.), pp. 136-152. Van Nostrand, Princeton, New Jersey. Koford, C. B. 1963b. Rank of mothers and sons in bands of rhesus monkeys. Science 141, 356-357. Koford, C. B. 1965. Population dynamics of rhesus monkeys on Cay0 Santiago. In “Primate Behavior” (I. DeVore, ed.), pp. 160-174. Holt, New York. Koford, C. B., Farber, P. A., and Windle, W. F. 1966. Twins and teratisms in rhesus monkeys. Folia Primatol. 4,221-226.

CARE AND EXPLOITATION OF PRIMATE INFANTS

155

Kortlandt, A. 1967. Experimentation with chimpanzees in the wild. In “Neue Ergebnisse der Primatologie” (D. Starck, R. Schneider, and H. J. Kuhn, eds.), pp. 208-224. Fischer, Stuttgart. Koyama,N. 1970. Changes in dominance rank and division of a wild Japanese monkey troop in Arashiyama. Primates 11, 335-390. Kummer, H. 1967. Tripartite relations in hamadryas baboons. In “Social Communication among Primates” (S. Altmann, ed.), pp. 63-71. Univ. of Chicago Press, Chicago, Illinois. Kummer, H. 1971. “Primate Societies: Group Techniques of Ecological Adaptation.” Aldine, Chicago, Illinois. Lahiri, R. K., and Southwick, C. H. 1966. Parental care in Macaca sylvana. Folk Primatol. 4, 257-264. Lancaster, J. 1971. Play-mothering: the relations between juvenile females and young infants among free-ranging vervet monkeys (Cercopithecus aethiops). Folk Primatol. 15, 161-1 82. Lehrman, D. S. 1961. Hormonal regulation of parental behavior in birds and infrahuman mammals. In “Sex and Internal Secretion” (W. C. Young, ed.), pp. 1268-1382. Wdbarns & Wilkins, Baltimore, Maryland. Loizos,C. 1967. Play behavior in higher primates: a review. In “Primate Ethology” (D. Morris, ed.), pp. 176-218. Aldine, Chicago, Illinois. Lorenz, R. 1970. Second generation bred in Goeldi’s monkeys, Callimico goeldi callmiconidae. Primates (Int. Zoo News) 17,7940. McCann,C. 1933. Notes on the colouration and habits of the white-browed gibbon or Hoolock (Hylobates hoolock hurl.). J. Bombay Natur. Hist. SOC.36(2), 395405. McCann, C. 1934. Observations on some of the Indian langurs. J. Bombay Natur. Hist. SOC. 36(3), 618628. MacRoberts, M. H. 1970. The social organization of Barbary apes (Macaca sylvana) on Gilbraltar. Amer. J. Phys. Anthropol. 33,8349. Marsden, N. M., and Vessey, S. H. 1968. Ailoption of an infant green monkey within a social group. Commun. Behav. Biol., Part A 2(6), 275-279. Mason, N. A. 1966. Social organization of the South American monkey Callicebus moloch: a preliminary report. Tulane Stud. Zool. 13, 23-28. Michael, R. P., and Walegalla, J. 1967. Ovarian hormones and the sexual behavior of the male rhesus monkey (Macaca mulatto) under laboratory conditions. J. Endocrinol. 36, 268. Michael, R. P., and Zumpe, D. 1970. Rhythmic changes in the copulatory frequency of rhesus monkeys (Macaca mulatto) in relation to the menstrual cycle and a comparison with the human cycle. J. Reprod. Fert. 21,199-201. Michael, R. P., Saayman, G., and Zumpe, D. 1967. Sexual attractiveness and receptivity in rhesus monkeys. Nature (London) 215,554-556. Mitchell, G. D. 1968. Attachment differences in male and female infant monkeys. Child Develop. 39,611620. Mitchell, G. D. 1969. Paternalistic behavior in primates.Psychol. Bull. 71(6), 399-417. Mitchell, G. D., and Brandt, E. M. 1972. Paternal behavior in primates. In “Primate Socialization’’ (F. Poirier, ed.), pp. 173-206. Random House, New York. Mitchell, G. D., Ruppenthal, G. C., Raymond, J., and Harlow, H. F. 1966. Long-term effects of multiparous and primiparous monkey mother rearing. Child Develop. 37(4), 781-791. Mitchell, G. D., Harlow, H. F., Griffin, G. A., and Moller, G. W. 1967. Repeated maternal separation in the monkey. Psychon. Sci. 8(5), 197-198. Mohnot, S. M. 1971. Some aspects of social changes and infant-killing in the Hanuman

156

SARAH BLAFFER HRDY langur, Presbytis entellus (Primates: Cercopithecidae), in Western India. Mammalia

35(2). 175-198. Moynihan, M. 1964. Some behavior patterns of Platyrrhine monkeys 1. The night monkey (Aorus trivirgatus). Smithson. Misc. Collect. 146(15). Napier, J., and Napier, P. 1967. “Handbook of Living Primates.” Academic Press, New York. Nishida, T. 1966.A sociological study of solitary male monkeys.Primates 7(2), 141-203. Ploog, D. W. 1967. The behavior of squirrel monkeys (Saimiri sciureus) as revealed by sociometry, bioacoustics and brain stimulation. In “Social Communication among Primates” (S. Altman, ed.). pp. 149-184. Univ. of Chicago Press, Chicago, Illinois. Poirier, F. E. 1968. The Nilgiri langur (Presbytis johnii) mother-infant dyad. Primates 9,

45-68. Poirier, F. E. 1969. The Nilgiri langur (Presbytis iohnii) troop: its composition, structure, formation and change. Folio Primatol. 10,2047. Pournelle, G. H.1966.Birth of a proboscis monkey. Zoonooz 39(3), 3-7. Rahaman, H., and Parthasarathy, M. D. 1962. Studies of the social behavior of bonnet monkeys. Primates 10,149-162. Rahm, U. 1967. Observations during chimpanzee captures in the Congo. In “Neue Ergebnisse der Primatologie” @. Starck, R. Schneider, and H. J. Kuhn, eds.). 195-207. Fischer, Stuttgart. Ransom, T. W., and Ransom, B. S. 1971.Adult male-infant relations among baboons (Papio anubis). Folia Primatol. 16,179-195. Ransom, T., and Rowell, T. E. 1972.Early social development of feral baboons. In “Primate Socialization” (F. Poirier, ed.), pp. 105-144. Random House, New York. Reynolds, V. 1968. Kinship and the family in monkeys, apes and man. Man 3(2), 209-223. Rheingold, H. 1969. The effect of a strange environment on the behavior of infants. In “Determinants of Infant Behavior” (B. M. Foss, ed.), Vol. IV. Methuen, London. Ripley, S. 1967. Inter-troop encounters among Ceylon gray langurs (Presbytis entellus). In “Social Communication among Primates” (S. Altmann, ed.). Univ. of Chicago Press, Chicago, Illinois. Ripley,S. 1970. Leaves and leaf monkeys. In “Old World Monkeys” (J. Napier and P. Napier, eds.), pp. 418-512. Academic Press, New York. Rosenblum, L. A. 1968. Mother-infant relations and early behavioral development in the squirrel monkey. In “The Squirrel Monkey” (L. A. Rosenblum and R W. Cooper, ed.), 207-234. Academic Press, New York. Rosenblum, L. A. 1971. Infant attachment in monkeys. In “The Origins of Human Social Relations” (R. Shaffer, ed.). Academic Press, New York. Rosenblum, L. A. 1972. Sex and age differences in response to infant squirrel monkeys. Brain Behav. Evol. 5, 3040. Rosenblum, L.A., and Kaufman, I.C. 1967. Laboratory observations of early mother-infant relations in pigtail and bonnet macaques. In “Social Communication among Primates” (S. Altman, ed.), pp. 3342. Univ. of Chicago Press, Chicago,Illinois. Rowell, T.E. 1961. Maternal behavior in non-maternal golden hamsters (Mesocriecetus auratus). Anim. Behav. 9,11-15. Rowell, T.E. 1963a. The social development of some rhesus monkeys (1961 seminar). In “Determinants of Infant Behavior” (B.M. Fms, ed.), VoL I1 pp. 35-49. Methuen, London. Rowell, T.E. 1963b. Behavior and female reproductive cycles of rhesus macaques. J. Reprod. Fertil. 6 , 193-203. Rowell, T.E. 1965. Some observations on a hand-reared baboon. In “Determinants of Infant Behavior” (B. M. Foss, ed.), VoL In, pp. 77-84.Methuen, London. Rowell, T. E. 1967. Female reproductive cycles and the behavior of baboons and rhesus

CARE AND EXPLOITATION OF PRIMATE INFANTS

157

macaques. I n “Social Communications among Primates” (S. Altmann, ed.), 15-32. Univ. of Chicago Press, Chicago, Illinois. Rowell, T. E., Hinde, R. A., and Spencer-Booth, Y. 1964. “Aunt”-infant interaction in captive rhesus monkeys. J. Anim. Behav. 12,219-226. Rudran, R. 1973. The reproductive cycles of two subspecies of purple-faced langurs (Presbytis senex) with relation to environmental factors. Folia Primatol. 19,4140. Rumbaugh, D. M. 1965a. Maternal care in relation to infant behavior in the squirrel monkey.Psycho1. Rep. 16,171-176. Rumbaugh, D. M. 1965b. The gibbon infant, Gabrielle. Zoonooz 38(12). Saayman, G. S. 197 1. Behavior of the adult males in a troop of free ranging chacma baboons (Papio ursinus). Folio Primatol. 15, 36-57. Sade, D. S. 1965. Some aspects of parentaffspring and sibling relations in a group of rhesus monkeys, with a discussion of grooming. Amer. J. Phys. Anthropol. 23, 1-17. Sade, D. S. 1967. Determinants of dominance in a group of free-ranging rhesus monkeys. In “Social Communication among Primates” (S. Altmann, ed.), pp. 99-114. Univ. of Chicago Press, Chicago, Illinois. Sauer, E. G. F. 1967. Mother-infant relationship in Galagos and the oral child transport among primates. Folia Primatol. 7, 127-149. Schaller, G. B. 1963. ‘The Mountain Gorilla: Ecology and Behavior.” Univ. of Chicago Press, Chicago, Illinois. Schaller, G . B. 1964. “The Year of the Gorilla.’’ Univ. of Chicago Press, Chicago, Illinois. Schenkel, R., and Schenkel-Hulliger, L. 1967. On the sociology of free-ranging colobus (Colobus guereza cuudatus Thomas 1885). In “Neue Ergebnisse der Primatologie” @. Stark, R. Schneider, and H. J. Kuhn, eds.), pp. 185-194. Fischer, Stuttgart. Schultz, A. H. 1969. “The Life of Primates.” Universe Books, New York. Seay, B. 1966. Maternal behavior in primiparous and multiparous rhesus monkeys. Folia Primatol. 4, 146-1 68. Simonds, P. E. 1965. The bonnet macaque in South India. In “Primate Behavior” (I. DeVore, ed.), pp. 175-196. Holt, New York. Simons, R. C., and Crawford, M. N. 1969. Determination of paternity in group-born pigtailed monkeys. Roc. Int. Congr. Primatol., 2nd, Atlanta, Ca. 1968, 1,254-260. Smuts, B. 1972. Natural Selection and Macaque Social Behavior. Senior Honors Thesis, Harvard Univ., Cambridge, Massachusetts. Southwick, C. H., Beg, M. A., and Siddiqui, M. R. 1965. Rhesus monkeys in North 1ndia.h DeVore, . ed.), pp. 111-159. Holt. New York. “Primate Behavior” (I Spencer-Booth, Y. 1968a. The behavior of group companions towards rhesus monkey infants. Anim Behav. 16,541-557. Spencer-Booth, Y. 1968b. The behavior of twin rhesus monkeys and comparisons with the behavior of single infants. Primates 9,7564. Spencer-Booth, Y. 1970. The relationship between mammalian young and conspecifics other than mothers and peers: a review. In “Advances in the Study of Behavior” (D. S. Lehrman, R. A. Hinde, and E. Shaw, eds.), Vol. 3, pp. 120-180. Academic Press, New York. Spencer-Booth, Y., and Hinde, R. A. 1967. The effects of separating rhesus monkey infants from their mothers for six days. J. ChildPsychol. Aychiat. 7,179-197. Struhsaker, T. T. 1967a. Behavior of vervet monkeys (Cercopithecus aethiops). University of Calif. (Berkeley). Publ. Zool. 82,l-74. Struhsaker, T.T. 1967b. Social structure among vervet monkeys (Cercopithecus aethiops). Behavior 24,83-119. Sugiyama, Y. 1965a. Behavioral development and social structure in two troops of Hanuman langurs (Ptesbytis entellus). Primates 6,213-247.

158

SARAH BLAFFER HRDY

Sugiyama, Y. 1965b. On the social change of Hanuman langurs (Presbytis entellus) in their natural conditions. Primates 6(3/4), 381-418. Sugiyama, Y. 1966. Artificial social change in a Hanuman langur troop (Presbytis entellus). Primates 7(1), 41-72. Sugiyama, Y. 1967. Social organization of Hanuman langurs. In “Social Communication among Primates” (S. Altman, ed.), pp. 221-236. Univ. of Chicago Press Chicago, Illinois. Teleki, G. 1973. “Predatory Behavior of Wild Chimpanzees.” Bucknell Univ. Press, Lewisburg, Pennsylvania. Thompson, N.S. 1967. Primate infanticide: a note and a request for information. Lab. Rimate Newslett. 6 , 18-19. Thorington, R. W., Jr. 1967. Feeding and activity of Cebus and Saimiri in a Colombian forest. In “Neue Ergebnisse der Primatologie” (D. Starck, R. Schneider, and H. J. Kuhn,eds.), pp. 180-184. Fischer, Stuttgart. Tinbergen, N. 1963. On aims and methods of ethology. Z. Tierpsychol. 20,410-433. Trivers, R. L. 1971. The evolution of reciprocal altruism. Quart. Rev. Biol. 46, 35-57. Trivers, R. L. 1972. Parental investment and sexual selection. In “Sexual Selection and the Descent of Man 1871-1971” (B. Campbell, ed.), pp. 136-179. Aldine, Chicago, Illinois. Trivers, R. L. 1974. Parent-offspring conflict. Amer. Zool. in press. Vandenbergh, J.G. 1963. Feeding activity and social behavior of the tree shrew, Tupaiu glis, in a large outdoor enclosure. Folia Primatol. 1, 199-207. Van den Berghe, L. 1959. Naissance d’un gorilla de montagne a la station de zoologie experimentale de Tshibati. Folia Sci. Afr. Cent. 4,81443. van LawickGoodall, J. 1965. Chimpanzees of the Gombe Stream Reserve. In “Primate DeVore, . ed.), pp. 425-473. Holt, New York. Behavior” (I van LawickGoodall, J. 1967. Mother-offspring relationships in free-ranging chimpanzees. In “Primate Ethology” (D. Morris, ed.) Aldine, Chicago, Illinois. van LawickCoodall, J. 1968. The behavior of free-living chimpanzees in the Gombe Stream Reserve. Anim Behav. Monogr. 1, 165-31 1. van LawickCoodall, J. 1971. “In the Shadow of Man.” Houghton, Boston, Massachusetts. Williams, G. C. 1972. “Adaptation and Natural Selection.” Princeton Univ. Press, Princeton, New Jersey. Wilson, A. P., and Bodkins, R. C. 1970. Evidence for seasonal variation in aggressive behavior by Macaca mulatta. Anim. Behav. 18,719-724. Wilson, A. P., and Vessey, S. H. 1968. Behavior of free-ranging castrated rhesus monkeys. Folb Primatol. 9, 1-4. Wilson, E. 0. 1971. “Insect Societies.” Harvard Univ. Press, Cambridge, Massachusetts. Wolfheim, J. H., Jensen, G. D., and Bobbitt, R. A. 1970. Effects of group environment on the mother-infant relationship in pigtailed monkeys. (Mamca nemestnna). Primates 11(2), 119-124.

Wooldridge, F. L. 1969. Behavior of the Abyssynian Colobus Monkey, Colobusguereza, in Captivity. M.A. Thesis, Univ. of Florida, Gainesville. Wooldridge, F. L. 1971. Colobus guereza: birth and infant development in captivity. Anim. Behav. 19,481-485. Yamada,M. 1963. A study of blood-relationship in the natural society of the Japanese macaque. Primates 4(3), 43-66. Yoshiba, K. 1968. Local and inter-troop variability in ecology and social behavior of common Indian Langurs. In “Primates’’ (P. Jay, ed.), pp. 217-242. Holt, New York. Zuckerman, S. 1932. “The Social Life of Monkeys and Apes.” Routledge & Kegan Paul, London.

Hypothalamic Mechanisms of Sexual Behavior, with Special Reference to Birds J. B. HUTCHISON MRC UNIT ON THE DEVELOPMENT AND INTEGRATION OF BEHAVIOUR UNIVERSITY SUB-DEPARTMENT MADINGLEY, CAMBRIDGE, ENGLAND

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

I. Introduction. 11. Localized Steroid Effects in the Brain .................... A. Copulatory Behavior . . . . . . . . . . . . . .............. B. Precopulatory Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Biochemical Factors in Androgen Action A. Uptake . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Metabolism.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C. Specificity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Variable Hypothalamic Sensitivity to Androgen A. Effects of Prolonged Androgen Deficit B. Sensitizing Effects of Androgen ..................... C. Environmental Factors and Androgen Action . . . . . . . . . . . . . V. Hypothalamic Androgen Concentration and the Structure of Courtship . VI. Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

I.

159 160 161 163 165 165 170 173 173 174 180 182 185 190 194

INTRODUCTION

Research into the influence of steroid hormones on brain mechanisms of sexual behavior has indicated two major effects: ( I ) in prenatal and perinatal development, these hormones may sexually differentiate the neural mechanisms that later integrate sexual behavior (Young, 1965; Goy, 1970a,b); and (2), in adulthood, steroid hormones act on these mechanisms so that sexual behavior is elicited by the appropriate sensory stimulation. A distinction can be drawn between indirect and direct modes of action. Thus, steroid hormones may influence peripheral sensory systems to modify brain functioning indirectly by altering afferent input or they may act directly on the brain itself. There is consider159

160

J. B. HUTCHISON

able psychophysical evidence (Beach and Levinson, 1950; Lehrman, 1955; Hinde et al., 1963) for the indirect peripheral effects of steroid hormones. However, apart from two studies showing that estradiol may increase the genital sensory field innervated by the pudendal nerve (Komisaruk et ai., 1972; Kow and Pfaff, 1973), there are as yet no relevant neurophysiological data. By contrast, numerous studies, reviewed later in this paper, have shown that discrete areas of the brain, especially the anterior hypothalamus and preoptic region, are directly influenced by sex steroids and closely linked with mechanisms underlying sexual behavior. Although there is some knowledge of the areas of the brain involved, neither the cellular systems nor the physiological processes whereby hormones influence the brain directly are understood. The main purposes of this paper are to review some recent studies on the action of androgen on hypothalamic mechanisms of male courtship behavior using intracerebral steroid implants in Barbary doves (Streptopelia rimria) and to consider the implications of an hypothesis derived from this work. This proposes that the sensitivity of the anterior hypothalamus to androgen, in relation to sexual behavior, is variable and depends on the endocrine state of the animal. To provide a framework for discussion of this hypothesis, it will be necessary to review present knowledge of the direct effects of intracerebral steroids on the hypothalamic mechanisms of sexual behavior in other birds and mammals and also to consider some recently discovered cellular effects of steroids that are relevant to the influence of hormones on brain mechanisms of behavior. In discussing the direct effects of hormones, the term activation appears later in the text. This term is currently used in the literature, and will be used here to represent the as yet unknown processes whereby steroid hormones influence the brain to facilitate patterns of reproductive behavior in animals receiving appropriate sensory stimulation.

11. LOCALIZED STEROID EFFECTS IN THE BRAIN

Much of the early evidence that gonadal steroids act directly on the brain came from studies of the effects of lesions, particularly in the anterior and posterior hypothalamus, on the initiation of male and female copulatory behavior by systemic androgen or estrogen (see review by Hams and Michael, 1964; Davidson, 1966b; Giantonio et ai., 1970). These lesions selectively eliminate sexual behavior without impairing the functioning of the hypophysiotrophic area controlling gonadotropin output or the secretion of gonadal hormones, suggesting that discrete localizable areas within the hypothalamus are both sensitive to steroid hormones and closely associated with the control of sexual behavior.

HYPOTHALAMUS AND SEXUAL BEHAVIOR

161

There are obvious difficulties in the interpretation of studies involving lesions to the hypothalamus. Although not affecting pituitary-gonad relationships, lesions may influence metabolism, and it becomes difficult to distinguish between the secondary effects of general physiological damage and the selective effects of these lesions on areas sensitive to steroids. To overcome this limitation, steroid hormones have been applied to the brain directly by means of fused crystalline implants designed to release hormone gradually in a localized area of the brain-an elegant technique originally developed by Harris et al. (1958). These workers were able to assess the probability that a synthetic estrogenic substance, stilbestrol, implanted into the brain of a female cat would have localizable effects on brain functioning or that it would diffuse from these solid implants into the systemic circulation. The method involved measuring the relative effects of different esters of stilbestrol on female copulatory behavior, uterine development, and vaginal cornification. The relative potency of hypothalamic implants of stilbestrol esters on copulatory behavior and peripheral target organs was found to be related to the length of the acid side chain. By manipulating the ester, the rate of absorption of stilbestrol from the implant and, therefore, diffusion from brain implants into the systemic circulation could be controlled (Harris and Michael, 1964). Their important finding that stilbestrol di-n-butyrate, implanted into the anterobasal hypothalamus, facilitated female copulatory responses to the male without affecting peripheral target organs, provided a basis for the many subsequent studies that have now been carried out using solid steroid implants.

A.

COPULATORY BEHAVIOR

Although the early work of Harris el al. (1958) indicated that a steroid-like substance would induce female copulatory behavior by direct action on the anterobasal hypothalamus but not elsewhere, it did not specifically confirm the hypothesis that normally occurring estrogens might have direct effects on discrete areas of the hypothalamus. However, Lisk (1962, 1967b) showed that solid implants of the normally occurring estrogen, estradiol-l7p, evoked female copulatory behavior in ovariectomized rats, and did so only when positioned in the anterior hypothalamus; this finding was later confirmed by Chambers and Howe (1968). There appear to be species differences in the location of the area sensitive to estrogen implants. Thus, the estrogen-sensitivearea of spayed rabbits is in the posterior rather than the anterior region of the hypothalamus, in the premamillary nuclei (Palka and Sawyer, 1966a). Because female copulatory behavior in rats is facilitated by the synergistic action of estradiol and progesterone, there has been interest in exploring the effects of intracerbral progesterone in ovariectomized females primed with small doses of estradiol. Progesterone implanted into the medial basal hypothalamus

162

J. B. HUTCHISON

evoked female copulatory behavior in ovariectomized rats primed with estradiol but was ineffective in other brain areas (Powers, 1972). However, implants of progesterone in the mesencephalic reticular formation of ovariectomized rats, not primed with estradiol, induced the display of female copulatory behavior (Ross ef al., 1971), suggesting the involvement of extra-hypothalamic areas in the activational effects of progesterone. The study of the direct effects of steroids on brain mechanisms of male copulatory behavior involves difficulties not present in female sexual behavior, namely, problems arising from the complexity of the relationships between the effects of androgen and those of early copulatory experience on the maintenance of sexual behavior after castration. Unlike copulatory behavior in the majority of female mammals, where there is a direct relationship between the display of copulatory behavior and the presence of circulating estrogen (Michael, 1965), or in some cases both progesterone and estrogen (Ciaccio and Lisk, 1971; Joslyn et al., 1971), male sexual behavior may continue to be displayed months after castration if adequate stimulus conditions are provided (Rosenblatt and Aronson, 1958; Cerall, 1963; Rosenblatt, 1965; Beach, 1970; Phoenix et al., 1973; Manning and McCill, 1974). The effects of intracerebral implants of androgen cannot be tested satisfactorily unless there is a significant and rapid loss of sexual behavior following the elimination of endogenous androgen. There is clearly considerable variation even in a single mammalian order, the Rodentia. Thus, in rats, components of male sexual behavior continue to be displayed to a female for at least 6 months after castration, although there are progressively increasing quantitative changes in behavior (Davidson, 1966a). By contrast, the male copulatory behavior of the Balb/c strain mice disappears rapidly after castration, irrespective of prior sexual experience, indicating a possible direct dependence of these behavior patterns on endogenous androgen (McCill, 1965). The first evidence for the direct activational influence of androgen on the mammalian brain came from studies on rats, in which an androgenic effect on a localized area of the hypothalamus was demonstrated. Crystalline testosterone propionate, implanted within an area extending from the medial preoptic nucleus caudally to the posterior hypothalamus, induced copulatory behavior in castrated male rats tested with females (Davidson, 1966b; Lisk, 1967a,b), but implants elsewhere in the brain were ineffective. Since the dosage threshold for the stimulation of growth of the seminal vesicles and prostate was lower than that for the activation of copulatory behavior (Davidson, 1966b), lack of a peripheral stimulatory effect on these target organs by hypothalamic implsnts of testosterone propionate provided evidence that the behavioral effects were due to a direct action on the brain and not to hormone that had leaked via the systemic circulatory system back to the brain. On the basis of this result, Davidson (1966b) suggested that there may be a network of testosteronesensitive neurons extending throughout the basal hypothalamic area. More re-

HYPOTHALAMUS AND SEXUAL BEHAVIOR

163

cently, the extent of this androgen-sensitive area has been examined critically using testosterone propionate implants with a more limited diffusion range. Implants in the preoptic and anterior hypothalamic areas were more effective than implants positioned further caudally. Johnston and Davidson (1972) concluded that the “androgen-sensitive structures regulating male sexual behavior in the rat are concentrated more in the anterior than the caudal hypothalamic region.” Androgen-sensitive areas of the brain associated with male copulatory behavior have also been localized in birds. This has been possible because, as in male mammals, the major androgenic steroid appears to be testosterone: this hormone has been detected in the blood plasma of pigeons (Rivarola etal., 1968) and doves (Hutchison and Katongole, 1975). Studies with intracerebral implants of testosterone propionate indicate that in capons the androgen-sensitive region associated with the activation of copulatory behavior is restricted to the preoptic area (Barfield, 1969); whereas in castrated male doves, androgen sensitivity associated with copulatory behavior is localized in both the preoptic and anterior hypothalamic areas (Barfield, 197 1). In male chicks, precocious copulation has been induced by testosterone propionate implants in the anterior hypothalamus (Gardner and Fisher, 1968), and inhibited by progesterone implants in the preoptic area (Meyer, 1972). Progesterone is known to antagonize the behavioral effect of androgen on the avian brain (Komisaruk, 1965). B. PRECOPULATORY BEHAVIOR

Can male sexual patterns other than copulatory behavior be elicited by means of androgen applied directly to the brain? Precopulatory patterns have not been studied to any extent in rodents, because courtship patterns, although probably important, are both brief and variable, and may be initiated by pheromones. In many species of birds, however, courtship consists of elaborate visual displays which are stereotyped in form and, within limits, predictable in sequence. The courtship of the male is usually a particularly conspicuous aspect of sexual behavior which, as Tinbergen (1965) has pointed out, occupies more time than the act of insemination. The causal factors underlying male courtship in birds are known from ethological studies to be complex. Hinde (1970) has summarized evidence for the view that many of the complexities of courtship behavior can be understood in terms of the hypothesis that it depends on conflicting tendencies to behave in incompatible ways toward the partner. Thus, the initial response of the male chaffinch (Fringilla coelebs) to the female is aggressive. If the female does not flee, the male shows courtship involving both aggressive and fleeing components‘ As the breeding season progresses, aggressive elements diminish and the courtship can be understood as a conflict between attempts to mate with the female and t o flee from her.

164

J.

B. HUTCHISON

It has been claimed that sexual and aggressive components of male courtship can be related to anatomically distinct areas of the brain (Barfield, 1965). Thus, in capons, unilateral implants of testosterone propionate in an area extending from the lateral diencephalon to the paleostriatum induced aggressive behavior alone; whereas unilateral implants in the preoptic area induced copulatory behavior in the absence of courtship waltzing and aggressive displays. Simultaneous activation of both areas by bilateral implants of testosterone resulted in the display of courtship waltzing. However, these experiments are difficult to interpret because the waltzing display may have been due to the elevated androgen levels within the brain, resulting from bilateral implants of testosterone propionate, rather than to the separable activational effects of testosterone on dual androgen-sensitive areas in the forebrain. This interesting hypothesis remains, therefore, to be confirmed. Androgen-sensitive mechanisms associated with courtship behavior have also been studied in the male Barbary dove. In this species, courtship behavior consists initially of a rapid alternation of aggressive displays (termed chasing and bowing), which cause the female to retreat, and nest-orientated behavior (termed nest-soliciting), when the male selects a potential site for the nest, causing the female to approach (see Hutchison, 1970a,b; Lovari and Hutchison, 1975). These courtship displays of the male decline and disappear within 20 days after castration and are rapidly reinstated by 300-pg intramuscular injections of testosterone propionate per day, indicating that male courtship depends on gonadal hormones and that testosterone may be the effective androgen (Hutchison, 1970b). The androgen-sensitive areas associated with courtship behavior have been studied by implanting testosterone propionate into the brain. Qualitatively normal courtship can be obtained only from implants in the preoptic and anterior hypothalamic areas (Hutchison, 1967). However, fragmentary courtship is sometimes obtained with implants in the area basalis, posterior hypothalamus, and ventral neostriatum intermediale-areas of the brain that are adjacent to the preoptic and anterior hypothalamic areas (Fig. 1). The effectiveness of implants in restoring courtship could, in fact, be related to their proximity to the steroidsensitive area and it seems likely that these fragmentary displays were due to low concentration of hormone diffusing to the preoptic and anterior hypothalamic areas (Hutchison, 1971). The general conclusion can be drawn that the preoptic and anterior hypothalamic areas of those mammals and birds that have been studied so far contain androgen-sensitive elements closely associated with copulatory behavior and, in the case of birds, with precopulatory behavior; the latter has yet to be studied in mammals. There appears so far to be no firm evidence implicating other brain areas in the mediation of androgenic effects on sexual behavior. Although aspects of female precopulatory behavior in birds, such as the early phases of nesting behavior in female budgerigars (Melsopsitrucus), are causally related to

HYPOTHALAMUS AND SEXUAL BEHAVIOR

165

estrogen (Hutchison, 1975), no attempts appear to have been made to establish whether estrogens have a direct influence on the hypothalamus to activate these patterns.

111. BIOCHEMICAL FACTORS IN ANDROGEN ACTION A.

UPTAKE

The technique of intracerebral implantation of steroids has contributed greatly to our knowledge of the extent of steroid-sensitive areas in the brain associated not only with sexual behavior but also with the feedback mechanisms controlling gonadotropic hormones (see review by Davidson, 1969). However, one of the major difficulties in this research has been the problem of measuring the rate of diffusion from implants of crystalline hormone within the brain. In the earliest work, Michael (1961) calculated the concentration of I4C-labeled diethylstilbestrol di-n-butyrate, liberated from solid 200-pg implants in the hypothalamus of cats, at various distances from the implant site, using radioautographic techniques. The concentration of labeled hormone was found to fall

FIG. 1. (a) Locations of labeled cells in the forebrain of the male chaffinch following systemic administration of test~sterone-~H. From Zigmond et al., 1972a.) (b) Locations of intracerebral implants of testosterone propionate in castrated male Barbary doves. Each symbol denotes the position of the tip of the implant in the coronal section and the type of behavior elicited. (0)Complete courtship, with aggressive and nest-orientated behavior; (e)incomplete courtship lacking in either aggressive behavior or nest-orientated behavior; (0) no courtship, behavior similar to a castrate. Abbreviations: CO, optic chiasma; FT, tractus frontothalamicus; HAM, nucleus hypothalamicus anterior medialis; HPM, nucleus hypothalamicus posterior medialis; LFB, lateral forebrain bundle; V, lateral forebrain ventricle; PALP, paleostriatum primitivum; P, pituitary; PO, nucleus preopticus medialis; SL, lateral septum; SO, nucleus supracopticus; SM,tractus septomesencephalicus;TOM,tractus opticus marginalia. scales represent 1 mm.

166

J. B. HUTCHISON

sharply with distance from the site of implanation until, at 800 pm, grain counts were not above background. A shell of tissue around the implant was subjected to a sustained high concentration of diffusing hormone, whereas tissue situated 1 mm or more from the implant received very little hormone. This diffusion range has been questioned by Palka et al. (1966), who implanted 6 , 7estradioL3H acetate of high specific activity into the hypothalamus of rabbits and found that radioactivity was detectable up t o 2 mm from the site of the implant. No studies have been made of the diffusion properties of solid implants of testosterone propionate, but there is n o reason to expect smaller diffusion ranges from testosterone than from estradiol implants. Because the rate of release of hormone from solid crystalline implants is proportional t o the surface area (Michael, 1965) the size of the implant is more likely to be critical in determining the concentration and diffusion range of the hormone than the molecular structure of the steroid itself. It can be argued that as the hypothalamus is a highly vascular region, solid implants of steroids break down the existing blood-brain barrier and distribute steroid t o the hypothalamic tissue in an entirely unphysiological manner. This is a perfectly valid objection. However, there is now considerable evidence that no effective blood-brain barrier exists for steroid hormones and that the whole steroid molecule enters hypothalamic cells from the diencephalic vascular system. Radioautography has been used extensively to locate estradiol-concentrating cells in the brain of the female rat (reviewed by Pfaff, 1971; Stumpf, 1971a). By using 6 , 7 e ~ t r a d i o l - l 7 & ~of H hi& specific activity in isotonic saline, injected intravenously into ovariectomized female rats, Stumpf (197 1b) has demonstrated that labeling of cells occurs over a widespread area of the ventral forebrain. E ~ t r a d i o l - ~was H retained and concentrated in nuclei of certain neurons but not in others, and not in the ependymal or glial cells which form, in many cases, a protective barrier between blood and nerve cells (Latjha and Ford, 1968). Higher densities of estradiol-concentrating neurons were to be found in the preoptic area, in the basal tuberal regions, and in the posterior hypothalamus. However, there is no doubt that in addition many areas of the central and caudal amygdala concentrate estradiol. The location of testosteroneconcentrating cells is less clear in male rats. Pfaff (1968) maintains that the regions having the highest densities of estradiol-concentrating cells tend also to accumulate radioactivity following injection of labeled testosterone. Testosteroneconcentrating cells have also been localized in the periventricular region of the preoptic-anterior hypothalamic area and especially in the medial preoptic area (McEwen et al., 1970b). Such cells have also been identified by Zigmond et al. (1972a) in the preoptic and septa1 areas, and in the midbrain nucleus intercolH licularis of castrated male chaffinches. The areas of dense t e s t ~ s t e r o n e - ~accumulation correspond well with the preoptic and anterior hypothalamic areas from which positive courtship responses were obtained by implants of testoster-

HYPOTHALAMUS AND SEXUAL BEHAVIOR

167

one propionate in castrated male doves (Fig. 1). There was also an accumulation of t e s t ~ s t e r o n e - ~in H the lateral septum, an area adjacent to the neostriatum intermediale from which fragmentary courtship patterns were obtained using implants. This suggests that the androgen-sensitive system associated with courtship behavior may extend into the neostriatum. However, further work in doves on the regional distribution of t e s t ~ s t e r o n e - ~ H uptake will have to be undertaken before this can be confirmed. Assignment of function to a particular androgen- or estrogen-concentrating area in the brain must be speculative at present and can only be done with reference to the effects of intracerebral steroid implants. However, it is becoming increasingly clear from using these implants that the hypothalamus contains two functional systems that may be quite discrete-an hypophysiotropic system associated with the feedback control of gonadotropin secretion and a steroid-sensitive system associated with the activation of behavior. The question that remains unresolved is whether these two systems overlap or occupy anatomically discrete areas. The evidence at hand suggests that there may be an anatomical separation of function for areas that concentrate estrogen in the brain of the female rat. Thus, the arcuate nucleus and basal tuberal areas of the caudal hypothalamus are associated with the feedback control of gonadotropin secretion (Lisk, 1960), whereas the preoptic area is associated with lordosis behavior (Lisk, 1962). Similarly, in the male rat, two androgen-sensitive systems may be distinguished anatomically. Johnston and Davidson (1972) have made the significant observation that implants of testosterone propionate in the medial basal hypothalamus inhibit gonadotropic secretion but have minimal activating effects on copulatory behavior, whereas implants of testosterone propionate in preoptic and anterior hypothalamic areas induce copulatory behavior but have no effect on gonadotropic secretion. Unfortunately, similar experiments have not yet been carried out in avian species. A question of major importance t o neuroendocrine studies of behavior, which remains unresolved, is whether hormone-concentrating cells in the anterior hypothalamus, detected by radioautography, correspond to the behavioral systems activated by hormonal implants. Although this question cannot be answered at present, there is fairly good anatomical agreement between those areas in the anterior hypothalamus that take up labeled hormone from the systemic circulation and the areas associated with sexual behavior that have been delimited by the hypothalamic implantation of steroids. Much neuroendocrine research is being devoted to the characterization of cellular “receptors” that “recognize” the chemical structure of the hormone. A greater part of this work has been concerned with peripheral target organs for steroid hormones, such as the seminal vesicle and prostate for androgen (Stern and Eisenfeld, 1969; Wilson and Gloyna, 1970; Mainwaring et al., 1973) and the uterus for estrogen (see review by Jensen and DeSombre, 1973), rather than

168

J. B. HUTCHISON

with the hypothalamus. For steroid hormones, these receptors are intracellular and consist of macromolecules or proteins (see reviews by McEwen et al., 1970c; Lisk, 1971; McEwen and Pfaff, 1973). The steroid hormone is thought to enter the target cell and bind t o specific receptor proteins in the cytoplasm and then to move to the cell nucleus, probably in combination with the receptor proteins. The hormone-receptor complex initiates changes in genomic function and RNA synthesis, which, in turn, modify the physiological characteristics of the target cell. Recently, considerable advances have been made in studying the binding of steroid hormones to receptors in brain cells. Both estradioL3H (Eisenfeld and Axelrod, 1965; Kato and W e e , 1967; McEwen and Pfaff, 1970) and teset al., 1970a) are concentrated in the hypothalamus and t ~ s t e r o n e - ~(McEwen H in the preoptic region within a relatively short period (1 hour) after injection into castrated male and female rats. By using cell fractionation procedures, McEwen et al. (1 972) have shown that the nuclear cell fractions of hypothalamic tissue contain the highest concentrations of radioactivity, most of which is associated with unchanged estradiol. Important steps in the further characterization of estradiol binding sites have come from observations that they are both steroid-specific and of limited capacity, that is, unlabeled hormone competitively reduces the concentration of labeled hormone in a particular brain region that is known to concentrate the hormone. Limited capacity binding of estradiol has been observed in the hypothalamus, preoptic area, and amygdala (McEwen and Pfaff, 1970). Limited capacity binding of testosterone has also been observed in the brain regions that concentrate estradiol. These areas appear t o bind less testosterone than estradiol (McEwen et al., 1970b). Soluble cytoplasmic receptors for testosterone have, however, been isolated from the hypothalamus of castrated male rats (Jouan et al., 1971). Cell fractionation studies of the uptake of testosterone have been more successul in birds. Hypothalamic cell nuclei accumulate 14 times as much t e s t ~ s t e r o n e - ~as H nuclear fractions from the cerebrum in castrated male doves and binding is of the saturable or limited capacity type, suggesting that hypothalamic cells of the male dove contain macromolecular or proteinous testosterone receptors (Zigmond et al., 1972b.). The time course of the uptake and binding of androgen to target cells within the hypothalamus may provide physiological data relevant to an understanding of the latency for the behavioral effects of hypothalamic implants of androgen; this is one aspect of the activation process that can be measured with some accuracy. Unfortunately, much of the data from birds and mammals involves spaced behavioral observations that preclude precise analysis of latency, but they do indicate a substantial delay in behavioral activation, ranging from 1 to several days after implantation of testosterone propionate. Delays of this order are also characteristic of the activation of female copulatory behavior by estrogen in rats (Lisk, 1962). In castrated male doves, the development of courtship is initially

169

HYPOTHALAMUS AND SEXUAL BEHAVIOR

similar with either intramuscular injection or intrahypothalamic implantation of testosterone propionate (Fig. 2). The majority of castrates begin to display courtship 2-3 days after the application of the hormone. However, a peak in the behavioral response occurs after 3 4 days in implanted castrates, whereas in I -SYSTEME

7P

2011

GE

H Y P O T H A L ~ KTP

BOWING

40

7~

c

80[ 60

SOLICITING NEST-

Ivv DAYS

FIG. 2. (a) Effects of intramuscular and intrahypothalamic testosterone propionate administration on castrated male doves. The results are expressed as the percentage of males (percent response) in each group that displayed courtship patterns on each day of testing. Three groups were compared: group I (N= 6), testosterone propionate (300 /&/day) was injected intramuscularly for 15 successive days and courtship was tested daily for 3 minutes/day beginning on the first day of injection; group I1 (N= l l ) , testosterone propionate implants (mean weight 38.5 pg) were implanted into the preopticanterior hypothalamic complex and courtship was tested daily for 3 minutes/day beginning on the first day after implantation (see Hutchison, 1971, for details of method); group I11 ( N =6), testosterone propionate implants were implanted into the preopticanterior hypothalamic complex and testing for courtship was initiated on the tenth day after implantation when the response to implants in group I1 had declined. The results are plotted with reference to the first day of testing for each group. (b) Effects of unilateral implants of testosterone propionate positioned in the preoptic areas of two castrated males (20/III and 19/IV, upper boxed diagram; implant weights 52.5 and 37 119)and in the anterior hypothalamus of a thud castrated male (31/V; lower boxed diagram; implant weight 28 pg). The durations of behavior displayed by each male on each postimplantation day of testing are shown.

170

J. B. HUTCHISON

systemically treated castrates levels of behavior rise progressively after treatment. This difference is presumably a consequence of the cumulative effect of high dosages of systemic testosterone: the implants by contrast probably produce only a brief pulse of testosterone. The transient nature of the courtship response is probably due to the effects of gliosis in masking diffusion from the implant (Hutchison, 1971). It seems unlikely that the binding of testosterone to receptors in the hypothalamus contributes to the delay in behavioral effects of the implants, for significant nuclear binding occurs within 1 or 2 hours of intravenous injection of labeled testosterone in castrated male doves (Zigmond er al., 1972b). Neither does it appear that the delay is due to effects of surgery masking the effects of the implant, because implantation operations have no significant effect on the courtship of castrates undergoing testosterone propionate therapy and tested daily for behavior (Hutchison, 1971). The delay in the behavioral effect of implanted steroid is more probably due t o a number of other factors that could include the rate of absorption of hormone from the implant, enzymatic cleavage of the radical from the testosterone molecule, or secondary effects on cell systems following steroid binding which are as yet a matter for speculation. One interesting possibility suggested by Zigmond (1975) is that steroid hormones induce the synthesis of enzymes involved in the metabolism of transmitter substances in cell bodies. On this view the latency between the initial cellular action of the hormone on steroid-sensitive cells and behavioral or physiological effect of the hormone may be due in part to the time taken to synthesize proteins in the neural cell bodies and transport them to the nerve terminals. This period may be as long as 6 days if the reserpine-induced stimulation of tyrosine hydroxylase in the rat hypothalamus can be taken as a valid model (Zigmond, 1975). Although the behavioral effects of testosterone propionate implants in castrated male doves are transient, the expression of courtship clearly requires the sustained action of testosterone propionate on the anterior hypothalamus. This could be demonstrated by testing for the behavioral effects of hypothalamic implants after the peak response t o implants normally occurs (Fig. 2) (Hutchison in preparation). Only a small percentage of castrates, initially tested for courtship on the tenth day after intrahypothalamic implantation with testosterone propionate, showed courtshp responses compared to the large percentage of implanted castrates tested on the first day after implantation. This indicates that there is little long-term “storage” of the effects of the implants on either the hypothalamic cells that bind testosterone or the extrahypothalamic mechanisms underlying courtship. B. METABOLISM

Apart from the binding of testosterone to cellular receptors, its metabolism may be of crucial importance for the activation of male sexual behavior. Testos-

HYPOTHALAMUS AND SEXUAL BEHAVIOR

171

terone is converted within the brain to a number of A4-reduced metabolites, notably by Sa-reductase and 3a-hydroxysteroid dehydrogenase enzymes that are probably associated with the cell endoplasmic reticulum or nucleus (Rommerts and van der Molen, 1971). Denef and McEwen (1972) have shown from in virro studies of testosterone metabolism in incubated slices of selected brain regions of male and female rats that the principal metabolite in all regions of the brain is Sa-dihydrotestosterone (Sa-androstan-17fl-ol-3-0ne). But androstenedione and androstanediol are also consistently formed. The pattern of conversion to dihydrotestosterone differs according to the brain area. Thus, in males, the highest conversion rate to dihydrotestosterone is in the midbrain, exceeding the cortex by 2 to 3 orders of magnitude. The midbrain is closely followed by the hypothalamus and thalamus. Conversion in the preoptic region, hippocampus, and cerebellum exceeds that in the cortex only slightly. There appear to be no regional differences in the formation of androstanediol and androstenedione (Denef er uf., 1973). Recently, there has been particular interest in 5a-dihydrotestosterone which has an androgenic potency equivalent to testosterone in stimulating growth of peripheral target organs such as the rat seminal vesicles and prostate (Wilson and Gloyna, 1970; Feder, 1971) and in a negative feedback effect on the release of pituitary luteinizing hormone. Significant conversion of labeled testosterone to dhydrotestosterone accompanies these stimulatory effects (Bruchovsky and Wilson, 1968; Robel et af., 1971; Bruchovsky, 1971), and dihydrotestosterone is accumulated in both cell nuclei and cytoplasm. This metabolite has also been detected in the hypothalamus and pituitary of rats (Jaffe, 1969; Stern and H Eisenfeld, 1971) and doves after systemic injection of t e s t o ~ t e r o n e - ~(Stem, 1972), but it is not clear from this work whether dihydrotestosterone is converted peripherally and transported to the brain or converted within the brain. Incubated slices of rat pituitary, amygdala, and cortex all convert testosterone to dihydrotestosterone (Kniewald et af., 1970) confirming that the Sa-reductase enzymes, essential for the conversion, are present in brain tissue. Although dihydrotestosterone is extremely effective in influencing pituitary function and growth of peripheral target organs, Scu-dihydrotestosterone propionate is far less effective than testosterone propionate in inducing male copulatory behavior in castrated male rats, whether injected systemically (Feder, 1971) or implanted into the hypothalamus (Johnston and Davidson, 1972), Intrahypothalamic implants of Sa-dihydrotestosterone or 5a-dihydrotestosterone acetate induced castrated male doves to show both aggressive and nestorientated courtship but at significantly lower levels than the courtship induced by implants of testosterone propionate (Table I). The courtship responses were similar to those evoked by testosterone implanted in the free alcohol form. Conceivably, the deficits in the behavioral effects of dihydrotestosterone implants are due to lack of the propionate radical. Experiments are being carried out to test this possibility, and, perhaps, dihydrotestosterone will be shown to be of more relevance to birds than mammals.

TABLE I DIFFERENTIAL EFFECTS OF INTRAHYPOTHALAMIC STEROID IMPLANTS ON MALE COURTSHIP a courtship chasing Response (%)' Peak duration(sec)d Latency (days)f Bowing Response(%)' Peak duration(sec)d Latency (days)f Nest-soliciting Response (%)c Peak duration (set$ Latency (days) f

TP

T

(41.2f3.4 pg)b

(5 1.0%.6 pg)b

100 37 (15-133) 2 (0-6)

57 21 (0-66) 1.5 (0-2)

36 0 (0-1 1) 1.5 (0-2) 100 64 (0-1 35) 3.5 ( 2 4 )

-

43 29 (0-150) 1.5 (0-2)

DHT (459.9~g)~

25 0 (0-14)' 5 (1-96 -

88 5 (049)h

9 (0-1 l ) e

DHTA (40*3.8p&

43 0 (0-18)e 7 (4-106 -

-

14 0 (0-1 )e 7 (7-8)e

EB (46.93.8~g)~

43 0 (0-6)' 3 (2-9) -

100 140 (20-167)h 2 (1-4)

ETP

100 33 (12-91) 2.5 (0-3) 50 5 (0-27) 5 (3-11)'

83 152 (0-170)h 1.5 (0-3)

'Implants of testosterone propionate (TP), testosterone (T), 50cdihydrotestosterone (DHT), 5Qdihydrotestosterone acetate (DHTA), and estradiol benzoate (EB) were positioned in the anterior hypothalamus of castrated male doves 30 days after castration. Exogenous testosterone propionate (ETP) was injected intramuscularly in saline (300c(g/day) into castrates beginning 30 days after castration; injections were continued for 15 successive days. Medians and ranges (in parentheses) of behavior are given. The results of each group were compared with those of the TP group; Mann Whitney U Test, two-tailed. hmplant weight f SEM. 'Percentage of males that showed a t least 5 seconds of a courtship pattern. dLongest of the daily durations of display of a pattern; the daily duration for each male is the sum of the durations of bouts of display of the pattern within the >minute observation period. 'p 0.002. fNumber of days before the first display of a pattern. gp 0.02. hp 0.05.

< < <

HYPOTHALAMUS AND SEXUAL BEHAVIOR C.

173

SPECIFICITY

The conversion of testosterone to further androgenic metabolites is, of course, not the only steroid conversion to take place in the brain. Androstenedione is aromatized to phenolic steroids, notably estrone, in the hypothalamus and limbic areas in male human fetuses (Naftolin et al., 1971) and in adult male and female rats (Naftolin et al., 1972). Although only a small percentage of available androstenedione is aromatized, probably less than 1.O%, the estrone produced may be of importance in maintaining the sexual patterns usually associated with androgen alone. This has been tested experimentally by treating castrates systemically with an androgen that cannot be aromatized to estrogen, such as Sa-dihydrotestosterone; the failure of this androgen to induce male copulatory behavior in castrated rats (McDonald e t a l . , 1970) may be due to lack of estrogen required to activate estrogen-sensitive mechanisms in the brain. The apparent lack of specificity of androgen and estrogen with respect to the activation of copulatory behavior can also be accounted for in terms of brain aromatization. Thus the female copulatory behavior, normally estrogen-dependent, can be induced by testosterone propionate implants in the posterior hypothalamus of spayed rabbits (Palka and Sawyer, 1966b); this may be due to significant proportions of the released androgen being aromatized to estrogen. Male courtship in doves can be induced by intrahypothalamic implants of estradiol as well as by testosterone, but the effects of the two hormones differ. Aromatization of testosterone to estradiol at the level of the hypothalamus may be particularly important in this species. Thus, anterior hypothalamic implants of estradiol monobenzoate not only induce higher levels of nest-soliciting, measured in terms of peak duration and total response period (Table I), but are also more effective than implants of testosterone propionate in restoring nestsoliciting to precastration levels. The difference in effect of these hormones is not restricted to nest-soliciting. Estradiol monobenzoate implants induce virtually no aggressive courtship; no bowing and very little chasing, whereas testosterone propionate implants restore aggressive courtship effectively. The results indicate a double dissociation, according to the definition of Weiskrants (1968), between the central effects of these steroids, suggesting that estradiol or some other estrogenic product of the aromatization of testosterone activates estrogensensitive mechanisms associated with nest-soliciting.

IV. VARIABLE HYPOTHALAMIC SENSITIVITY TO ANDROGEN For many years a distinction has been drawn between the phasic patterns of gonadotropic activity inducing the output of ovarian hormones in female mammals and the tonic secretion of male gonadotropin and androgen secretion.

174

J. B. HUTCHISOW

Recently, it has become clear that testicular activity is very variable. Plasma levels of luteinizing hormone in bulls show short-term fluctuations that are closely correlated to changes in plasma testosterone levels (Katongole et al., 1971). Stimuli provided by the female, which normally evoke male copulatory behavior, also elevate plasma luteinizing hormone and testosterone rapidly (Saginor and Horton, 1968; Haltmeyer and Eik-Nes, 1969). These short-term fluctuations in androgen level can be contrasted with the prolonged seasonal changes in androgen level characteristic of many temperate species of mammal such as the ferret (fitonus vulgaris) (Herbert, 1971), whose pattern of pituitary luteinizing hormone secretion and testicular secretion of androgen is regulated by seasonal changes in day length. Photoperiodic regulation of testicular activity is also characteristic of many species of birds (Menaker and Keatts, 1968; Lofts et al., 1970). Although the precise photoperiod that is stimulatory varies with species and is related to latitude of habitat and breeding season, in most cases, short day lengths of 6 hours cause testicular atrophy, whereas rapid testicular development occurs in male birds exposed to day lengths longer than 12 hours. Long photoperiods imposed on male quail (Coturnix colurnix) that have been subjected to a series of short photoperiods stimulate rapid neuroendocrine events. Thus, the surge in secretion of luteinizing hormone that induces testicular growth can occur within 1 day of exposing male quail to a long photoperiod (Follett and Farner, 1966). The question arises as to whether the sensitivity of the hypothalamus to the activating effects of androgens remains stable during long-term changes in plasma androgen level. More specifically, is behavioral responsiveness to androgen as high during periods of low endogenous androgen level as it is during periods of high androgen level? This question can be answered experimentally by studying the effects of androgen therapy under conditions of prolonged androgen deficit imposed either by gonadectomy or by manipulation of the photoperiod. A.

EFFECTS OF PROLONGED ANDROGEN DEFICIT

It has been known for some time that the type of sexual behavior induced by androgen therapy in castrated males may be qualitatively similar to the behavior shown before castration. In guinea pigs (Riss and Young, 1954) and in rats (Larsson, 1967), there is a positive correlation between precastration and posttreatment copulatory behavior induced by testosterone propionate. Similarly, in male doves, there is a strong probability that the courtship evoked by androgen therapy will be similar structurally to precastration courtship: males that displayed the aggressive components of courtship in the absence of nest-orientated courtship before castration behave similarly afterward if treated systemically or

HYPOTHALAMUS AND SEXUAL BEHAVIOR

175

implanted intrahypothalamically with testosterone propionate (Hutchison, 1970b, 1971). A further question, however, is whether sexual behavior is restored in quantitative terms to precastration levels by androgen therapy. Davidson and Bloch (1969) found that the dosage of testosterone propionate, required to resfore ejaculatory patterns to precastration levels in castrated male rats treated immediately after daily tests had shown the disappearance of sexual behavior, was significantly higher than the minimal dosage required to muinruin the display of ejaculatory behavior. At the end of a 24-day period of daily therapy and testing, the behavioral responses of castrates had stabilized such that the dosage difference was 50 pg of testosterone propionate per day. One explanation of this finding would be that the sensitivity to androgen of mechanisms underlying male ejaculatory patterns may change after castration. Such a decrease in androgen sensitivity has also been postulated for the sexual accessory glands of male rats after castration (Clar et ul., 1967; also quoted in Davidson and Bloch, 1969). Although suggestive of decreasing sensitivity to exogenous androgen, Davidson and Bloch’s experiment does not show whether the decline in effectiveness of androgen is due to (a) changes in peripheral sensory structures such as penile papillae known to be androgen-sensitive, ( b ) changes in peripheral metabolism of androgen or rate of transport of hormone to the brain, or (c) the responsiveness to androgen of brain mechanisms underlying sexual behavior after androgen has reached the brain. This problem is difficult to study with reference to male copulatory behavior in mammals, because abnormalities in the behavior of castrates may be due in part to deficiencies in sensory feedback from peripheral receptors such as penile papillae or in the functioning of spinal reflex mechanisms coordinating copulatory behavior. Both require the influence of androgen (Beach and Levinson, 1950; Hart, 1967) which in long-term castrates may not adequately reverse the effects of prolonged androgen deficits. An androgen deficit may well cause a decline in the sensitivity of peripheral sensory structures to androgen, comparable to the diminishing sensitivity of the seminal vesicles in long-term castrated rats. In many species of birds, the courtship patterns, although androgendependent, are visually and auditorily mediated and, therefore, probably depend less on the androgenic sensitization of tactile receptors. Two observations suggest that the responsiveness of brain mechanisms underlying male courtship to androgen declines after castration: first, courtship is difficult to restore to precastration levels even with dosages of testosterone propionate as large as 300 pg/day; second, in a sample of 40 castrated male doves, a proportion (29%) showed no courtship response to intramuscular testosterone propionate injected daily for 10 days, although the remainder all responded to therapy within 1 to 3

176

J. B. HUTCHISON

days. When the daily dosage was doubled (600 pg/day), all of the castrates that had previously failed to respond began to display courtship (Fig. 3) Hutchison, unpublished observations). This result suggests that a decline in responsiveness to intramuscular testosterone propionate occurs more rapidly in some individuals than others. On this view, the threshold for the activation of courtship by androgen would be higher in the group that did not respond to 300 pg of testosterone propionate per day. The significant question is whether the differential decline in responsiveness to testosterone propionate is due to differences in peripheral metabolism and distribution of hormone or to factors associated with the brain mechanisms of courtship. To answer this problem, an experiment was carried out to establish whether the responsiveness of brain mechanisms underlying male courtship behavior declines in the prolonged absence of circulatory androgen, by measuring the effectiveness of intrahypothalamic implants of testosterone propionate in restoring courtship a t different periods after castration. Males were brought in from outside aviaries in spring (April) for a preliminary period of 3 months during which they were illuminated for 13 hours/day. At the end of this period each male was tested with females to assess the courtship behavior (the testing procedure was described in detail previously; Hutchison, 1970b). In view of the differences in effect of hypothalamic implants on males that show only aggres(b)

(0)

PEAK DURATION (MEDIAN SEC) 0

PEAK DURATION (MEDIAN SEC)

I (71%) 30kgTP- C.B.NS

n (29%)300~gTP-WC,B BOWING NESTSOLICITING

I

ANDROGEN THERAPY

PRECASTRATION

FIG. 3. Differential responses to intramuscular testosterone propionate in castrated male doves. A total of 50 castrates (1) were injected for 10 successive days with testosterone propionate (TP)and tested for courtship for 3 minutes/day. The sample was separated into two groups according to whether the testosterone propionate induced the display of chasing (C), bowing (B), and nest-soliciting (NS) (group I, 71% of total sample) or had no courtshipinducing effects at all (group II,29% of total sample). On the eleventh day after initiation of therapy with 300 pg of testosterone propionate, group I1 was injected with 600 pg of testosterone propionate and tested for courtship. Injections and tests were continued for a further 9 successive days and chasing, bowing, and nestsoliciting were initiated in all cases. There were no significant differences between the peak durations of courtship induced in groups 1 and 11; there were also no differences between the peak durations of courtship induced in groups I and 11: there were no differences between the groups in precastration peak durations of courtship@). Peak duration is the longest of the daily durations of display of a pattern; the daily duration is the sum of the duration of bouts of display of a pattern within a 3-minute observation period.

HYPOTHALAMUS AND SEXUAL BEHAVIOR

177

sive courtship and on those that show full courtship consisting of both the aggressive and nest-orientated components before castration (Hutchison, 197 l), only males that showed full courtship were selected. The males were randomly assigned to four groups (I-IV) and castrated. At this stage, the photoperiod used for groups 1-111 was reduced from 13 hours/day to 8.5 hours/day, but maintained at 13 hours/day for group IV. These photoperiods approximated midwinter and spring day lengths in Cambridge. The purpose of shortening the photoperiod was to reduce gonadotropin secretion and the high plasma levels of gonadotropins thought to result from castration (Nalbandov, 1967) and thereby to prevent both regeneration of testicular tissue and any possible effects of high concentrations of gonadotropins on the mechanisms controlling courtship (Davis, 1957). In group IV, the 13-hour/day photoperiod was maintained throughout the experiment to control for any possible effects of the reduced photoperiod in group 111. All males received implants of testosterone propionate into the preoptic or anterior hypothalamic areas. Males of groups I and I1 received implants on the fifteenth and thirtieth day after castration, respectively; males of groups I11 and IV received implants on the ninetieth day after cast ration. There were clear differences between the behavioral responses of the groups. Whereas implants were highly effective in restoring courtship in 15-day castrates, their effectiveness with respect to bowing was diminished in the 30-day castrates. In the 90-day castrates maintained on a 8.5-hour/day photoperiod, implants were almost completely ineffective in initiating the display of courtship (Hutchison, 1969, 1974a,b). A percentage of 90-day castrates on a 13-hour/day photoperiod showed chasing and nest-soliciting, but only at a low level (Fig. 4). Taken together the results indicated that the behavioral effects of testosterone propionate implants in the hypothalamus were inversely related to the duration of the period between castration and implantation. The differences between the groups were not due to differences in (a) precastration behavior; ( b ) position of implant in the brain (Fig. 4) (Hutchison, in preparation); ( c ) extent of gliosis in the tissue surrounding the implant which may have slowed diffusion of hormone; (d) body weight which may have indicated gross metabolic changes in long-term castrates; and (e) the period of visual isolation experienced by the long-term castrates, since visual isolation for an equivalent period before castration had no effect on courtship. Because testosterone propionate implants, with a limited diffusion range, exert their maximum effects in inducing male courtship behavior on the preoptic and anterior hypothalamic areas (Hutchison, 1971), it is very likely that the decline in effectiveness of these implants is due to a functional change in the hypothalamus itself or in associated brain areas involved in the androgenic activation of male courtship behavior. The immediate question raised by these results is whether the focus for changes in brain function responsible for the ineffectiveness of hypothalamic implants lies within the hypothalamus itself or

178

J. B. HUTCHISON

whether it involves extrahypothalamic brain mechanisms underlying courtship. It can be argued that a prolonged androgen deficit may cause a functional change in extrahypothalamic systems, so that they either fail to respond t o the testos(b) CHASING

12 10

- 100

00

- 80

06

-60

04

- 40

d::

02 0

BOWING

@

06

2

04

,

0

z

5L"

0.2

0

0 NEST-SOL CITING

1.2

I

1 0

xx)

08

80

06

60

0.4

40

0.2

20

0

0

I FIG. 4. (a) Positions of intrahypothalamic implants of testosterone propionate in castrated doves maintained on an 8.S-hour/day photoperiod and implanted IS (group I), 30 (group 11), or 90 (group 111) days after castration, and males maintained on a 13-hour/day photoperiod implanted 90 days after castration (group IV). (0)Chasing, bowing, and nestsoliciting; (@) chasing and nest-soliciting, but no bowing displayed; (@ chasing and bowing, but no nest-soliciting displayed; (a)only chasing displayed; re) only nest-soliciting displayed; (0)no response. Behavioral terminology from Hutchison (1970a). (b) The relationship between precastration and postimplantation courtship is expressed as a restoration ratio: postimplantation peak durationlprecastration peak duration (cross-hatched bars). Medians and ranges are shown. The percentage response is the number of castrates that displayed at least 5 seconds of a courtship pattern expressed as a percentage (open bars).

HYPOTHALAMUS AND SEXUAL BEHAVIOR

179

terone-induced activity of hypothalamic cells or inhibit the response of these cells to androgen. There is, however, some justification for implicating the hypothalamus alone. Because the activating effects of testosterone on courtship are mediated by cells in this brain region, changes in the uptake and retention properties of testosterone in these cells may be responsible for the behavioral deficits in long-term castrates implanted with testosterone propionate. Is there any evidence that testosterone receptor cells of the anterior hypothalamus change their steroid-binding properties after castration? There is no conclusive answer to this question as yet although we are studying the problem experimentally. Recently, on the basis of biochemical studies in rats of an inverse relationship between peak uptake of estradioL3H and the period between ovariectomy and intravenous injection of the hormone (McCuire and Lisk, 1969), Lisk (1 97 1) has suggested that estradiol receptor molecules in the hypothalamus become inactivated in the prolonged absence of circulating estrogen. This interesting study suggests that the steroid-binding properties of the hypothalamus may change as a consequence of the prolonged absence of endogenous estrogen. Unfortunately, this result does not show conclusively that brain mechanisms were involved: the labeled estradiol was injected intravenously, and delays in the peak uptake may have been due to peripheral factors rather than to the steroidbinding properties of hypothalamic cells. However, one may suggest tentatively that one factor responsible for the ineffectiveness of testosterone implants in the anterior hypothalamus of long-term castrates may have been the increased degradation of the hypothalamic “receptor” macromolecules that bind testosterone. Very little appears to be known of the turnover of the receptor macromolecules that bind steroids. It may well be that in the absence of endogenous androgen after castration, hypothalamic-binding protein may be degraded gradually over a period of days and not be replaced. Apart from the possible effects on testosterone binding in the hypothalamus, the metabolism of testosterone may be affected by the prolonged androgen deficit. As mentioned above, the hypothalamus of castrated male doves concentrates Sa-dihydrotestosterone; a metabolite of testosterone that may play a part in mediating the effects of testosterone on target cells. Castration changes Sa-reductase activity and thereby decreases conversion of testosterone to dihydrotestosterone in peripheral target organs, such as the sebaceous gland of male hamsters (Takayasu and Adachi, 1972), and similar effects may occur in the hypothalamus of male doves. So far, no other physiological parameter of the influence of testosterone on brain cells has been studied in doves. However, studies, particularly on the cellular action of estradiol in rats (reviewed by McEwen and Pfaff, 1973), have demonstrated hormone-induced increases in RNA of the brain, protein synthesis (Salaman, 1970), and dopamine levels (Fuxe etal., 1970; Barthwal et al., 1971). Although the biochemical factors mentioned above may play an important part in the declining sensitivity of the brain to exogenous androgen, changes in

180

J. B. HUTCHISON

behavioral responsiveness t o intrahypothalamic testosterone may also be a consequence of declining auditory contact or the lack of opportunity to display courtship for a long period when endogenous androgen levels are low. B.

SENSITIZING EFFECTS OF ANDROGEN

If the ineffectiveness of hypothalamic implants of testosterone propionate in long-term castrated doves is due to degradation of hypothalamic steroid receptors, the question arises as to whether or not the sensitivity of the hypothalamic mechanisms to androgen is changed irreversibly. The data presented in Section IV,A indicate that the pulse of testosterone provided by a hypothalamic implant of testosterone propionate is insufficient to counteract the changes that have occurred in brain mechanisms mediating the effects of androgen. 1s it possible that increasing levels or a sustained high level of androgen can reverse the effects of a long-term androgen deficit and “sensitize” hypothalamic cells so that courtship can be reactivated? The formulation of this problem implies that there are two processes involved in the influence of androgen on hypothalamic mechanisms of male sexual behavior: precursov sensitization, which may be defined as a regenerative process whereby androgen reverses the effects of prolonged androgen deficit on brain mechanisms of sexual behavior to establish the preconditions for activation, which has been defined in Section I in terms of a facilitatory influence on neural mechanisms underlying pat terns of sexual behavior in animals receiving appropriate stimulation. Experiments were carried out to study the process of sensitization by treating 90-day castrates with high daily doses of testosterone propionate (300 pg/day) and comparing the behavioral effects with those in similarly treated 30-day castrates. The display of male courtship was restored in 90-day castrates, indicating that the effects of prolonged androgen deficit can be reversed. However, the latency to the initial display of courtship was 3 4 days, significantly longer than the 1-2 day latency observed in the 30-day group. Thus, there seems to be a direct relationship between the latency to the initiation of courtship and the period between castration and the onset of treatment; a result that may be understood on the hypothesis that in the long-term absence of endogenous androgen, precursory sensitization of the anterior hypothalamus by continuous influence of testosterone is necessary before the activating effects of the hormone are mediated by the brain. However, an experiment involving the use of systemic treatments of androgen cannot conclusively demonstrate a postulated change in brain sensitivity to androgen because of the possible changes in the peripheral metabolism and transport of hormone t o the brain. For a conclusive statement, high levels of androgen would have to be sustained by direct perfusion of the hormone to the hypothalamus. Unfortunately, this is not yet feasible technically. Circumstantial evidence for the precursory sensitizing effects of androgen was provided, however, by a different approach. This involved the application of a

18 1

HYPOTHALAMUS AND SEXUAL BEHAVIOR

primary implant of androgen to the anterior hypothalamus followed by a secondary implant positioned close to the primary implant after the response to the primary implant had disappeared. The prediction was made that if precursory sensitization of the hypothalamus does occur, the secondary dosage would be more effective in activating courtship than the primary dosage. In carrying out this experiment the bilateral representation of the hormone-sensitive areas (Lisk, 1967b; Barfield, 1969) could be exploited. In doves, these areas are separated by less than 1.0 mm so that an implant in one area is likely to affect cells in the adjacent areas by diffusion. Castrates were implanted with a testosterone propionate implant into the right nucleus hypothalamicus anterior medialis close to the midline, and the behavioral response to the implant assessed. A second implant was positioned close to the first implant, 12 days later, near the left nucleus hypothalamicus anterior medialis. There were no differences between the first and second implants in the number of males that responded to females or in the levels of behavior initiated. However, the latencies to initiation of bowing and nest-soliciting were significantly shorter following the second implantation than the first, and the total period of response to the second implant was significantly longer (Table 11). These results suggest that precursory sensitization TABLE I1 SENSITIZING EFFECTS OF INTRAHYPOTHALAMIC TESTOSTERONE ON MALE COURTSHIP a Chasing

I

Courtship Peak duration (sec) Total response period (days)

*

Response latency (days) Latency peak response (days)f

52 (24-97) 8.5

I1

I

33

7.5

Bowing I1

(9-80) (0-19)

(1-10)

7 (6-9)

0 (0-1) 3.5 (0-5)

(0-3) 2.5 (1-8)

0

1 (0-5) 2.5 (1-16) 4.5 (1-16)

Nest-soliciting I I1

12.5 (0-108)

98 99 (0-154) (38-166)

5 (0-7)' 0.5

5.5 (0-9)

9 (4-10)d

(0-4)C

2.5 (0-6)

0.5 (0-2)e

4 (0-5)

5.5 (0-8)

3.5 (0-7)d

"An initial implant of testosterone propionate (I) was positioned in the region of the right nucleus hypothalamicus anterior medialis of each male (N = 6) 30 days after castration. A further implant (11) was positioned 12 days later in the region of the left nucleus hypothalamicus anterior medialis as closely as possible to the previous implant. Medians and ranges (in parentheses) of behavior are given. Also the results of statistical comparisons between Groups I and I1 are included (Mann-Whitney U Test, two-tailed). bNumber of days from the first to the last display of a pattern. cp 0.002. dp 0.02. ep
< <

182

J. B. HUTCHISON

of the brain mechanisms underlying courtship may occur and that this effect may have its focus in the preoptic-anterior hypothalamic complex. However, interpretation of the sequential application of steroid hormones to the brain is subject to the reservation that testosterone from the secondary implant might have acted additively with the residual hormone from the primary implant. That the behavioral response from the primary implant had disappeared entirely when the secondary implant was positioned would argue against this. Nevertheless, it is possible that levels of hormone released from the primary implant were below threshold for courtship but sufficient to act additively with the secondary implant. Further work is being undertaken to examine this point. Davidson and Bloch (1969) and Davidson (1972) have demonstrated a doseresponse relationship for the effects of systemic androgen on the latency to the restoration of male copulatory behavior in rats. Thus, in long-term castrates, only 13 days of intramuscular testosterone propionate at a high dosage level (400 pg) was required for restoration of copulation and ejaculation; at least 21 daily doses of a lower dose level of intramuscular testosterone propionate (100 pg) was required before copulatory behavior was restored. It would appear that a more prolonged sensitization of the anterior hypothalamus is required in rats than in male doves. But, as stated above, androgenic activation of mammalian copulatory behavior is complicated by the peripheral effects of androgen on the penis. Relatively long periods of treatment with high doses of exogenous androgen may be required to “prime” peripheral sensory systems involved in mediating input to the brain. C. ENVIRONMENTAL FACTORS AND ANDROGEN ACTION

Although the effects of a prolonged deficit in endogenous androgen on the steroid-binding properties of hypothalamic cells may be a principal factor in the decline in sensitivity of the hypothalamus to implanted testosterone, environmental factors such as photoperiod or social stimuli may also play a significant role. These may well act via extrahypothalamic mechanisms to block the response of the anterior hypothalamus to implanted testosterone propionate or to inhlbit the response of the extrahypothalamic mechanisms associated with courtship behavior that normally respond to the activation of the hypothalamus by androgen. Of these environmental factors, it seemed most likely that photoperiod might have influenced the hypothalamic activation of courtship behavior, because implants of testosterone propionate were more effective in initiating the aggressive components of courtship in long-term castrates maintained on a 13-hour/day photoperiod (Hutchison, 1 9 7 4 ~ )than castrates maintained on an 8.5-hour/day photoperiod (see Section IV,A). The differences between the two groups could have been due to at least three influences of photoperiod: ( I ) the direct effects

HYPOTHALAMUS AND SEXUAL BEHAVIOR

183

of higher gonadotropin levels on the brain in the 13-hour/day group, in view of the increased aggressiveness of starlings (Davis, 1957) and possibly doves (Vowles, personal communicaton) induced by systemic treatment with luteinizing hormone; (2) the effects of low gonadotropin level in the 8.5-hourlday group on testosterone uptake in the hypothalamus, in view of the decrement in testosterone uptake by the preoptic area of the rat caused by hypophysectomy (McEwen et al., 1970a); (3) the direct effects of long photoperiod on brain mechanisms mediating aggressive behavior, in view of the observation that photoperiod is critical for the initiation of nest-building in female canaries by systemic estradiol treatment, an effect thought to be independent of gonadotropin level (Steel and Hinde, 1972). In another experiment, an attempt was made to establish whether or not photoperiod influences the activation of courtship behavior by hypothalamic implants. The rationale behind the experiment was to lower gonadotropin levels and, consequently, testicular activity by maintaining males on short day lengths for a period before castration. Then the behavioral effects of hypothalamic implants of testosterone propionate were to be assessed in the castrates with low gonadotropin levels. This experiment was based on the assumption that the elevation of gonadotropin level normally induced by the castration would be lower in males maintained on a short photoperiod (Wilson and Follett, 1974; Hinde et al., 1974). As yet there is no supporting physiological evidence for this notion involving measurement of gonadotropin levels in doves. To establish behavioral homogeneity, males that showed only aggressive courtship patterns were used and, consequently, the conclusions derived from this experiment may apply only to the activation of the aggressive components of courtship by testosterone. These males were randomly assigned to a group maintained on a long day length regime of 14 hours/day (LD group) and to a group on a short day length of 6 hours/day (SD group). After 1 month, males of both groups were again tested for courtship and then castrated. There was a pronounced photoperiodic effect on both testicular weights and courtship display: the SD males clearly showed less courtship and had smaller testes (Table 111). After a further 30 days, males of both groups received hypothalamic implants of testosterone propionate. A higher percentage of LD males displayed chasing than was observed in the SD males, and, as shown in Table 111, the LD group displayed chasing during a greater number of daily tests than the SD group. The results indicate that there were no differences between LD and SD groups in the latencies and durations of chasing. If it can be assumed that levels of gonadotropin differed immediately prior to castration and subsequently between the LD and SD groups, the overall similarity in behavioral effects of implants suggest that gonadotropin level or photoperiod have little influence on the level of aggressive courtship behavior induced by testosterone. However, there may be a direct influence of photoperiod or gonadotropin level to prolong the action of testosterone. This effect has still t o be confirmed.

TABLE 111 PHOTOPERIOD AND AGGRESSIVE COURTSHIP RESPONSES TO INTRAHYPOTHALAMIC IMPLANTS OF TESTOSTERONE' Testes wt. (gm) Group N

12

SD LD

14

Left

0.04 (0.02-0.13) 0.45 (0.09-0.74)b

Right

0.03 (0.02-0.14) 0.58 (0.14-0.81)b

Precastration peak durations (sec)

I 85 (36-1 66) 100 (36-161)

I1

% Response

21

58

(040) 104

(1-179)'

71

Postimplantation peak duration (sec)

16 (043) 14.5 (0-77)

Response latency (days)

Total response period (days)

1.5 (0-3)

2.5 (0-9)d

'Two groups of males were tested for courtship (precastration I) and then subjected t o either 6 hours of light/day (SD group) or 13 hours of light/day (LD group) for 30 days. Courtship was reassessed at the end of this period (precastration 11). Each male was castrated and received an intrahypothalamic implant of testosterone propionate 30 days after castration. Medians and ranges (in parentheses) of testes weights and behavior are given. Only aggressive courtship patterns were displayed and the behavioral data refers t o chasing alone. The results of statistical comparisons between SD and LD groups are included with the LD results (Mann-Whitney U Test, two-tailed).

< <

bp 0.002. 'p 0.02. dp <0.05.

HYPOTHALAMUS AND SEXUAL BEHAVIOR

185

The general conclusion can be drawn that in the long-term absence of endogenous androgen, caused by castration, there is a decline in the responsiveness of hypothalamic and associated brain mechanisms of male courtship behavior to testosterone in doves. The question arises as to whether a similar decline in responsiveness occurs in male doves undergoing a seasonal decline in testicular activity during autumn and winter. The results described above indicate that under laboratory conditions there is a pronounced photoperiodic effect of short day length on testicular weight. A similar, although less pronounced, decline in testicular weight occurs in males sampled during the autumn and winter (Davies, 1965). To determine whether a seasonal decline in the responsiveness of brain mechanisms of courtship occurs, males that had undergone some testicular atrophy were brought into the laboratory in November, tested for courtship, castrated, and 30 days later implanted intrahypothalamically with testosterone propionate. No courtship was shown before castration. Although the majority of castrates responded to the implants, their response consisted almost entirely of low levels of chasing; only a single male showed bowing, and nest-soliciting was not displayed (Table IV). This preliminary result suggests that there may well be a seasonal decline in the responsiveness to androgen of the brain mechanisms of male courtship behavior. There appears to be no detailed study reported in the literature that might support this hypothesis. But, an observation by Lincoln et al. (1972) is very relevant. These workers noticed that there can be a seasonal decline in responsiveness of the rutting behavior of the red deer stag (Cervus elaphus) to exogenous androgen. Thus, castrated stags, implanted with testosterone propionate, went into normal rut during the rutting season (December). However, when these stags were subsequently implanted with testosterone propionate during the season of sexual quiescence (April-June), there were visible signs of testosterone being present in the circulatory system (velvet and hard horn on the antlers), but no rutting behavior was observed. This would suggest that, although the sensitivites of peripheral androgen-sensitive structures are unchanged during the period of sexual quiescence, the threshold of activation of central mechanism of rutting behavior by androgen may be elevated.

V. HYPOTHALAMIC ANDROGEN CONCENTRATION AND THE STRUCTURE OF COURTSHIP In the preceding section, evidence that attributes changes in sexual behavior to changes in the reactivity of the brain mechanisms mediating the behavior was reviewed. A further problem concerns the degree to which changes in sexual behavior may be due to alterations in the concentration of androgen affecting the hypothalamus. There is both circumstantial and experimental evidence that the type of court-

TABLE IV SEASONAL DIFFERENCES IN THE EFFECTS OF INTRAHYPOTHALAMICIMPLANTS' % Response

Subjects

Chasing

Bowing

Peak duration Nestsoliciting

Chasing ~~

Winter males (40.7 f3.8 pg)' Summer males (41.2 k3.4 pg)b

70

17

100

36

uMedians and ranges (in parentheses) of behavior are given. hnplant weights ~ S E M . 'Statistical comparison (Mann-Whitney U Test, two-tailed);p

< 0.05.

-

100

Bowing Nest-soliciting

_______

13 (5-20)' 37 (15-133)

0 (0-3) 0 (0-11)

-

64 (0-135)

187

HYPOTHALAMUS AND SEXUAL BEHAVIOR

ship displayed by a male dove depends on the concentration of androgen in the anterior hypothalamus. Thus, there are consistent differences between the rates of decline of courtship patterns after castration (Hutchison, 1970b). The aggressive components of courtship decline within 1 to 3 days of castration, whereas the nest-orientated components may be displayed for as long as 10-15 days. Since it can be assumed that endogenous androgens are metabolized and disappear from the peripheral plasma within minutes after castration, the aggressive behavior, which disappears first, may depend more on relatively high concentrations of androgen in the hypothalamus and on the residual effects of the androgen than does nest-orientated behavior. The view that the concentration of hormone in the anterior hypothalamus is related to the type of behavior displayed can be tested by manipulating hypothalamic androgen levels directly. Testosterone can be elevated to various concentrations in the hypothalamus by the use of solid implants of differing surface areas, or the effects of testosterone on the hypothalamus can be reduced by introducing an antagonist to the action of testosterone. Using the first of these experimental approaches, the effects of three types of implants with differing surface areas were compared: ( I ) spherical hgh-diffusion implants (55-85 pg); (2) spherical medium-diffusion implants (25-55 pg); and ( 3 )low-diffusion implants (hormone contained in bore of 26-gauge stainless steel tubing). The lowdiffusion implants resulted in courtship behavior in which the nest-orientated components were restored to precastration levels, but the aggressive components were almost absent (Fig. 5) (Hutchison, 1970a, 1975). By contrast, both high- and medium-diffusion implants restored courtship display,

CHASING

BOWING

NESTSOLICITING

CHASING

BOWING

NEST SOLICITING

FIG. 5. Effects of testosterone propionate implants with different surface areas placed in the anterior hypothalamus and preoptic regions. (HD) Highdiffusion (MD) mediumdiffusion, and (LD) lowdiffusion implants. Medians and ranges are shown. (**)p<0.002. (*) p 0.05 (Mann-Whitney U Test, two-tailed).

<

188

J. B. HUTCHISON

but a larger proportion of males with high-diffusion implants displayed aggressive courtship for longer durations than males with medium-diffusion implants. In the second of these experimental approaches, the antagonistic effects of progesterone on androgen action in doves (Erickson el al., 1967; Hutchison, 1974a) was employed to reduce the effectiveness of testosterone acting on the hypothalamus. The antagonistic effects of progesterone on androgen action are presumably mediated at the level of the hypothalamus, because the behavioral effects of intrahypothalamic implants of testosterone propionate in castrated male doves are suppressed by systemic progesterone (Hutchison, 1974a). The prediction was made that as progesterone concentration in the hypothalamus increased, selectively blocking the effects of testosterone, the aggressive components of courtship would decline relative to the nest-orientated behavior. This was found to be the case in males treated with progesterone and testosterone propionate (300 pg of each hormone per day). The effects on courtship were very similar to those obtained by castration. Thus, aggressive behavior declined rapidly and disappeared within 3 to 6 days of the start of progesterone treatment (Fig. 6 ) . Nest-orientated behavior continued t o be displayed by the majority of males until the tenth to twentieth day after initial treatm'ent and then disappeared. This result suggests that the aggressive and nest-orientated components of courtship are differentially affected by progesterone. However, given sufficiently large doses of progesterone, both components of courtship disappear. Strutting, which normally occurs between bouts of aggressive and nest-orientated courtship, was not significantly affected by the combined testosterone and progesterone therapy, suggesting that the decline in the courtship components was selective and not due to the adverse effects of progesterone and the general metabolism or activity of the treated animals. Taken together, the results of the. two methods of manipulation of hypothalamic testosterone level are consistent with the hypothesis that when effective concentrations of testosterone in the anterior hypothalamus are high, both aggressive and nest-orientated behavior will be displayed, whereas when testosterone concentrations are low, aggressive behavior will be absent. It can be suggested that the anterior hypothalamus differentiates between testosterone concentrations by means of a threshold system organized so that mechanisms in the brain associated with aggressive behavior have a higher androgenic activation threshold than those associated with nest-orientated behavior. At present, it is unclear whether the focus for this threshold system lies within the hypothalamus or elsewhere in the brain. A cellular mechanism could form the basis for such a threshold system if the hypothalamus is principally involved. Steroid uptake of hypothalamic cells may be differentially sensitive to androgen concentration. Thus, certain cells may have a greater affinity for binding androgen than other androgen-sensitive cells. The linear relationship between brain tissue, including H and Pfaff, 1970) hypothalamus, and plasma uptake of t e s t ~ s t e r o n e - ~(McEwen

189

HYPOTHALAMUS AND SEXUAL BEHAVIOR

and estradiol (McEwen et al., 1972) in male and female rats indicates that a mechanism of this sort may operate physiologically, but experiments have yet to be carried out in which dosages of labeled hormone fall within physiological limits. In view of these differences in threshold of sensitivity to androgen between mechanisms of aggressive and nest-orientated courtship, the question arises as to whether the relationships between the sensitivity thresholds remain stable in the long-term absence of circulating androgen. This would appear to be the case in castrates implanted with testosterone propionate 90 days after castration, because nest-orientated courtship, which has the lower activation threshold, was elicitable from some 90-day castrates. Aggressive courtship, particularly bowing, I

I I

CHASING

BOWING

I

I

STRUTTING

I

I I I , 0'

4

(MOM TP)

, I , , 8 f

12

,

I

16

1

I

M DAY5

FIG. 6. Effects of progesterone on androgen-induced courtship activity in doves. Three groups of castrates were treated intramuscularly with 300 ccp testosterone propionate for 8 successive days. On the ninth day, group (TP +PROG) received an intramuscular injection of 300 pg of testosterone propionate and an equivalent dosage of progesterone, both dosages homogenized in saline; group (TP +%PROG) received 300 ccp of testosterone propionate and 150 pg of progesterone; group (TP h a l i n e ) received 300 /.& of testosterone propionate and saline vehicle alone. Treatments were continued for a further 20 days.

190

J. B. HUTCHISON

was virtually absent. However, of the sexually inactive males, tested when endogenous androgen levels were presumably at their lowest (Section IV,C), the majority displayed the aggressive components of courtship. There was no nestorientated courtship (see Table IV). This result would indicate that the thresholds of responsiveness of the aggressive components of courtship to androgen are lower than those for nest-orientated components in this group. The threshold relationships of these males would appear to be the reverse of the 90-day groups that had displayed complete courtship consisting of both aggressive and nestorientated components prior to castration. Is there any seasonal change in the structure of courtship of male doves? So far we have no quantified evidence for this, but males brought into the laboratory from aviaries containing heterosexual groups of birds in early spring appear to display a preponderance of aggressive courtship. By contrast, males brought in later in summer will readily display both components of courtship. This observation is consistent with the seasonal transition from aggressive to more typically sexual courtship patterns in the dhaffinch and other species of birds (Hinde, 1953, 1970). If a similar seasonal transition in behavior does occur in male doves, it can be suggested that two processes occur in the male during seasonal reproductive development. First, the lengthening photoperiod results in the gonadotropic induction of rising endogenous androgen levels, which reconstitute the steroid-binding mechanism, resulting in the activation of the brain mechanisms of courtship by testosterone. Second, there is a reversal in the thresholds of androgenic activation of the brain mechanisms underlying courtship with the result that there is a transition in the balance of male courtship from aggressive to nest-orientated responses. The validity of the hypothesis will depend on whether there is a seasonal difference in the structure of the courtship of the male that is independent of other variables such as seasonal variation in the courtship of the female.

VI.

SUMMARY AND CONCLUSIONS

By using intrahypothalamic implants of testosterone propionate, an inverse relationship can be demonstrated between the behavioral effects of the implants and the period between castration and implantation, indicating that hypothalamic or other brain mechanisms associated with the hormonal activation of male courtship behavior become less sensitive to androgen with time after castration. A direct relationship is found, however, between the duration of the postcastration period and the latency to the activation of courtship by systemic testosterone propionate therapy, suggesting that in long-term castrates androgenic sensitization of the brain occurs before the hypothalamic and associated brain areas can mediate the effects of androgen on courtship behavior. There

HYPOTHALAMUS AND SEXUAL BEHAVIOR

191

may be a direct effect of photoperiod or gonadotropin level on the activation of brain mechanisms underlying the aggressive components of courtship to androgen. Thus, the effects of intrahypothalamic implants of testosterone propionate are more prolonged with respect to the aggressive components of courtship in long-term castrates maintained on long photoperiods. By using intrahypothalamic implants of testosterone propionate, it is also possible to demonstrate that the aggressive components of courtship require higher hypothalamic concentrations of androgen for their activation than nestorientated components, suggesting that mechanisms within the hypothalamus or associated brain mechanisms underlying male courtship behavior are differentially sensitive to androgen: mechanisms associated with aggressive behavior have a higher threshold of sensitivity to androgen than those associated with nest-orientated behavior. The relationships between these thresholds of activation of the courtship patterns by androgen do not appear to change as a consequence of prolonged androgen deficit in castrates. However, a reversal of the activation thresholds does appear to occur in males that have undergone a seasonal decline in endogenous androgen level. There is now some understanding of the cellular effects of steroid hormones in the hypothalamus, particularly with regard to binding of steroids to intracellular macromolecules. Certain, as yet unspecified, cells in localized areas of the hypothalamus, appear to be targets for steroids. It is very likely that binding of androgen to steroid-specific target cells forms the initial stage of the process of the activation of male sexual behavior. If this is so, and as yet there is no direct evidence linking steroid-binding with mechanisms of behavior, then the diminishing effectiveness of intrahypothalamic implants with time after castration may be a function of the changing steroid-binding properties of cellular receptors within the hypothalamus. Similarly, the latency before reappearance of courtship in castrates treated with high systemic doses of testosterone propionate may be a function of the reconstitution of these steroid receptors. These conclusions have direct application to castrated males, but the question remains whether they can be applied to intact males undergoing a seasonal fluctuation in endogenous androgen level. Although there is only preliminary evidence at present, it is conceivable that in males subject to seasonal changes in androgen level, there may be a correlated change in the capacity of the hypothalamus to respond to the activating effects of androgen. Significant changes in the level of circulating androgen occur not only in male animals with a seasonal reproductive capacity, but also during ontogeny. Many studies have indicated that androgen is particularly important at perinatal (Goy, 1970a) and pubertal (Larsson, 1967) stages in the development of brain mechanisms underlying male sexual behavior. With the discovery of hypothalamic steroid receptors and their probable relationship with the activating effects of androgen on behavior, it may be possible to phrase questions concerning the

192

J. B. HUTCHISON

development of individual differences in the expression of adult male sexual behavior more clearly. Does the development of particular hypothalamic thresholds of sensitivity to circulating androgen determine what type of adult behavior an individual will display? In doves (Hutchison, 1970a), as in other species, individual male sexual behavior varies quantitatively. Although individual behavioral differences are probably due in part to differences in experience and learning during development, it is possible that the ways in which increasing levels of gonadal steroids influence the hypothalamus during development are critical in determining the individual’s later responsiveness to the activational effects of testosterone. Since perinatal steroid treatment of female rats has been shown to affect the development of hypothalamic steroid-binding molecules in adulthood (Flerko et al., 1969; McGuire and Lisk, 1969; McEwen and Pfaff, 1970, 1973), and because a maturational sequence can be traced in the development of hypothalamic estradiol receptors (Kato, 1972), it is possible that androgen level may be important at critical perinatal and pubertal stages of development in determining the adult activation threshold of the system, by affecting the generation of steroid receptors in the hypothalamus. This review is concerned mainly with variables that may affect the initial hypothalamic stage of the activation of brain mechanisms of sexual behavior, but, what of other brain areas? Leaving aside the anatomy of the interconnections between the hypothalamus and other brain areas, which are little understood from a functional viewpoint (Raisman and Field, 1971) but are obviously of vital importance for understanding the process of hormonal activation, two types of hypothesis can be put forward (Fig. 7). First, it is theoretically feasible, using a “wiring diagram” analogy of brain function, that mechanisms integrating patterns of sexual behavior have incorporated into them separate steroid-sensitive “hypothalamic loops.” These loops would contain separable steroid-sensitive cell populations specifically associated with each behavior pattern. The evidence to support this type of hypothesis might be that electrostimulation of the anterior hypothalamus or preoptic area induces components of sexual behavior. For instance, Roberts et al. (1967) were able to induce male opossums to mount receptive females by means of electrostimulation of the preoptic area. Mounting components of copulation have also been elicited from male rats by electrostimulation of the anterior hypothalamus (Vaughan and Fisher, 1962). In pigeons (Akerman, 1966), components of male courtship have been elicited by electrostimulation of the preoptic-anterior hypothalamic complex. It could be argued that the electrostimulation mimics in some way the activating effects of androgen on hypothalamic loops associated with each behavior pattern. The second hypothesis (Fig. 7), and in my view a more likely one, is that the hypothalamus contains a steroid-sensitive cell population that is not specifically associated with particular behavior patterns, but whose collective response to hormone concentration differentially affects the extrahypothalamic systems

193

HYPOTHALAMUS AND SEXUAL BEHAVIOR

involved in the integration of sexual behavior. These cells could provide information on hormonal type and concentration within the hypothalamus which is relayed to each extrahypothalamic system directly involved in the integration of sexual behavior. The hypothalamus does contain a multiplicity of chemoreceptors involved, for example, in osmoregulation and thermoregulation. The concept of steroid-sensitive cells capable of monitoring changes in level of circulating steroids is important not only in understanding steroid-sensitive behavioral systems but also the feedback effects of sex steroids on cells of the hypophysiotropic area in the hypothalamus. As Raisman and Field (1971) have pointed out, “to explain the versatility with which the output of specific releasing factors can be modified in response to appropriate feedback stimuli, the cells of the peri-cellular neuro-secretory system should be responsive to changes in level of circulatory target gland hormones.” In support of this view, Davidson (1 969) has suggested from experimental work that the basomedial hypothalamus or median eminence of male rats contains a receptor system that responds to increases in testosterone concentration by decreasing gonadotropin secretion,

15cll A

/

/ TIME

FIG. 7. Two possit threshold mechanisms in the activation of sexual behavior by androgen. (a) Specified, androgen-sensitive cell populations (SCI,SCz, SC3,. .,n),with different thresholds of sensitivity to hypothalamic androgen concentration, specifically associated with brain systems ( S l , Sz. S3, . . .,n) integrating separate units of behavior. (b) An unspecified, androgensensitive cell population (C) not specifically associated with brain systems ( S , , Sz, S 3 , . . .,,,), but whose output differentially affects these brain systems according to hypothalamic androgen concentration.

.

194

J. B. HUTCHISON

and to decreases in concentration of testosterone by increasing secretion of gonadotropic hormones. Separate receptors for the feedback action of high and low concentrations of hormones have also been proposed for corticoids (Ganong, 1963) and for thyroxin (Sinha and Meites, 1965). Smelik (1974) has proposed that the focus for the feedback control of ACTH, which is probably in the hypothalamus, contains “rate-sensitive” cells that respond to rising levels of corticosterone and “level-sensitive’’ cells responding to equilibrium levels of this hormone. Whether cells involved in the feedback control of steroids also play a role in the activation of sexual behavior remains to be established. The current evidence, based on differences in the localization of the effects of hypothalamic implants, suggests that the areas associated with androgenic feedback effects and sexual behavior may be separable. Acknowledgments The work reported in this paper has been supported by the Science Research Council and the Medical Research Council. It is a pleasure to acknowledge the helpful criticism of Professor Robert Hinde, F.R.S., Dr. Rosemary Hutchison, and Dr. Richard Zigmond on earlier drafts of the manuscript. I am also grateful to Mr. Les Bardon for technical help. The preparation of this paper was carried out partly while I was a Visiting Scientist at the Institute of Animal Behavior, Rutgers University. I am indebted to the late Professor D.S. Lehrman and to Professor Jay Rosenblatt for the facilities made available to me and for discussion of the work. References Akerman, B. 1966. Behavioural effects of electrical stimulation in the forebrain of the pigeon. 1. Reproductive behaviour. Behaviour 26, 323-338. Barfield, R. J. 1965. Induction of aggressive and courtship behavior by intracerebral implants of androgen in capons. Amer. Zool. 5 , 203. (Abstr.) Barfield, R. J. 1969. Activation of copulatory behavior by androgen implanted into the preoptic area of the male fowl. Horm. Behuv. 1, 37-52. Barfield, R. J. 1971. Activation of sexual and aggressive behavior by androgen implanted into the male ring dove brain. Endocrinology 89, 1470-1476. Barthwal, J. P., Gupta, T. K., Gupta, M. L., and Bhargava, K. P. 1971. Role of catecholamines in the central actions of female sex hormones. Jup. J. Phurmacol. 21, 1 4 . Beach, F. A. 1970. Coital behavior in dogs: V1. Long-term effects of castration upon mating in the male. J. Comp. Physiol. Psychol. 70(3), Part 2. Beach, F. A., and Levinson, C. 1950. Effects of androgen on the glans penis and mating behavior of castrated male rats.J. Exp. Zool. 114, 159-168 Bruchovsky, N. 1971. Comparison of the metabolites formed in rat prostate following the in vivo administration of seven natural androgens. Endocrinology 89, 1212-1 222. Bruchovsky, N., and Wilson, J. D. 1968. The conversion of testosterone to S&androstan17@l-3~neby rat prostate in vivo and in vitro. J. Biol. Chem. 243, 2012-2021. Chambers, W. F., and Howe, G. 1968. A study of estrogen-sensitive hypothalamic centers using a technique for rapid application and removal of estradiol. Proc. Soc. Exp. Biol. Med. 128, 292-294. Ciaccio, L. A., and Lisk, R. D. 1971. Hormonal control of cyclic estrus in the female hamster. Amer. J. Physiol. 221,936-942.

HYPOTHALAMUS AND SEXUAL BEHAVIOR

195

Clar, J., Massons, J. M., and Robuste, T. 1967. Accion de 10s androgenos sobre prostata y vesiculas seminales en la rata macho castrada. Rev. Espan. Fisiol. 23, 115-1 16. Davidson, J. M. 1966a. Characteristics of sex behaviour in male rats following castration. Anirn. Behav. 14, 266-272. Davidson, J. M. 1966b. Activation of the male rat’s sexual behavior by intracerebral implantation of androgen. Endocrinology 79, 783-794. Davidson, J. M. 1969. Feedback control of gonadotrophon secretin. In “Frontiers in Neuroendocrinology” (W. F. Ganong and L. Martini, eds.), pp. 345-388. Oxford Univ. Press, London and New York. Davidson, J. M. 1972. Hormones and reproductive behavior. In “Reproductive Biology” (H. Balin and S. Glasser, eds.), pp. 877418. Excerpta Med. Found., Amsterdam. Davidson, J. M., and Bloch, G. J . 1969. Neuroendocrine aspects of male reproduction. Biol. Reprod. 1,67-92. Davies, S . J. J. F. 1965. Studies of the Behaviour of Certain Streptopelia Doves and Their Hybrids. Ph.D. Thesis, Cambridge Univ., England. Davis, D. E. 1975. Aggressive behavior in starlings. Science 126,253. Denef, C . , and McEwen, B. S . 1972. Regional and sex differences in the metabolism of testosterone in the rat brain. Proc. Int. Congr. Endocrinol., 45th; Exerpta Med. Found. Int. Congr. Ser. 256, Abstr. No. 302. Denef, C., Magnus, C., and McEwen, B. S. 1973. Sex differences of testosterone metabolism in rat pituitary and brain. J. Endocrinol. 59,605421. Eisenfeld, A. J., and Axelrod, J. 1965. Selectivity of estrogen distribution in tissues. J. Pharmacol. Exp. Ther. 150,469475. Erickson, C. J., Bruder, R. H., Komisaruk, B. R., and Lehrman, D. S. 1967. Selective inhibition by progesterone of androgen-induced behavior in male ring doves (Streptopelia risoria). Endocrinology 8 1 , 3 9 4 5 . Feder, H. H. 1971. The comparative actions of testosterone propionate and 5&androstan17/3-01-3-one propionate on the reproductive behaviour, physiology and morphology of male rats. J. Endocrinol. 51, 241-252. Flerko, B., Mess, B., and Illei-Donhoffer, A. 1969. On the mechanism of androgen sterilization. Neuroendocrinology 4, 164-169. Follett, B. K., and Farner, D. S. 1966. The effect of the daily photoperiod on gonadal growth, neuro-hypophysical hormone content and neurosecretion in the hypothalamohypophysial system of the Japanese quail. (Coturnix coturnix japonica). Gen. Comp. Endocrinol. 7, 1 11-1 24. Fuxe, K., Hokfelt, T., and Nilsson, 0.1970. Castration. sex hormones, and tuberoinfundibular dopamine neurons. Int. Rev. Neurohiol. 13,93-126. Ganong, W. F. 1963. The central nervous system and the synthesis and release of adrenocorticotropic hormone. In “Advances in Neuroendocrinology” (V. Nalbandov, ed.), pp. 92-149. Univ. of Illinois Press, Urbana. Gardner, J. W., and Fisher, A. E. 1968. Induction of mating in male chicks following preoptic implantation of androgen. Physiol. Behav. 3,709-712. Gerall, A. A. 1963. An exploratory study of the effect of social isolation variables on the sexual behavior of male guinea pigs. Anim. Behav. 11,274-282. Giantonio, G., Lund, N., and Gerall, A. 1970. Effect of diencephalic and rhinencephalic lesions on the male rat’s sexual behavior. J. Comp. Physiol. Psychol. 73, 3846. Goy, R. W. 1970a. Experimental control of psychosexuality. Phil. Trans. Roy. SOC.London, Ser. B 259, 149-162. Goy, R. W. 1970b. Early hormonal influences on the development of sexual and sex-related behavior. In “The Neurosciences: Second Study Program” (F. 0. Schmitt, ed.), pp. 196-207. Rockefeller Univ. Press, New York.

196

J. B. HUTCHISON

Haltmeyer, G. L., and Elk-Nes, K. B. 1969. Plasma levels of testosterone in male rabbits following copulation. J. Reprod. Fert. 19, 273-277. Harris, G. W., and Michael, R.P. 1964. The activation of sexual behaviour by hypothalamic implants of 0estrogen.X Physiol. (London) 171, 275-301. Harris, G. W., Michael, R. P., and Scott, P. P. 1958. Neurological site of action of stilboestrol in eliciting sexual behaviour. Neurol. Basis Behav., Ciba Found. Symp., 1957 p. 236. Hart, B. L. 1967. Testosterone regulation of sexual reflexes in spinal male rats. Science 155, 1283-1284. Herbert, J . 1971. The role of the pineal gland in the control by light of the reproductive cycle of the ferret. Pineal Gland, Ciba Found. Symp. pp. 303-327. Hinde, R. A. 1953. The conflict between drives in the courtship and copulation of the Chaffinch. Behaviour 5, 1-31. Hinde, R. A. 1970. “Animal Behaviour: A Synthesis of Ethology and Comparative Psychology,” 2nd Ed. McGraw-Hill, New York. Hindc, R. A., Bell, R. Q., and Steel, E. 1963. Changes in sensitivity of the canary brood patch during the natural breeding season. Anim. B e h v . 11,553-56. Hinde, R. A., Steel, E., and Follett, B. K. 1974. Effect of photoperiod on oestrogen-induced nest-building in ovariectomized or refractory female canaries (Serinus canarius). J. Reprod. Fertil. 4 0 383-399. Hutchison, J. B. 1967. Initiation of courtship by hypothalamic implants of testosterone propionate in castrated doves (Streptopelia risoria). Nature (London) 216,59 1-592. Hutchison, J. B. 1969. Changes in hypothalamic responsiveness to testosterone in male Barbary doves (Streptopelia risoria). Nature (London) 222, 176-177. Hutchison, J . B. 1970a. Influence of gonadal hormones on the hypothalamic integration of courtship behaviour in the Barbary dove. J. Reprod. Fert., Suppl. 11, 1 5 4 1 . Hutchison, J . B. 1970b. Differential effects of testosterone and oestradiol on male courtship in Barbary doves (Streptopelia risoria). Anim. Behav. 18,41-52. Hutchison, J. B. 1971. Effects of hypothalamic implants of gonadal steroids on courtship behaviour in Barbary doves (Streptopelia risoria). J. Endocrinol. 5 0 , 9 7 4 13. Hutchison, J. B. 1974a. Differential hypothalamic sensitivity to androgen in the activation of reproductive behaviour. In “The Neurosciences: Thud Study Volume” (0.Schmidt and F. G. Worden, eds.), pp. 593-597. MIT Press, Cambridge, Massachusetts. Hutchison, J. B. 1974b. Post-castration decline in behavioural responsiveness to intrahypothalamic androgen in doves.Brain Res. 81, 169-181. Hutchison, J. B. 1974c. Effect of photoperiod o n the decline in behavioural responsiveness to intrahypothalamic androgen in doves (Streptopelia risoria). J. Endocrinol. 6 3 , 583-584. Hutchison, J. B., and Katongole, C. B. 1975. Plasma testosterone in courting and incubating male Barbary doves fstreptopelia risoria1.J. Endocrinol. 65. 275-276. Hutchison, R E. 1975. Influence of oestrogen on the initiation of nesting behaviour in female budgerigars. J. Endocrinol. 64,417-428. Jaffe, R. B. 1969. Testosterone metabolism in target tissues: hypothalamic and pituitary tissues of the adult rat and human foetus, and immature rat epiphysis. Steroids 14, 483-498. Jensen, E. V., and DeSombre, E. R. 1973. Estrogen-receptor interaction. Science 182, 126-133. Johnston, P., and Davidson, J. M. 1972. lntracerebral androgens and sexual behavior in the male rat. Horm. Behav. 3,345-357.

HYPOTHALAMUS AND SEXUAL BEHAVIOR

197

Joslyn, W. E., Feder, H. H., and Goy, R. W. 1971. Estrogen conditioning and progesterone facilitation of lordosis in guinea pigs. Physiol. Behav. 7,477482. Jouan, P., Samperez, M., Thieulant, M.L., and Mercier, L. 1971. Etude du recepteur cytoplasmique de la (1 , z - ~ H )testosterone dans l’hypophyse anterieure et I’hypothalamus du rat. J. Steroid Biochem. 2,223-236, Kato, J. 1972. Maturation of oestradiol receptors in female rat hypothalamus and the onset of puberty. Proc. Int. Congr. Endocrinol., 4th; Exerpta hied. Found Int. Congr. Ser. 256, Abstr. Kato, J., and Villee, C. A. 1967. Preferential uptake of estradiol by the anterior hypothalamus of the rat. Endocrinology 80,567-575. Katongole, C . B., Naftolin, F., and Short, R. V. 1971. Relationship between blood levels of luteinizing hormone and testosterone in bulls, and the effects of sexual stimulation. J. Endocrinol. 50,457466. Kniewald, Z., Massa, R., and Martini, L. 1970. The transformation of testosterone into dihydrotestosterone by the anterior pituitary and the hypothalamus. In Third Znternational Congress on HormonalSteroids. Excerpta Med. Int. Congr. Ser. 210,59. Komisaruk, B. R. 1967. Effects of local brain implants of progesterone on reproductive behavior of ring doves. J. Comp. Physiol. Psychol. 51, 32-36. Komisaruk, B. R., Adler, N. T., and Hutchison, J. B. 1972. Genital sensory field: enlargement by estrogen treatment in female rats. Science 178, 1295-1298. Kow, L.-M., and Pfaff, D. W. 1973. Estrogen effect on pudendal nerve receptive field size in the female rat. Anat. Rec. 175,362-363. Larsson, K. 1967. Testicular hormone and developmental changes in mating behavior of the male rat. J. Comp. Physiol. Psychol. 63, 223-230. Latjha, A., and Ford, D. H. 1968. Brain barrier systems.&ogr. Brain Res. 29, 1-559. Lehrman, D. S. 1955. The physiological basis of parental feeding behavior in the Ring Dove (Streptopelia risoria). Behaviour 7, 24 1-286. Lincoln, G. A., Guinness, F., and Short, R. V. 1972. The way in which testosterone controls the social and sexual behavior of the Red Deer Stag (Cervus elaphus). Horm. Behav. 3, 375-396. Lisk, R. D. 1960. Estrogen-sensitive centers in the hypothalamus of the rat. J. Exp. 2001. 145,197-205. Lisk, K. D. 1962. Diencephalic placement of estradiol and sexual receptivity in the female rat. Amer. J. Physiol. 203,493496. Lisk, R. D. 1967a. Sexual behavior: hormonal control. In “Neuroendocrhology” (L. Martini and W.F. Ganong, eds.), Vol. 2, pp. 197-239. Academic Press, New York. Lisk, R. D. 1967b. Neural localization for androgen activation of copulatory behavior in the male rat. Endocrinology 80,754-761. Lisk, R. D. 1971. The physiology of hormone receptors. Amer. 2001.11,755-767. Lofts, B., Follett, B. K., and Murton, R. K. 1970. Temporal changes in the pituitarygonadal axis. Mem. SOC.Endocrinol. 18,545-575. Lovari. S., and Hutchison, J. B. 1975. Behavioural transitions in the reproductive cycle of Barbary doves (Streptopelia risoria L.)Behaviour 5 3 , 126-150. McDonald, P., Beyer, C., Newton, F., Brian, B., Baker, R., Jan, H. S., Sampson, C., Kitching, P., Greenhill, R., and Pritchard, D. 1970. Failure of Scldihydrotestosterone to initiate sexual behavior in the castrated male rat. Nature (London) 227,964-965. McEwen, B. S., and Pfaff, D. W. 1970. Factors influencing sex hormone uptake by rat brain regions. 1. Effects of neonatal treatment, hypophysectomy, and competing steroid on estradiol uptake. Brain Res. 21, 1-16.

198

J. B. HUTCHISON

McEwen, B. S., and Pfaff, D. W. 1973. Chemical and physiological approaches to neuroendocrine mechanisms. Attempts at integration. In “Frontiers in Neuroendocrinology” (W. F. Ganong and L. Martini, eds.), pp. 267-335. Oxford Univ. Press, London and New York. McEwen, B. S., Pfaff, D. W., and Zigmond, R. E. 1970a. Factors influencing sex hormone uptake by rat brain regions. 11. Effects of neonatal treatment and hypophysectomy on testosterone uptake. Brain Res. 21, 17-28. McEwen, B. S., Pfaff, D. W., and Zigmond, R. E. 1970b. Factors influencing sex hormone uptake by rat brain regions. 111. Effects of competing steroids on testosterone uptake. Brain Res. 21,29-38. McEwen, B. S., Zigmond, R. E., Azmitia, E. C., and Weiss, J. M. 1970c. Steroid hormone interaction with specific brain regions. In “Biochemistry of Brain and Behavior” (R. E. Bowman and S. P. Datta, eds.), pp. 123-167. Plenum, New York. McEwen, B. S., Zigmond, R. E., and Cerlach, J. L. 1972. Sites of steroid binding and action in the brain. In “Structure and Function of the Nervous System” (G. H. Bourne, ed.), pp. 205-291. Academic Press, New York. McGill, T. E. 1965. Studies of the sexual behavior of male laboratory mice: effects of genotype, recovery of sex drive, and theory. In “Sex and Behavior” (F. A. Beach, ed.), pp. 76-88. Wiley, New York. McCuire, J. L., and Lisk, R. D. 1969. Localization of estrogen receptors in the rat hypothalamus. Neuroendocrinology 4,289-295. Mainwaring, W. I. P., Mangan, F. R., Wilce, P. A., and Milroy, E. G. P. 1973. Androgens 1.-A review of current research on the binding and mechanism of action of androgenic steroids, notably Scldihydrotestosterone. Advan. Exp. Med. Biol. 36, 197-231. Manning, A., and McCill, T. E. 1974. Neonatal androgen and sexual behavior in female house mice. Horm. Behav. 5,19-31. Menaker, M., and Keatts, H. 1968. Extraretinal light perception in the sparrow. 11. Photoperiodic stimulation of testis growth.Boc. Nut. Acad. Sci. U S . 60, 146-151. Meyer, C. C. 1972. Inhibition of precocial copulation in the domestic chick by progesterone brain implants. J. Comp. Physiol. Psychol. 79,8-12. Michael, R. P. 1961. An investigation of the sensitivity of circumscribed neurological areas to hormonal stimulation by means of the application of estrogens directly to the brain of the cat. Reg. Neurochem.; Reg. Chem., Physiol. Pharmacol. Nerv. Syst., Proc. Int. Neurochem. Symp., 4th, Varenna, Italy, I960 pp. 465480. Michael, R. P. 1965. Oestrogens in the central nervous system. Brit. Med. Bull. 2 1 , 8 7 9 0 . Naftolin, F., Ryan, K. J., and Petro, Z. 1971. Aromatization of androstenedione by the diencephalon. J. Clin. Endocrinol. Metab. 33, 368-370. Naftolin, F., Ryan, K. J., and Petro, Z. 1972. Aromatization of androstenedione by the anterior hypothalamus of adult male and female rats. Endocrinology 90.295-298. Nalbandov, A. V. 1967. Releasing factors and LH in the plasma of intact and hypophysectomised chickens. In “Reproduction in the Female Mammal” (G. E. Lamming and E. C. Amoroso, eds.), p. 243. Butterworth, London. Palka, Y.S., and Sawyer, C. H. 1966a. The effects of hypothalamic implants of ovarian steroids on oestrous behavior in rabbits. J. Physiol. (London) 185,251-269. Palka, Y. S., and Sawyer, C. H. 1966b. Induction of estrus behavior in rabbits by hypothalamic implants of testosterone. Amer. J. Physiol. 21 1,225-228. Palka, Y.S., Ramirez, V. D., and Sawyer, C. H. 1966. Distribution and biological effects of tritiated estradiol implanted in the hypothalo-hypophyseal region of female rats. Endocrinology 78,478-499. Pfaff, D. W. 1968. Autoradiographic localization of radioactivity in rat brain after injection of tritiated sex hormones. Science 161, 1355-1356.

HYPOTHALAMUS AND SEXUAL BEHAVIOR

199

Pfaff, D. W. 1971. Steroid sex hormones in the rat brain: Specificity of uptake and physiological effects. In “Steroid Hormones and Brain Function” (C. H. Sawyer and R. A. Gorski, eds.), pp. 103-1 12. Univ. of California Press, Berkeley. Phoenix, C. H., Slob, A. K., and Goy, R. W. 1973. The effects of castration and replacement therapy on sexual behavior of adult male rhesuses. J. Comp. Physiol. Psychol. 84, 472-482. Powers, J. B. 1972. Facilitation of lordosis in ovariectomized rats by intracerebral progesterone implants. Brain Rex 48, 31 1-325. Raisman, G., and Field, P. M. 1971. Anatomical considerations relevant to the interpretation of neuroendocrine events. In “Frontiers in Neuroendocrinology” (L. Martini and W. F. Ganong, eds.), pp. 3-36. Oxford Univ. Press, London and New York. Riss, W., and Young, W. C. 1954. The failure of large quantities of testosterone propionate to activate low drive male guinea pigs. Endocrinology 54, 232-235. Rivarola, M. A., Snipes, C. A., and Migeon, C. J. 1968. Concentrations of androgens in systemic plasma of rats, guinea pigs, salamanders and pigeons. Endocrinology 82, 115-1 21. Robel, P., Lasnitzki, I., and Baulieu, E. E. 1971. Hormone metabolism and action: testosterone and metabolites in prostate organ culture. Biochemie 53,8197. Roberts, W. W., Steinberg, M. L., and Means, L. W. 1967. Hypothalamic mechanisms for sexual, aggressive and other motivational behaviors in the opossum. (Didelphis virginiana). J. Comp. Physiol. Psychol. 64, 1-15. Rommerts, F. F. G., and van der Molen, H. J. 1971. Occurrence and localization of 5& steroid reductase, 3& and 17bhydroxysteroid dehydrogenases in hypothalamus and other brain tissues of the male tat. Biochim. Biophys. Acta 248,489-502. Rosenblatt, J. S. 1965. Effects of experience on sexual behavior in male cats. In “Sex and Behavior” (F. A. Beach, ed.), pp. 416-439. Wiley, New York. Rosenblatt, J. S., and Aronson, L. R. 1958. The decline of sexual behavior in male cats after castration with special reference to the role of prior sexual experience. Behaviour 12, 285-338. Ross, J., Claybaugh, C., Clemens, L. G., and Gorski, R. A. 1971. Short latency induction of estrous behavior with intracerebral gonadal hormones in ovariectomised rats. Endocrinology 89, 32-38. Saginor, M., and Horton, R. 1968. Reflex release of gonadotropin and increased plasma testosterone concentration in male rabbits during copulation. Endocrinology 82, 627430. Salaman, D. F. 1970. RNA synthesis in the rat anterior hypothalamus and pituitary: Relation to neonatal androgen and the oestrous cycle. J. Endocrinol. 48, 125-127. Sinha, D., and Meites, J. 1965. Effects of thyroidectomy and thyroxine on hypothalamic content of thyrotrophin releasing factor and pituitary content of thyrotrophin in rats. Neuroendocrinology 1,4-14. Smelik, P. G. 1974. Neuroendocrinologie de I’axe corticotrope (Brainadrenal interactions) (INSERM, ed.). Steel, E., and Hinde, R. A. 1972. The influence of photoperiod on oestrogenic induction of nest building in canaries. J. Endocrinol. 55 265-278. Stern, J. M. 1972. Androgen accumulation in hypothalamus and anterior pituitary of male ring doves; influence of steroid hormones. Gen. Comp. Endocrinol. 18,439-449. Stern, J. M., and Eisenfeld, A. J. 1969. Androgen accumulation and binding to macromolecules in seminal vesicles: Inhibition by cyproterone. Science 166,233-235. Stern, J. M., and Eisenfeld, A. J. 1971. Distribution and metabolism of ’H-testosterone in castrated male rats; effects of cyproterone, progesterone and unlabelled testosterone. Endocrinology 88, 1 1 1 7-1 125.

200

J. B. HUTCHISON

Stumpf, W. E. 1971a. Autoradiographic techniques and the localization of estrogen, androgen and glucocorticoid in the pituitary and brain. Amer. Zool. 11, 725-739. Stumpf, W. E. 1971b. Hypophysiotrophic neurons in the periventricular brain: Typography of estradiol concentrating neurons. In “Steroid Hormones and Brain Function” (C. H. Sawyer and R. A. Gorski, eds.), pp. 103-1 12. Univ. of California Press, Berekeley. Takayasu, S., and Adachi, K. 1972. The in vivo and in vitro conversion of testosterone to 17phydroxy-5ol-androstan-3~ne (dihydrotestosterone) by the sebaceous gland of hamsters. Endocrinology 9 0 , 7 3 4 0 . Tinbergen, N. 1965. Some recent studies of the evolution of sexual behaviour. In: “Sex and Behavior” (F. A. Beach, ed.), pp. 1-33, Wiley, New York. Vaughan, E., and Fisher, A. E. 1962. Male sexual behavior induced by intracranial electrical stimulation. Science 137,758-760. Weiskrants, L. 1968. In “Analysis of Behavioral Change” (L. Weiskrantz, ed.), p. 418. Harper, New York. Wilson, F. E., and Follett, B. K. 1974. Plasma and pituitary luteinizing hormone in intact and castrated tree-sparrows (Spizellu arboreu) during a photo-induffid gonadal cycle. R/l,Gen. Comp. Endocrinol. Wilson, J. S., and Gloyna, R. E. 1970. The intranuclear metabolism of testosterone in the accessory organs of reproduction. Recent Progr. Horm. Res. 26,309-330. Young, W. V. 1965. The organization of sexual behavior by hormonal action during prenatal and larval periods in Vertebrates, In “Sex and Behavior” (F. A. Beach, ed.), pp. 89-108. Wiley, New York. Zigmond, R. E. 1975. Binding, metabolism anal action of steroid hormones in the central nervous system. In “Handbook of Psychopharmacology” (L. L. Iversen, S. D. Iversen, and S. H. Snyder, eds.), Vol. 1, pp. 239-328. Plenum, New York. Zigmond, R. E., Nottebohm, F., and Pfaff, D. W. 1972a. Distribution of androgenconcentrating cells in the brain of the chaffinch. Proc. Int. Congr. Endocrinol., 4th; Exerpta Med. Found. Int. Congr. Ser. 256, Abstr. No. 340. Zigmond, R. E., Stern, J. M., and McEwen, B. S. 1972b. Retention of radio-activity in cell nuclei in the hypothalamus of the ring dove after injection of 3H-testosterone. Gen. Comp. Endocrinol. 18,450453.

Sex Hormones. Regulatory Behaviors. and Body Weight GEORGEN .WADE DEPARTMENT OF PSYCHOLOGY UNIVERSITY OF MASSACHUSETTS AMHERST. MASSACHUSETTS

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I1 Influence of Sex and Reproductive Condition . . . . . . . . . . . . . . . . A SexDifferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B Reproductive Condition . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Activating Effects of Sex Hormones: Conadectomy andReplacement Therapy in Adults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A . Males . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Females . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV Site and Mechanism of Action of Estradiol and Progesterone A SiteofAction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Mechanism of Action: A Lipostatic Hypothesis . . . . . . . . . . . . C. Interaction with Brain Monoaminergic Systems . . . . . . . . . . . . V Development of Responsiveness to Ovarian Steroids and Effects of Lactation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI Sex Differences in Neuroendocrine Regulation of Body Weight . . . . . . A. Organizing Effects of Perinatal Hormones B Sex Differences in Hypothalamic Control of Body Weight . . . . . . VII . Hormonal Effects on Taste Preferences and Dietary Self-Selection A. Hormones and Taste Preferences for Nonnutritive Solutions . . . . . B Selection of Dietary Protein . . . . . . . . . . . . . . . . . . . . . . . VIII . Hormones and Weight Regulation in Nonrat Species . . . . . . . . . . . . A. Other Rodents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. Ruminants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . C Primates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IX . Conclusions and Directions for Future Research . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

.

.

. .

.......

.

. .

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

. . .

201 203 203 204 207 207 210 215 215 225 234 237 243 243 248 253 253 258 260 261 262 263 264 267

I . INTRODUCTION Body weight is carefully regulated in most laboratory animals and in human beings by a variety of physiological and behavioral control mechanisms (see 20 I

202

GEORGE N. WADE

Balagura, 1973; Code, 1967; Hervey, 1969; Morgane, 1969). Although shortterm errors do occur in this regulation, long-term regulation is amazingly accurate, and adults of most mammalian species gain weight only very slowly with age (Hervey, 1969). It is obvious that careful weight regulation is essential to the health and well-being of these organisms. Although weight is carefully regulated, body weight set-point' is not immune to physiological and environmental fluctuations. A variety of factors, including diet palatability (Corbit and Stellar, 1964), environmental temperature (Brobeck, 1960), relative levels of various metabolic hormones (Woods ef al., 1974), and the opportunity to exercise in running wheels (Leshner, 197 l), affect body weight level in rats. Of course, several kinds of neurological damage may also raise or lower body weight set-point (Hoebel and Teitelbaum, 1966; Powley and Keesey, 1970). Recently, it has become increasingly apparent that gonadal steroids should be added to the list of factors affecting behavioral and physiological regulation of body weight (Wade, 1972). As discussed below, sex and reproductive condition often have very dramatic effects on the behaviors affecting body weight and composition in a wide variety of species, although most of the experimental work has been performed with rats. This work indicates that many fluctuations in behavior and body weight are due to fluctuations in steroid secretion by the gonads. Body weight in rats (and other species) is a product of the balance between energy intake and expenditure. If intake exceeds expenditure, calories are stored as fat; when output is chronically higher than caloric consumption, the body's energy stores are mobilized, and body weight drops. Caloric intake is simply determined by food consumption. Since various foodstuffs vary in caloric content and nutritive value, what a rat eats may be nearly as important as how much it eats so far as weight regulation is concerned. However, what and how much a rat eats are easily measured. Animals expend calories in a wide variety of ways, but the two principal sources of energy loss (and the two most affected by behavior) are exercise and heat loss to the environment. In rats, voluntary exercise is typically measured in 'Throughout this paper the term set-point is used to describe the level at which rats maintain their body weights. The set-point concept is a convenient way to describe the changes in body weight following hormonal manipulations. Strictly speaking, this term may not be an accurate description of what is actually happening, since we do not know whether rats really do regulate their weights about a neurally determined set-point. What appears to be regulation about a set-point could merely be a product of a number of forces acting upon regulatory behaviors and metabolic processes. The set-point may just be a compromise among these various forces.

HORMONES AND BODY WEIGHT

203

running wheels (hchter, 1922). Although changes in running wheel activity are often more dramatic than in other activity measures, fluctuations in various indices of activity tend to correlate very well with one another during various hormonal states (Finger, 1969; Jennings, 1971), so that wheel running is likely a reasonable (although perhaps exaggerated) measure of day-to-day exercise. Leshner (1969) provides an extensive discussion of the regulatory role of voluntary activity . Rats also vary their behavior in order t o control heat loss to the environment. If given an opportunity to do so, rats build nests to conserve heat. Thermoregulatory nest-building increases when rats are placed in a cold environment or after thyroidectomy or hypophysectomy, when physiological heat production is impaired (Richter, 1956). Cold rats can also be trained to bar-press for various heat sources, but the effects of sex hormones on operant responding for heat have not been studied as yet. Although it has been known for centuries that sex and reproductive condition affect body weight in human being and carcass size and quality in domesticated animals, systematic studies of the influence of hormones on behavioral regulation of body weight have been carried out only within the last SO-60 years. All of the body weight-affecting behaviors (food intake, dietary self-selection, voluntary exercise, and thermoregulatory behavior) are affected by gonadal steroids.

11. A.

INFLUENCE OF SEX AND REPRODUCTIVE CONDITION

SEX DIFFERENCES

I

Males of most, but not all, mammalian species are both larger and heavier than their female counterparts (Tanner, 1962). Male rats are very slightly, but significantly, heavier than females at birth and this sexual dimorphism is greatly magnified in adulthood (King, 1915). One reason for this dimorphism is that adult males eat more (Wang, 1924a) and exercise less (Hitchcock, 1925; Wang et al., 1925) than adult females. Although Kinder (1927) found no obvious sex difference in thermoregulatory nest-building, Hainsworth (1967) has reported differences in thermoregulatory saliva spreading. However, sex differences in heat loss have received little experimental attention as yet. Since it is clear that males consume more food and burn up fewer calories exercising than females, it is not surprising that they weigh more. The hormonal basis for this sex difference will be considered later in the context of the organizing effects of steroids (Section

j

VI,A).

I

Male rats do not show regular, systematic day-to-day fluctuations in sex hor-

204

GEORGE N. WADE

mone secretion as females do. Thus, it is not surprising that males d o not exhibit any noncircadian, rhythmic fluctuations in eating (Wang, 1924a), activity (Slonaker, 1924b), or nest-building (Kinder, 1927). There are some nonrhythmic changes in these behaviors in males, but these are not likely related to blood hormone levels. B.

REPRODUCTIVE CONDITION

In female rats, there are very dramatic changes in ovarian hormone secretion during the various reproductive states such as puberty, estrous cycles, pregnancy, lactation, and pseudopregnancy. These fluctuations in steroid secretion are accompanied by equally dramatic changes in behavior and body weight (Table I). Rhythmic changes in behavior are obvious during estrous cycles. At proestrus,2 activity increases while food intake and nest-building decrease (Kinder, 1927; Slonaker, 1924a; Wang, 1923). Since energy expenditure (wheel-running activity and heat loss) increases and intake decreases, proestrous rats lose weight. TABLE I INFLUENCE OF HORMONAL STATUS ON FOOD INTAKE, THERMOREGULATORY NEST-BUILDING, VOLUNTARY EXERCISE, RESPONSIVENESS TO NONNUTRITIVE SACCHARIN AND QUININE SOLUTIONS, AND SELECTION OF DIETARY PROTEIN IN ALBINO RATS" Reproductive conditionb

Food intake

Nestbuilding

Voluntary exercise

Taste responsiveness

High High Moderate Low Low Low Low Low Low Low Low

High HighC High Low Low Low Low Low Low

Low

Moderate

Low

High

Proestrous 9 Spayed 9 + EB Diestrous Spayed Spayed + P Pregnant 9 Pseudopregnant $ I Lactating 9 Prepubertal Prepubertal + EB Prepubertal + P

Low Low Moderate High High High High Very high High High High

Low

Male

Moderate

Moderate

8

9

f

Moderate High High High Very high High

Protein intake

Low High High High High High

nSee text for references. bEB = estradiol benzoate; P = progesterane. CSmaUquantities of progesterone may be necessary. 21n the litcrature, confusion abounds as to how the various stages of the estrous cycle should be labeled. In this paper, proestrus is defined as the period of darkness during which estrous behavior occurs as well as the light period immediately preceding estrous behavior. Estrus refers to the 24-hour period following proestrus; the other days of the cycle are referred to as diestrus.

HORMONES AND BODY WEIGHT

205

The increase in wheel running is especially striking. In proestrous females running scores of 15,000 to 20,000 revolutions per 24 hours are typical, and Slonaker (1 924a) reported that one proestrous female ran 44,640 revolutions (over 38 miles) in a day! These behaviors are highly correlated with the occurrence of estrous behavior. Changes in eating, activity, and nest-building occur almost exclusively during the 8-1 2 hours when females are sexually receptive (Kinder, 1927; Slonaker, 1924a; Wang, 1923), which suggests a common endocrine basis for all of the behavioral changes. At diestrus this behavioral pattern is reversed. Eating and nest-building increase (Kinder, 1927; Slonaker, 1924a; Wang, 1923), wheel-running activity drops sharply (Slonaker, 1924b; Wang, 1923), and the females gain weight. As Kennedy and Mitra (1963a) have pointed out, female rats do not match energy intake and expenditure over short periods of time, but they do match over an estrous cycle. Thus, “the rat oscillates about a state of energy balance rather than preserving it all the time” (Kennedy and Mitra, 1963a). In addition to the changes in nest-building behavior, the body temperature of female rats fluctuates with the estrous cycle. Earlier investigators reported that body temperature was lowest at proestrus in cycling rats (Brobeck et al., 1947; McLean and Coleman, 1971). We have been unable to replicate this finding. In fact, our own data indicate that there appears to be a significant increase in the colonic temperatures of female rats at proestrus (Fig. 1) (Marrone el al., 1974). At proestrus, temperatures averaged about 0.5’ C higher than during diestrus; estrous temperatures were about midway between proestrous and diestrous

PM AM

FIG. 1. Effects of estrous cycles and gonadectomy on colonic temperatures of rats. Temperatures of intact females were measured both 2 hours before Iights-out (AM) and at the middle of the dark period (PM). (From Marrone et d.,1974.)

206

GEORGE N. WADE

values. At present we are unable to explain the apparently contradictory findings reported by the various laboratories. It is not likely that the proestrous hyperthermia is entirely due to increased locomotor activity (although it may be a contributing factor), since there is not a good correlation between steroid treatments that increase body temperature and those that stimulate running wheel activity in ovariectomized rats (see Section 111,B). Finally, proestrous hyperthermia may explain the drop in nestbuilding seen at this time. If the rats are hyperthermic (because of hormonal changes), a drop in nest-building would allow the rats to lose some of this heat. During pregnancy and pseudopregnancy there is an exaggeration of the behaviors seen at diestrus (Table I). Immediately following mating, there is a sharp drop in voluntary exercise, and eating and nest-building rise (Kinder, 1927; Slonaker, 1924b; Wang, 1923, 1924b), causing an increased weight gain. These behavioral changes persist throughout lactation and return to premating levels only after pups are weaned and estrous cycles resume. If rats become pseudopregnant, rather than pregnant, the diminished activity and elevated eating and nest-building last approximately 2 weeks-until estrous cycles resume. It is likely that the enhanced nest-building during pregnancy and pseudopregnancy is thermoregulatory, rather than maternal, in nature. Raising the environmental temperature during pregnancy or pseudopregnancy prevents the rise in nest-building, but these females do build maternal nests immediately following parturition (Kinder, 1927). Brobeck el al. (1947) reported an increase in body temperature during pseudopregnancy. Behaviorally, prepubertal females are quite similar to pregnant, pseudopregnant, or lactating adults (Table I). If food intake is computed on a body weight basis, weanling females eat nearly twice as much as adults-comparable to hypothalamic hyperphagia (Kennedy, 1957). In addition, weanlings are almost completely inactive in running wheels (Slonaker, 1924a; Wang, 1923). Prepubertal females are physically capable of running in an activity wheel, since they do run when food-deprived (Gentry, unpublished observations; Kennedy, 1964); they simply choose not t o run. Finally, prepubertal females build larger, more elaborate nests than adults d o (Kinder, 1927). Around the time of puberty there is a sharp increase in locomotor activity (Slonaker, 1924a; Wang, 1923), nest-building drops gradually (Kinder, 1927), there is a relative decrease in food intake (Wade and Zucker, 1970a), and weight gain decreases. To summarize, proestrous rats expend more calories than they consume and lose weight. This pattern is reversed at diestrus, females gain weight, and the female rat oscillates around a state of energy balance during estrous cycles. During pregnancy, pseudopregnancy, lactation, and preceding puberty, the &estrous pattern of behaviors is exaggerated, and there is a rapid weight gain.

207

HORMONES AND BODY WEIGHT

111.

ACTIVATING EFFECTS OF SEX HORMONES:GONADECTOMY AND REPLACEMENT THERAPY IN ADULTS

These early descriptive studies of the 1920s strongly suggested a causal relation between sex hormones and the behaviors affecting body weight, but because none of the sex steroids had been chemically isolated, identified, and synthesized at that time, it was not possible to test this proposition. Since that time pure sex steroids have become readily available, and it has been a relatively simple, if time-consuming, task to examine the effects of various sex steroids on behavioral regulation of body weight. A.

MALES

Castration of adult male rats causes decreases in food intake, weight gain, and locomotor activity (Fig. 2) (Hoskins, 1925; Kakolewski et al., 1968; Wang et ul., 1925). Treatment of adult castrates with low doses of testosterone propionate increases food intake and weight gain to levels not different from those of intact males, but treatment with very high testosterone doses (more than 1 mglday) reduces eating and weight gain below that of untreated castrates (Hervey and

SUIGERV

qo+...o...o oop..o...n...Q'" oop-o...n...Q'" Q,,,o..O"'O"" 0'

,~.;u:.,g::..,....-'

1

-6

I

,....,,,**...** ....W*"*".*

o . . . . . . * . . . . . *

9 I

0

,..o-o

R.0"

I

6

1

I

12

I

18

I

I

24

L

30

I

I

36

I

42

I

~

40

I

54

I

I

I

~

~

~

60

1 IME (days) FIG. 2. Effects of gonadectomy on food intake and body weight of male an and .d female rats. Gonadectomies or sham operations were performed as indicated by the arrow. (From Wade, Marrone, Gentry, and Roy, unpublished data, 1974.)

~

208

GEORGE N. WADE

Hutchison, 1973; Rubimtein and Solomon, 1940, 1941; Slob, 1972). Although testosterone propionate restores precastration levels of activity in castrated males (Asdell et al., 1962; Stern and Murphy, 1971), it is not unlikely that it is a metabolite of testosterone that is actually stimulating wheel-running. In some androgendependent peripheral tissues, Sa-androstan-l7fl-o1-3-one (dihydrotestosterone), a metabolite of testosterone, is the physiologically active androgen (Williams-Ashman and Reddi, 1971), and it has been suggested that testosterone must be aromatized to an estrogen by brain tissues in order to activate male copulatory behavior (McDonald et al., 1970). The rat hypothalamus does contain the enzymes necessary to aromatize androgens (Naftolin et al., 1972). There is some suggestion that estrogens can very effectively stimulate running in castrated male rats W c h t e r and Hartman, 1934; Wang et al., 1925), and Asdell et al. (1962) claimed that estradiol benzoate was more effective in stimulating male running activity than testosterone propionate was. These data raised the possibility that testicular androgens were being aromatized by the brain prior to stimulating activity. Roy (Roy and Wade, 1975a) has attempted to test this hypothesis by comparing the effectiveness of estradiol benzoate, testosterone propionate, and the nonaromatizable androgen, dihydrotestosterone propionate (Ryan, 1960), in activating voluntary exercise in castrated male rats. He has also examined the effects of the nonsteroidal antiestrogen, MER-25, on estradiol- and testosterone-induced running. It is clear from this work that 1Opg estradiol benzoate/day was far more effective than 1 mg testosterone propionate/day, and 1 mg dihydrotestosterone propionatelday was completely ineffective in stimulating activity (Fig. 3). It was also found that an antiestrogen, MER-25 (10 mglday), significantly attenuated the activity induced by either estradiol benzoate or testosterone propionate (Fig. 4). Because (a) estradiol is more effective than testosterone, ( b) dihydrotestosterone, a nonaromatizable androgen, had no effect, (c) MER-25, an antiestrogen, significantly reduced testosterone-induced activity, and (d) cyproterone acetate, an antiandrogen, had n o inhibitory effect on androgen-induced running (Stern and Murphy, 1971), it is very likely that aromatization of testosterone plays a significant role in the activation of voluntary exercise by the testes. However, although this principle may be valid for exercise and copulatory behavior in male rats, there are some doubts as to its applicability to other species (Feder et al., 1974a). Whereas it is apparent that testosterone therapy reverses the decreases in activity, eating, and weight gain that follow castration, it is not obvious that all of these effects are actually mediated by testosterone itself. It is likely that the androgen must be aromatized before stimulating running wheel activity, but estrogens probably d o not play an important role in the stimulation of eating and body weight by testosterone. In fact, aromatization may counteract the effects of testosterone on eating and body weight.

209

HORMONES AND BODY WEIGHT

0

EB

LI

TP DHTP OIL

TWO-DAY MEANS

FIG. 3. Effect of estradiol benzoate (EB), testosterone propionate (TP), dihydrotestosterone propionate (DHTP), and sesame oil vehicle on running wheel activity in castrated male rats. (From Roy and Wade, 1975a.)

n

0 x

$ 6 -

h

z

g

5 -

o

+ MER-25

EB EB

A 3

0

2 4E

0

i

TP

A TP

v)

+ MER- 2 5

0

OIL

>

h

t

2 3I0

a

d2-

A A

gz

0

W

z

I -

z

3 E

0

1 1

, ,

2

4

1 1 6

~ ~ 8 TWO-DAY

1 1 10

~ ~

MEANS

1 1 I2

~ ~ 14

1 1 16

~ ~

iof 1011 10 day1

~ ~

FIG. 4. Effect of estradiol benzoate (EB) and testosterone propionate (TP), alone or in combination with the antiestrogen, MER-25,o n running wheel activity in castrated male rats. (From Roy and Wade, 1975a.l

~ ~

" "

210

GEORGE N. WADE

It was noted above that eating and body weight are stimulated more by low doses of testosterone propionate than by high doses. Thomas Gentry has replicated and extended these findings. A low dose of testosterone propionate (200 pglday) stimulated both eating and weight gain in castrated male rats, whereas higher doses (1 or 2 mg/day) had no effect on eating and actually reduced weight gain (particularly during prolonged treatments-2-6 weeks). These data suggest that with the higher doses of testosterone propionate significant amounts of the androgen are being aromatized, and i t is the estrogens that may be causing the weight loss. To test this possibility, Gentry treated castrated male rats with four doses (200 pg, 1 mg, 2 mg, or 20mglday) of the nonaromatizable androgen, dihydrotestosterone propionate, and measured eating and weight gain. None of the four doses of dihydrotestosterone propionate was as effective as 200 pg of testosterone propionate in stimulating eating and weight gain in castrated males, but none decreased body weight. Finally, Gentry found that the weight-reducing actions of high doses (1 mg/day) of testosterone propionate were prevented by concurrent injections of progesterone (5 mg/day). Castrated males treated with lmg testosterone propionate -!- 5 mg progesterone gained nearly as much weight as those treated with 200 pg testosterone propionate alone. In female rats progesterone effectively attenuates the weightdepressing effects of estradiol benzoate (Section 111,B). Of course, progesterone alone has n o effect on body weight in castrated male rats. Unlike testosterone, the four doses of dihydrotestosterone propionate were equally effective in stimulating eating and weight gain. These data are certainly consistent with our hypothesis that with very high doses of testosterone significant amounts of aromatization occur and counteract the anabolic actions of testosterone. They also indicate that testosterone does not have to be reduced to dihydrotestosterone t o stimulate weight gain (Gentry and Wade, 1975). B.

FEMALES

The effects of gonadal hormones on behavioral regulation of body weight are more striking and somewhat more complicated in female rats than in males. Ovariectomy of adult female rats increases food intake and nest-building, decreases voluntary exercise, and markedly accelerates weight gain (Fig. 2) (Kakolewski etal., 1968; Rosenblatt, 1967; Stotsenburg, 1913; Wang, 1923). Note that these behavioral changes are qualitatively and quantitatively similar to those occurring during pregnancy, pseudopregnancy, lactation, and preceding puberty (see Table I). These data suggest that so far as weight-regulating behaviors are concerned noncycling females are functionally ovariectomized (Wade, 1972). Withdrawal of ovarian estradiol is probably responsible for the behavioral changes following spaying. Daily treatment with physiological doses of estradiol

HORMONES AND BODY WEIGHT

21 1

benzoate is sufficient to restore preovariectomy levels of eating, voluntary exercise, and body weight (Stern and Murphy, 1972; Tarttelin and Gorski, 1973; Wade, 1975; Zucker, 1969, 1972). Estradiol increases activity and decreases eating and body weight (see Table I). Although the constant daily doses of estradiol benzoate stimulate activity and inhibit appetite, there is no noncircadian cyclic fluctuation in these behaviors, suggesting that the cyclicity observed in intact females is due to rhythmic changes in steroid secretion, rather than to some endogenous cyclicity in responsiveness to steroids (Gerall e t al., 1973). Attempts to induce cyclic behavior with exogenous steroids have met with mixed success (Kennedy, 1964; Tarttelin and Gorski, 1973). Treatment of ovariectomized rats with a wide range of doses of progesterone (0.5-10 mdday) has no effect on eating, activity, o r body weight, but it does depress activity and stimulate eating and weight gain when given to intact female rats (Galletti and Hopper, 1964; Hervey and Hervey, 1966; Roberts et al., 1972; Rodier, 1971; Ross and Zucker, 1974; Wade, 1975;Zucker, 1969). In addition, progesterone given to ovariectomized, estradiol-treated rats can attenuate or completely block the actions of estradiol on behavior and body weight in a dose-dependent fashion (Roberts etal., 1972; Rodier, 1971; Ross and Zucker, 1974; Wade, 1975; Zucker, 1969). Thus, the following data have led us (Wade, 1972; Wade and Zucker, 1969a; Zucker, 1972) to hypothesize that estradiol is the principal ovarian steroid affecting behavioral regulation of body weight in rats and that the principal role of progesterone is simply to attenuate the actions of estradiol: ( I ) estradiol benzoate alone reverses the ovariectomy-induced changes in eating, activity, and body weight, whereas progesterone has no effect on any of these measures in ovariectomized rats; (2) progesterone inhibits activity and stimulates eating and weight gain in intact or spayed, estrogen-treated rats; (3)progesterone and ovariectomy cause quantitatively and qualitatively similar changes in behavior and in body weight and composition (Galletti and Hopper, 1964; Hervey and Hervey, 1966; Rodier, 1971 ;Wade, unpublished data); (4) the effects of progesterone and ovariectomy are not additive, suggesting a common mode of action (Galletti and Hopper, 1964; Hervey and Hervey, 1966). These data suggest that high plasma progesterone titers functionally ovariectomize female rats by inhibiting the actions of estradiol in the brain and perhaps by inhibiting estradiol secretion in intact females. These two possible modes of action are not mutually exclusive, of course. Antiestrogenic actions of progesterone on peripheral tissues and estrous behavior are welldocumented in rats (e.g., Courrier, 1951; Powers and Zucker, 1969; Rothchild, 1965). It has been suggested that high plasma progesterone titers may competitively inhibit uptake of tritiated estradiol by rat brain tissues (Anderson and Greenwald, 1969; Lisk, 1974), although this effect is at least somewhat difficult t o demonstrate (Wade, unpublished data) and awaits further clarification.

212

GEORGE N.WADE

The fluctuations in regulatory behaviors, body weight, and ovarian steroid secretion during estrous cycles, pregnancy, and pseudopregnancy are consistent with our hypothesis that levels of food intake and voluntary exercise are determined by plasma estradiol-to-progesterone ratios (estradiol availability). Blood estradiol levels peak on the morning of proestrus (Hori et al., 1968; Yoshinaga etal., 1969), just before activity peaks and eating drops. Ter Haar (1972) has shown a close correspondence between blood estradiol levels and hour-to-hour changes in food consumption during the estrous cycle. At diestrus, estradiol-toprogesterone ratios are much lower (Hori et al., 1968; Hashimoto et al., 1968; Uchida el al., 1969), activity declines, and eating and body weight rise. This, of course, raises a question as to the role of the preovulatory progesterone peak (Feder et al., 1968) in the control of these regulatory behaviors. Stem and Zwick (1972) have shown that ovariectomy of cycling females just after the proestrous estradiol peak, but prior t o the preovulatory progesterone surge, does not alter their activity peak. These data suggest that estradiol, but not progesterone, is essential for the surge in running. This dissociates hormonal control of running wheel activity and estrous behavior, since preovulatory progesterone is essential for the occurrence of sexual receptivity (Powers, 1970). Similar experiments have not been performed with eating behavior. It is possible that the progesterone peak contributes to the decrease in running and the increase in eating on the day following proestrus. During pregnancy and pseudopregnancy, ovarian estradiol secretion is very low, plasma progesterone titers are elevated (Hashimoto et al., 1968; Yoshinaga etal., 1969), and levels of eating and exercise are very similar to those after ovariectomy (Table I). We have suggested previously that these extremely low estradiol-to-progesterone ratios may cause a functional ovariectomy so far as food intake and locomotor activity are concerned (Wade and Zucker, 1969a): the high progesterone levels may be sufficient to block completely the actions of the small amounts of circulating estradiol. Although estradiol and progesterone have rather striking effects on regulatory behaviors and body weight in rats and these measures are highly correlated with estradiol availability, this does not necessarily mean that these are actually the active steroids in target tissues. Some of our uptake work with radioactive steroids suggests that both estradiol and progesterone are metabolized to other compounds in the brain. After injection of ovariectomized rats with tritiated estradiol, substantial quantities of estrone are found in the brain in addition to estradiol (Feder et al., 1974b), but little was known about the effects of estrone on energy balance in rats. Similarly, rat brain extensively metabolizes progesterone. Just 4 hours after subcutaneous injection of tritiated progesterone, only about 25% of the brain radioactivity remains as unmetabolized progesterone (Wade et al., 1973). Karavolas and Herf (1971) have demonstrated that rat hypothalamic tissue reduces progesterone to 5a-pregnane-3,20-dione and other Sa-

213

HORMONES AND BODY WEIGHT

reduced metabolites in vitro. Again, nothing was known about the effects of these metabolites on eating and body weight. Recently, it has been shown that both estradiol and progesterone are more effective than their principal metabolites in affecting eating and body weight (Wade, 1975). Two micrograms of estradiol benzoate per day is more effective than 20 pg estrone benzoate in depressing food intake and body weight in ovariectomized rats (Fig. 5). Similarly, while 0.5 mg progesterone/day significantly increased eating and body weight in ovariectomized, estradiol-treated rats, 1 mg/day of 5a-pregnane-3P-ol-20-oneand Sa-pregnane3,20dione were completely without effect. Therefore, it is unlikely that either estradiol or progesterone must be converted to an active metabolite in the target tissues. Perhaps the significance of the estradiol and progesterone metabolism in the brain is simply that the metabolites are not active forms and have little effect on behavior (Wade, 1975). Our knowledge of the activational effects of sex steroids on thermoregulation is amazingly sparse considering the widespread use of basal body temperatures as a means of detecting ovulation in women. I am aware of no published studies on the effects of estradiol or progesterone administration on thermoregulatory behavior in rats, although it has been reported that nest-building increases after ovariectomy (Rosenblatt, 1967). In addition, we have recently found that gonadectomy causes a significant decrease in colonic temperatures of rats of both sexes (Fig. 1) (Marrone et ul., 1974). Work done primarily in human beings has led to the widely accepted hypothesis that progesterone is thermogenic (for reviews, see Kappas and Palmer, 1963; Rothchild, 1969). However, as Rothchild (1969) points out, labeling proges-

460 440

-

^o

420

Cl

-

Oil

g

380

2 0

*---- Estronr *-o-*

400

360

*

Estrodlol

340 2

1

1

1

1

1

1

1

1

4

6

8

I0

12

14

16

I8

1

1

20

22

1

1

24 26

I

1

28

30

SUCCESSIVE 3-DAY PERIODS

FIG. 5. Food intake (top) and body weight (bottom) of ovariectomized rats injected with estradiol benzoate (2 Wday), estrone benzoate, or sesame oil vehicle from day 7 through day 33. Estrone dose was increased from 2 to 20 &day as indicated by arrow. (From Wade, 1975.)

214

GEORGE N.WADE

terone “thermogenic” may be somewhat premature, since there is little evidence suggesting just how progesterone raises body temperature. It could just as well be acting to decrease heat loss rather than increasing heat production. In any event, the work on hormonal control of body temperature in women is less than definitive to say the least. One widely cited abstract suggests that progesterone raises and that “estrogen” lowers body temperature in ovariectomized rats, but n o mention is made of the magnitude of the temperature changes or of the steriod doses necessary to induce them (Nieburgs and Greenblatt, 1948). The temperature-raising effect of progesterone in ovariectomized rats has been confirmed (Freeman et al., 1970; Marrone et al., 1974). As little as 0.5 mg progesterone/day significantly raised body temperature in ovariectomized rats. One milligram or 5 mg progesterone/ day were roughly equivalent in raising rectal temperature (mean increase = 0.8OC) and were approximately twice as effective as 0.5 mg (mean increase = 0.4OC). The effects of all three doses were transient, and rectal termperatures returned to baseline within 2 weeks, in spite of continued progesterone treatment (Marrone etal., 1974). However, it is unlikely that progesterone alone accounts for the fluctuations in colonic temperature during the estrous cycle. Marrone has found that treatment of ovariectomized rats with three different doses of estradiol benzoate (1, 3, or 10 pg/day) also raises colonic temperature in a dose-dependent fashion (Marrone etal., 1974a). However, unlike progesterone, the lower dose of estradiol benzoate (1 pg) was substantially more effective in raising colonic temperature than were the higher doses. This finding contrasts with the earlier suggestions that estrogens lower body temperature in rabbits (Brown et al., 1970) and rats (Nieburgs and Greenblatt, 1948). However, these data indicating that low (and probably “physiological”) doses of both estradiol and progesterone raise colonic temperature in ovariectomized rats are certainly consistent with our finding that colonic temperature rises at proestrus (Section 11, A), just after blood estradiol and progesterone titers peak (Hashimoto et al., 1968; Yoshinaga et al., 1969). As mentioned previously, it is unlikely that the proestrous activity peak is totally responsible for the proestrous hyperthermia, although it may be a contributing factor. Progesterone, which has no effect on locomotor activity in ovariectomized rats (Rodier, 1971), significantly raises body temperature. In addition, high doses of estradiol benzoate (10 pg/day) are more effective than lower doses (1 pg/day) in stimulating running wheel activity in ovariectomized rats, but the lower dose was more effective in raising colonic temperature. It is clear that a great deal of additional research is needed before we can specify the hormonal factors affecting thermoregulation in rats. We have found it difficult even to replicate some of the older work in this field (Marrone et al.,

HORMONES AND BODY WEIGHT

215

1974). Perhaps hormonal effects on body temperature require very narrowly defined environmental conditions to be evident, which may help to account for the irritating lack of progress in this area. Note that during the fluctuations in estradiol availability, energy intake and expenditure do not vary independently of one another. Rather, food intake consistently is inversely related t o activity and heat loss (see Table I). During times of high estradiol availability, energy expenditure exceeds intake, and this pattern is reversed during low estradiol availability. The consistent, although negatively related, coordination among energy balance-regulating behaviors has led several authors to suggest that similar or identical neuroendocrine mechanisms may control these several behaviors (Brobeck et ul., 1947; Kinder, 1927; Rothchild, 1967). In fact, in an extremely insightful review of a widely divergent literature, Rothchild (1967) has suggested that a single neurological change accounts for a wide variety of steroid-induced changes in reproductive physiology and behavior. Briefly, he has hypothesized that progesterone acts to inhibit the ventromedial hypothalamus, disinhibiting the lateral hypothalamus. The disinhibition of the lateral hypothalamus, in turn, causes an increase in appetite and prolactin secretion and inhibits heat loss, sexual receptivity, locomotor activity, and maximal-rate luteinizing hormone secretion. Although a number of data that do not seem consistent with this model have subsequently appeared, this was a remarkable attempt to integrate a wide variety of hormonerelated phenomena. Unfortunately, as is shown in the following, it is not possible to attribute all of these functions to a single neural system.

IV. SITE AND MECHANISM OF ACTION OF ESTRADIOL AND PROGESTERONE A.

SITE OF ACTION

In 1947, Brobeck et al. noted the rhythmic fluctuations in eating, exercise, and body temperature during estrous cycles and pseudopregnancy and suggested that one might consider the possibility that one or more of the hormones in question act upon the hypothalamic cells responsible for this regulation in such a way as to direct the overall energy exchange now towards the side of energy storage, now towards the side of energy expenditure. Whether this is indeed the case cannot be decided at the present time, but the problem appears to be one which lends itself to further experimental study.

The Brobeck et al. (1947) hypothesis that sex hormones act directly on hypothalamic neurons to alter regulation of energy balance seems to have been

216

GEORGE N. WADE

well ahead of its time, for, although their idea did, indeed, lend itself to further study, 16 years elapsed until additional data were published on this problem. In an exciting series of papers, Kennedy and Mitra (1963a,b,c; Kennedy, 1964) attempted to integrate much of the work suggesting that similar hypothalamic sites controlled gonadotropin release, eating behavior, estrous behavior, and voluntary exercise. Kennedy and Mitra reported that they could abolish some of the estrogenic effects on behavior with electrolytic lesions of specific hypothalamic loci. Ventromedial hypothalamic lesions seemed to abolish voluntary exercise ; neither estradiol benzoate, amphetamine, nor underfeeding stimulated exercise in lesioned rats. All of these treatments are effective in neurologically intact rats. Because he found n o treatment that increased exercise in rats with ventromedial hypothalamic lesions, Kennedy (1964) suggested that this brain region was not directly sensitive t o estradiol, but represented an integrative area where a variety of stimuli converged t o induce running. On the other hand, some lesions in the rostral hypothalamus selectively abolished estradiol-induced running. Animals with these lesions ran in response to amphetamine or underfeeding but not in response to exogenous estradiol benzoate. Kennedy (1964) concluded that the rostral hypothalamus contained estrogen-sensitive neurons that relay facilitatory s t i m d for activity to the ventromedial hypothalamus. He did not speculate as to where ovarian hormones might act t o affect eating or thermoregulation. Colvin and Sawyer (1969) provided tests of the Kennedy hypothesis by examining the effects of bilateral intracerebral implants of a 10% mixture of estradiol benzoate in cholesterol on locomotor activity in ovariectomized rats. They explored a wide variety of mid- and forebrain sites. As Kennedy predicted, positive sites where estradiol stimulated activity were located in the rostral basal diencephalon, whereas placements in the ventromedial hypothalamus were consistently negative. However, a variety of other positive sites was also found, including dorsal posterior hypothalamus and rostral midbrain. Colvin and Sawyer summarized by suggesting that their positive sites followed the course of the medial forebrain bundle. Latencies t o increase running in the responding animals averaged around 6 days. Intracranial implants of cholesterol alone were uniformly negative. At about the same time, Wade and Zucker ( 1 9 7 0 ~ )observed the effects of unilateral diencephalic implants of estradiol benzoate on both voluntary exercise and food intake in ovariectomized rats. We found that the estradiol benzoate implanted in the vicinity of the ventromedial hypothalamus significantly depressed food intake within 12 hours of application (Fig. 6). This effect was similar in magnitude to the changes in eating that follow systemic estradiol injection, and it lasted only as long as the estradiol was left in the brain (3 days)

217

HORMONES AND BODY WEIGHT

open symbols

< 10'/o

circle

ha If - shaded

10- 20%

triangle

FIG. 6.Coronal sections of the rat diencephalon indicating the location of hormone implants and their effectiveness in decreasing food intake. The response to progesterone (P) is indicated by the form of the symbol: circles, a decrease 10%; triangles, a 10-20% decrease, squares, a decrease 20%. Response to estradiol benzoate (EB) is indicated by degree of shading, regardless of the form of the symbol: a blank symbol indicates a decrease of 10%; a half-shaded symbol a decrease of 10-20%; and a black symbol a decrease of 20%. (From Wade and Zucker, 1970c.)

>

<

<

>

218

GEORGE N. WADE

(Fig. 7). In these same animals, stimulation with either testosterone or cholesterol was without effect. Estradiol placements in the lateral hypothalamus, anterior hypothalamus, and medial preoptic area were generally ineffective in altering eating. This effect has recently been replicated in both sexes (Beatty, OBriant & Vilberg, 1974; Jankowiak and Stern, 1974). On the other hand, estradiol implants in the medial preoptic area had n o effect on eating, but increased running wheel activity within 24 hours. Estradiol in the lateral hypothalamus-medial forebrain bundle and the ventromedial hypothalamus were ineffective, as was cholesterol in the medial preoptic area. Activity induced by the estradiol implants began to decline as soon as the estradiol was removed from the brain (after 4 days), but activity did not return to preimplant levels for about 3 weeks. Thus, estradiol benzoate implanted in the vicinity of the ventromedial hypothalamus depressed eating without affecting activity. Estradiol benzoate in the medial preoptic area stimulated voluntary exercise without altering eating. None of the responsive rats showed any evidence of sexual receptivity. There seems to be some disagreement about the distribution of neural sites where estradiol can act to stimulate activity. Both Colvin and Sawyer (1969) and Wade and Zucker ( 1 9 7 0 ~ )found positive sites in the medial preoptic area and negative sites in the ventromedial hypothalamus. But Colvin and Sawyer found additional activity-stimulating sites throughout the course of the medial fore-

$1

n

gast, I

:

26

24 22

71 I

I 2

l 3

l 4

I

1

5

6

1 7

DAYS

1 8

1 1 1 1 9101112

FIG. 7 . Mean food intake before, during, and after stimulation of the ventromedial hypothalamus with estradiol benzoate (EB) in ovariectomized rats. Hormone was in the brain during days 6, 7 , and 8. Rats showed an average decrease of (A) 10-20% and (B) >20%. (From Wade and Zucker, 1970c.)

HORMONES AND BODY WEIGHT

219

brain bundle, whereas Wade and Zucker found no evidence for positive sites in the lateral hypothalamus-medial forebrain bundle complex. It is not inconceivable that the wider distribution of facilitatory sites found by Colvin and Sawyer could be due to intracerebral diffusion of the estradiol. In their experiment, the estradiol benzoate was distributed over a wider surface area (bilateral 23-gauge cannulas) than in the Wade and Zucker work (unilateral 27-gauge cannulas). The 6-day latency reported by Colvin and Sawyer (versus less than 24 hours in the Wade and Zucker) to increase activity might represent the time needed for the estradiol to diffuse t o a site of action. However, this latter hypothesis is not supported by the abundance of ineffective placements in the ventromedial hypothalamic area, which is closer t o the medial preoptic area than many of the effective loci. There are some additional data that may suggest that estradiol acts on a relatively restricted area, perhaps the medial preoptic area (or rostral basal hypothalamus), to enhance activity. First, autoradiographic experiments indicate that the medial preoptic area has a relatively high density of estradiolconcentrating neurons, whereas the medial forebrain bundle generally does not (Pfaff and Keiner, 1973). Second, Kennedy (1964) reported that lesions restricted to the rostral hypothalamus, selectively abolished estradiol-induced running, which should not happen if the estradiol acts on a wide variety of neural loci. Third, Stem and Jankowiak (1972) have found that bilateral actinomycin D implants restricted to the anterior hypothalamus-preoptic area prevented the increase in activity which normally follows systemic estradiol injection with no evidence of general toxicity to the animals (Fig. 8). Although these data seem t o be consistent with the notion of a rather restricted site of

11 11111111111 4 8 I2162024283236404448 Days

FIG. 8. Running wheel activity of ovariectomized female rats given systemic injections o f 15 clg estradiol benzoate (EB)/day. Half of the rats (X) had actinomycin D implanted in the anterior hypothalamus-preoptic area on days 25-34 and the cocoa butter vehicle on days 35-45. The other half (0) had cocoa butter implanted on days 25-34 and actinomycin D o n days 35-45. (From Stern and Jankowiak, 1972.)

220

GEORGE N. WADE

action for estradiol, they are by no means conclusive, and clarification of this point awaits further experimentation. One point that clearly emerges from these lesion and chemical stimulation studies is that the highly correlated changes in eating, running wheel activity, and sexual receptivity cited previously cannot be due to hormone action on a common neural substrate as was suggested by Rothchild (1967) and others. Estradiol placements that stimulated activity had no effect on appetite; placements inhibiting food intake had n o effect on running; and none of the placements induced lordosis (Wade and Zucker, 1 9 7 0 ~ ) I. n addition, although medial preoptic estradiol implants have been reported to stimulate both running and sexual receptivity (Lisk, 1962), lesions in this area abolish estradiol-induced running (Kennedy, 1964) but facilitate induction of sexual receptivity (Powers and Valenstein, 1972b). Thus, although these behaviors are highly coordinated during a variety of reproductive conditions, these fluctuations are almost certainly due to hormonal actions on separate neural substrates. Finally, these data are not consistent with the competing behavior models sometimes invoked t o explain the pattern of behavioral changes at proestrus. For example, it has been suggested that proestrous rats eat less because they are “busy” running or copulating; nor can changes in one behavior merely be a consequence of changes in another (e.g., females lose their appetite because of “arousal” or “sexual excitement”). I t is quite clear that the various behaviors characteristic of proestrous rats can be manipulated independently. The finding that estradiol benzoate placement in the vicinity of the ventromedial hypothalamus decreases eating and body weight (Beatty et ul., 1974; Jankowiak and Stern, 1974; Wade and Zucker, 1970c) is particularly intriguing, since it has been known for some time that neurons in or around the ventromedial hypothalamus act to restrain eating and body weight in a variety of species, (e.g., Hoebel, 1971). These data raise the possibility that estradiol exerts its effect on eating by exciting, either directly or indirectly, hypothalamic neurons that normally restrain eating and/or body weight. The anorega of proestrus (or other times of high estradiol availability) could be explained by estrogenic stimulation of the ventromedial hypothalamic area. Conversely, perhaps the overeating after estradiol withdrawal is a result of decreasing ventromedial hypothalamic activity. That is, ovariectomy may be somewhat akin to a small ventromedial hypothalamic lesion. There are some similarities between the effects of ventromedial hypothalamic lesions and ovariectomy on the eating behavior of female rats. After ventromedial lesions, female rats increase their meal size in both the light and dark phases of the daily light cycle, but although the number of meals eaten in the light also rises, there is a slight drop in the frequency of nighttime meals (Balagura and Devenport, 1970). Kenney and Mook (1974) have reported that following ovariectomy rats also eat larger meals in both the light and the dark.

HORMONES AND BODY WEIGHT

I

22 1

However, whereas all three of their ovariectomized groups decreased their meal frequency at night, two out of the three groups seemed t o eat more frequent meals when the lights were on. Thus, to the extent that comparisons are possible, the meal pattern data are not inconsistent with the possibility that estradiol withdrawal and ventromedial hypothalamic lesions may have a common mode of action. Kenney and Mook (1974) suggest that ovariectomy acts primarily to impair the onset of satiety during a meal. A recent paper by Drewett (1974) indicates that the opposite pattern occurs at proestrus; meal size decreases at the time when eating is reduced. Thus, estradiol withdrawal (ovariectomy) increases meal size, and rising plasma estradiol levels (proestrus) decrease meal size. Another similarity between the effects of ovariectomy and ventromedial hypothalamic lesions on eating is that both are time-limited; that is, rats only overeat for a restricted period of time. The hyperphagia then subsides and weight stabilizes (Fig. 2) (Hoebel and Teitelbaum, 1966; Mook et ul., 1972; Tarttelin and Gorski, 1973; Wade, 1975). (This point will be discussed more fully in Section IV,B.) Not surprisingly, there are several difficulties for the hypothesis that estradiol affects eating via the ventromedial hypothalamus. For example, after ventromedial lesions, rats may be extremely finicky, refusing to eat unpalatable food (Teitelbaum, 1955), but ovariectomy reduces responsiveness to tastes (Wade and Zucker, 1969b, 1970b; Zucker, 1969). This contradiction is easily resolved if it can be assumed that overeating and finickiness are two separate and not causally related aspects of the ventromedial hypothalamic syndrome, as Corbit and Stellar (1964) have suggested. In fact, Margules (1970a,b) has suggested that there are at least two neurochemically distinct systems in the ventromedial hypothalamus: one mediating cues related t o satiety and one mediating responsiveness to tastes. Both systems would be destroyed by lesions, but estradiol withdrawal might have opposite effects on the two systems (increasing eating and decreasing taste responsiveness). Perhaps the greatest difficulty for the hypothesis that estradiol acts on the ventromedial hypothalamus to depress eating and body weight are the several published (Beatty et al., 1975; King and Cox, 1973; Montemurro, 1971) and unpublished (Finger and Mook, 1971; Kennedy, 1969) reports that systemic injections of estradiol benzoate or diethylstilbestrol depress eating and/or body weight in rats and/or mice with ventromedial hypothalamic lesions. Of course, if the ventromedial hypothalamus were the only target site for estradiol, lesions of this region should abolish the appetite- and body weight-depressing actions of estradiol, but they do not. These data suggest that estradiol can act on other brain loci to affect eating and raise some question as to whether estradiol normally acts via the ventromedial hypothalamus at all (King and Cox, 1973). Both King and Cox (1973) and Beatty et ul., (1975) found that ovariectomy increased, whereas estradiol benzoate injections decreased, both eating and body

222

GEORGE N. WADE

weight in rats with ventromedial hypothalamic lesions, indicating that the animals could still respond to estradiol. However, it was clear that the neurologically intact rats demonstrated significantly greater increases in weight gain and food intake after ovariectomy than did the lesioned females (Beatty et al., 1975). Similarly, exogenous estradiol benzoate depressed body weight more in the unlesioned ovariectomized females than in the animals with ventromedial lesions (Beatty etal., 1975, King and Cox, 1973). The exogenous estradiol benzoate also depressed eating more in the intact than in the lesioned females, but this difference fell short of statistical significance (Beatty e f ul., 1975). Finally, an abstract by Nance and Gorski (1973) suggests that under some circumstances, ventromedial hypothalamic lesions can completely abolish the effects of estradiol benzoate on both eating and body weight in ovariectomized rats. Thus, ventromedial hypothalamic lesions appear to attenuate the effects of estradiol on eating and body weight even though they may not completely abolish the effects. (See Beatty et ul., 1975, for an informative discussion of this work.) One shortcoming of many of these reports is the absence of a detailed histological analysis of the lesions. Our own data (Portnoy, Wade, Ralph, and Balagura, 1973 unpublished data) indicate that if lesions of varying sizes are made in the ventromedial hypothalamus of ovariectomized rats, the rats with the larger lesions are substantially less responsive to estradiol than the rats with smaller lesions. Therefore, hypothalamic lesion size may be a crucial variable in any attempts to alter responsiveness t o hormones with brain lesions. From these results it seems reasonable to conclude that while the ventromedial hypothalamus may not be the site of action of estradiol to depress eating and body weight, it is very likely that the ventromedial hypothalamus is a site of estradiol action. Actually, it is not surprising that estradiol could act at other neural loci, since a wide variety of brain regions have been reported to take up and retain radioactive estradiol as well as affect eating behaviors and body weight. Some obvious examples include corticomedial amygdala, septa1 area, and the olfactory bulbs (Larue and LeMagnen, 1970; Pfaff and Keiner, 1973; Singh and Meyer, 1968; White and Fisher, 1969). However, the medial amygdala probably is not a crucial site of estradiol action on eating and body weight. Powers (1969 unpublished data) found that estradiol benzoate implanted in this region in either satiated or food-deprived ovariectomized rats was without effect on eating. Similarly, corticomedial amygdala lesions do not appear to abolish responsiveness to estradiol in ovariectomized rats (Cox and King, 1974). Additional hormone implant studies will be necessary t o specify just where estradiol can act in the brain to affect eating behavior. It is not surprising that estradiol may act on several neural loci to affect eating, since it appears that there might be other brain areas (as yet largely unspecified) outside the ventromedial hypothalamus that act to restrain eating

HORMONES AND BODY WEIGHT

223

and body weight. Following ventromedial hypothalamic lesions, rats do not overeat and gain weight indefinitely; eventually both eating and weight gain return to prelesion levels (Hoebel and Teitelbaum, 1966). Some other brain area(s) might be restraining body weight; perhaps estradiol acts on the other brain area(s) that take over some of the functions of the ventromedial hypo thalamus. Although some progress has been made in specifying where estradiol might act to stimulate voluntary exercise and inhibit food intake, very little progress has been made in identifying the output pathways connecting these estradiolsensitive neurons to the “final common pathways” for the behaviors. From Kennedy’s (1964) work, it seems as though estrogenic excitation of the rostra1 diencephalon stimulates activity via the ventromedial hypothalamus. Where the pathway leads from there is not clear. It probably does not depend on connections between the ventromedial hypothalamus and the lateral hypothalamus, since bilateral knife cuts between these two regions cause no permanent changes in the voluntary exercise of female rats (Sclafani, 1971). Similarly, the integrity of the lateral hypothalamus cannot be critical for estrogenic effects on eating, since ovaries seem to restrain body weight just as effectively in rats that have “recovered” from lateral hypothalamic damage as they do in neurologically intact females (Harrell and Balagura, 1975). Additional experiments utilizing selective brain lesions or knife cuts, such as those of Kennedy and Mitra, of Gold, and of Sclafani, could be very helpful in mapping these pathways. Although estradiol can act on the brain to alter energy balance-controlling behaviors, this certainly does not exclude the possibility that ovarian steroids might act on nonneural target tissues t o affect body weight regulation. It is clear that estradiol can alter secretion of anterior pituitary hormones and, thus, the secretions of their target glands. (Coyne and Kitay, 1969; D’Angelo and Fisher, 1969; McCann et al., 1968). It is also clear that pituitary and target gland hormones have important effects on behavioral regulation of body weight (Kennedy and Mitra, 1963a; Pfaff, 1969; Richter, 1956; Stem, 1970; Stevenson and Franklin, 1970; Wade, 1974). However, although it is possible that fluctuations in anterior pituitary function may contribute to the behavioral effects of estradiol, the anterior pituitary is not essential for these effects. Wade and Zucker (1970a; Wade, 1974) found that estradiol benzoate depressed eating in ovariectomized-hypophysectomized weanling rats, and Stern and Jankowiak (1973) found that estradiol benzoate also stimulated running wheel activity in ovariectomized-hypophysectomized adult rats. Therefore, estradiol can still modulate eating and voluntary exercise in the absence of the pituitary gland. Although the anterior pituitary may not be necessary for gonadal effects on behavioral regulation of energy balance (which is the subject of this review), it would be a serious mistake to underestimate the contribution of physiologicalmetabolic factors in gonadal effects on body weight. In fact, sometimes ovarian

224

GEORGE N. WADE

hormones alter the body weight of rats with no apparent effect on behavior. In weanling rats, estradiol and progesterone may lower and raise body weight, respectively, without affecting food intake (Ross and Zucker, 1974; Wade, 1974; Zucker, 1972). We have also found that, if rats are kept in warm environments (around 80°F), they do not overeat following ovariectomy, but they d o gain more weight than intact controls. However, they do.seem to take longer to gain the weight than rats in cooler environments that d o overeat (Portnoy and Wade, 1972 unpublished data). On the other hand, the behavioral changes induced by hormone treatments are not necessarily sufficient to alter body weight by themselves. Roy has compared body weights of three groups of ovariectomized rats: one group given oil injections and ad libitum access t o food; one group given estradiol benzoate injections and ad libitum food; and a final group given oil injections but pair-fed with the estradiol-treated animals. The estradiol-injected group reduced their food intake and lost weight, but the pair-fed, oil-treated rats did not lose any weight, in spite of their restricted food ration. It is apparent that, although behavioral changes undoubtedly contribute to the fluctuations in body weight during various reproductive states, the behavioral changes, in and of themselves, are likely neither necessary nor sufficient to affect body weight. The widespread metabolic effects of gonadal steroids (Aschkenasy, 1959; Salhanic et ul., 1969) are not to be underestimated. Thus far nothing has been mentioned about the neural site of action of progesterone or about where any of the gonadal steroids might act to affect thermoregulation-this is not simply an oversight. With the exception of the report by Jankowiak and Stern (1974) that dorsomedial hypothalamic progesterone implants increase food intake, virtually nothing is known about these problems. It has been suggested (Wade, 1972) that, since the medial preoptic area contains a high density of estradiol-concentrating neurons (F'faff and Keiner, 1973) and is an important neural region for physiological and behavioral thermoregulation (Corbit, 1970; Gale, 1973), perhaps this is a site where ovarian steroids may act to affect thermoregulation. Recent attempts to test this hypothesis have yielded generally negative results. Estradiol implanted in the preoptic area of ovariectomized rats had no effect on rectal temperatures, but there was some suggestion that progesterone implants in this region might raise colonic temperatures (Marrone e f al., 1974). In summary, estradiol may act on a relatively restricted region in the basal anterior diencephalon to stimulate voluntary exercise without affecting other estradioldependent behaviors. Estradiol may act on the area of the ventromedial hypothalamus (and perhaps on the other as yet unspecified brain sites that are functionally similar to the ventromedial hypothalamus) to depress food intake without affecting other estradioldependent behaviors. Because of certain similarities between the eating behaviors of rats following ovariectomy or ventromedial hypothalamic lesions, it was further suggested that, so far as eating

HORMONES AND BODY WEIGHT

225

behavior is concerned, estradiol withdrawal may be somewhat akin to a small reversible lesion of the ventromedial h y p o t h a l a m u ~ . ~ B.

MECHANISM OF ACTION: A LIPOSTATIC HYPOTHESIS

Experiments demonstrating that estradiol implanted in the diencephalon stimulates exercise and reduces eating in ovariectomized rats independent of the adenohypophysis could indicate that ovarian steroids act directly on neurons that are part of the neural systems controlling regulatory behaviors and that these regulatory behaviors then alter body weight. However, an alternative hypothesis has recently been advanced which may provide a more adequate explanation of a variety of data. Separate laboratories have suggested independently, and nearly simultaneously, that ovarian steroids do not affect regulatory behaviors directly, but, rather, they may act on the brain to alter the set-point about which body weight is regulated (Mook etal., 1972; Wade, 1972). The changes in behavior are seen as attempts to bring body weight into line with this new set-point. A variety of data consistent with this view have subsequently been published. The effects of hormone withdrawal or replacement on food intake are transient, lasting only until a new body weight is reached, but the effects on body weight are permanent, lasting as long as the animal's hormonal condition is held constant. For example, after ovariectomy, rats overeat and gain weight, but after several weeks the hyperphagia and accelerated weight gain subside (see Fig. 2). Food intake then does not differ significantly from preovariectomy levels, although body weight may remain 20-25% above that of intact controls (Mook etal., 1972; Tarttelin and Gorski, 1973). Treatment of ovariectomized rats with estradiol benzoate reverses this pattern; rats undereat until body weight drops to 3Although the role of the ventromedial hypothalamus in estrogenic control of eating and body weight has been stressed, I do not want to place undue emphasis on this brain region, since it is abundantly clear that the ventromedial hypothalamus is but a part of a widely distributed neural regulatory system (cf. Grossman, 1968). It has been suggested that the hyperphagia and obesity following ventromedial hypothalamic lesions may be due to damage t o the ventral noradrenergic bundle passing through this region rather than damage to the ventromedial nucleus per se (Ahlskog and Hoebel, 1973;Gold, 1973). The hyperphagia and obesity induced by ventral bundle lesions are prevented by hypophysectomy, but removal of the pituitary gland has no effect on these changes after electrolytic lesions of the ventromedial hypothalamus (Ahlskog et al., 1974; Valenstein el al., 1969). This finding has two important implications. First, it implies that there is more than one hypothalamic system restraining eating and body weight. Also, it suggests that estrogenic suppression of eating and body weight is probably not mediated by the ventral noradrenergic bundle, since estradiol affects these measures in hypophysectomized rats (Wade, 1974; Wade and Zucker, 1970a). In any event, the important point is that, although the ventromedial hypothalamus is only a portion of the neural weight-regulating system, it is probably one of the parts of this system where estradiol acts to influence eating behavior and body weight.

226

GEORGE N. WADE

preovariectomy levels. In fact, lowering of regulated body weight appears to be proportional to the amount of circulating estradiol, so long as these estradiol titers approximate physiological levels (Wade, 1975) (Fig. 9). Food intake then rises, and body weight levels off at a new lower set-point (Mook et al., 1972; Tarttelin and Gorski, 1973; Wade, 1975) for as long as the estradiol treatment continues. If estradiol treatments are withdrawn, rats respond just as after ovariectomy; a transient hyperphagia causes a permanent elevation of body weight (Wade, 1975) (See Fig. 4). Therefore, estradiol seems t o affect eating behavior by lowering the body weight set-point; estradiol withdrawal appears t o raise body weight set-point. The changes in eating behavior are viewed as being secondary t o altered body weight set-point. Although the effects of hormonal manipulations on eating are transient, lasting only until body weight is realigned, the effects on locomotor activity appear to be permanent. After ovariectomy, rats are permanently hypoactive; activity does not return when body weight stabilizes. Conversely, unlike the hypophagia, the increase in activity after estradiol replacement therapy outlasts the period of weight loss (Mook et d.,1972). Thus, although activity levels are always inversely related t o food intake (see Table I) and undoubtedly contribute t o the fluctuations in body weight, it is likely that estradiol directly stimulates voluntary exercise, and the changes in activity are not a consequence of a shift in body weight set-point. These results are consistent with our report that estradiol acts on separate neural loci to affect eating and exercise (Wade and Zucker, 1970~). The possibility that the primary effect of estradiol is t o lower the body weight set-point is consistent with the hypothesis that a principal site of estradiol action is the ventromedial hypothalamus. For some time, Kennedy (1953, 1967,1969) has maintained that one function of this brain region is to monitor body weight

-3 = 0 \

w

*

z2

28

440

24

420

20

4 0 0 I-

16

380

12

360

8

340

4

320

I

e

w

c)

3

n 0

^o

n

I

2

3

4

5

8

7

0

9

1011

2

l Z I 3 1 4 I S

SUCCESSIVE 2- DAY PERIODS

FIG. 9. Food intake (top) and body weight (bottom) of ovariectomized rats injected with 0.5, 2.0, and 5.0 /.fg estradiol benzoate (EB)/day or with sesame oil vehicle. Injections were given from day 7 through day 30. (From Wade, 1975 .)

HORMONES AND BODY WEIGHT

227

(or fat content) and to adjust feeding behavior so that body fat levels are maintained within a restricted range. This lipostatic theory suggests, “It is apparent that the ventromedial nucleus in the normal rat restrains the accumulation of surplus fat independently of any possibly existing central control of the lean body mass” (Kennedy, 1969). This lipostatic interpretation of ventromedial hypothalamic function is supported by a variety of data. Perhaps the most convincing is the work of Hoebel and Teitelbaum (1966). Following ventromedial hypothalamic lesions rats overeat only until a certain amount of weight is gained; then the hyperphagia subsides, and the growth curve is parallel to, but higher than, that of intact rats. If the obese rats are starved until they reach their prelesion bod) weights, the hyperphagia is reinstated when ad libitum feeding is resumed but lasts only until the obesity is restored. Conversely, if the obese rats are made “superobese” by daily force feeding, they voluntarily undereat (once the force feeding is terminated) until they return to their obese set-point. Finally, if neurologically intact rats are made obese before lesioning, they are not hyperphagic following brain damage. Thus, rats with ventromedial hypothalamic lesions seem to be perfectly capable of regulating their body weights; they merely regulate at a higher weight level (or body fat content) than unlesioned rats. The ventromedial hypothalamic lesion seems to raise the body weight set-point, and the rats overeat in order to align body weight with this new set-poin t . The several similarities between the eating behaviors of ovariectomized and ventromedial hypothalamic-lesioned rats have already been mentioned. Perhaps, then, estradiol acts on the ventromedial hypothalamic lipostat to lower the regulated body weight. Consistent with this possibility are the changes in carcass composition that follow manipulation of gonadal hormones in female rats. The increase in body weight following ovariectomy or progesterone treatment is mainly due t o increased fat deposition. As discussed previously, these two treatments produce similar and nonadditive effects on body weight and composition (Galletti and Klopper, 1964; Hervey and Hervey, 1966, 1967; Leshner and Collier, 1973). During pregnancy there is an increase in carcass fat content, similar t o that during progesterone treatment (Hervey et al., 1967). Finally, Kennedy (1969) has suggested that the principal source of weight loss during estradiol treatment is lipolysis. Thus, ovarian hormones, like the ventromedial hypothalamus, seem to be of some importance for the regulation of the body’s fat stores. As Mook et al. (1972) point out, it is important to note that, whereas blood estradiol levels may dictate a particular body weight level, they can do SO o d y within certain limits. Even though rats overeat and gain weight after either ovariectomy or ventromedial hypothalamic lesions, the changes in these measures following ovariectomy are relatively modest compared with the effects of the lesions. After ventromedial hypothalamic lesions, a doubling of body weight is not uncommon, but the body weight of ovariectomized rats rarely increases

228

GEORGE N. WADE

by more than 25-30%. There seems to be an upper limit beyond which body weight cannot shift following just ovarian steroid withdrawal. Conversely, there seems t o be a lower body weight limit below which body weight cannot be forced by estradiol treatment. Neither reasonable (3 pg/day) (Tartellin and Gorski, 1973) nor unreasonable (200 &day) (Hervey and Hervey, 1965) doses of estradiol benzoate will force body weight significantly below that of untreated intact female rats. Nor does estradiol treatment affect eating and body weight of ovariectomized rats that are already regulating at a low body weight because of prior adrenalectomy (Redick e t a l . , 1973) or because of neonatal underfeeding (Zucker, 1972). Redick et ul. (1973) have shown that estradiol treatment will decrease both eating and body weight in ovariectomized-adrenalectomized rats that have been made mildly obese by giving them an especially attractive diet. They conclude by suggesting‘ that “estradiol suppresses feeding only in the face of actual or impending obesity. It probably affects the system(s) concerned with the long-term regulation of body weight; but it does not act directly on the mechanisms which mobilize or inhibit feeding.” Thus, although estradiol could act upon a hypothalamic lipostat to lower body weight set-point, it appears that body weight (or more likely fat content) cannot be driven below that of intact cycling females. Conversely, estradiol withdrawal induces only a limited obesity-no more than a 25-30% rise in body weight. Estradiol may cause only a “fine tuning” of lipostatic regulation. If estradiol alters eating behavior by lowering the neural set-point for body weight, how, then, does progesterone exert its effects on these measures? Because progesterone did not seem to affect eating and body weight in the absence of estradiol (see Section III,B for references), we reasoned that it exerted its appetite- and weight-stimulating actions by acting as an antiestrogen (Wade, 1972; Wade and Zucker, 1969a)-perhaps by inhibiting uptake of estradiol by the brain (Anderson and Greenwald, 1969; Lisk, 1974). It has recently been shown that concurrent injections of progesterone attenuate the effects of estradiol benzoate on eating and body weight (Fig. 10). In fact, if sufficient progesterone is given it can completely block the lowering of body weight set-point by estradiol (Wade, 1975). However, Roberts el ul. (1972) have suggested an alternative interpretation: if estradiol acts directly on the brain to lower body weight set-point, perhaps progesterone also can act on the brain independently of estradiol t o raise body weight set-point. It was suggested that the reason that ovariectomized rats do not respond t o progesterone is because they are already obese; that is, if estradiol cannot depress eating and body weight in already lean rats, perhaps progesterone cannot raise weight above the obesity induced by ovariectomy. Roberts et ul. (1972) tested their hypothesis by treating ovariectomizedadrenalectomized rats with progesterone and observing the effects on eating and

229

HORMONES AND BODY WEIGHT

body weight. This preparation is particularly appropriate for this sort of experiment since it has been shown that adrenalectomy either prevents or reverses the effects of ovariectomy on body weight (Grunt, 1964; Mook et al., 1972; Stem et al., 1974). They found that treatment of ovariectomized-adrenalectomized rats with 5.0 mg progesteronelday caused a significant increase in food intake and body weight (Roberts et al., 1972). In this preparation there is, of course, no circulating estradiol, so that the progesterone could not be affecting energy balance by antagonizing the actions of estradiol. This finding has recently been replicated (Ross and Zucker, 1974). It was also shown that the rapid weight gain during progesterone treatment was transient; body weights plateau parallel to, but higher than, oil-treated controls, suggesting that progesterone was directly raising body weight set-point. Roberts et al. noted that treatments that restore responsiveness to progesterone in ovariectomized rats (adrenalectomy or estradiol treatment) all prevent the usual postovariectomy obesity. Perhaps, then, progesterone does directly raise body weight set-point (independent of estradiol), but only in the absence of a preoccurring weight gain. One difficulty with the work in ovariectomized-adrenalectomized rats is the extremely high dose (5.0 mg/day) of progesterone that has been used (Roberts et al., 1972; Ross and Zucker, 1974). This is approximately 20 times the dose of progesterone reported to facilitate lordosis in 100%of ovariectomized estradiolprimed rats (Powers and Valenstein, 1972a). Therefore, it is difficult to determine whether the effects of progesterone on eating and body weight in ovariectomized-adrenalectomized rats are of any physiological significance or whether they are merely artifacts of unusually high progesterone doses. If the absence of a preoccurring weight gain is all that is necessary to induce responsiveness to progesterone in ovariectomized rats, then the method used to

-8

28

440

P 24

420

,o 20 W

2 5

400

C

I

380

l6 12

360

4 ~

0

0

340

2

LL

320

4

I

2

3

4

5

6

7

8

9

1 0 1 1 l 2 I 3 I 4 I 5

SUCCESSIVE 2-DAY PERIODS

FIG. 10. Food intake (top) and body weight (bottom) of ovariectomized rats injected with sesame oil vehicle, 2.0 pg estradiol benzoate (EB)/day, or with estradiol benzoate plus 0.5, 1.0, or 2.0 mg progesterone @)/day. Injections were given from day 7 through day 30. (From Wade, 1975.)

230

GEORGE N. WADE

prevent the weight gain (adrenalectomy or estradiol treatment) should not affect the rats’ responsiveness to exogenous progesterone. If, on the other hand, the principal action of progesterone is to interfere with the actions of estradiol, then estradiol-treated ovariectomized rats might be more responsive to progesterone than ovariectomized-adrenalectomized rats. This possibility was recently tested in an experiment in which the effects of several doses of progesterone on eating and body weight were compared in two groups of rats whose postovariectomy weight gain was prevented either by concurrent adrenalectomy or by daily injections of 2.0 pg estradiol benzoate (Wade, 1975). It was clear that the estradiol-treated rats were more responsive to the progesterone than those that were adrenalectomized. Less than 0.5 rng progesterone/ day significantly increased eating and body weight in the ovariectomized, estradiol-treated females, whereas progesterone doses of less than 5 .O mg/day were without effect on either measure in the ovariectomized-adrenalectomized rats (Wade, 1975). Therefore, the antiestrogenic effects of progesterone are clearly evident with plasma hormone titers far below those necessary to stimulate eating and weight gain in the absence of estradiol. This work seems to indicate that the principal action of progesterone on energy balance-regulating behaviors is to decrease estradiol availability or effectiveness in the central nervous system. Although these data do not rule out the possibility that endogenous progesterone could possibly directly raise the body weight set-point under certain circumstances, the extremely high dosages necessary to elicit this effect suggest that it may occur only during times of very high progesterone secretion, such as during late pregnancy (Hashimoto et al., 1968). However, even during pregnancy it is not clear whether the overeating and obesity are due to the high plasma progesterone titers, the very low levels of circulating estradiol (Yoshinaga et al., 1969), or both. Ross and Zucker (1974) have recently suggested an alternative interpretation of the effects of progesterone on ovariectomized-adrenalectomized rats. They indicate that progesterone might mimic the effects of adrenal corticosteroids and permissively increase the food intake of adrenalectomized rats simply by improving their general health. They point out that progesterone, which restores the weight gains of ovariectomized-adrenalectomized rats to levels characteristic of ovariectomized females, is effective in prolonging the survival of adrenalectomized rats (Greene et al., 1939). Relevant to this point are the reports that adrenocorticosteroids (like progesterone) also permit normal weight gain in ovariect omized-adrenalec tomize d females (Rodie r , 1973; Tarttelin and Gorski , 1973). On the other hand, in nonadrenalectomized rats, exogenous corticosterone only depresses eating and body weight (Stevenson and Franklin, 1970). We have also found that progesterone stimulates running wheel activity in ovariectomized-adrenalectomized rats (Fig. 11) (Marrone et al., 1975), just as corticosterone does in adrenalectomized males (Leshner, 1971). This activity-

S

b

WEIGHT G A I N (grams 1

8 S VI ,

I

S I

,

Y

FOOD INTAKE (gramsldoy)

,

"0 i

RUNNING WHEEL ACTIVITY (rovolutionr / d a y )

3

f

::

2 -;.

m m

232

GEORGE N.WADE

Progesterone could exert its adrenocorticosteroid-like effects by being converted to a corticosteroid by extra-adrenal enzymes (it is a precursor of the adrenal hormones) or by acting directly as a weak adrenocorticoid. It is known, for example, that corticosterone can substitute for progesterone in facilitating lordosis more effectively than the progesterone metabolite, Sa-pregnane-3,20dione, in guinea pigs (Wade and Feder, 1972b). These data certainly raise the possibility that “the increases in eating and body weight [progesterone] produces may, therefore, reflect generally improved health and not specific changes in the central energy balance system” (Ross and Zucker, 1974). The only data I am aware of that may be inconsistent with this possibility is the report by Jankowiak and Stern (1974) that progesterone implanted in the dorsomedial hypothalamus of ovariectomized-adrenalectomized rats stimulates eating and weight gain. Unless there is substantial systemic leakage of the progesterone from the implants, it is not obvious how this treatment could be improving the animals’ general health. Perhaps a better way to test the Roberts etal. hypothesis would be to find another way to prevent the postovariectomy obesity than by adrenalectomy and then inject progesterone. Possibly rats that have been undernourished during infancy would be appropriate, since they d o not seem to become obese after adult ovariectomy (Zucker, 1972). Another interesting implication of the Ross and Zucker work is that progesterone may be able to block the peripheral metabolic actions of estradiol independently of its antiestrogenic effects on behavior. Hervey and Hervey (1966b) reported that progesterone increased body weight in intact rats even when they were pair-fed with vehicle-treated controls, but Ross and Zucker (1974) were unable to find any evidence for a weight-promoting effect of progesterone in ovariectomized rats pair-fed with oil-treated controls. An obvious interpretation of these divergent results is that progesterone can attenuate the metabolic effects of ovarian estradiol in intact females in the absence of any effects on eating. Thus, estradiol may affect eating behavior indirectly by acting on the brain (perhaps the ventromedial hypothalamus) to lower the set-point about which body weight (or fat content) is regulated in a dose-dependent fashion. Withdrawal of estradiol, by any means, raises body weight set-point. Progesterone can attenuate or completely block the set-point lowering effects of estradiol, again in a dose-dependent fashon, perhaps by decreasing the availability or effectiveness of estradiol in crucial brain sites. It has also been suggested that progesterone might directly raise body weight set-point independent of its antiestrogenic actions, but this effect is somewhat more difficult to elicit and may be due to nonspecific actions of pharmacological doses of progesterone. The possibility that estradiol acts on the brain t o lower body weight set-point has a very exciting implication if i t can be assumed that this effect is not limited to laboratory rats, an assumption that is not without justification (see Section VIII). That overweight is a persistent and pervasive health problem in many societies goes without question (Balagura, 1973; Mayer, 1969), and an especially

HORMONES AND BODY WEIGHT

233

impressive aspect of much of this obesity is its resistance to any form of permanent treatment. The long-term prognosis for m y “cure” for obesity has been exceptionally discouraging. Recently, Nisbett (1972) has advanced an hypothesis which may in part account for this impressive lack of success in treating human obesity. He suggests that many persons are biologically programmed t o be obese; because of either a genetic predisposition to overweight or, because of overnutrition during early development, their brains defend a higher body weight set-point than lean persons. If this is, in fact, the case, treatments of the symptoms of obesity (overeating, underactivity, inappropriate meal patterns, etc.) are ultimately destined to failure if body weight set-point remains elevated. On the other hand, an ideal (and seemingly painless) treatment would be one that lowers the neural set-point for body weight; then the appropriate behavioral adjustments should follow naturally and voluntarily. The problem, of course, is finding a treatment that lowers body weight setpoint. If estradiol acts in human beings to lower set-point as it does in rats, then it could be used for weight control. An estrogenic weight-reducing compound could be especially useful, since the effectiveness of estradiol in depressing food intake and in dictating a low body weight in rats seems to be independent of diet nutritional value or palatability (Jennings, 1973; Redick and Mook, 1973). However, it is obvious that the widespread physiological actions of estradiol would preclude its use in human beings, especially in men. The only hope for this sort of treatment for obesity would be if there were some way of selectively stimulating the neural estrogen “receptors” that control body weight without stimulating any of the other estrogen-sensitive sites in the body. Intracerebral placement of estradiol would not be an appropriate methods, for several obvious reasons. However, if the neural estrogen “receptor” for weight regulation were t o have some unique property that differentiated it from all other estrogen receptors in the body, it might be possible t o find some chemical that stimulated it (and lowered body weight set-point) without affecting the other estrogen-sensitive systems. That this may, in fact, be the case is suggested by some work examining the effects of some antiestrogens on regulatory behaviors in rats. The nonsteroidal compound, MER-25, is consistently antiestrogenic on a variety of physiological end points in a number of species. In addition, MER-25 appears to have almost no estrogenic, androgenic, antiandrogenic, progestational, antiprogestational, or gonadotropic activities with regard t o these end points (Lerner et al., 1958). Compound MER-25 could exert its antiestrogenic actions by competing with endogenous or exogenous estradiol for target-organ receptor sites (Jensen et d.,1972). Similarly, MER-25 seems to be antiestrogenic toward behaviors, since it can block estradiol-induced sexual receptivity (Meyerson and Lindstrom, 1968) and locomotor activity (Roy and Wade, 1975a). In contrast to its nearly pure antiestrogenic effects on other measures,

234

GEORGE N. WADE

MER-25 appears to act as an estrogen on eating and body weight. Injection of male or female gonadectomized rats with MER-25 reduces both eating and body weight. Just as with estradiol, the effects of MER-25 on eating seem to be transient, but the weight-reducing effects last as long as the MER-25 treatments continue. It is extremely unlikely for several reasons that these appetite- and weight-reducing effects of MER-25 are due to a nonspecific toxicity: (1) the changes in eating are transient, whereas the effect on body weight is lasting; (2) MER-25 does not suppress base-line running wheel activity (in non-estrogentreated rats), nor does it attenuate starvation-induced activity, as it might if the animals were simply ill; (3) concurrent progesterone injections attenuate the effect of MER-25 on eating and body weight, just as they do with estradiol; (4) implants of MER-25 in the ventromedial hypothalamus depress eating in ovariectomized rats, just as estradiol does; (5) we have been unable to establish conditioned gustatory aversions (Garcia e t d . , 1974) using MER-25 as the unconditioned stimulus; and (6) female rats appear to be more responsive to MER-25 than males (Roy and Wade, 1975b). It is quite possible, therefore, that the neural estrogen “receptor” affecting body weight does differ from other estrogen-sensitive systems in the body: MER-25 is solely estrogenic with regard to body weight regulation but antiestrogenic in other physiological and behavioral systems. These estrogenic actions of MER-25 may be shared by some other steroid antagonists, including the antiandrogen, cyproterone acetate (Vilberg et al., 1974), and the antiestrogen, CI-628 (Powers, 1974 personal communication). It is obvious that compounds such as MER-25 or CI-628 would not be appropriate for weight control, since their antiestrogenic side effects would be just as objectionable in most cases as the estrogenic effects of estradiol. It seems as though an ideal weight-reducing drug would be one that acts on the neural estrogen-sensitive mechanisms to lower body weight set-point but has neither estrogenic nor antiestrogenic effects in other hormone-sensitive systems. I doubt that any such compound exists at the present time. C. INTERACTION WITH BRAIN MONOAMINERGIC SYSTEMS

Although a lipostatic hypothesis is an attractive and convenient way to describe the effects of ovarian steroids on eating and body weight, it is only a description of the data and in and of itself provides little information about how steroids might interact with the brain to affect energy balance. At best, the notion of shifts in body weight set-point helps to bring some order to our thinking about a variety of phenomena and suggests some reasonable approaches for additional research. At worst, this sort of pseudoexplanation could lead to a false impression that we understand more than we really do and serve to stifle creative research. The distinction between labeling a phenomenon and explaining it should remain clear in our minds.

HORMONES AND BODY WEIGHT

235

Actually, work on the mechanisms of action of steroids in the central nervous system is still in its infancy and lags far behind similar work in hormone-sensitive peripheral tissues (e.g., O’Malley and Means, 1974). In peripheral tissues there is a great deal of evidence suggesting that steroids act on the genome to alter genetic transcription. To the extent that comparisons are possible, it is not inconceivable that ovarian hormones have a similar mechanism of action in the brain. An extensive discussion of the neurochemical actions of steroids is beyond the scope of this paper, but the interested reader is referred to several excellent recent reviews (McEwen ef ul., 1970, 1972, 1974). One very exciting possibility is that steroids might affect behavior by altering the synthesis or activity of one (or more) of the several monoamines hypothesized to serve as neurotransmitters in the central nervous system. Indeed, although the current literature is somewhat confusing and controversial, it seems clear that ovarian hormones d o alter brain monoamine metabolism, particularly that ofnorepinephrine (Anton-Tay and Wurtman, 1971; Wurtman, 1971). It has been apparent for some time that manipulation of hypothalamic noradrenergic activity has very striking effects on eating behavior of rats (see Hoebel, 1971, for a particularly cogent review). One current theory suggests that there are cY-adrenergic-receptive cells in the perifornical and medial hypothalamus (highest density in the paraventricular nucleus) where norepinephrine acts to enhance eating behavior. Opposing this (Y feeding system are padrenergicreceptive cells in the perifomical and lateral hypothalamus where norepinephrine acts t o inhibit food intake (Leibowitz, 1972). This is a very attractive theory, with many data supporting it, and it is easy to imagine how estradiol might inhibit food intake by either inhibiting the a-adrenergic “feeding” system or by stimulating the 0-adrenergic “satiety” system. However attractive it may be, this model has not gone without challenge. Margules (1970a) has suggested that ceadrenergic agonists inhibit, rather than facilitate, food intake when placed in the medial hypothalamus. Furthermore, whether a-adrenergic agonists stimulate or inhibit eating may vary as a function of the light-dark cycle (Margules etul., 1972). Clearly, the action of norepinephrine on feeding behavior is far from simple. As indicated above, ovarian hormones do affect hypothalamic norepinephrine metabolism. During the estrous cycle, anterior hypothalamic norepinephrine varies cyclically. At proestrus the absolute level of norepinephrine is high; it drops just after ovulation to its lowest values on the day of estrus; and it rises again during metestrus and diestrus (Stefano and Donoso, 1967). After ovariectomy there is an increase in absolute levels, synthesis, and turnover of hypothalamic norepinephrine (Anton-Tay ef al., 1969, 1970; Anton-Tay and Wurtman, 1968; Bapna el ul., 1971; Donoso and Stefano, 1965). These effects of ovariectomy can be reversed by huge doses of estradiol benzoate and progesterone. Two recent experiments have attempted to integrate this information and indicate how estradiol acts on the brain to alter noradrenergic activity and, thus,

236

GEORGE N. WADE

eating behavior (Simpson and DiCara, 1973; Stern and Zwick, 1973). Unfortunately, they arrive at opposite conclusions. Simpson and DiCara (1 973) suggest that high blood titers of estradiol inhibit eating by decreasing levels of hypothalamic dopamine-0-hydroxylase, the enzyme that catalyzes the synthesis of norepinephrine from dopamine. They find that noradrenergic stimulation of the anterior hypothalamus increases food intake of female rats regardless of blood estradiol levels (estrus, diestrus, ovariectomized with and without estradiol replacement), but that dopaminergic stimulation increases eating only in the absence of high estradiol levels (diestrus and ovariectomized without estradiol replacement). They speculate that estradiol inhibits the dopamine-to-norepinephrine conversion, and, thus, eating behavior. However, this theory would seem to have some difficulty accounting for the increased eating at estrus relative to proestrus, since hypothalamic norepinephrine is higher at proestrus (Stefan0 and Donoso, 1967). Stem and Zwick (1973) suggest essentially the opposite hypothesis, that is, increased brain norepinephrine inhibits eating and stimulates activity. They found that intraventricular injections of either 1 fig estradiol benzoate or large doses of norepinephrine (100-250 pg) inhibited eating and stimulated activity of ovariectomized rats in the dark. In addition, they report that phentolamine, an a-adrenergic antagonist, attenuated and imipramine, which blocks norepinephrine reuptake, enhanced the effects of both the estradiol and norepinephrine on activity. In agreement with Margules er d.,(1972), they suggest that high hypothalamic norepinephrine inhibits food intake. But, then, why does ovariectomy increase both eating and hypothalamic norepinephrine? Given the multitude of potential pitfalls along the way, it is little wonder that no consensus has arisen from the attempts to correlate blood estradiol levels with hypothalamic norepinephrine and eating behavior. First, the functional significance of the changes in hypothalamic norepinephrine levels and turnover with changes in steroid secretion are not really clear. Do increased brain levels indicate enhanced synthesis and activity or decreased release and activity? Does increased turnover indicate an increase in synaptic release and activity or merely an increase of intracellular metabolism independent of synaptic release? Second, the anatomical resolution in most of the studies correlating steroids and brain monoamines is really inadequate for any conclusions about specific behaviors. Typically, no more than two or three pieces of hypothalamus are examined, even though anatomically overlapping monoaminergic neurons in any one of these areas may affect a variety of functions including eating, activity, thermoregulation, estrous behavior, and control of hypothalamic releasing hormones. There is no way of knowing which of these functions is being affected by the monoamine fluctuations. Finally, just how norepinephrine affects eating behavior has not yet been resolved. Whether norepinephrine increases or decreases eating depends on a variety of factors including anatomical locus, stage of the light-dark cycle, and the investigator.

HORMONES AND BODY WEIGHT

237

V. DEVELOPMENT OF RESPONSIVENESS TO OVARIAN STEROIDS AND EFFECTS OF LACTATION In a number of ways the behavior of prepubertal female rats resembles that of estradiol-deprived (ovariectomized, pregnant, or pseudopregnant) adult females. Weanling females are almost completely inactive in running wheels (Kennedy and Mitra, 1963b). These animals are able to run, since it has been shown that they increase their activity when food-deprived (Gentry and Wade, 1973 unpublished data; Kennedy, 1964); they simply choose not t o exercise. At the time of puberty activity increases. Prepubertal females, like estradiol-deprived adults, are hyperphagic. In fact, if eating is computed on a body weight basis, weanlings eat nearly twice as much as adults (Kennedy, 1957). Immature rats also build larger nests and gain weight more rapidly than sexually mature rats (Kinder, 1927; Slob, 1972). Around puberty eating, nest-building, and weight gain drop gradually. The simplest explanation for the similarities between weanling and estradioldeprived adults, of course, would be that the prepubertal ovary secretes insufficient estradiol to affect behavior and restrain body weight. However, Weisz and Gunsalus (1973) suggest that prepubertal female rats have higher plasma estradiol levels than adults do. An alternative (and not mutually exclusive) hypothesis is that the estradiol-sensitive neurons in the brain are not sufficiently mature t o respond to estradiol. This latter hypothesis is not likely to be correct for several reasons. Plapinger and McEwen (1973; Plapinger, McEwen, and Clemens, 1973) have reported that the hypothalamic cytoplasmic and nuclear estradiol-binding systems are mature well before puberty. Also, it is likely that the hypothalamic neurons controlling the adenohypophysis are hypersensitive to estradiol before puberty (Ramirez and McCann, 1963; Smith and Davidson, 1968). A third possibility is that whether or not the prepubertal ovary is secreting any estradiol, perhaps some other hormone or metabolic factor is interfering with neural responsiveness to estradiol. The preponderance of the available data is consistent with this third possibility. Ovarian hormones d o not restrain eating or body weight prepubertally. Ovariectomy at birth or at the time of weaning does not affect either measure until the time intact controls reach puberty. Then the ovariectomized animals eat and weigh more than the intacts (Grunt, 1964; Slob, 1972; Wade and Zucker, 1970a). However, there is no particular significance t o the relation between puberty and ovarian restraint of body weight, since induction of precocious puberty does not cause a precocious slowing of growth (Wade and Zucker, 1970a). This is an important point indicating that the neural changes causing puberty and a slowing of growth are not causally related. Rather, i t is likely that both effects are caused by a third factor-attainment of a certain minimum body weight (see below).

238

GEORGE N. WADE

The neural mechanisms controlling eating and body weight to not respond to exogenous hormones prepubertally. Injection of ovariectomized rats with suprathreshold doses (for adults) of estradiol benzoate does not suppress eating until approximately 40 days of age (Fig. 12). This is about the same time ovarian estradiol first restrains eating in intact females (Wade and Zucker, 1970a). Similarly, progesterone injections do not stimulate food intake in intact weanlings as they do in adults (Ross and Zucker, 1974). This makes sense: if estradiol does not restrain eating, progesterone cannot block its effects. However, estradiol and progesterone do affect body weight in weanlings, to some extent. Estradiol benzoate depresses body weight before i t has any effect on eating, and progesterone increases weight gain in intact weanlings without affecting eating (Ross and Zucker, 1974; Wade, 1974; Zucker, 1972). These effects are relatively modest when compared to the effects seen in adults, and they very likely represent direct effects of the steroids on metabolic processes cited previously. In addition, the effect of prepubertal estradiol on body weight may not be reversible (Wade, 1969 unpublished data). Why doesn’t estradiol inhibit food intake in weanlings as it does in adults? We have suggested previously that the ventromedial hypothalamus does not restrain eating and weight gain in immature rats. If estradiol inhibits eating by acting on the ventromedial hypothalamus (or other brain sites restraining body weight), then it should have no effect in an animal with an impotent ventromedial hypothalamus (Wade, 1974; Wade and Zucker, 1970a). Prepubertal rats are hyperphagic (intake computed on a body weight basis), and they eat as much as adults with ventromedial hypothalamic lesions H 0-0 0-0 0-0

SPAYED 0 INTACT8 SPAYED 8 SPAMD $

t 8

t

dl days Jo-50 oil Qp 30-50

EB W S30-50 $ EB dm 40-50

FIG. 12. Food intake of immature ovariectomized rats treated with 1.0 ccg estradiol benzoate (EB)/day or with sesame oil vehicle. A group o f intact females treated with the oil vehicle was included for comparison. Duration of treatment is indicated. (From Wade and Zucker, 1970a.)

HORMONES AND BODY WEIGHT

239

(Kennedy, 1957). In addition, neither electrolytic lesions of the ventromedial hypothalamus nor parasaggital knife cuts between the ventromedial and lateral hypothalamus (both of which induce overeating and obesity in adult rats) causes any additional hyperphagia in immature female rats (Bernardis, 1966; Bernardis and Skelton, 1965-1966; Gold and Kapatos, 1975; Kennedy, 1957). This is not simply due to a ceiling effect; weanlings do increase their food intake when placed in the cold (Teitelbaum etul., 1969). Rats lesioned as weanlings d o overeat and outgain unlesioned animals, but only after they reach 7-8 weeks of age, the time the intact females reach puberty (Gold and Kapatos, 1975). Kennedy (1957) has suggested that the postpubertal overeating and weight gain is simply a continuation of the juvenile pattern. Thus, the ventromedial hypothalamus only restrains eating and weight gain postpubertally. It is very likely that pituitary hormones are either directly or indirectly (through their metabolic effects) responsible for the lack of hypothalamic restraint of eating and body weight in immature rats. Hypophysectomy of weanling rats depresses eating and slows growth. Rapid weight gain and hyperphagia can be restored in these hypophysectomized weanling either by ventromedial hypothalamic lesions or by daily injections of growth hormone (Goldman et ul., 1970; Han, 1967; Kurtz etul., 1972), neither of which has any effect in intact weanlings. These effects of ventromedial hypothalamic lesions and growth hormone in immature rats are not additive, suggesting a common mode of action: inhibition of the ventromedial hypothalamus. How does growth hormone inhibit the ventromedial hypothalamus? One obvious possiblity is that growth hormone acts directly on the neurons in this region to inhibit their activity. I am aware of no data bearing directly on this hypothesis, although it is clear that other hormones, such as insulin or estradiol, can act directly on the ventromedial hypothalamus (Debons el ul., 1969; Wade and Zucker, 1 9 7 0 ~ ) .An alternative (and not mutually exclusive) possibility is that growth hormone causes certain metabolic changes that feed back to affect hypothalamic activity. Kennedy (1967, 1969) has suggested that, in the weanling rat, lipostatic regulation by the ventromedial hypothalamus is fully mature. However, because growth hormone stimulates rapid growth of the lean body mass and inhibits lipogenesis, there is little accumulation of body fat. Because there is little body fat in weanling rats, the ventromedial hypothalamic lipostat is not triggered until after puberty when growth slows and body fat begins to accumulate. Thus, growth hormone masks ventromedial hypothalamic restraint of eating by eliminating the normal satiety signals t o the neural lipostat. If this is the case, it is no wonder that estradiol, which only causes a ‘‘fine tuning” of lipostatic regulation (Section IV,B), is not effective in an animal with little body fat. This hypothesis predicts that if hypophysectomy (and increased fat deposition) restore lipostatic control, then estradiol should depress eating in hypophysec-

240

GEORGE N. WADE

tomized weanlings. This is, in fact, the case. Treatment of ovariectomizedhypophysectomized weanlings with a dosage of estradiol benzoate (1 pg/day) that is ineffective in nonhypophysectomized weanlings immediately depressed both eating and body weight (Wade and Zucker, 1970a). If growth hormone is really the pituitary factor responsible for the prepubertal refractoriness to estradiol, then it should be possible to restore the refractoriness in hypophysectomized weanlings with exogenous growth hormone. Again, this is the case. The effects of estradiol benzoate on eating in hypophysectomized-ovariectomized weanling rats were completely blocked by concurrent injections of bovine growth hormone (Wade, 1974). These results raise the question: Why does pituitary growth hormone inhibit responsiveness to estradiol in weanlings but not in adults? It had been suggested previously that growth hormone secretion might decline at about the time estradiol begins to influence food intake (Kurtz etul., 1972; Wade and Zucker, 1970a). More recent evidence indicates that this is clearly not the case: plasma growth hormone levels rise sharply at puberty (Dickerman el ul., 1972). Whether or not plasma growth hormone levels drop, there is a decreasing responsiveness to the actions of growth hormone with increasing age and/or body weight (Emerson, 1955). A reduced metabolic responsiveness to growth hormone, rather than lower plasma levels, could be the reason that adult females respond to estradiol but weanlings do not. To summarize, it is hypothesized that pituitary growth hormone masks lipostatic regulation of eating and body weight in immature rats by abolishing the normal satiety signals to the brain. Because estradiol alters eating by way of this lipostatic mechanism, it is ineffective in weanlings. Hypophysectomy abolishes and growth hormone restores the refractoriness of the neural eating system t o estradiol. The work of Zucker (1972) is consistent with this hypothesis. He reared female rats in litters of 3, 9, or 15 pups. This resulted in rats of a wide range of body weights at weaning. The rats were then divided into three groups of high, medium, and low body weights. They were ovariectomized at 23 days, and half the animals were treated with 1 pg estradiol benzoatelday on days 30-55. Although the three groups exhibited undereating in response to the estradiol at dissimilar ages (the heavier rats responded earlier), hypophagia attributable to the estradiol did occur at a similar body weight in all three groups. Each of the estradiol-treated groups began to eat significantly less than its respective control group when it reached approximately 160 gm. Therefore, it is likely that the attainment of a certain minimum body weight (or body fat content) is a prerequisite for the appetite-suppressing actions of estradiol, rather than the attainment of any particular chronological age (Zucker, 1972). By using the same technique, Kennedy and Mitra (1963b) have shown that puberty also is more closely related to body weight than to chronological age in rats. It is likely, then,

HORMONES AND BODY WEIGHT

24 1

that puberty and eating responsiveness to estradiol are temporally associated, not because of a causal relation between the two, but because they are both triggered by the attainment of a certain body weight. The refractoriness of the prepubertal neural feeding system is probably restricted to stimuli, such as estradiol, that affect food intake via the lipostatic mechanisms in the ventromedial hypothalamus. Amphetamine, insulin, and environmental temperature all affect eating in prepubertal rats (Lytle et al., 1971; Teitelbaum et al., 1969; Wade and Zucker, 1970a), but none of these stimuli acts via the ventromedial hypothalamus. It is likely that amphetamine and insulin act via the lateral hypothalamus, whereas thermal stimuli act on the anterior hypothalamus-preoptic area to influence eating (Booth, 1968; Carlisle, 1964; Epstein and Teitelbaum, 1967; Hamilton and Brobeck, 1964). The neural system controlling voluntary exercise is also refractory to exogenous estradiol prepubertally. Even very large doses of estradiol benzoate (up to 100 pg/day) do not stimulate wheel running in ovariectomized weanlings (Gentry and Wade, 1974 unpublished data; Kennedy, 1964; Porterfield and Stern, 1974). The refractoriness of the neural exercise system may also be due to the actions of pituitary growth hormone. Porterfield and Stem (1974) report that hypophysectomized-ovariectomized weanlings increase their locomotor activity in response to exogenous estradiol benzoate before intact weanlings do. In addition, this responsiveness to estradiol injections can be delayed by concurrent injections of bovine growth hormone (1 mglday). These results may mean that growth hormone can act directly on the brain to inhibit responsiveness to estradiol independently of any effects on lipostatic mechanisms. Other antagonisms between estradiol and growth hormone have been noted previously (Josimovich et al., 1967; Schwartz et al., 1969). Note, however, that the inhibition of running by growth hormone did not last indefinitely; activity eventually increased in spite of continuing growth hormone treatment (Porterfield and Stern, 1974). We, too, have had little luck in inhibiting estradiol-induced running with growth hormone in adult rats (Gentry and Wade, 1973 unpublished data). Finally, Rothchild (1969) has reported that progesterone does not raise body temperature in prepubertal rats as it does in adults. The physiological basis for this refractoriness to progesterone is, as yet, unspecified. In conclusion, prepubertal female rats behave and gain weight much like estradiol-deprived adults, not because of a lack of estrogenic stimulation, but because they are unresponsive to estradiol. This refractoriness to estradiol is probably due to pituitary growth hormone which may act on peripheral metabolic processes to alter the chemical feedback signals to the brain and interfere with neural restraint of eating behavior. It is interesting t o note that prolactin, which has many properties in common with growth hormone, may interfere with hypothalamic restraint of food intake during lactation and abolish responsiveness to estradiol, just as growth hormone

242

GEORGE N.WADE

does in weanling. Lactating females are hyperphagic, eating at least as much as nonlactating females with ven tromedial hypothalamic lesions. In addition, just as in weanling, ventromedial hypothalamic lesions do not induce additional hyperphagia or obesity in lactating females. Lesioned females do not become obese until lactation is terminated (Kennedy, 1953). This lactation-induced inhibition of ventromedial hypothalamic restraint of appetite also seems t o abolish responsiveness to exogenous estradiol. Treatment of lactating females with up to 5.0 pg estradiol benzoate/day does not significantly depress either food intake or body weight (Fleming, 1974 personal communication; J. M. Stern, 1974 personal communication). This refractoriness to estradiol benzoate is not due to high levels of plasma progesterone, since ovariectomized lactating rats do not respond to estradiol benzoate with a decrease in food intake or body weight (J. M. Stern, 1974 personal communication); it is likely that this refractoriness is partially due to prolactin. How could prolactin induce a refractoriness to estradiol in the neural feeding system? Prolactin probably does not act directly upon the brain to prevent the effects of estradiol benzoate. Treatment of ovariectomized virgin rats with prolactin has no effect on estrogenic suppression of food intake or body weight, indicating that the prolactin is probably not interfering with the estradiol activity directly (Fleming, 1974 personal communication). It is possible that the postparturient metabolic effects of prolactin (lactation) interfere with hypothalamic restraint of appetite and responsiveness t o estradiol, since a large amount of energy must be diverted into milk production. Ota and Yokoyama (1967a,b) have reported a hgh correlation between levels of lactation and food consumption in intact and ovariectomized postparturient rats; females with large litters eat more than females nursing small litters. This tremendous energy drain could induce the hyperphagia and mask hypothalamic restraint. Consistent with this possibility are data suggesting that estradiol benzoate treatments that interfere with prolactin secretion and lactation will also depress food intake (Fleming, 1974 personal communication; J. M. Stern, 1974 personal communication). Similarly, ventromedial hypothalamic lesions that interrupt lactation also induce an immediate obesity rather than a delayed obesity which can be produced by lesions that do not interrupt lactation (Kennedy, 1953). Thus, either growth hormone or prolactin appears to be able to act on metabolic processes of rats to mask hypothalamic restraint of food intake and, thereby, interfere with the appetite-depressing actions of estradiol. Note that there only seem to be certain times when growth hormone and prolactin have these effects. Immature rats are most responsive to growth hormone, whereas lactating females are highly responsive to prolactin. Adult cycling females are not especially responsive to either.

HORMONES AND BODY WEIGHT

243

VI. SEX DIFFERENCES IN NEUROENDOCRINE REGULATION OF BODY WEIGHT A.

ORGANIZING EFFECTS OF PERINATAL HORMONES

Males of most mammalian species are larger and heavier than their female conspecifics (Kakolewski et al., 1968; Tanner, 1962). This sex difference is also evident in rats; males eat and weigh more and exercise less than females in adulthood (see Section 11,A). What are the hormonal factors responsible for these striking differences in behavior and body weight? The most obvious explanation could simply be that, in adulthood, males and females secrete different hormones and that these different hormones are responsible for the sexual dimorphisms. Males secrete androgens that stimulate eating and weight gain but that are rather ineffective in stimulating exercise. On the other hand, females secrete large quantities of estrogens that depress eating and body weight and stimulate voluntary exercise (Section 111). Consistent with the possibility that the sexual dimorphism is determined by the activating effects of adult hormones is the report that the body weights of male and female rats differ only very slightly prior to 4-6 weeks of age, about the time the ovary begins to restrain body weight (Slob, 1972). Through 4 weeks of age the body weights of the two sexes are virtually identical; then the growth rate of the females slows and body weights diverge with age (Fig. 13). Thus, before the gonads are able to affect body weight (see Section V), there is almost no sex difference in body weight. However, although postpubertal gonadal secretions undoubtedly contribute to the sex differences in body weight, other factors must also be operating. Whereas adult gonadectomy may abolish the sex differences in carcass composition (males typically have a higher fat content) (Leshner and Collier, 1973), it does not abolish the sex difference in body weight (Kakolewski et al., 1968; Leshner and Collier, 1973). There is, of course, an obvious explanation for this inability to a b o h h the adult weight difference: by the time rats reach adulthood there are substantial differences in skeletal size that cannot be readily altered by gonadectomy. However, this is still only a partial explanation. Slob (1972) has shown that gonadectomy at 21 days of age, before there are any sex differences in body weight or skeletal size, attenuates, but does not abolish, the adult sex difference in body weight. Thus, a sexual dimorphism does develop in the absence of any postpubertal activating hormones, indicating that, although adult hormones do influence body weight, additional factors are also operating. What factors other than adult hormone secretions could be responsible for the sexual dimorphism in body weight? One factor could be genetic differences, since there is some sex difference in body weight at birth (King, 1915; Slob,

244

GEORGE N. WADE

1972). However, it is unlikely that genetic or prenatal factors play a significant role in determining the sex differences in body weight (see p. 246). Another possibility is that testicular hormones, secreted during the early postnatal period, are acting on the central nervous system and/or other nonneural tissues to alter 9 1

200

1 .

-

c . d In. 81 #day 21 ln.131 # day 1 (n.141

r

t 9

0-Q

0

'1 "1

0

2

L

6

8

10

(n= 91 day 21 (n=111 day 1 In=lZl

16

1L

12

wight velocity

b

0

0

2

4

6

8

eoy

i

lO121416 ogm in wm&r

FIG. 13. Body weight and growth rate of male and female rats left intact (3or gonadecor 21 (a) days of age. (From Slob, 1972.) tomized at 1 (9

HORMONES AND BODY WEIGHT

245

the adult body weight set-point and adult responsiveness to activating hormones (Beatty et al., 1970; Bell and Zucker, 1971; Slob, 1972). An overwhelming body of evidence now suggests that many, if not all, sex differences in behavior and physiological functions are determined at least in part by the presence or absence of testicular androgens during a particular stage early in development. Since this field has been more than adequately reviewed (for a variety of viewpoints see, e.g., Beach, 1971; Gorski, 1971; Goy, 1970; Harris, 1964; Phoenix et al., 1967), I will attempt only a brief summary here. Early in development, the rat’s nervous system is inherently feminine, no matter what the animal’s genetic sex. In the absence of any androgenic stimulation the brain will continue to develop in a feminine pattern, will control cyclic pituitary gonadotropin secretion, and will show a typical feminine responsiveness to hormonal stimulation in adulthood. A feminine nervous system can be induced in genetically male rats by depriving them of perinatal androgens, usually by castration immediately after birth. On the other hand, if the developing rat’s brain is exposed to androgens sometime between the late prenatal period and approximately the tenth postnatal day, it will be masculinized and defeminized. In adulthood this masculine brain can only support tonic pituitary hormone secretion and is much less responsive to ovarian hormones than a nonmasculinized brain. Genetic female rats can be masculinized by a single injection of testosterone (or estradiol) shortly after birth. These permanent, irreversible effects of androgens early in development have been termed organizing effects, and stand in contrast to the transient, reversible activating effects of steroids secreted in adulthood (Phoenix et al., 1967). Therefore, it is quite possible that neonatal testicular secretions could also alter the neural substrates for body weight and regulatory behaviors, regardless of the hormones secreted in adulthood. Consistent with this possibility are the several reports that injection of female rats with testosterone propionate within the first week after birth increases body weight in adulthood when compared with vehicle-treated littermates (Beatty et al., 1970; Bell and Zucker, 1971; Slob, 1972; Swanson and van der Werff ten Bosch, 1963; Valenstein, 1968). However, these data by themselves do not necessarily demonstrate that neonatal testosterone organizes body weightregulating mechanisms. It is entirely possible that all these neonatal androgen injections are doing is to suppress ovarian functioning in adulthood; that is, perhaps the androgenized females weigh more simply because they are estrogendeprived as adults. To demonstrate conclusively that neonatal hormones do organize body weight-regulating mechanisms, it must be shown that neonatal hormonal manipulations affect adult body weight in the absence of activating hormones in adulthood. This has been done. Slob (1972) has found that whereas gonadectomy at 21 days of age (before the sexes diverge) does not prevent the adult sexual dimorphism in body weight, gonadectomy on the day of birth abolishes the adult sex difference in body

246

GEORGE N.WADE

weight (see Fig. 13). Thus, the gonads are doing something between birth and 21 days of age that results in a later divergence in body weight. Furthermore, it is the neonatal testis, and not the ovary, that is responsible for the sex difference. There is n o difference between day 1 and day 21 ovariectomized females in adult body weight, but males castrated on day 1 are significantly lighter than day 2 1 castrates in adulthood (Slob, 1972). An important point emerging from this work is that neonatal gonadectomy abolishes adult sex differences in body weight, indicating that neither genetic nor prenatal environmental factors are sufficient to cause the adult sexual dimorphism. If neonatal castration reduces male body weights, will neonatal testosterone injections increase female body weights in the absence of ovaries? Both Slob (1972) and Bell and Zucker (1971) have found that a single neonatal injection of testosterone propionate significantly increases the adult body weights of female rats ovariectomized just after birth. The results of these experiments demonstrate that neonatal androgens can organize adult body weight-regulating mechanisms independently of adult activating hormones. Of course, we have seen that adult hormones d o also play a role in determining the sex difference, so that gonadal hormones both organize and activate sex differences in body weight in rats. There may be a third way in which gonadal steroids contribute to sexual dimorphisms in body weight: by an interaction between organizing and activating influences (i.e ., neonatal hormones may organize adult responsiveness t o the activating hormones). It is very clear from the work on copulatory behavior that male and female rats are not equally responsive to activating hormones and that this sex difference is determined by neonatal hormones (see Beach, 1971, for a discussion of this point). Bell and Zucker (1971), Slob (1972), and Beatty et al. (1970) have examined the effects of neonatal hormonal manipulations (gonadectomy and/or various steroid treatments) on adult responsiveness t o steroids. There seems t o be a general consensus that neonatal exposure to testosterone enhances the weightpromoting effects of low doses of testosterone propionate given to gonadectomized adult rats of either genetic sex. Although it is clear that neonatal androgen treatment increases body weight in female rats, this treatment does not abolish responsiveness to ovarian hormones. Adult ovariectomy increases body weight in rats treated with testosterone neonatally, indicating that the ovary does restrain weight gain in these animals. However, Bell and Zucker (1971) have found that, whereas nearly all of their neonatal treatment groups lost weight during adult injections of a mixture of estradiol benzoate and progesterone, animals exposed t o testosterone just after birth lost significantly less weight than nonandrogenized rats. These experiments indicate that neonatal hormone exposure does organize adult responsiveness to the body weight effects of adult activating hormones; animals exposed to androgens in infancy are less responsive to the weight-depressing actions of ovarian hormones than are animals not androgenized as neonates.

HORMONES AND BODY WEIGHT

247

In summary, there are three ways in which gonadal steroids act to determine the sex differences in adult body weight in rats. First, neonatal testicular secretions organize body weight-controlling mechanisms to enhance adult body weight independently of adult steroidogenesis. Second, hormones secreted in adulthood act either to promote (testosterone) or to inhibit (estradiol) further weight gain, i.e., activating effects. Finally, there is an interaction between the organizing and activating effects. The neonatal hormonal environment determines the adult responsiveness to ovarian and testicular hormones. It is clear that neonatal hormones have organizing effects on adult body weight, but are the sex differences in behavior organized similarly? Neonatal androgen exposure decreases voluntary activity levels of rats and attenuates adult running responses to exogenous estradiol. Gerall (1967) reported that a single injection of 1250 pg testosterone propionate at 5 days of age significantly reduced adult running wheel activity when rats were tested with their ovaries. In addition, Gerall et al. (1972) have shown that neonatal injection of 5, 10, or 1250 pg testosterone propionate reduces running wheel activity in a dosedependent fashion, when rats are ovariectomized and injected with exogenous estradiol benzoate as adults. Harris (1964) has indicated that, when both groups are given ovarian transplants, neonatally castrated male rats show higher levels of running wheel activity than males castrated as adults. Therefore, i t appears that naonatal exposure to androgens greatly reduces adult running responses to either endogenous or exogenous estradiol. In contrast to the several published reports of organizing effects of steroids on body weight in rats, I am awaR of only one systematic study of the organizing and activating effects of sex hormones on eating behavior (Bell and Zucker, 197 1). From this work it appears as though the sex difference in eating behavior is little affected by the neonatal hormonal environment in the absence of activating hormones. Bell and Zucker found that, whereas intact males ate more than intact females, gonadectomized adults of both sexes displayed intermediate levels of food intake that were not affected by neonatal hormone manipulations (gonadectorny and/or androgen treatment). Animals exposed t o neonatal androgens weighed, but did not eat, more than nonandrogenized animals in the absence of activating hormones. The neonatal hormonal environment did not seem to have much effect on the eating responses of male rats when they were given exogenous hormones in adulthood. Testosterone propionate, which stimulated weight gain more in neonatally androgenized rats, did not differentially affect food intake in males, regardless of neonatal treatment. Also, the estradiol benzoate + progesterone, which depressed body weight more in the nonandrogenized groups, seemed to be equally effective in inhibiting food intake in all neonatal treatment groups in males. By contrast, the eating responses of female rats to activating hormones were dramatically affected by neonatal androgen injections. Females treated neonatally with oil decreased their food intake significantly more than neonatally

248

GEORGE N. WADE

androgenized females when treated with estradiol benzoate + progesterone as adults (Bell and Zucker, 1971). We have found that male rats castrated as adults and neonatally androgenized females show significantly smaller decreases in food intake than do nonandrogenized females when given estradiol benzoate alone as adults (Gentry and Wade, 1975). Why should the eating responses of females be affected by neonatal androgen treatments when those of males apparently are not? One possibility is that, in the Bell and Zucker experiment, all of their males had already been at least partially masculinized by endogenous testosterone prior t o any neonatal manipulations. According to their procedure, the rats were castrated “within the first 24 hr after birth.” However, as Thomas and Gerall(l969) point out, significant masculinization is occurring just hours after birth. These few hours of androgen exposure may have been sufficient to alter the adult eating respones of the males. Another possibility is that when Bell and Zucker were comparing the responsiveness of the various neonatal treatment groups to ovarian hormones, they injected a combination of estradiol benzoate and progesterone. It has been shown that if lordosis is used as a behavioral end point, neonatal androgenization reduces behavioral responsiveness to both estradiol and progesterone. In fact, the effect on responsiveness to progesterone may be the more dramatic of the two (Clemens et al., 1970). Therefore, the possibility remains that so far as eating behavior is concerned the male rats exposed to androgens neonatally might be less responsive to both the appetite-depressing action of estradiol and the appetite-stimulating action of progesterone. If both estradiol and progesterone are given simultaneously, the two effects might cancel one another, and it would appear as though the androgenized and nonandrogenized males are equally responsive t o ovarian steroids. This possibility remains to be tested. In conclusion, it is clear that the sex differences in adult rats’ body weights are due to both organizing and activating effects of gonadal hormones. The neonatal environment directly affects adult body weight-regulating mechanisms and also alters adult responsiveness to activating hormones. The activating hormones secreted in adulthood then act to exaggerate these sex differences. B. SEX DIFFERENCES IN HYPOTHALAMIC CONTROL OF BODY WEIGHT

Since the brain seems to defend different body weight set-points and compositions in male and female rats, reports that hypothalamic lesions produce different effects on eating and body weight in the two sexes should not be especially startling. Cox etal. (1969) found that lesions of the ventromedial hypothalamus caused greater hyperphagia and increase in weight gain in female rats than in males.

HORMONES AND BODY WEIGHT

249

Following ventromedial hypothalamic lesions, females showed a dramatic hyperphagia (around a 70% increase in eating) and increased weight gain, but males were only moderately hyperphagic (about a 20-30% increase) and showed n o weight gain when compared with unlesioned controls. These data have essentially been reproduced using a more palatable high-fat diet (Cox et al. used ground Purina rat chow) (Rehovsky and Wampler, 1972). Gold (1970) has also found that a combination of a high-fat diet and bilateral parasaggital knife cuts between the ventromedial and lateral hypothalamus resulted in greater overeating and weight gain than Cox et al. (1969) found with electrolytic lesions. However, the sex difference persisted; females showed a greater hyperphagia and weight gain relative to unoperated controls than the males did. Therefore, although there may be some differences in interpretation of the results (Gold, 1970; Rehovsky and Wampler, 1972), it is clear that either ventromedial hypothalamic lesions or parasaggital knife cuts produce a greater hyperphagia and weight gain in female than in male rats on a variety of diets when lesioned animals are compared to unlesioned controls (or to their own preoperative data). What is the physiological basis for this sex difference in response to ventromedial hypothalamic lesioning? In experiments in which comparisons are easily made (Cox et al., 1969; Gold, 1970), it is apparent that the sex differences in body weight gain are much more impressive in unlesioned rats than in lesioned rats. Before ventromedial hypothalamic lesions (or in sham-operated controls), males eat more and gain weight faster than females. After lesioning, males and females seem t o gain at nearly the same rate. In the Cox et ul. experiment, males showed no increase in weight gain after lesioning, but females increased their rate of weight gain to match that of the males. In the Gold experiment, the knife cuts accelerated the males’ weight gain, but the females increased weight even more so that they were gaining at approximately the same rate as the lesioned males. Therefore, i t appears as though there is n o sex difference in weight gain following ventromedial hypothalamic damage; the sexual dimorphism in response to lesioning is due to the abolition of the prelesioning sex difference in weight gain. If this is the case, perhaps we should rephrase the problem: Why is there a sex difference in prelesion body weight, and why does a ventromedial hypothalamic lesion abolish this difference? I hope that the first half of this question was answered, at least in part, in the previous section. Males seem to be heavier than females because of both organizing and activating actions of sex steroids (see Section V1,A). But why should ventromedial hypothalamic lesions abolish both the organizing and activating influences of sex hormones on body weight gain? Let us first consider the activating effects. Valenstein et al., (1969) suggested that ventromedial hypothalamic lesions may be partially gonadectomizing rats by decreasing pituitary gonadotropin release. In fact, they did find some

250

GEORGE N.WADE

evidence of gonadal atrophy (suggesting decreased hormone output) following their lesions. If this is the case, then a reduction in circulating estradiol should enhance weight gain, whereas a decrease in testicular androgen output should decrease weight gain. A partial gonadectomy could certainly contribute to the sex difference. Whether or not the lesions of the ventromedial hypothalamus decrease circulating hormone titers, these lesions could also interfere with the activating effects of steroids by removing some of their neural sites of action. Recall that estradiol implanted in the ventromedial hypothalamus depressed eating and body weight (Beatty etal., 1974; Porterfield and Stern, 1974; Wade and Zucker, 1970c) and that lesions of the ventromedial hypothalamus attenuated the weight-depressing actions of estradiol (Beatty er al., 1975; King and Cox, 1973). Thus, the lesions could be (partially) functionally ovariectomizing the females by removing a neural site of action of estradiol. Thus, there are at least two ways ventromedial hypothalamic lesions could interfere with the activating effects of sex hormones and contribute to the attenuation of the sex difference following lesioning: by causing gonadal atrophy and/or by removing some of the neural target sites for estradiol. If an attenuation of the activating effects is really a contributing factor, then longterm gonadectomized rats should show a significantly reduced sex difference in response to ventromedial hypothalamic lesions. Conadectomy substantially reduces the normal sex difference in body weight in adult rats. Reducing the prelesioning sex differences should also attenuate the relative changes following lesioning, since the similar postlesion weight gains would be compared with more comparable base-line values. Although some effort has been made in this direction (Valenstein er ul., 1969), the appropriate experiments remain to be done. Ventromedial hypothalamic lesions could also abolish the sex differences in hypothalamic regulation of body weight organized by neonatal hormone exposure. It has been shown that neonatal exposure to androgens permanently raises adult body weight independently of activating hormones (see Section VI,A), and Beatty et ul. (1970) have argued that testosterone might act on hypothalamic weight-regulating systems to produce this effect. It would not be unreasonable to suppose that the ventromedial hypothalamus might be a site of neonatal androgen action (Gorski, 197 1). Perhaps, then, ventromedial hypothalamic lesions dictate a high (and similar) level of body weight in both males and females. Consistent with this possibility is the report that neonatally androgenized females, whose prelesion body weights are intermediate between those of normal males and females, show a response to ventromedial hypothalamic lesions that is intermediate t o those of normal males and females (Valenstein, 1968). However, because these rats were tested with their gonads intact, it is not possible to rule out the possibility that the androgenized females were different from the normal females simply because their ovaries were secreting different activating hormones prior to lesioning.

HORMONES AND BODY WEIGHT

25 1

To summarize, it has been suggested that male and female rats respond differently to ventromedial hypothalamic lesions, not because there is a sex difference in weight gain after lesioning, but because they start from different body weight set-points (levels of weight gain). Males are heavier (and gain faster) than females. Therefore, when similar postlesion weight gains are compared with different baselines, females exhibit a greater change. There are several ways in which ventromedial hypothalamic lesions could interfere with the organizing and activating effects of sex steroids that are responsible for the prelesion sex differences. Pfaff (1969) has very cleverly suggested another cause for the sex difference in hypothalamic hyperphagia and obesity-one that is independent of gonadal steroids. He noted that growth hormone injections stimulated weight gain in hypophysectomized male rats, but they were relatively ineffective in females. On the other hand, daily injections of prolactin stimulated eating and weight gain more in hypophysectomized females than in males. It has been known for some time that lesions of the medial basal hypothalamus decrease pituitary growth hormone secretion and also disinhibit pituitary secretion of prolactin (McCann et al., 1968). Thus, after ventromedial hypothalamic lesions, the drop in plasma growth hormone levels could inhibit weight gain in males without having much effect in females. On the other hand, the postlesioning increase in plasma prolactin should increase the eating and weight gain of females without affecting the males. Perhaps, then, these changes in pituitary hormone secretion are responsible for the sex differences (Pfaff, 1969). Although this hypothesis seems to be quite reasonable, I am not as yet convinced that this is really the basis for the sex difference in hypothalamic obesity. It has been exceptionally difficult, if not impossible, to influence the eating and weight gain with exogenous growth hormone or prolactin in nonhypophysectomized rats (Fleming, 1974 penonal communication; Gentry and Wade, 1974 unpublished data). In addition, i t appears as though hypophysectomized female rats gain weight nearly as fast as nonhypophysectomized females after ventromedial hypothalamic lesions (Valenstein et ul., 1969). Pfaffs hypothesis would predict a substantial difference between the two groups. Clearly, more research is necessary to evaluate this hypothesis. There also appears to be a sex difference in the effects of lateral hypothalamic lesions in rats. Powley and Keesey (1970) reported that male rats that had “recovered” from lateral hypothalamic lesions chronically regulated their body weight at a point below that of unlesioned control males. For this and other reasons, they suggested that lateral hypothalamic lesions permanently lower the body weight set-point, just as ventromedial hypothalamic lesions seem to raise the body weight set-point permanently (Hoebel and Teitelbaum, 1966). However, several laboratories have failed to observe this phenomenon in lesioned female rats; after lateral hypothalamic lesions, female rats regulate their body weight at levels identical to those of unoperated controls (Cox and

252

GEORGE N. WADE

Kakolewski, 1970; Harrell and Balagura, 1975; Mufson and Wampler, 1972). Why did Powley and Keesey (1970) find a chronically reduced body weight after lateral hypothalamic lesions in male rats whereas other researchers did not reproduce these findings in females? Mufson and Wampler (1972) rejected the possibility of a sex difference in response to lesions and instead suggested that the Powley-Keesey phenomenon was simply due to finickiness induced by the lesions. They suggested that “recovered” lateral hypothalamic-lesioned rats are extremely finicky, and, because Powley and Keesey maintained their rats on a relatively unappealing diet (Purina rat chow and tap water), their animals were underweight because of hypophagia and dehydration. To support their claim, Mufson and Wampler point out that both their rats, given a very palatable high-fat diet and sweetened water, and those of Cox and Kakolewski, given a very attractive (to rats), moist cat and dog food, showed no chronic weight reduction. However, it is very clear that the Mufson and Wampler hypothesis simply does not account for the data. Cox and Kakolewski (1970) indicated that males lesioned and maintained exactly like their females responded as in the Powley and Keesey experiment; their body weight was chronically lowered even on a very palatable diet. Also, HarreIl and Balagura (1975) have recently replicated the Cox and Kakolewski results in females (no lowered body weight after “recovety”) maintained on Purina rat chow and tap water, which is the same diet as that used by Powley and Keesey. Finally, Keesey and Boyle (1973) have found that lateral hypothalamic-lesioned males are no more responsive t o adulteration of their diets than are unlesioned males. It seems clear that there is a true sex difference in weight regulation after lateral hypothalamic lesions that is quite independent of diet. Another possible explanation for this sex difference is that, perhaps, males can afford to lose some weight after a lateral hypothalamic lesion but that females cannot; that is, perhaps we are seeing a basement effect in females but not in males. It is well known that male rats have a higher carcass fat content than females do (Leshner and Collier, 1973). Perhaps it is this extra fat that is permanently lost in males after lateral hypothalamic lesions. Females could already be regulating at a minimum weight and they might have no excess fat to shed following a lesion. If this is the case, adding some body fat to females by prior ovariectomy (Leshner and Collier, 1973) should give them something t o lose, and ovariectomized females should show a chronic weight reduction after lateral hypothalamic lesions, just as males do. However, Harrell and Balagura (1975) have recently shown that ovariectomized females behave just as intacts do following lateral hypothalamic lesions. There is no permanent change in body weight-they weigh just as much as unlesioned ovariectomized females. Thus, the inability of lean intact females to lose additional weight cannot explain the sex difference.

HORMONES AND BODY WEIGHT

253

The sex difference in weight regulation following lateral hypothalamic lesions might be a consequence of a sex difference in thermoregulation after lesioning. Wampler (1974) has suggested that the lowered body weights following lesioning in males might be very sensitive to environmental temperature; slightly lowering room temperature seems to restore body weights to prelesion values. Harrell and deCastro (1974 personal communication) have noted that male rats are hyperthermic (by approximately l0C) for at least 2 weeks following lesioning. This hyperthermia may cause a chronic undereating or at least prevent the hyperphagia necessary to restore prelesion body weights in male rats. On the other hand, female rats do not seem to be hyperthermic following the same kinds of lesions. Perhaps the absence of a hyperthermia permits the females to overeat following the postlesion weight loss and to retum their weights to prelesion levels. In conclusion, it is clear that there is a sex difference in body weight response to lateral hypothalamic lesions. After lesioning, male rats exhibit a permanently lowered body weight, whereas females do not. The lowered body weight in males is not an artifact of finickiness, since the sex differences have been reproduced using a variety of diets. Nor can the females' failure to lose weight permanently be due to a shortage of expendable fat, since obese ovariectomized females show no evidence of lowered body weight after lesioning. A final possibility is that the sex difference in weight regulation might be secondary to a sex difference in thermoregulation following lateral hypothalamic lesions. This hypothesis remains to be tested.

VII. HORMONAL EFFECTS ON TASTE PREFERENCES AND DIETARY SELF-SELECTION A.

HORMONES AND TASTE PREFERENCES FOR NONNUTRITIVE SOLUTIONS

By now it should be apparent that gonadal hormones affect how much rats eat. These same hormones also seem to modulate the taste preferences of rats and selection of dietary components. Thus, sex hormones could alter body weight and composition in rats by influencing both how much and what the animals eat. In 24-hour, two-bottle preference tests with distilled water, adult female rats exhibit greater preferences for sweet solutions than do adult males. This sex difference exists for both nutritive (glucose) and nonnutritive (saccharin) sweets (Valenstein etul., 1967b) and has been replicated in several strains of rats (Valenstein et ul., 1967a; Wade and Zucker, 1969a). Zucker (1969) has shown that this sex difference in saccharin preference is primarily due to the stimu-

254

GEORGE N. WADE

latory actions of ovarian hormones. Whereas castration of adult male rats produces only small increases in saccharin preference, postpubertal ovariectomy substantially diminishes the saccharin preferences of females. High saccharin preferences can be restored in ovariectomized females by daily injections of a combination of estradiol benzoate (0.5 pglday) and progesterone (0.5 mglday). Neither estradiol benzoate nor progesterone alone in a wide range of doses is effective in altering saccharin preferences (Zucker, 1969). The sexual dimorphism in adult saccharin preference is also determined in part by organizing actions of neonatal steroids. A single injection of testosterone propionate on the fifth day of life significantly reduces the saccharin preferences of female rats in adulthood (Wade and Zucker, 1969b). In addition, estradiol progesterone treatments, which enhance the saccharin preference of ovariectomized females, are completely ineffective in male rats castrated as adults (Zucker, 1969). Thus, the sex difference in saccharin preference in adult rats is

+

: z

12-

w P

-

10-

W 0 :

-

n

I -

5 ; I

s

V

V

2

4 -

I

2 -

\WEANING

PRE-PREFNANCY

PREGNANCY

DAYS

LACTATION

POST-LACTATKWI

FIG. 14. Saccharin preferences [saccharin (S) intake/water 0 intake] of female rats during pregnancy and lactation. (From Wade and Zucker, 1969a.)

255

HORMONES AND BODY WEIGHT

jointly determined by the activating effects of ovarian hormones in adulthood and by the organizing actions of neonatal androgens. In addition to stimulating saccharin preferences in adult females, ovarian hormones seem t o enhance responsiveness to aversive quinine solutions. Rats ovariectomized postpubertally will drink significantly more o f a very bitter 0.0075% solution of quinine sulfate than intact females will. Treatment of progesterone mixture, ovariectomized females with the estradiol benzoate iwhich stimulates saccharin intake, also restores responsiveness to quinine. As with saccharin intake, neither estradiol benzoate nor progesterone alone had any effect on quinine aversion in ovariectomized rats (Wade and Zucker, 1970b). Ovarian hormones appear to enhance responsiveness to both palatable and aversive nonnutritive solutions, stimulating both saccharin preference and quinine aversion. The changes in hormone secretion during various reproductive states also influence responsiveness to tastes (see Table I). Pregnant and pseudopregnant rats exhibit very dramatic decreases in saccharin preference and quinine aversion-just as can be shown with ovariectomized females (Figs. 14 and 15) (Wade and Zucker, 1969a, 1970b). Also, there is no sex difference in saccharin preference prior to puberty; both males and females reject saccharin. At about the

\/ 1

I

UNMATED PREGNANT o--O PREGNANT-NO WNlNE 0--0 PSNDOPREWT 0--0

t t

0

1

1

1

'

1

1

4

S

6

7

8

9

1011

DAYS

SINCE

'

1

12

MATING

FIG. 15. Fluid intake of pregnant, pseudopregnant, and unmated rats as a function of quinine adulteration. Quinine sulfate (0.0075%) was added to the water supply on days indicated by arrows. (From Wade and Zucker, 1970b.)

256

GEORGE N. WADE

time the females reach puberty their saccharin preferences increase, and the sex difference appears (Fig. 16) (Wade and Zucker, 1969a). There is a remarkable correlation between the responsiveness of rats to tastes and levels of various energy balance-regulating behaviors (see Table I). The only exception to this relation appears to be that taste preferences do not fluctuate with the estrous cycle but the other behaviors do (Wade, 1968 unpublished data). With this one exception, it is clear that rats demonstrating high food intake and weight gain and low levels of voluntary exercise (males and ovariectomized, pregnant, pseudopregnant, lactating or prepubertal females) are also relatively unresponsive to nonnutritive tastes. This very high correlation among the various behavion suggests that they have a common endocrine basis. We have hypothesized that, if estradiol secretion (or the plasma estradiol-to-progesterone ratio) is within a sufficiently high rang,as during the estrous cycle, then saccharin preference and quinine aversion are enhanced. If estradiol availability is reduced (as a consequence of low secretory rate or of high plasma proges-

w*. ..- _. F E Y A L C S

F 8 a

>

Y4LES

,/ /

W

W

: LL

/"

so20

ID

-

FRST ESTRUS

FIG. 16. Saccharin preference [saccharin (S) intakelwater ov) intake] and total fluid intake (saccharin -k water) of male and female rats as a function of age. Mean age at fiist estrous behavior is indicated. (From Wade and Zucker, 1969a.)

HORMONES AND BODY WEIGHT

257

terone levels), then responsiveness to tastes is reduced (Wade, 1972; Wade and Zucker, 1969a, 1970b). Thus, estradiol appears to be the principal ovarian hormone regulating nonnutritive taste preferences, and progesterone has secondary facilitatory and inhibitory effects. This hypothesis is certainly consistent with the changes in taste responsiveness during pregnancy and pseudopregnancy and following ovariectomy. In addition to decreased estradiol availability another factor may be operating in prepubertal females: an insensitivity to estradiol. Recall that estradiol does not affect eating or exercise in female rats until they reach a certain body weight (see Section V). A similar insensitivity to estradiol could be responsible, at least in part, for the diminished taste responsiveness in weanlings. Consistent with this possibility is the demonstration that induction of precocious puberty (and adult hormone secretion) in female rats does not induce a precocious elevation of saccharin preference (Wade and Zucker, 1969a). These data also indicate that attainment of puberty is not sufficient to enhance saccharin intake. Rather, the onset of a preference for saccharin solution appears to be more closely correlated with body weight than with puberty or chronological age in female rats (Wade, unpublished data). These results suggest that for taste preferences, as with eating and voluntary exercise, onset of responsiveness to ovarian hormones may be triggered by the attainment of a certain body weight or composition. Although these experiments point out that ovarian steroids do affect the taste preferences of rats, they tell us little or nothing about how the hormones might be effecting these changes. There are, of course, several possibilities. The high correlations between taste preferences and the other behaviors could indicate that the fluctuations in taste preferences are simply an artifact of changes in one of the other behaviors. For example, high saccharin preferences occur in rats with low food intake (intact, cycling females) but not in rats that eat a great deal (males and noncycling females). It has been shown that food deprivation increases the saccharin intake of rats (Valenstein, 1967; Wade and Zucker, 1969b), so that the intact, cycling females might be drinking more saccharin solution simply because they are eating less. However, it is possible to dissociate the changes in eating and taste preferences. For example, treatment of ovariectomized rats with estradiol benzoate alone depresses food intake without affecting taste preferences (Zucker, 1969). In addition, prevention of the usual postmating increase in food intake does not affect the decrease in saccharin preference during pregnancy (Wade and Zucker, 1969a). Finally, food intake, but not taste preferences, fluctuates with the estrous cycle. Therefore, the changes in eating and taste preferences for nonnutritive solutions cannot be causally related. Similar arguments would apply to the correlations between taste preferences and other hormone-related behaviors. A somewhat similar hypothesis has been advanced by Marks (1974) to account for the changes in “stimulus reactivity.” Marks argues that responsive-

258

GEORGE N. WADE

ness to tastes is determined by body weight, with a high body weight dictating a low responsiveness to tastes. The changes in taste preference following hormone treatments are viewed as being secondary to the changes in body weight. There does, indeed, seem to be a correlation between body weight and taste preferences, since rats with low responsiveness to tastes (males and pregnant, pseudopregnant, or ovariectomized females) do tend to be heavier than intact cycling females. However, Marks’s (1974) data often fail to demonstrate a close relation between body weight and responsiveness to tastes. In addition, treatment of ovariectomized females with estradiol benzoate has been shown to reduce body weight without affecting taste preferences (Wade and Zucker, 1970b; Zucker, 1969). Body weight and taste preferences cannot be directly related. A third possibility is that ovarian hormones could act on peripheral taste receptors to alter taste sensitivity. Although it is known that sex hormones can directly affect taste receptors (Hoshishima, 1967), this hypothesis remains to be tested. Perhaps the most appealing possibility is that ovarian hormones could act directly on the brain to alter the reinforcing properties of the various tastes. Many of the limbic and hypothalamic sites where lesions alter responsiveness to tastes also take up and retain estradiol and progesterone. Some notable examples include ventromedial hypothalamus, olfactory bulbs, septa1 area, and medial amygdala (Beatty and Schwartzbaum, 1967; Gesell and Fisher, 1968; Kemble and Schwartzbaum, 1969; Pfaff and Keiner, 1973; Teitelbaum, 1955; Wade and Feder, 1972a). Perhaps estradiol and progesterone act on one or more of these neural sites to alter taste preferences. This hypothesis could be tested by intracerebral hormone implant studies. To summarize, it is clear that sex steroids influence preferences and aversions of rats for nonnutritive taste factors. Ovarian hormones enhance the responsiveness to tastes of female rats, and during times of low estradiol availability (pregnancy, pseudopregnancy, following ovariectomy) saccharin preference and quinine aversion are reduced. The prepubertal depression in saccharin intake may be due to a weight-related lack of responsiveness to hormones. Although it is clear that the changes in taste preference are not secondary to fluctuations in body weight or other hormone-related behaviors, we do not know how or where ovarian steroids act to influence taste preferences. One appealing (and testable) hypothesis is that these hormones act on hypothalamic or limbic sites to alter the reinforcing properties of tastes. B.

SELECTION OF DIETARY PROTEIN

So what if sex hormones affect responsiveness to saccharin and quinine solutions? Real rats are not likely to be given a choice between saccharin or quinine and distilled water. What is the biological significance, if any, of these hormone-

HORMONES AND BODY WEIGHT

259

dependent sex differences and fluctuations in responsiveness to nonnutritive tastes? We (Wade, 1972; Wade and Zucker, 1969a) have suggested that these changes in taste preferences may represent a hedonic mechanism that influences self-selection of dietary components by rats during various reproductive states. It is well known that, if rats are placed in a cafeteria situation and allowed to self-select their diets, they are able to choose a diet that is compatible with normal growth and health. It has also been demonstrated that if selecting rats are subjected t o various environmental or endocrinological manipulations, they are able to make adaptive changes in their dietary self-selection patterns (e.g., Richter, 1956). More recently, Leshner and his colleagues have examined the dietary selection patterns of rats given a choice between a carbohydrate diet (0% protein) and a high-protein diet (45% protein) under conditions of varying nutritional requirements (Collier et al., 1969; Leshner, 1972; Leshner and Collier, 1973; Leshner e t a l . , 1971, 1972; Leshner and Walker, 1973). In these experiments they found that rats would vary their protein and carbohydrate intakes separately to meet different nutritional needs. For example, if male rats’ energy requirements are increased by allowing them access to running wheels or by placing them in a cold environment, they increase their carbohydrate intake without altering protein consumption (Collier et ul., 1969; Leshner et ul., 1971). This is certainly adaptive, since carbohydrates are a ready source of energy. Sex and reproductive status also affect protein-carbohydrate choice (see Table I). Adult male rats select a higher proportion of their diet as protein than do adult cycling females. This sex difference is primarily attributable to the suppressive actions of ovarian hormones on protein intake. Postpubertal ovariectomy increases the proportion of the diet selected as protein by females to levels equivalent to those of intact males, but castration of adult males seems to have no effect on protein-carbohydrate selection (Leshner and Collier, 1973). Prepubertal female rats select a greater proportion of their diet as protein than do adult cycling females. In addition, protein intake increases during pregnancy and pseudopregnancy, whereas carbohydrate intake remains unchanged (Leshner et al., 1972; Richter, 1956; Richter and Barelare, 1938). However, there is no fluctuation in dietary self-selection during the estrous cycle (Leshner and Collier, 1973). Once again, these differences in protein-carbohydrate selection seem to make sense. The animals that choose a higher proportion of protein in their diets (males and ovariectomized, pregnant, pseudopregnant, or prepubertal females) are all growing more rapidly than intact cycling females. The increased protein intake would be expected to facilitate the laying-down of new tissues in these rats. There is one aspect of the dietary self-selection work that does not seem to make sense at first. Most of the animals that show a higher dietary protein selection than intact cycling females (males and ovariectomized, pregnant, or pseudopregnant females) have higher carcass fat proportions than the cycling

260

GEORGE N. WADE

females (Galletti and Klopper, 1964; Hervey and Hervey, 1966; Leshner and Collier, 1973). Why should increased protein selection be correlated with high carcass fat content? A closer inspection of the data reveals that these rats increase their proportion of dietary protein by increasing protein intake without reducing carbohydrate intake. Thus, they are eating at least as much carbohydrate as cycling females (Leshner and Collier, 1973; Leshner et al., 1972). However, none of these groups is nearly as active as intact females (see Section 111). Perhaps these animals store the calories ingested as carbohydrate as fat rather than using them as an energy source for exercise as is done by more active females. In this way similar carbohydrate intakes could result in very different carcass fat levels, independent of dietary protein consumption. How is this selection of dietary protein determined? Note the virtually perfect inverse correlation between responsiveness to nonnutritive tastes and protein intake (see Table I). Animals that are highly responsive to tastes select significantly less protein in their diets than animals that are less responsive to saccharin and quinine. Also note that, during estrous cycles, neither protein intake nor taste preferences change, even though all of the other behaviors d o fluctuate. For example at proestrus, total food intake drops, but the proportion selected as protein does not change (Leshner and Collier, 1973). This extremely high correlation between taste responsiveness and protein selection suggests that the two might be causally related. Perhaps the hormonerelated taste preferences have evolved as a hedonic mechanism to regulate protein intake during the varying conditions of nutritional requirement that accompany changes in reproductive status. Proteins do not seem t o be highly preferred foods for rats. The increased finickiness of intact cycling females should suppress protein intake. During the postweaning period, pregnancy, or in males, the reduced finickiness should permit an increased intake of the nonpreferred protein that is required for growth.

VIII. HORMONES AND WEIGHT REGULATION IN NONRAT SPECIES Although gonadal regulation of body weight has been studied most extensively in rats, sex hormones affect regulatory behaviors and body weight in a wide variety of mammalian species, ranging from rodents to primates. A literature survey by Kakolewski et ul. (1968) and my own informal survey reveals that gonadectomy alters body weight regulation in at least guinea pig, rats, hamsters, mice, dogs, cats, cattle, sheep, goats, swine, rhesus monkeys, baboons, and human being. Most commonly gonadectomy increases weight gain in females but decreases it in males; however, there are interesting exceptions to this rule (see below). In the following, I shall not attempt an extensive cataloging of all the difference species in which sex hormones affect body weight and the manner

HORMONES AND BODY WEIGHT

26 1

in which they do so. Rather, I hope to discuss a few of the more interesting examples of gonadal regulation of body weight at three different phylogenetic levels: rodents, ruminants, and primates. A.

OTHER RODENTS

It is very likely that gonadal control of body weight is similar to that of the rat in a variety of rodents. Castration of male mice and guinea pigs inhibits weight gain, whereas ovariectomy of females increases body weight (Slob et ul., 1973; Wright and Turner, 1973), suggesting that in mice and guinea pigs, as in rats, estradiol inhibits and androgens promote weight gain. Adult male mice and guinea pigs are heavier than their female counterparts. As in rats, treatment of female guinea pigs with testosterone propionate early in development increases body weight and reduces the adult sex difference, so that the gonads may both organize and activate the sex differences in body weight (Slob etul., 1973). Finally, damage to the ventromedial hypothalamus in mice by gold thioglucose induces greater increases in body weight among females than males (Sanders et al., 1973; Wright and Turner, 1973). However, there is a very interesting exception to this ratlike control of body weight among the various rodents: the golden hamster. Adult female hamsters weigh and eat more than adult males (Kowalewski, 1969; Swanson, 1967; Zucker etul., 1972). Ovariectomy of adult or weanling female hamsters has no significant effect on body weight, but castration of males increases body weight until they become not significantly different from females. Treatment of gonadectomized adults with estrone or estradiol benzoate has no effect on either eating or body weight in males and females (Kowalewski, 1969; Swanson, 1968; Zucker et ul., 1972). It could be argued that insufficient estradiol benzoate was used in these experiments, since hamsters are much less responsive to estradiol than other rodent species (Feder et al., 1974b)-probably because of reduced neural affinity for the steroid. However, not even 10 pg estradiol benzoatelday, more than enough to induce sexual receptivity, had any effect on eating and body weight (Feder etul., 1974b; Zucker etul., 1972). On the other hand, treatment of adult gonadectomized male hamsters with testosterone propionate depresses both food intake and body weight, but it is ineffective in females (Zucker et uf., 1972). In addition, it has been reported that treatment of hamsters with high doses of progesterone (5 mg/day) increases both food intake and body weight. Progesterone seems to be more effective in females than in males (Swanson, 1968; Zucker et ul., 1972). However, for several reasons, it is very likely that this effect of progesterone is largely a pharmacological artifact and is of little significance to the intact hamster. First, progesterone doses less than 5 mg/day (a huge dose for a hamster) are not effective in altering eating and body weight. Second, and most important, changes in progesterone secretion during estrous cycles and

262

GEORGE N. WADE

pregnancy are typically not accompanied by fluctuations in food intake (Lukaszewska and Greenwald, 1970;Zucker et al., 1972). It appears, therefore, that androgens, rather than ovarian steroids are the important activating hormones in hamsters, so far as eating and body weight are concerned. Gonadectomy increases the body weight of males but leaves females unaffected. Testosterone treatment reduces the body weights of males but not of females; treatments with physiological doses of ovarian hormones are without effect in either sex. It seem as though testosterone has much the same effect in hamsters as estradiol does in rats. It would be interesting t o see whether testosterone-treated hamsters undereat and lose weight only until they reach a new body weight set-point, as do estradiol-treated rats. It is very likely that gonadal hormones organize as well as activate the sex difference in adult body weight. Exposure of female hamsters to testosterone during the first 10 days of life significantly reduces adult body weight (Gottlieb et al., 1974). In contrast to the very dramatic rat-hamster species differences in hormonal regulation of eating and body weight, the two species may be very similar with regard to gonadal effects on activity and taste preferences. If female hamsters are given access t o running wheels, their voluntary exercise fluctuates with the estrous cycle just as it does in rats. Activity is highest when the females are in heat. In addition, running wheel activity is substantially reduced during pregnancy and pseudopregnancy (Richards, 1966), just as in rats. These similarities suggest that the ovarian hormones might have identical effects on voluntary exercise in rats and hamsters. Hamsters also exhibit sex differences in saccharin preferences similar to those of rats. Females consume more saccharin than males. As in rats, the sex difference seems to be due to the stimulatory effects or ovarian hormones on saccharin preference. Ovariectomy of adult females significantly reduces saccharin preference, but castration of adult males has no substantial effect (Zucker et ul., 1972). These data from rodent species other than rats indicate that, although hormonal effects on body weight and regulatory behaviors may be widespread, generalizations from one species to another should be undertaken only with caution. B.

RUMINANTS

A great deal of research has been done on the effects of sex steroids on body weight and carcass quality in domestic ruminants, since factors affecting meat production and quality have widespread economic implications. Much of this work has been extensively and admirably reviewed (Baile and Forbes, 1974;Hafs et al., 1971). Until recently, diethylstilbestrol, a very potent nonsteroidal estrogen, was

HORMONES AND BODY WEIGHT

263

widely used by growers of cattle and sheep to improve meat production. In ruminants, treatment with diethylstilbestrol increases growth and improves meat quality. Although little is really known about how diethylstilbestrol exerts these effects, it has been suggested that the increased growth is due to a stimulation of pituitary growth hormone release. Both diethylstilbestrol and growth hormone have been reported to stimulate eating and weight gain, and there is some evidence indicating that diethylstilbestrol treatment increases pituitary and plasma growth hormone levels (see Baile and Forbes, 1974; Hafs et al., 1971, for more extensive discussions of this point and for documentation). Changes in endogenous hormone secretion appear to affect food intake in cows and ewes. Food intake of ewes drops at estrus, and the increased estrogen secretion during late pregnancy is also accompanied by a decrease in food intake in cows and ewes (Forbes, 1971; Tarttelin, 1968). These data might mean that endogenous estradiol may inhibit eating in sheep and cattle. Consistent with this possibility are the reports that exogenous estradiol treatments depress food intake in cows (Muir er al., 1972), ewes (Tarttelin, cited in Forbes, 1974), and wethers (Forbes, 1972). In addition, the appetite-depressing action of estradiol in cows can be reversed by progesterone (Muir el al., 1972). The recent work by Forbes (1974) indicates that ovarian steroids could affect eating in ruminants by a direct action on the brain. Injection of estradiol benzoate into the lateral cerebral ventricles of wethers increased food intake at low doses (10-20 pg) but decreased eating at higher doses. Concurrent intraventricular injection of progesterone attenuated the appetite-stimulating effects of low doses of estradiol benzoate (Forbes, 1974). Thus, i t is clear that gonadal steroids have significant effects on eating and body weight in ruminants. However, these actions of hormones may be very different, depending on the species, dosage, and form of the hormone given. C.

PRIMATES

Finally, although the order has not been studied to any great extent, i t appears as though gonadal steroids affect body weight and regulatory behaviors in a variety of primate species, including human being. There have been several reports indicating that food intake fluctuates with the menstrual cycle in rhesus monkeys and baboons (Gilbert and Gillman, 1956; Krohn and Zuckerman, 1938). John Czaja (cited in Goy and Resko, 1972) found that there was an abrupt drop in the percentage of the rhesus monkeys that ate their full daily ration of food midway between menstrual periods. This is the time when endogenous estradiol levels are rising rapidly or are at a maximum. Czaja also found that a single injection of estradiol benzoate (60 pg) produced a marked and significant decrease in food intake of ovariectomized rhesus monkeys within 24 hours. Injections of progesterone alone were ineffective in altering the food intake of ovariectomized rhesus females. Thus, ovarian hormones may affect eating behavior in nonhuman primates much as they do in rats.

264

GEORGE N. WADE

In human beings there are obvious sex differences in food intake and body weight, but the role of gonadal steroids in determining these sex differences is unspecified, as yet. It is clear that body temperature and, perhaps, body thermostatic set-point (Cunningham and Cabanac, 1971) fluctuate with the menstrual cycle. These body temperature fluctuations have been attributed t o “thermogenic” actions of endogenous progesterone during the luteal phase of the cycle (see Section 11,A). Gonadal hormones might also be of some consequence for the regulation of human alimentary preferences. Pangborn (1959) has noted that women have lower identification thresholds for sweet substances and in general are more sensitive to tastes than men are. Taste preferences also fluctuate with the menstrual cycle and pregnancy (Seifrit, 1961; Smith and Sauder, 1969; Suvorova, 1950; Thorn et ul., 1938; Wright and Crow, 1973). Wright and Crow (1973) have reported fluctuations in women’s pleasantness ratings for sucrose solutions with the menstrual cycle. Following a glucose meal, the sugar solutions were rated as being less pleasant, but this shift did not occur as rapidly around the time of ovulation. Smith and Sauder (1969) have also reported changes in food cravings during menstrual cycles. Cravings for certain foods, especially sweets, increased in the premenstrual period. Of course, there is n o shortage of anecdotal reports of unusual food preferences during pregnancy (RWade, 1971 and 1974 personal communications). It is not at all obvious that these fluctuations in taste preferences are due to hormones rather than to social and environmental influences on attitudes re. lating to sex and reproductive condition. These social-environmental influences are extremely important for the differentiation of gender identity and for the development of attitudes toward reproductive behaviors and physiology (e.g., Money and Ehrhardt, 1973). However, Nisbett and Gurwitz (1970) have reported that newborn baby girls are more responsive to sweetness and are less willing to work for their formula than baby boys are. These data suggest that sex and/or hormones may influence human alimentary behavior independent of any learned influences.

IX. CONCLUSIONS AND DIRECTIONS FOR FUTURE RESEARCH Gonadal hormones have dramatic and far-ranging effects on the behavioral regulation of energy balance in a wide variety of species. Over the past 50 years a great deal of descriptive research and simple endocrinological approaches have given us a good idea of how regulatory behaviors and body weight fluctuate with variations in reproductive status and which of the hormones are responsible for these fluctuations. But our work has really just begun. For example, we have almost n o knowledge of what the hormones are doing to the brain to effect

HORMONES AND BODY WEIGHT

265

these changes in body weight regulation. Clearly, a great deal of work remains to be done before we can begin to understand these hormone-behavior interactions. In female rats, estradiol is the principal ovarian steroid influencing behavioral regulation of energy balance. It is likely that estradiol acts directly on the brain to lower the set-point of a neural lipostatic mechanism. Lowering the body weight set-point reduces food intake which, in tum, lowers body weight (or the proportion of the total body mass devoted to fat stores). The changes in eating behavior following hormonal manipulations are seen as attempts to align body weight with the new set-point. A very likely site of action of estradiol on eating and body weight is the ventromedial hypothalamus, and estradiol may simply cause a “fme tuning” of the lipostatic control mechanisms in this part of the brain. This, of course, does not exclude the possibility of other neural sites of action. In contrast to the estrogenic effects on eating, estrogenic stimulation of voluntary exercise does not seem to be secondary to changes in body weight set-point . Progesterone also affects weight regulation in female rats. However, the principal effect of progesterone seems to be to attenuate or block the body weightregulating actions of estradiol. It is difficult to demonstrate effects of progesterone in the absence of estradiol. An exciting implication of this set-point hypothesis is that if this phenomenon is not restricted to rats (not an unreasonable assumption; see Section VIII), an estrogen-like compound could be used to lower body weight and combat obesity in human being. This approach would be feasible only if the neural estradiolsensitive system for weight regulation could be manipulated without affecting the other estradiol-sensitive systems in the body. Recent work with antiestrogens suggests that this may be possible. Perhaps an ideal weight-control drug would be one that acts on neural weight-regulating systems to lower body weight set-point without being estrogenic or antiestrogenic in other hormone-sensitive systems. Although a lipostatic hypothesis is an attractive and convenient way to describe the effects of estradiol and progesterone on eating and body weight, it is only a description and tells us little about what the steroids axe doing to neural tissues to change their functioning. We actually know very little about the biochemical mechanism of action of estradiol in neural tissues. Two principal approaches have been taken to study this problem. One has been to examine the uptake, binding, and genomic effects of estradiol in an attempt to compare the neural mechanism of action with that in peripheral tissues. Another approach has been to study interactions between sex hormones and brain neurotransmitters. This latter approach is plagued with numerous difficulties (see Section IV,C), but, in my opinion, it has a tremendous potential for creative, intelligent research. I hope that a great deal of progress can be made in studying hormone-neurotransmitter-eating behavior interactions in the near future.

266

GEORGE N. WADE

In addition to this ignorance of the biochemical effects of steroids on brain cells, there are other problems in the area of hormone action on the brain. For example, even though it has been demonstrated that hormones implanted in specific brain regions can affect eating or exercise, little is known about the neuroanatomical distribution of the hormone “receptors” for each behavior, or about the neural pathways activated once the estradiol reaches the brain4 Similarly, progesterone attenuates behavioral responsiveness to estradiol, but little is known about how or where it acts. Does progesterone act on the brain to inhibit estradiol responsiveness? Does it act on the same neurons that estradiol does? No clear answer has emerged as yet. Food intake, voluntary exercise, thermoregulatory behavior, taste preferences, and sexual receptivity all show highly correlated fluctuations during various reproductive states (see Table I). These data may mean that there are common endocrine factors underlying these changes, but they definitely do not mean that the various behaviors bear a cause-and-effect relation. It is very simple to dissociate these behaviors by hormonal manipulations. For example, estradiol implants in the ventromedial hypothalamus can depress food intake without affecting exercise or sexual receptivity, whereas placements in the anterior hypothalamus-preoptic area stimulate running wheel activity without affecting eating or lordosis. These behaviors can fluctuate independently, and competing behavior models cannot account for the coordinated pattern of behavioral changes. There are sex differences in body weight regulation in many species. In rats this sex difference is attributable to both activating effects of hormones secreted postpubertally and organizing effects of perinatal androgens. In adults the ovary secretes hormones that reduce body weight, whereas testicular androgens increase body weight. The presence or absence of androgens perinatally also influences adult body weight independent of any activating influences. Animals exposed to androgens neonatally are heavier as adults than nonandrogenized rats. Furthermore, exposure to neonatal androgens can also alter the responsiveness of the adult weight-regulating system to activating hormones. In spite of a great volume of research on the problem, there is no obvious consensus as to how neonatal androgens exert their effects on the nervous system. Sex hormones have some rather striking effects on preferences of rats for both rewarding and aversive nonnutritive tastes. It has been suggested that these 40ne interesting and testable possibility is that estradiol might affect eating and body weight by indirectly altering pancreatic insulin secretion. It is well known that the brain, particularly the ventromedial hypothalamus, can control insulin secretion (Woods and Porte, 1974). More recently evidence has begun t o accumulate suggesting that a hypersecretion of insulin is the cause of the overeating and obesity that follow ventromedial hypothalamic lesions (for reviews, see Bray, 1974; Woods et ul., 1974). Perhaps, then, estradiol acts via the ventromedial hypothalamus to depress insulin secretion and, thus, eating and body weight. If this is, in fact, the case, then ovarian hormones should not affect eating and body weight in diabetic rats maintained with controlled amounts of exogenous insulin.

HORMONES AND BODY WEIGHT

267

hormone-related taste preferences may represent a hedonic mechanism that has evolved to regulate the protein intake of rats during varying conditions of nutritive need. There is a nearly perfect inverse correlation between responsiveness to tastes and selection of dietary protein. However, virtually nothing is known about how sex hormones might alter dietary self-selection. We d o not know whether or not hormones even act on the brain to alter taste preferences. Although the adaptive advantage of altering protein intake t o meet needs imposed by reproductive status seems obvious, why total food intake and voluntary exercise of rats should be linked to gonadal steroid secretion is not so obvious. Is there any adaptive advantage to having voluntary exercise and body weight regulation tied to blood sex hormone levels? Although these questions cannot be answered with any certainty, it is easy to speculate. For example, an increase in locomotor activity at proestrus could increase a female’s chances of encountering a male and becoming impregnated. Also, it is obvious that an active mobile female rat is more attractive to males than an inactive female, which also increases the likelihood of her becoming pregnant. A lowered level of activity before puberty or during pregnancy would make more calories available for growth or for increasing the body’s energy stores prior to lactation and would decrease chances of predation. It is more difficult to understand why food intake should fluctuate with estrous cycles. Perhaps food intakeis linked to sex hormone secretion just to ensure that eating and energy stores will increase during pregnancy, and the fluctuations during estrous cycles are simply a by-product of this association. Acknowledgments Preparation of this paper was supported in part by National Institutes of Health grant NS-10873. My research that is cited in this paper has been supported at various times by grants HD-02982 (to Irvmg Zucker), HD-04467 (to Harvey Feder), and NS-10873 from the National Institutes of Health. Drs. William Beatty, Alison Fleming, Lindy Harrell, J. Bradley Powers, and Judith Stern have generously allowed me to cite their unpublished data. I am very grateful to Alison Fleming, Tom Gentry, Babs Marrone, Larry Morin, Ed Roy, and Irv Zucker for their many helpful and constructive criticisms of the earlier draft of this paper. Finally, I am much indebted to Nancy Zygmont for her cheerful assistance. References Ahlskog, J. E., and Hoebel, B. G. 1973. Overeating and obesity from damage t o a noradrenergic system in the brain. Science 182, 166-169. Ahlskog, J. E., Hoebel, B. G., and Breisch, S. T. 1974. Hyperphagia following lesions of the hypothalamic noradrenergic pathway is prevented by hypophysectomy. Fed. Proc., Fed. Amer. Soc. Exp. Biol. 33,463. Anderson, C. H., and Greenwald, G. S. 1969. Autoradiographic analysis of estradiol uptake in the brain and pituitary of the female rat. Endocrinology 85, 1160-1 165. Anton-Tay,F., and Wurtman, R. J. 1968. Norepinephrine: Turnover in rat brains after gonadectomy. Science 159 1245.

268

GEORGE N. WADE

Anton-Tay,F., and Wurtman, R. J. 1971. Brain monoamines and endocrine function. In “Frontiers in Neuroendocrinology, 1971” (L. Martini and W. F. Ganong, eds.), pp. 45-66. Oxford Univ. Press, London and New York. Anton-Tay, F., Pelham, R. W., and Wurtman, R. J. 1969. Increased turnover of 3Hnorepinephrine in rat brain following castration or treatment with ovine follicle stimulating hormone. Endocrinologv 84 1489-1492. Anton-Tay, F., Anton, S. M., and Wurtman, R. J. 1970. Mechanism of changes in brain norepinephrine metabolism after ovariectomy. Neuroendocrinology 6,265-273. Aschkenasy-Lelu, P., and Aschkenasy, A. 1959. Effects of androgens and oestrogens on the metabolism of proteins and the growth of tissues. World Rev. Nun. Diet. 1, 29-60. Asdell, S. A., Doorenbal, H., and Sperling, G. A. 1962. Steroid hormones and voluntary exercise in rats. J. Reprod. Fert. 3, 26-32. Baile, C. A., and Forbes, J. M. 1974. Control of feed intake and regulation of energy balance in ruminants. Physiol Rev. 54,160-214. Balagura, S . 1973 “Hunger, A Biopsychological Analysis.” Basic Books, New York. Balagura, S., and Devenport, L. D. 1970. Feeding patterns of normal and ventromedial hypothalamic lesioned male and female rats. J. Comp. Physiol. Psychol. 71,357-364. Bapna, J., Neff, N. H., and Costa, E. 197 1. A method for studying norepinephrine and serotonin metabolism in small regions of rat brain: Effect of ovariectomy on amine metabolism in anterior and posterior hypothalamus. Endocrinology 89,1345-1349. Beach, F. A. 1971. Hormonal factors controlling the differentiation, development, and display of copulatory behavior in the ramstergig and related species. In “The Biopsychology of Development” (E. Tobach, L. R. Aronson, and E. Shaw, eds.), pp. 249-296. Academic Press, New York. Beatty, W. W., and Schwartzbaum, J. S. 1967. Enhanced reactivity t o quinine and saccharin solutions following septa1 lesions in the rat. Psychon. Sci. 8,483-484. Beatty, W. W., Powley, T. L., and Keesey, R. E. 1970. Effects of neonatal testosterone injection and hormone replacement in adulthood on body weight and body fat in female rats. Physiol. Behav. 5, 1093-1098. Beatty, W. W., O’Briant, D. A., and Vilberg, T. R. 1974. Suppression of feeding by intrahypothalamic implants of estradiol in male and female rats. Bull. Psychon. SOC. 3 , 213-274.

Beatty, W. W., O’Briant, D. A., and Vilberg, T. R. 1975. Effects of ovariectomy and estradiol injections on food intake and body weight in rats with ventromedial hypothalamic lesions. Pharmacol. Biochem. Behav. 3,539-544. Bell, D. D., and Zucker, 1. 197 1. Sex differences in body weight and eating: Organization and activation by gonadal hormones m the rat. Physiol. Behav. 7,27-34. Bernardis, L. L. 1966. Development of hyperphagia in female rats with ventromedial hypothalamic lesions placed at four different ages. Experientia 22, 593. Bernardis, L. L.,and Skelton, F. R. 1965-1966. Growth and obesity following ventromedial hypothalamic lesions placed in female rats at four different ages. Neuroendocrinology 1,265-275.

Booth, D. A. 1968. Amphetamine anorexia by direct action on the adrenergic feeding system of the rat hypothalamus. Nature (London) 217,869-870. Bray, G. A. 1974. Endocrine factors in the control of food intake. Fed. Proc., Fed. Amer. SOC. EXP.Bid 33,1140-1 146. Brobeck, J. R. 1960. Food and temperature. Recent Progr. Horm. Res. 16,439-466. Brobeck, J. R., Wheatland, M., and Strominger, J. L. 1947. Variations in regulation of energy exchange associated with estrus, diestrus, and pseudopregnancy in rats. Endocrinology 40,65-72.

HORMONES AND BODY WEIGHT

269

Brown, R. W., Gander, G. W., and Goodale, F. 1970. Estrogen and cortisone: Effects on thermoregulation in the female rabbit.Proc. SOC.Exp. Biol. Med. 134, 1-4. Carlisle, H. J. 1964. Differential effects of amphetamine on food and water intake in rats with lateral hypothalamic lesions. J. Comp. Physiol. Psychor 58,47-54. Clemens, L. G., Shryne, J., and Gorski, R. A. 1970. Androgen and development of progesterone responsiveness in male and female rats. Physiol. Behav. 5,673-678. Code, C. F., ed. 1967. “Handbook of Physiology, Sec. 6: Alimentary Canal, Vol. I: Food and Water Intake.” Amer. Physiol. SOC.,Washington, D.C. Collier, G., Leshner, A. I., and Squibb, R. L. 1969. Dietary self-selection in active and non-active rats. Physiol. Behav. 4,79-82. Colvin, G. B., and Sawyer, C. H. 1969. Induction of running activity by intracerebral implants of estrogen in ovariectomized rats. Neuroendocrindogy 4, 309-320. Corbit, J. D. 1970. Behavioral regulation of body temperature. In “Physiological and Behavioral Temperature Regulation” (J. D. Hardy, ed.), pp. 777-801. Thomas, Springfield, Illinois. Corbit, J. D., and Stellar, E. 1964. Palatability, food intake and obesity in normal and hyperphagic rats. J. Comp. Physiol. PsychoL 5 8 , 6 3 6 7 . Courrier, R. 195 1. Interactions between estrogens and progesterone. Vitam. Horm.,(New York) 8, 179-214. Cox, V. C., and Kakolewski, J. W. 1970. Sex differences in body weight regulation in rats following lateral hypothalamic lesions. Commun. Behav. Biol., 5, 195-197. Cox, V. C., and King, J. M. 1974. The effects of estradiol on food intake and weight in ovariectomized rats with amygdaloid lesions. Physiol. Psychol. 2, 371-373. Cox, V. C., Kakolewski, J. W., and Valenstein, E. S. 1969. Ventromedial hypothalamic lesions and changes in body weight and food consumption in male and female rats. J. Comp. PhysioL Psychol. 67, 320-326. Coyne, M. D., and Kitay, J. I. 1969. Effect of ovanectomy on pituitary secretion of ACTH. Endocrinology 85,1097-1102. Cunningham, D. J., and Cabanac, M. 1971. Evidence from behavioral thermoregulatory responses of a shift in setpoint temperature related to the menstrual cycle. J. Physiol. (Paris) 63,236-238. D’Angelo, S. A., and Fisher, J. S. 1969. Influence of estrogen on the pituitary-thyroid system of the female rat: Mechanisms and loci of action. Endocrinology 84, 117-122. Debons, A. F., Krimsky, I., From, A., and Cloutier, R. J. 1969. Rapid effects of insulin on the hypothalamic satiety center. Amer. J. Physiol. 217, 1114-1118. Dickerman, E., Dickerman, S., and Meites, J. 1972. Influence of age, sex, and estrous cycle on pituitary and plasma GH levels in rats. Growth and Growth Horm.: Proc. Int. Symp. Growth Horm., I971 Excerpta Med. Found. Int. Congr. Ser. No. 244, 351-364. Donoso, A. O., and Stefano, F. J. E. 1965. Sex hormones and the concentration of noradrenaline and dopamine in the anterior hypothalamus of castrated rats. Experientia 23,665-666.

Drewett, R. F. 1974. The meal patterns of the oestrous cycle and their motivational significance. Quart. J. Exp. Psychol. 26,489-494. Emerson, J.D. 1955. Development of resistance to growth promoting action of anterior pituitary growth hormone. Amer. J. Physiol. 181, 390-394. Epstein, A. N., and Teitelbaum, P. 1967. Specific loss of hypoglycemic control of feeding in recovered lateral rats. Amer. J. Physiol. 213, 1159-1 167. Feder, H. H., Resko, J. A , and Goy, R. W. 1968. Progesterone levels in the arterial plasma of pre-ovulatory and ovariectomized rats. J. Endocrinol. 41,563-569. Feder, H. H., Naftolin, F., and Ryan, K. J. 1974a. Male and female sexual responses in male

270

GEORGE N. WADE

rats given estradiol benzoate and S&androstan-l7/301-3-one propionate. Endocrinology 94,136-141. Feder, H. H., Siegel, H., and Wade, G. N. 1974b. Uptake of (6, 7-3H) estradiol-170 in ovariectomized rats, guinea pigs and hamsters: Correlation with species differences in behavioral responsiveness to estradiol. Brain Res. 71,93-103. Finger, F. W. 1969. Estrus and general activity in the rat. J. Comp. Physiol. Psychol. 68, 461-466. Finger, F. W., and Mook, D. G. 197 1. Basic drives. Annu. Rev. Psychol. 22, 1-31. Forbes, J. M. 1971. Physiological changes affecting voluntary food intake in ruminants. Proc. Nutr. Soc. 30, 135-142. Forbes, J. M. 1972. Effects of oestradiol-17P on voluntary food intake in sheep and goats. J. Endocrinol. 52, viii-ix. Forbes, J. M. 1974. Feeding in sheep modified by intraventricular estradiol and progesterone. Physiol. Behav. 12,741-747. Freeman, M. E., Crissman, J. J., Louw, G. N., Butcher, R. L., and Inskeep, E. K. 1970. Thermogenic action of progesterone in the rat. Endocrinology 8 6 , 7 17-720. Gale, C.C. 1973. Neuroendocrine aspects of thermoregulation. Annu. Rev. Physiol. 35, 391-430. Galletti, F., and Klopper, A. 1964. The effect of progesterone on the quantity and distribution of body fat in the female rat. Acta Endocrinol. (Copenhagen) 46, 379-386. Garcia, J., Hankins, W.G., and Rusiniak, K. W. 1974. Behavioral regulation of the milieu interne in man and rat. Science 185,824-831. Gentry, R. T., and Wade, G. N. 1975. Sex differences in sensitivity of food intake, body weight, and running wheel activity to ovarian steroids in rats. J. Comp. Physiol. Psychol. in press. Gentry, R. T., and Wade, G. N. 1975. Androgenic control of food intake and body weight in male rats. J. Comp. Physiol. Psychol. in press. Gerall, A. A. 1967. Effects of early postnatal androgen and estrogen injections on the estrous activity cycles and mating behavior of rats. Anat. Rec. 157,97-104. Gerall, A. A., Stone, L. S., and Hitt, J. C. 1972. Neonatal androgen depresses female responsiveness to estrogen. Physiol. Behav. 8, 17-20. Gerall, A. A., Napoli, A. M., and Cooper, V. C. 1973. Daily and hourly estrous running in intact, spayed, and estrone implanted rats. Physiol. Behau. 10,225 -229. Gesell, C., and Fisher, G. L. 1968. Cafferine aversion and saccharin preference in rats without olfactory bulbs. Physiol. Behau. 3,523-525. Gilbert, C., and G i m a n , J. 1956. The changing pattern of food intake and appetite during the menstrual cycle of the baboon (Papio urninus) with a consideration of some of controlling endocrine factors. S. Afr. J. Med. Sn'. 21,75-88. Gold, R. M. 1970. Hypothalamic hyperphagia: Males get just as fat as females. J. Comp. Physiol. Psychol. 71, 347-356. Gold, R. M. 1973. Hypothalamic obesity: The myth of the ventromedial nucleus. Science 182,488-490. Gold, R. M., and Kapatos, G. 1975. Delayed hyperphagia and increased body length after hypothalamic knife cuts in weanling rats. J. Cornp. Physiol. Psychol. 88,202-209. Goldman, J. J., Schnatz, J. D., Bernardis, L. L. and Frohman, L. A. 1970. Adipose tissue metabolism of weanling rats after destruction of ventromedial hypothalamic nuclei: Effect of hypophysectomy and growth hormone. Metab., Clin. Exp. 19,995-1005. Gorski, R. A. 197 1. Gonadal hormones and the perinatal development of neuroendocrine

HORMONES AND BODY WEIGHT

27 1

function. I n “Frontiers in Neuroendocrinology, 1971” (L. Martini and W.F. Ganong, eds.), pp. 237-290 Oxford Univ. Press, London and New York. Gottlieb, H., Gerall, A. A., and Thiel, A. 1974. Receptivity in female hamsters following neonatal testosterone, testosterone propionate, and MER-25. Physiol. Behuv. 12, 61-68. Goy, R. W. 1970. Experimental control of psychosexuality. Phil. Trans. Roy. Soc. London, Ser. B 259, 149-162. Goy, R. W.,and Resko, 1. A. 1972. Gonadal hormones and behavior of normal and pseudohermaphrodite nonhuman female primates. Recenf Progr. Horm. Res. 28,707-733. Greene, J . , Wells, J., and Ivy, A. 1939. Progesterone will maintain adrenalectomized rats. Proc. SOC.Exp. Biol. Med. 40,83-86. Grossman, S . P. 1968. Hypothalamic and limbic influences on food intake. Fed. Proc., Fed. Amer. SOC.Exp. Biol. 27, 1349-1360. Grunt, J. A. 1964. Effects of adrenalectomy and gonadectomy on growth and development in the rat. Endocrinology 75,446-451. Hafs, H. D., Purchas, R. W., and Pearson, A. M. 1971. A review: Relationships of some hormones to growth and carcass quality of ruminants. J. Anim. Sci. 3 3 , 6 4 7 1 . Hainsworth, F. R. 1967. Saliva spreading, activity, and body temperature regulation in the rat. Amer. J. Physiol. 212, 1288-1292. Hamilton, C. L., and Brobeck, J. R. 1964. Food intake and temperature regulation in rats with rostra1 hypothalamic lesions. Amer. J. Physiol. 207,291-297. Han, P. W. 1967. Hypothalamic obesity in rats without hyperphagia. Trans. N. Y.Acud. Sci. 30,229-243. Harrell, L. E., and Balagura, S. 1975. The influence of ovarian hormones on the recovery period following lateral hypothalamic lesions. J. Comp. Physiol. Psychol. 88, 194201. Harris, G. W. 1964. Sex hormones, brain development and brain function. Endocrinology 75,627-648. Hashimoto, I., Henricks, D. M., Anderson, L. L., and Melampy, R. M. 1968. Progesterone and pregn-4en-20&01-3-one in the ovarian venous blood during various reproductive states in the rat. Endocrinology 82, 333-341. Hervey, E., and Hervey, G. R. 1966. The relationship between the effects of ovariectomy and of progesterone treatment on body weight and composition in the female rat. J. Physiol. (London) 87,44P45P. Hervey, E., and Hervey, G. R. 1967. The effects of progesterone on body weight and composition in the rat. J. Endocrinol. 37, 361-384. Hervey, E., Hervey, G. R, and Berry, P. M. 1967. A comparison between the changes in body weight and composition in female rats during pregnancy and during progesterone treatment. J. Endocrinol. 38, iii-iv. Hervey, G. R. 1969. Regulation of energy balance. Nufure (London) 222,629-631. Hervey, G. R., and Hervey, E. 1965. Interaction of the effects of oestradiol and progesterone on body weight in the rat. J. Endocrinol. 33, ix-x. Hervey, G. R, and Hutchinson, I. 1973. The effects of testosterone on body weight and composition in the rat. J. Endocrinol. 57, xxiv. Hitchcock, F. A. 1925. Studies in vigor V. The comparative activity of male and female albino rats. Amer. J. PhysioL 75,205-210. Hoebel, B. G. 1971. Feeding: Neural control of intake. Annu. Rev. Phvsiol. 33. 5 33-568.

272

GEORGE N. WADE

Hoebel, B. G., and Teitelbaum P. 1966. Weight regulation by normal and hypothalamic hyperphagic rats. J. Comp. Physiol. Aychol. 61, 189-193. Hori, T., Ide, M., and Miyake, T. 1968. Ovarian estrogen secretion during the estrous cycle and under the influence of exogenous gonadotropins in rats. Endocrinol. Jap. 15, 2 15-222. Hoshishima, K. 1967. Endocrines and taste. In “The Chemical Senses and Nutrition” (M. R. Kare and 0. Maller, eds.), pp. 139-200. Johns Hopkins Press, Baltimore, Maryland. Hoskins, R. G. 1925. Studies on vigor 11. The effect of castration on voluntary activity. Amer. J. Physiol. 72,324-330. Jankowiak, R., and Stern, J. J. 1974. Food intake and body weight modifications following medial hypothalamic hormone implants in female rats. Physiol. Behav. 12,875-879. Jennings, W. A. 197 1. Total home-cage activity as a function of the estrous cycle and wheel runningin the rat.Psychon. Sci 22, 164-165. Jennings, W. A. 1973. Estrus anorexia: Single-tube intake and barpress rate in the albino rat. Physiol. Psychol. 1, 369-273. Jensen, E. V., Jacobson, H. I., Smith, S., Jungblut, P. W., and DeSombre, E. R 1972. The use of estrogen antagonists in hormone receptor studies. Horn.Antagonists, Gynecol. Invest. 3,108-123. Josimovich, J. B., Mintz, D. H., and Finster, J. L. 1967. Estrogenic inhibition of gorwth hormone-induced tibial epiphyseal growth in hypophysectomized rats. Endocrinology 81,1428-1430. Kakolewski, J. W., Cox, V. C., and Valenstein, E. S. 1968. Sex differences in body-weight changes following gonadectomy of rats. Psychol. Rep. 22,547454. Kappas, A., and Palmer, R. H. 1963. Selected aspects of steroid pharmacology. Pharmucol. Rev. 15,123-167. Karavolas, H. J., and Herf, S.M. 1971. Conversion of progesterone by rat medial basal hypothalamic tissue to 5&pregnane-3,2O-dione.Endocrinology 89,940-942. Keesey, R. E., and Boyle, P. C. 1973. Effects of quinine adulteration upon body weight of Comp. Physid. PsychoL 84,3846. LH-lesioned and intact male rats. .l Kemble, E. D., and Schwartzbaum, J. S. 1969. Reactivity to taste properties of solutions following amygdaloid lesions. Physiol. Behav. 4,98 1-985. Kennedy, G. C. 1953. The role of depot fat in the hypothalamic control of food intake in the rat.Proc. Roy. SOC,Ser. B 140,578-592. Kennedy, G.C. 1957. The development with age of hypothalamic restraint upon the appetite of the rat. J. Endocrinol. 16,9-17. Kennedy, C . C. 1964. Hypothalamic control of the endocrine and behavioral changes associated with oestrus in the rat. J. Physid. (London) 172, 383-392. Kennedy, G.C. 1967. Ontogeny of mechanisms controlling food and water intake. In “Handbook of Physiology, Sec. 6: Alimentary Canal, Vol. I: Control of Food and Water Intake” (C. F. Code, ed.), pp. 337-35 1. Amer. Physiol. SOC.,Washington, D.C. Kennedy, G. C. 1969. Interactions between feeding behavior and hormones during growth. Ann. N.Y. Acad. Sci. 157, 1049-1061. Kennedy, C. C., and Mitra, J. 1963a Hypothalamic control of energy balance and the reproductive cycle in the rat. J. Physiol. (London) 166, 395407. Kennedy, G. C., and Mitra, J. 1963b. Body weight and food intake as initiating factors for puberty in the rat. J. Physiol. (London) 166,408418. Kennedy, G. C., and Mitra, J. 1963c. Spontaneous pseudopregnancy and obesity in the rat. J. Physiol. (London) 166,419424.

HORMONES AND BODY WEIGHT

273

Kenney, N. J., and Mook, D. G. 1974. Effects of ovariectomy on meal pattern in the albino rat. J. Comp. Physiol. Psychol. 87, 302-309. Kinder, E. F. 1927. A study of the nest-building activity of the albino rat. J. Exp. Zool. 47, 117-161. King, H. D. 1915. The growth and variability in the body weight of the albino rat. Anar. Rec. 9,75 1-776. King, J. M., and Cox, V. C. 1973. The effects of estrogens on food intake and body weight following ventromedial hypothalamic lesions. Physiol. Psychol. 1,26 1-264. Kowalewski, K. 1969. Effect of pre-pubertal gonadectomy and treatment with sex hormones on body growth, weight of organs and skin mllagen of hamsters. Acta Endocrinol. (Copenhagen) 6 1 , 4 8 4 6 . Krohn, P. L., and Zuckerman, S. 1938. Water metabolism in relation to the menstrual cycle. J. Physiol. (London) 88, 369-378. Kurtz, R. G., Rozin, P., and Teitelbaum, P. 1972. Ventromedial hypothalamic hyperphagia in the hypophysectomized weanling rat. J. Comp. Physiol. Psychol. 80, 19-25. Larue,C., and LeMagnen, J. 1970. Effect of the removal of the olfactory bulbs upon hyperphagia and obesity induced in rats by VMH lesion. Physiol. Behav. 5,509-513. Leibowitz, S . F. 1972. Central adrenergic receptors and the regulation of hunger and thirst. Res. Publ. Ass R e s New. Ment. Dis.50, 327-358. Lerner, L. J., Holthaus, F. J., and Thompson, C. R. 1958. A non-steroidal estrogen antagonist 1-@-2diethylaminoethoxyphenyl~1-phenyl-2p-methoxyphenyl ethanol. Endocrinology 63,295-318. Leshner, A. L 1969. The Adrenals and Activity. Ph.D. Thesis, Rutgers Univ., New Brunswick, New Jersey. Leshner, A. I. 1971. The adrenals and the regulatory nature of running wheel activity. Physiol. Behav. 6,551-558. Leshner, A. I. 1972. The effects of adrenalectomy on protein-carbohydrate choice. Psychon. Sci. 27,289-290. Leshner, A. I., and Collier, G. 1973. The effects of gonadectomy on the sex differences in dietary self-selection patterns and carcass composition of rats. PhysioL Behaw. 11, 671-676. Leshner, A. I., and Walker, W. A. 1973. Dietary self-selection, activity and carcass composition of rats fed thiouraciL Physiol. Behav. 10, 373-378. Leshner, A. I., CoIlier, G. H., and Squibb, R. L. 1971. Dietary self-selection at cold temperatures. Physiol. Behav. 6, 1-3. Leshner, A. I., Siegel, H. I., and Collier, G. 1972. Dietary self-selection by pregnant and lactating rats. physwl. Behav. 8, 15 1-154. Lisk, R. D. 1962. Diencephalic placement of estradiol and sexual receptivity in the female rat. Amer. J. Physid. 203,493-496. Lisk, R. D. 1974. Cyclicity of estrogen retention and biphasic effects of progesterone on estrogen retention m female rats. Fed Proc., Fed. Amer. SOC.Exp. BioL 33,267. Lukaszewka, J. H., and Greenwald, G.S. 1970. Progesterone levels in the cyclic and pregnant hamster. Endocrinology 86, 1-9. Lytle, L. D., Moorcroft, W. H., and Campbell, B. A. 1971. Ontogeny of amphetamine anorexia and insulin hyperphagia in the rat. J. Comp. PhysioL Aychol. 77,388-393. McCann, S.M., Dhariwal, A.P. S., and Porter, J.C. 1968. Regulation of the adenohypophysis. Annu. Rev. Physiol. 30,589-640. McDonald, P., Beyer, C., Newton, F., Brien, B., Baker, R., Tan, H. S., Sampson, C., Kit-

274

GEORGE N. WADE

ching, P., Greenhill, R., and Pritchard, D. 1970. Failure of 5(Yilihydrotestosterone to initiate sexual behavior in the castrated male rat. Nature (London) 227,764-965. McEwen, B. S., Zigmond, R; E., Azmita, E. C., and Weiss, J. M. 1970. Steroid hormone interactions with specific brain regions. In “Biochemistry of Brain and Behavior” (R. E. Bowman and S. P. Datta, eds.), pp. 123-167. Plenum, New York McEwen, B. S., Zigmond, R. E., and Gerlach, J. L. 1972. Sites of steroid binding and action in the brain. In “Structure and Function of Nervous Tissue” (G. H. Bourne, ed.), Vol. 5 , pp. 205-291. Academic Press, New York. McEwen, B. S., Denef, C. J., Gerlach, J. L., and Plapinger, L. 1974. Chemical studies of the brain as a sterioid hormone target tissue. In “The Neurosciences: Third Study Program” (F. 0. Schmitt and F. G. Worden, eds.), pp. 599-620. MIT Press, Cambridge, Massachusetts. McLean, J. H., and Coleman, W. P. 1971. Temperature variation during the estrous cycle: active vs. restricted rats. Psychon. Sci. 22, 179-180. Margules, D. L. 1970a. Alpha-adrenergic receptors in hypothalamus for the suppression of feeding behavior by satiety. J. Comp. Physiol. Psychol. 73, 1-12. Margules, D. L. 1970b. Beta-adrenergic receptors in the hypothalamus for learned and unlearned taste aversions. J. Comp. Physid. Psychol. 7 3 , 13-21. Margules, D. L., Lewis, M. J., Dragovich, J. A, and Margules, A. S. 1972. Hypothalamic norepinephrine: Circadian rhythms and the control of feeding behavior. Science 178, 640442. Marks, H. E. 1974. Reactivity to different saccharin concentrations as a function of testing procedure and alterations in body weight of intact and oophorectomized female rats. Physiol. Behav. 12,29-38. Marrone, B. L., Gentry, R T., and Wade, G. N. 1974. Gonadal hormones and body temperature in rats: Effects of estrous cycles, castration and steroid replacement. Physiol. Behov. (submitted for publication). Marrone, B. L., Roy, E. J., and Wade, G. N. 1975. Progesterone stimulates running wheel activity in adrenalectomized-ovariectomized rats. Horrn. Behav. 6, 231-236. Mayer, J. 1969. “Overweight.” Prentice-Hall, Englewood Cliffs, New Jersey. Meyerson, B. J., and Lindstrom, L. 1968. Effects of an oestrogen antagonist ethamoxytriphetol (MER-25) on oestrous behavior in rats. Acta Endocrinol. (Copenhagen) 59, 4148. Money, J., and Ehrhardt, A. A. 1973. “Man and Woman, Boy and Girl.” Johns Hopkins Press, Baltimore, Maryland. Montemurro, D. G. 1971. Inhibition of hypothalamic obesity in the mouse with diethylstilbestrol. Can. J. Physiol. Aharmacol. 49,554-558. Mook, D. G., Kenney, N. J., Roberts, S., Nussbaum, A. I., and Rodier, W. I., III. 1972. Ovarian-adrenal interactions in regulation of body weight by female rats. J. Comp. Physiol. PsychoL 81, 198-21 1. Morgane, P. I., ed. 1969. Neural regulation of food and water intake. Ann. N. Y. Acad. Sci. 157,531-1216. Mufson, E. J., and Wampler, R. S. 1972. Weight regulation with palatable food and liquids in rats with lateral hypothalamic lesions. J. Comp. Physiol. Psychd. 80,382-392. Muir, L. A., Hibbs, J. W., Conrad, H. R., and Smith, K. L. 1972. Effect of oestrogen and progesterone on feed intake and hydroxyproline excretion following induced hypocalcaemia in cows. J. Daily Sci. 55,1613-1620. Naftolin,F., Ryan, K. J., and Petro, Z. 1972. Aromatization of androstenedione by the anterior hypothalamus of adult male and female rats. Endocrinology 90,295-298. Nance, D. M., and Gorski, R. A. 1973. Effects of VMH lesions on estrogen suppression of

HORMONES AND BODY WEIGHT

2 75

body weight and food intake. Int. Congr. SOC. Psychoneuroendocrinol., 45th. Berkeley, Calif. Nieburgs, H. E., and Greenblatt, R.B. 1948. The role of the endocrine glands in body temperature regulation. J. Clin. Endocrinol. Metab. 8, 622-623. Nisbett, R. E. 1972. Hunger, obesity, and the ventromedial hypothalamus.Psycho1. Rev. 79, 433453. Nisbett, R. E., and Gurwitz, S. B. 1970. Weight, sex, and the eating behavior of human newborns. J. Comp. Physiol. Psycho.! 73,245-243. O’Malley, B. W., and Means, A. R. 1974. Female steroid hormones and target cell nuclei. Science 183,610-620. Ota, K., and Yokoyama, A. 1967a. Body weight and food consumption of lactating rats: Effects of ovariectomy and of arrest and resumption of suckling J. Endocrinol. 38, 25 1-261. Ota, K., and Yokoyama, A. 1967b. Body weight and food consumption of lactating rats nursing various sizes of litters. J. Endocrinol. 38,263-268. Pangborn, R. M. 1959. Influence of hunger on sweetness preferences and taste thresholds. Amer. J. Clin. Nutr. 7,280-287. Pfaff, D. W. 1969. Sex differences in food intake changes following pituitary growth hormone or prolectin injections. Proc. Annu. Conv. Amer. Psychof. Ass., 77th, New York 211-212. Pfaff, D. W., and Keiner, M. 1973. Atlas of estradiol-concentrating cells in the central nervous system of the female rat.X Comp. Neurol. 151, 121-158. Phoenix, C. H., Goy, R. W., and Young, W. C. 1967. Sexual behavior: General aspects. In “Neuroendocrinology” (L. Martini and W. F. Ganong, eds.), Vol. 2, pp. 163-196. Academic Press, New York. Plapinger, L., and McEwen, B. S. 1973. Ontogeny of estradiol-binding sites in rat brain. 1. Appearance of presumptive adult receptors in cytosol and nuclei. Endocrinology 93, 1119-1128. Plapinger, L., McEwen, B. S., and Clemens, L. E. 1973. Ontogeny of estradiol-binding sites in rat brain. 11. Characteristics of a neonatal binding macromolecule. Endocrinology 93,1129-1 139. Porterfield, A. L., and Stem, J. J. 1974. Growth hormone and the refractoriness of the prepubertal activity system to estradiol in the rat. Physid. Psychol. 2,23-25. Powers, J. B. 1970. Hormonal control of sexual receptivity during the estrous cycle of the rat. Physiol. Eehav. 5,831-835. Powers, J. B., and Valenstein, E. S. 1972a. Individual differences in sexual responsiveness t o estrogen and progesterone in ovariectomized rats. Physio.! Behav. 8,673-676. Powers, J. B., and Valenstein, E. S. 1972b. Sexual receptivity: Facilitation by medial preoptic lesions in female rats. Science 175,1003-1005. Powers, J. B., and Zucker, L 1969. Sexual receptivity in pregnant and pseudopregnant rats. Endocrinology 84,820-827. Powley, T . L., and Keesey, R. E. 1970. Relationship of body weight to the lateral hypothalamic feeding syndrome. J. Comp. Physiol. Psychol. 70,25-36. Powley, T . L., and Opsahl, C. A. 1974. Ventromedial hypothalamic obesity abolished by subdraphragmatic vagotomy. Amer. J. Physiol. 226,25-33. Ramirez, V. D., and McCann, S . M. 1963. Comparison of the regulation of luteinking hormone (LH) secretion in immature and adult rats. Endocrinology 72,452-464. Redick, J. H., and Mook, D. G. 1973. Estrogen-induced suppression of food intake and body weight: Independence of diet palatibility. Proc. Annu. Meet. East. Psychol. Ass., 44th, Washington, D.C.

276

GEORGE N. WADE

Redick, J. H., Nussbaum, A. I., and Mook, D. G. 1973. Estradiol induced suppression of feeding in the female rat: Dependence on body weight. PhySiol. Behav. 10, 543-547. Rehovsky, D. A., and Wampler, R. S. 1972. Failure to obtain sex differences in development of obesity following ventromedial hypothalamic lesions in rats. J. Comp. Physiol. Psychol. 78,102-112. Richards, M. P. M. 1966. Activity measured by running wheels and observation during the oestrous cycle, pregnancy and pseudopregnancy in the golden hamster. Anim. Behau. 14,450-458. Richter, C. P. 1922. A behavioristic study of the activity of the rat. Comp. Psychol. Monogr. i, No. 2. Richter, C. P. 1956. Self-regulatory functions during gestation and lactation. In “Gestation” (C. A. Wee, ed.), pp. 11-93. Josiah Macy, Jr., Found., New York. Richter, C. P., and Barelare, B. 1938. Nutritional requirements of pregnant and lactating rats studies by the self-selection method. Endocrinology 23, 15-24. Richter, C. P., and Hartman, C. G. 1934. The effect of injection of amniotin on the spontaneous activity of gonadectomized rats. Amer. J. Physiol. 108,136-143. Roberts, S., Kenney, N. J., and Mook, D. G. 1972. Overeating induced by progesterone in the ovariectomized, adrenalectomized rat. Horm. Behau. 3,267-276. Rodier, W. I., 111. 1971. Progesteroneestrogen interactions in the control of activity-wheel running in the female rat. X Comp. Physiol. Psychol. 74,365-373. Rodier, W. I., 111. 1973. The effect on body weight of corticosterone replacement in adrenalectomized-ovariectomized rats. ~ O C Annu. . Meet. East. Psychol. Ass., 44th. Washington, D.C. Rosenblatt, J. S. 1967. Nonhormonal basis of maternal behavior in the rat. Science 156, 1512-1514. Ross, G. E., and Zucker, I. 1974. Progesterone and the ovarian-adrenal modulation of energy balance in rats. Horn Behau. 5 , 4 3 4 2 . Rothchild, 1. 1965. Interrelations between progesterone and the ovary, pituitary, and central nervous system in the control of ovulation and the regulation of progesterone secretion. Vitam. Horn. (New York) 23,209-327. Rothchild, I. 1967. The neurologic basis for the anovulation of the luteal phase, lactation and pregnancy. In “Reproduction in the Female Mammal” ( C . E. Lamming and E. C. Amoroso, eds.), pp. 30-54. Plenum, New York. Rothchild, I. 1969. The physiologic basis for the temperature raising effect of progesterone. In “Metabolic effects of gonadal hormones and contraceptive steroids” (H. A. Salhanic, D. M. Kipnis, and R. L. van de Wiele, eds.), pp. 668-675. Plenum, New York. Roy, E. J., and Wade, G. N. 1975a. Role of estrogens in androgen-induced spontaneous activity in rats. J. Comp. Physiol. 89,573-579. Roy, E. J., and Wade, G. N. 1975b. Estrogenic effects of an antiestrogen, MER-25, on eating and body weight in rats. J. Comp. Physiol. Psychol. in press. Rubinstein, H. S., and Solomon, M. L. 1940. Growth-stimulating effect of testosterone propionate. Proc. SOC.Exp. B i d . Med. 44,442-443. Rubinstein, H. S., and Solomon, M. L. 1941. The growth depressing effect of large doses of testosterone propionate in the castrate albino rat. Endocrinology 28,112-1 14. Ryan, K. J. 1960. Estrogen formation by the human placenta: studies on the mechanism of steroid aromatization by mammalian tissue. Acta Endocrinol. (Copenhagen), Suppl. 51,697. Salhanic, H.A., Kipnis, D. M., and van de Wiele, R. L., eds. 1969. “Metabolic Effects of Gonadal Hormones and Contraceptive Steroids.” Plenum, New York.

HORMONES AND BODY WEIGHT

277

Sanders, M., Lakey, J. R., and Singh, D. 1973. Sex differences in hyperphagia and body weight gains following goldthioglucose-induced hypothalamic lesions in mice. Physiol. Psychd 1,237-240. Schwartz, E., Wiedemann, E., Simon, S., and Schiffer, M. 1969. Estrogenic antagonism of administered growth hormone. J. Clin. Endocrinol. Metab. 29,1176-1 181. Sclafani, A. 197 1. Neural pathways involved in the ventromedial hypothalamic lesion syndrome in the rat. J. Comp. Physioi. Psychol. 77,70-96. Seifrit, E. 1961. Changes in beliefs and food practices in pregnancy. J. Amer. Diet. Ass. 39, 455-466. Simpson, C. W., and DiCara, L. V. 1973. Estradiol inhibition of catecholamine elicited eating in the female rat. Pharmacol., Biochem Behav. 1,413419. Singh, D., and Meyer, D. M. 1968. Eating and drinking by rats with lesions of the septum and ventromedial hypothalamus. J. Comp. Physioi. Psychol. 65, 163-168. Slob, A. K. 1972. Perinatal Endocrine and Nutritional Factors Controlling Physical and Behavioral Development in the Rat. Ph.D. Thesis, Rotterdam Med. Fac., Rotterdam. Slob, A. K., Goy, R. W., and van der Werff ten Bosch, J. J. 1973. Sex differences in growth of guinea-pigs and their modification by neonatal gonadectomy and prenatally administered androgen. J. Endocrinol. 58,ll-19. Slonaker, J. R. 1924a. The effect of pubescence, oestruation and menopause on the voluntary activity in the albino rat. Amer. J. Physiol. 68,294-315. Slonaker, J. R 1924b. The effect of copulation, pregnancy, pseudopregnancy and lactation on the voluntary activity and food consumption of the albino rat. Amer. J. Physiol. 71, 362-394. Smith, E. R, and Davidson, J. M. 1968. Role of estrogen in the cerebral control of puberty in female rats. Endocrinology 82, 100-108. Smith, S. L., and Sauder, C. 1969. Food cravings, depression, and premenstrual problems. Psychosom Med. 31,28 1-287. Stefano, F. J. E., and Donoso, A. 0. 1967. Norepinephrine levels in the rat hypothalamus during the estrous cycle. Endocrinology 81, 1405-1406. Stern, J. J. 1970. The effects of thyroidectomy on the wheel running activity of female rats. Physiol. Behav. 5, 1277-1279. Stem, J. J., and Jankowiak, R. 1972. Effects of actinomycin-D implanted in the anterior hypothalamic-preoptic region of the diencephalon on spontaneous activity in ovariectomized rats. J. Endocrinol. 55,465466. Stem, J. J., and Jankowiak, R. 1973. No effect of neonatal estrogenic stimulation or hypophysectomy on spontaneous activity in female rats. J. Comp. Physiol. Psychol. 8 5 , 4 0 9 4 12. Stem, J. J., and Murphy, M. 1971. The effects of cyproterone acetate on the spontaneous activity and seminal vesicle weight of male rats. J. Endocrinol. 50,441443. Stem, J. J., and Murphy, M. 1972. The effects of thyroxine and estradiol benzoate on wheel running activity in female rats. Physioi. Behav. 9,79-82. Stem, J. J., and Zwick, G. 1972. Hormonal control of spontaneous activity during the estrous cycle of the rat.Psychol. Rep. 30,983-988. Stern, J. J., and Zwick, G. 1973. Effects of intraventricular norepinephrine and estradiol benzoate on weight regulatory behavior in female rats. Behov. Biol. 9,605-612. Stem, J. J., Porterfield, A. L., and Krupa, R. J. 1974. Endocrine interactions in the regulation of body weight by female rats. J. Comp. Physid. Psychol. 86,926-929. Stevenson, J. A. F., and Franklin, C. 1970. Effects of ACTH and corticosteroids in the regulation of food and water intake. Progr. Brain Res. 32, 141-152.

278

GEORGE N. WADE

Stotsenburg, J. M. 1913. The effect of semi-spaying young albino rats (Mus Norwegicus Albinus) on the growth in body weight and body length. Anut. Rec. 7, 183-194. Suvorova, N. M. 1950. The state of gustatory sensitivity in normal and pathological pregnancy. Akushetstvo Ginek. 6 , 3 3 4 0 . Swanson, H. H. 1967. Effects of pre- and post-pubertal gonadectomy on sex differences in growth, adrenal and pituitary weights of hamsters. J. Endocrinol. 39,555-564. Swanson, H. H. 1968. Effect of progesterone on the body weight of hamsters. J. Endocrinol.41, xiii. Swanson, H. H., and van der Werff ten Bosch, J. J. 1963. Sex differences in growth of rats and their modification by a single injection of testosterone propionate shortly after birth. J. Endocrinol. 26,197-207. Tanner, J. M. 1962. “Growth at Adolescence.” Blackwell, Oxford. Tarttelin, M.F. 1968. Cyclical variatons in food and water intakes of ewes. J. Physiol. (London) 195,29P-31P. Tarttelin, M. F., and Gorski, R. A. 1973. The effects of ovarian steroids on food and water intake and body weight in the female rat. Actu Endocrinol. (Copenhagen} 72, 55 1-568. Teitelbaum, P. 1955. Sensory control of hypothalamic hyperphagia. J. Comp. Physiol. Psychol. 48, 156-163. Teitelbaum, P., Cheng, M.-F., and Rozin, P. 1969. Development of feeding parallels its recovery after hypothalamic damage. J. Comp. Physiol. Psychol. 67,430441. ter Haar, M. B. 1972. Circadian and estrual rhythms in food intake in the rat. Horn Eehov. 3,213-219. Thomas, C. N., and Gerall, A. A. 1969. Effect of hour of operation on feminization of neonatally castrated male rats. Psychon. Sci. 16, 19-20. Thorn, G. W., Nelson, R R, and Thorn, D. W. 1938. Study of mechanism of oedema associated with menstruation. Endocrinology 22, 155-163. Uchida, K., Kadowaki, M., and Miyake, T. 1969. Ovarian secretion of progesterone and 2Ockhydroxypregn4en-3-one during rat estrous cycle in chronological relation to pituitary release of luteinking hormone. Endocrinol. Jup. 16,227-237. Valenstein, E. S. 1967. Selection of nutritive and non-nutritive solutions under different conditions of need. J. Comp. Physiol. Psychol. 63,429-433. Valenstein, E. S. 1968. Steroid hormones and the neuropsychology of development. In “The Neuropsychology of Development” (R. L. Isaacson, ed.), pp. 1-39. Wiley, New York. Valenstein, E. S., Cox, V. C., and Kakolewski, J. W. 1967a. Further studies of sex differences in taste preferences with sweet solutions.Aychol. Rep. 20, 1231-1234. Valenstein, E. S., Kakolewski, J. W., and Cox, V. C. 1967b. Sex differences in taste preferences for glucose and saccharin solutions. Science 156,942-943. Valenstein, E. S., Cox, V. C., and Kakolewski, J. W. 1969. Sex differences in hyperphagia and body weight following hypothalamic damage. Ann. N.Y. Acud. Sn‘. 157, 1030-1048. Vilberg, T. R., Revland, P. B., Beatty, W. W., and Frohman, L. A. 1974. Effects of cyproterone acetate on growth and feeding in rats. Phurmucol., Eiochem. Behuv. 2, 309-3 16. Wade, G. N. 1972. Gonadal hormones and behavioral regulation of body weight. Physiol. Eehuv. 8,523-534. Wade, G. N. 1974. Interaction between estradiol-17fl and growth hormone in control of food intake in weanling rats. J. Comp. Physiol. Psychd. 86,359-362. Wade, G. N. 1975. Some effects of ovarian hormones on food intake and body weight in female rats. J. Comp. Physiol. Psychol. 88. 183-193. Wade, G, N., and Feder, H. H. 1972a. (1,2-’H) Progesterone uptake by guinea pig brain and uterus: Differential localization, time-course of uptake and metabolism, and effects of age, sex, estrogen-priming, and competing steroids. Bruin Res. 45,525-543.

HORMONES AND BODY WEIGHT

279

Wade, G. N., and Feder, H. H. 1972b. Effects of several pregnane and pregnene steroids on estrous behavior in ovariectomized, estrogen-primed guinea pigs. Physiol. Behav. 9, 773-775. (b) Wade, G. N., and Zucker, I. 1969a. Hormonal and developmental influences on rat saccharin preferences. J. Comp. Physiol. Psychol. 69,29 1-300. Wade, G. N., and Zucker, I. 1969b. Taste preferences of female rats: Modification by neonatal hormones, food deprivation, and prior experience. Physiol. Behav. 4, 9 35-94 3. Wade, G. N., and Zucker, I. 1970a. Development of hormonal control over food intake and body weight in female rats. J. Comp. Physid. Psychol. 70,213-220. Wade, G. N., and Zucker, 1. 1970b. Hormonal modulation of responsiveness to an aversive taste stimulus in rats!khysioL Behav. 5,269-273. Wade, G. N., and Zucker, I. 1970c. Modulation of food intake and locomotor activity in female rats by diencephalic hormone implants. 1 Comp. Physid. Psychol. 72, 328-336. Wade, G. N., Harding, C. F., and Feder, H. H. 1973. Neural uptake of ( w 3 H ) progesterone in ovariectomized rats, guinea pigs, and hamsters: Correlation with species differences in responsiveness. Bruin Res. 61, 351-367. Wampler, R. S. 1974. “Normal” regulation of body weight in male rats with laterial hypothalamic lesions. €’roc. Annu. Conv. East. Psychol. Ass., 4 9 h , Philadelphia, Pa. Wang, G. H. 1923. The relation between ‘spontaneous’ activity and oestrous cycle in the white rat. Comp. Psychol. Monogr. ii, No. 6. Wang, G. H. 1924a. Age and sex differences in the daily food-intake of the albino rat. Amer. J. Physiol. 71,729-735. Wang, G. H. 1924b. The changes in the amount of daily food-intake of the albino rat during pregnancy and lactation. Amer. J. Physiol. 71,735-741. Wang, G. H., Richter, C. P., and Guttmacher, A. F. 1925. Activity studies on male castrated rats with ovarian transplants and correlation of the activity with the histology of the grafts. Amer. J. Physiol. 73,581-599. Weisz, J., and Gunsalus, P. 1973. Estrogen levels in immature female rats: True or spuriousovarian or adrenal? Endocrinology 93, 1057-1065. White, N. M., and Fisher, A. E. 1969. Relationship between the amygdala and hypothalamus in the control of eating behavior. Physiol. Behav. 4, 199-208. William*Ashman, H. G., and Reddi, A. H. 1971. Actions of vertebrate sex hormones. Annu. Rev. Physiol. 33,31-82. Woods, S.C., and Porte, D. 1974. Neural control of the endocrine pancreas. Physiol. Rev. 54,596-619. Woods, S. C., Decke, E., and Vasselli, J. R. 1974. Metabolic hormones and regulation of body weight.Psycho1. Rev. 8 1 , 2 6 4 3 . Wright,P., and Crow, R. A. 1973. Menstrual cycle: Effect on sweetness preferences in women. Horm. Behav. 4,387-391. Wright, P., and Turner, C. 1973. Sex differences in body weight following gonadectomy and goldthioglucose injections in mice. Physiol. Behav. 11, 155-159. Wurtman, R. J., ed. 1971. Brain monoamines and endocrine function. Neurosci. Res. Program, Bull. 9,172-297. Yoshinaga, K., Hawkins, R. W., and Stocker, J. F. 1969. Estrogen secretion by the rat ovary in vivo during the estrous cycle and pregnancy. Endocrinology 85, 103-1 12. Zucker, 1. 1969. Hormonal determinants of sex differences in saccharin preference, food intake and body weight. Physiol. Behav. 4,595-602. Zucker, I. 1972. Body weight and age as factors determining estrogen responsiveness in the rat feeding system. Behav. Biol. 7,527-542. Zucker, I., Wade, G. N., and Ziegler, R. 1972. Sexual and hormonal influences on eating, taste preferences, and body weight of hamsters. Physiol. Behav. 8, 101-11 1.

This Page Intentionally Left Blank

Subject Index estradiol and progesterone in interaction with brain monoaminergic systems, 234-236 mechanism of action of, 225-234 site of action of, 215-225 hormonal effects on food selection and, 25 3-260 hypothalamic, 248-253 lactation and, 237-242 organizing effects of perinatal hormones in, 243-248 in primates, 263-264 reproductive condition and, 204-206 responsiveness to ovarian steroids and, 237-242 in rodents, 261-262 in ruminants, 262-263 sex differences in, 203-204 Brain estradiol and progesterone in body weight regulation and, 215-236 mechanisms of sexual behavior in, 159-200 biochemical factors in androgen action, 165-173 hypothalamic androgen concentration, 185-190 hypothalamic sensitivity to androgen, 173-1 85 localized steroid effects in, 1 6 6 1 6 5

A Adoption aunting and, 128-1 30 by male, 107-108 Agonistic buffering, by male, 108-1 10 Androgen action of environmrntal factors and, 182-185 metabolism and, 170-172 specificity and, 173 uptake and, 165-170 prolonged deficit of, 174-180 sensitizing effects of, 180-182 structure of courtship and, 185-190 Attraction, infant care and, 142-145 Aunting, 120-142 adoption and, 128-130 benefits for mother-infant pair, 130-133 incompetence, kidnapping, and aunting to death, 125-128 infant independence and, 133-1 36 learning to mother and, 122-125 preferred and available aunts and infants, 137-142 status benefits and, 136-137 Aversions to poison, in rats, 34-38 specific hungers and, 42-43 taste and, 49-50 Avoidance of predators, social transmission of, 87-88

C Calorie hunger, in rats, 33-34 Caretaking, see Infants Chicken, food selection in, 51-52 Conditioning, in social transmission, 88-92 Conspecific caretaking, see Infants Copulatory behavior, steroid effects in brain and, 161-163 Courtship, hypothalamic concentration of androgen and, 185-190 Culture, food selection and, 62-67

B Baby-sitting, by male, 106-107 Behavior, see also specific behaviors developmental determinants of classification in terms of, 8-12 inirial, 2-8 relevant experience, 12-1 7 Behavioral transmission, see Social transmission Behavioral units, 4 4 Birds, social transmission of vocalizations in, 88 Body weight regulation, 201-279 activating effects of sex hormones in females, 210-215 in males, 207-210

D Developmental determinants behavior classification in terms of, 8-12 initial, 2-8 relevant experience, 12-17 281

282

SUBJECT INDEX

Domestication, food selection and, 50-51

E Environment, infant care and, 145-148 Estradiol interaction with brain monoaminergic systems, 234-236 mechanism of action of, 225-234 site of action of, 215-225 Ethnicity, food selection and, 5 8 4 2 Experience, relevant, 12-17 Exploitation of infant, by male, 104-1 18

F Feeding behavior, social transmission of, 84-87 Food selection, 21-76 in chickens, 51-52 in generalists, 27 hormonal effects on, 253-258 in humans, 52-53 biological factors in, 53-56 culture and, 62-67 ethnic-racial differences in, 58-62 specific hungers and, 56-57 in rats, 27-29 calorie hunger and, 33-34 domestication and, 50-51 poison avoidance and, 34-38 sodium hunger and, 30-33 specific hungers and, 38-49 tasteaversion learning and, 49-50 water hunger and, 29 in specialists, 24-27

selection of dietary protein and, 258-260 taste preferences and, 253-258 Humans, food selection in, 52-53 biological factors in, 53-56 culture and, 62-67 ethnic-racial differences in, 58-62 specific hungers and, 56-57 Hunger (s) for calories, 33-34 for sodium, 30-33 specific in humans, 56-57 in rats, 38-49 for water, 29 Hypothalamus androgen concentration in, 185-1 90 body weight regulation and, 248-253 sensitivity to androgen, 173-1 85 I

Incompetence, aunting and, 125-1 28 Infanticide aunting and, 125-128 by male, 110-1 1 3 Infants, conspecific caretaking of, 101-158 aunting, 120-142 male care vs. exploitation, 104-1 18 nurture vs. abuse, 118-120 selective pressures on infant, 142-148 K Kidnapping, aunting and, 125-1 28

L G

Generalists, food selection by, 27 rats, 27-51 Gonadectomy, body weight regulation and in females, 210-215 in males, 207-210

H Hormones, see also Steroids; specific

hormones in body weight regulation activating effects of, 207-215 in nonrat species, 260-264 organizing effects of, 243-248

Lactation, body weight regulation and, 237-242 Lactose intolerance, 59-62 Learning of maternal behavior, 122-125 social, 88-92

M Male-infant interactions, 104-1 18 adoption, 107-108 agonistic buffering, 108-1 10 baby-sitting, 106-107 degree of relationship in, 113-118 infanticide, 110-1 13

SUBJECT INDEX protection and rescue, 105-106 Maternal behavior, learning of, 122-125 Metabolism, of androgen, 170-172 N

Natal coats, infant care and, 142-145 Novelty, specific hungers and, 4 1 4 2

0 Omnivores, food selection in, 27 P Phenylthiocarbamide, ability to taste, 58-59 Phylogeny, infant care and, 145-148 Poison avoidance, in rats, 34-38 Precopulatory behavior, steroid effects in brain and, 163-165 Predator avoidance, social transmission of, 87-88 Predatory behavior, social transmission of, 84-87 Preferences specific hungers and, 4 2 4 3 taste, hormonal effects on, 253-258 Primates, body weight regulation in, 263-264 infant care in, 101-158 aunting and, 120-142 male care vs. exploitation, 104-118 nurture vs. abuse, 118-120 selective pressures on infant and, 142-148 Progesterone interaction with brain monoaminergic systems, 234-236 mechanism of action of, 225-234 site of action of, 215-225 Protection, by male, 105-106 Protein, hormonal effects on selection of, 258-260

R Race, food selection and, 58-62 Rats

283

body weight regulation in, see Body weight regulation food selection by, 27-29 calorie hunger and, 33-34 domestication and, 50-5 1 poison avoidance and, 34-38 sodium hunger and, 30-33 specific hungers and, 3 8 4 9 bte-aversion learning and, 49-50 water hunger and, 29 Reproductive condition, body weight regulation and, 204-206 Reproductive success, learning to mother and, 125 Rescue, by male, 105-106 Rodents, see also Rats body weight regulation in, 261-262 Ruminants, body weight regulation in, 262-263

s Sex differences, in body weight regulation, 203-204 neuroendocrine, 243-25 3 Sex hormones, see also specific hormones

activating effects in body weight regulation, 207-215 Sexual behavior, 159-200 biochemical factors in androgen action and, 165-173 hypothalamic androgen concentration and, 185-190 hypothalamic sensitivity to androgen and, 173-185 localized steroid effects in brain and, 160-165 Social transmission, 77-100 of bud vocalizations, 88 of feeding and predatory behavior, 84-87 learning and conditioning paradigms for, 88-92 of predator avoidance, 87-88 of spatial utilization, 82-84 terminology for, 92-95 Sodium hunger, in rats, 30-33 Spatial utilization, social transmission of, 82-84 Specialists, food selection in, 24-27

284

SUBJECT INDEX

Specific hungers in humans, 56-57 in rats, 38-49 Specificity, see also Developmental determinants of androgen, 173 Steroids, see also Hormones;specific hormones in body weight regulation, 237-242 copulatory behavior and, 161-163 precopulatory behavior and, 163-165

T Taste hormonal effects on preferences for, 253-258 of phenylthiocarbamide, 58-59 Thiamine-specific hunger, in rats, 38-41, 46 W

Water hunger, in rats, 2 9

A 6 E l

c a 0 9

E O F C H 1 J

1 2 3 4 5