Philosophical Perspectives on Depiction (Mind Association Occasional)

Philosophical Perspectives on Depiction MIND ASSOCIATION OCCASIONAL SERIES This series consists of occasional volumes...

Author: Catharine Abell | Katerina Bantinaki

156 downloads 1305 Views 2MB Size Report

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

Report copyright / DMCA form

DOWNLOAD PDF

Philosophical Perspectives on Depiction

MIND ASSOCIATION OCCASIONAL SERIES This series consists of occasional volumes of original papers on predeﬁned themes. The Mind Association nominates an editor or editors for each collection, and may cooperate with other bodies in promoting conferences or other scholarly activities in connection with the preparation of particular volumes. Publications Ofﬁcer: M. A. Stewart Secretary: M. Fricker Recent volumes in the series include: Identity and Modality Edited by Fraser MacBride Values and Virtues Aristotelianism in Contemporary Ethics Edited by Timothy Chappell Truthmakers The Contemporary Debate Edited by Helen Beebee and Julian Dodd Dispositions and Causes Edited by Toby Handﬁeld Empiricism, Perceptual Knowledge, Normativity, and Realism Essays on Wilfrid Sellars Edited by Willem A. deVries

Philosophical Perspectives on Depiction Edited by Catharine Abell and Katerina Bantinaki

1

1

Great Clarendon Street, Oxford ox2 6dp Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide in Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With ofﬁces in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries Published in the United States by Oxford University Press Inc., New York © the several contributors 2010 The moral rights of the authors have been asserted Database right Oxford University Press (maker) First published 2010 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this book in any other binding or cover and you must impose the same condition on any acquirer British Library Cataloguing in Publication Data Data available Library of Congress Cataloging in Publication Data Data available Typeset by Laserwords Private Limited, Chennai, India Printed in Great Britain on acid-free paper by MPG Books Group, Bodmin and King’s Lynn ISBN 978–0–19–958596–0 10 9 8 7 6 5 4 3 2 1

Acknowledgements Most of the chapters in this collection were presented, in an earlier form, at a conference organized by the editors at the University of Manchester in May 2007 and funded by the Mind Association, the British Society of Aesthetics, and the Analysis Trust. We are grateful to these institutions for their support. The editors would particularly like to express their sincere gratitude to the Mind Association for their support and encouragement during the planning and preparation of the collection. We would also like to thank the philosophers and graduate students who attended the conference, whose wholehearted interest in depiction encouraged us to proceed with this volume. Finally, we would like to thank the chapters’ authors for contributing them to this collection, for their hard work and willingness to engage with one another’s arguments, and for their patience during the process of putting this volume together.

This page intentionally left blank

Contents List of Figures and Table Notes on the Contributors Introduction Catharine Abell and Katerina Bantinaki

ix xi 1

1. Pictorial Diversity John Kulvicki

25

2. Picture This: Image-based Demonstratives Dominic McIver Lopes

52

3. The Epistemic Value of Photographs Catharine Abell

81

4. Depictive Seeing and Double Content John Dilworth

104

5. Picture Perception as Twofold Experience Katerina Bantinaki

128

6. Inﬂected Pictorial Experience: Its Treatment and Signiﬁcance Robert Hopkins

151

7. Inﬂected and Uninﬂected Experience of Pictures Bence Nanay

181

8. Seeing Things in Pictures John H. Brown

208

Index

237

This page intentionally left blank

List of Figures 2.1. Grey square and black circle 2.2. Grey circle and black square

56 56

2.3. The phi phenomenon

57

2.4. The phi phenomenon with change in colour

57

2.5. Titchener–Ebbinghaus Illusion

61

2.6. Two ﬁsh

62

2.7. Poker chip on a table

63

2.8. Pablo Picasso, Portrait of Fernande Olivier (1906). Drypoint, 16.2 × 11.8 cm. Worcester Art Museum, Worcester, Massachusetts, Sarah C. Garver Fund and anonymous gifts © Estate of Pablo Picasso (Paris)/SODRAC (Montr´eal) 2009

70

2.9. Georges Seurat, Eiffel Tower (1889). Oil on panel, 24.1 × 15.2 cm

71

2.10. Three shots of a rolling ball

72

6.1. Rembrandt, Jan Cornelisz. Sylvius (c.1646), sketch for the posthumous etching. Pen and brown ink, with white heightening, 28.5 × 19.5 cm. © Trustees of the British Museum

153

6.2. Rembrandt, Jan Cornelisz. Sylvius (1646), posthumous etching, 27.8 × 18.8 cm. © Trustees of the British Museum

162

8.1. Alfred Stevens, Study of a Nude Figure, red chalk on buff paper, 16.5 × 16.4 cm. Reproduced by permission of The Barber Institute of Fine Arts, University of Birmingham

211

List of Table 2.1. Three shots of a rolling ball

72

This page intentionally left blank

Notes on the Contributors Catharine Abell is a philosopher at the University of Manchester. Her research interests lie predominantly in aesthetics, particularly in the philosophy of the representational arts. They include the nature and value of depictive representation; its role in communication and its relation to linguistic representation; expression and interpretation in the representational arts; and the phenomenon of style. Katerina Bantinaki worked at the University of Manchester before moving to the University of Crete, where she is currently a lecturer in aesthetics and the philosophy of art. Her research interests lie predominantly in the philosophy of art. She is particularly interested in philosophical issues relevant to the perception and interpretation of works of art, the value of works of art, and their expressive import. John H. Brown has been a member of the Philosophy Department at the University of Maryland at College Park since 1963, specializing in aesthetics and philosophy of art, including philosophy of art history. His recent publications include ‘Connoisseurship: Conceptual and Epistemological Fundamentals’, in Perspectives on Connoisseurship of Chinese Painting, edited by Jason C. Kuo (2008); and articles entitled ‘Beauty’ and ‘Abstract Art’ in the Routledge Encyclopedia of Philosophy (1998), revisions of which will appear in the online edition (forthcoming). Other papers, a number of which apply digital technology to aesthetic problems in visual art, may be accessed on his website, . John Dilworth is a professor of philosophy at Western Michigan University. His book The Double Content of Art was published in 2005. He is also the author of numerous articles in aesthetics, philosophy of mind, and philosophy of language, copies of which are available via his website at . Robert Hopkins is a philosopher at the University of Shefﬁeld. His research interests include the nature and value of pictorial representation, as well as many topics in the philosophy of art and philosophy of mind more generally. His chapter in this volume forms part of a series on the varieties of pictorial experience and their value, both aesthetic and epistemic. John Kulvicki received his Ph.D. in 2001 from the University of Chicago, and worked at Washington University in St Louis and Carleton University in Ottawa

xii

notes on the contributors

before moving to Dartmouth College in 2004, where he is currently Assistant Professor of Philosophy. He works on the philosophy of art and the philosophy of perception. Dominic McIver Lopes teaches philosophy at the University of British Columbia. He is the author of Understanding Pictures, Sight and Sensibility: Evaluating Pictures, and A Philosophy of Computer Art, as well as a number of articles on aesthetics, including pictorial representation. Bence Nanay is Assistant Professor of Philosophy and Adjunct Assistant Professor of Biology at Syracuse University, and Visiting Assistant Professor of Philosophy at the University of British Columbia in the Spring semesters. He has published articles on philosophy of mind, philosophy of biology, and aesthetics in various journals such as the Philosophical Quarterly, Philosophy and Phenomenological Research, Philosophical Studies, Analysis, Synthese, Monist, and the Canadian Journal of Philosophy. His edited collection Perceiving the World is forthcoming at Oxford University Press in 2010.

Introduction CATHARINE ABELL AND KATERINA BANTINAKI

Depiction is a distinctive form of representation, characteristic of ﬁgurative paintings, drawings, and photographs. Pictures are representations that depict their objects. We do not have to look far for examples. They leer at us from advertising hoardings, televisions, and computer screens. They grace the walls of art galleries and the pages of fashion magazines. While the primary function of some pictures is aesthetic, this is not true of all, or indeed most, pictures. The pictures that accompany ﬂat-packed furniture are not beautiful, but are essential if we are to assemble sheets of veneered chipboard into usable beds, desks, and drawers. Photographs of smiling politicians may help them to secure our trust and thereby our vote. The photographs that accompany newspaper reports of disasters in far-off places convey their scale and effects in a way the reports alone could not. Although depiction plays as important a role as language in contemporary culture and communication, its function is relatively poorly understood. While the philosophy of language has long been considered a philosophical discipline in its own right, the philosophy of depiction is usually thought of, when it is thought of at all, as a sub-discipline in aesthetics. This is like conﬂating the philosophy of language with the philosophy of literature. Although there are many interesting issues concerning the aesthetics of pictures, at least as many non-aesthetic issues fall within its domain. Its central concerns—what it is for something to depict an object, what sorts of things pictures can depict, how interpreters are able to work out what pictures depict, and how pictures affect their perceivers—are important for understanding how pictures function in both aesthetic and non-aesthetic contexts.

2 catharine abell and katerina bantinaki The philosophy of depiction draws on research in other areas of philosophy, most notably the philosophy of mind and language, as well as aesthetics. It also has the capacity to contribute to our understanding of other areas. An adequate understanding of the nature of depiction may improve our understanding of both mental representation and scientiﬁc theories. Adequate accounts of what pictures are and how they are interpreted may help us to grasp the role that pictures play in mathematical and scientiﬁc reasoning. An understanding of how they function communicatively may also assist the development of a general theory of human communication. This volume aims to further debate across a wide range of issues in the philosophy of depiction and to do so in a way that helps to make clear how those issues bear on our understanding of both the aesthetic and nonaesthetic roles that pictures play. We hope that this introductory chapter will serve two purposes. Firstly, we aim to provide an accessible overview of both the main issues in the philosophy of depiction and the main positions contemporary philosophers hold with regard to those positions. Secondly, we aim to situate the various chapters in this volume in relation to these issues and positions. Thus, we hope that this Introduction will provide both useful background knowledge for those unfamiliar with the topics addressed by the chapters that follow, and a tool to assist subsequent reﬂection on their wider implications.

1. The Project of Deﬁnition 1.1. A Taxonomy of Approaches Philosophical accounts of depiction aim, ﬁrst and foremost, to answer the metaphysical question of what it is for one thing to depict another. That is, they aim to provide individually necessary and jointly sufﬁcient conditions for being a picture of some object. There are four main contemporary philosophical approaches to this issue: (1) structural, (2) phenomenological, (3) recognitional, and (4) resemblance-based. 1. Structural accounts take the most important metaphysical problem in the philosophy of depiction to be that of distinguishing depiction from language and other non-depictive systems of representation. They therefore seek to identify features that jointly sufﬁce to make a representation of some object

introduction

3

depictive. Since they are interested only in explaining distinctions between different kinds of representation, and not in explaining distinctions between depictive representations and things that are not representations, they can appeal to features that pictures have in virtue of being representations. They attempt to identify structural features that are individually necessary and jointly sufﬁcient for a representation to be depictive. Structural features are features that representations have in virtue of their relationship to other members of the system of representation to which they belong. Nelson Goodman was the ﬁrst to propose a structural account of depiction (Goodman 1976). He argued that depictive systems of representations are those that are both analog and relatively replete. A representational system is analog if it is both syntactically dense (any complete ordering of the system’s members is such that, between any two representations, there is a third) and semantically dense (any complete ordering of things denoted by the system’s syntactically distinct members is such that, between any two denotations, there is a third). It is replete if, in comparison with other systems, relatively many of its members’ features are relevant to determining what they represent. John Kulvicki has recently developed an alternative structural account (Kulvicki 2006). He improves on Goodman’s notion of analogicity, and adds a further structural feature, transparency, to those Goodman identiﬁed. He notes that a picture surface comprised of a circular mark could depict either a sphere, a circular disk seen from front on, or an oval dish seen from an oblique angle, and so on. A picture’s bare bones content, Kulvicki argues, incorporates all the properties that all its possible objects have in common. Pictures, he argues, instantiate their own bare bones contents. Bare bones content is opposed to ﬂeshed out content, which is what a picture actually depicts. Although a picture of another picture in the same system will generally differ in ﬂeshed out content from the latter picture (for example, a photograph of a photograph of a building will generally depict a photograph, not a building), Kulvicki argues that both will have the same bare bones content. Transparency consists in the fact that any picture of another picture in the same system will share the latter’s bare bones content. In Chapter 1 in this volume, Kulvicki continues the project of identifying structural features that distinguish depictive representational systems from linguistic and other non-depictive systems of representation. He identiﬁes a further feature which concerns, not the relations between different pictures

4 catharine abell and katerina bantinaki in a single depictive system, but rather the relations that different depictive systems of representation bear to one another. He notes that there is a variety of different depictive systems, distinguished both by the features of their constituent members and by the interpretative signiﬁcance of those features. There are various methods of depicting an object—for example, one can produce a picture of a lemon by photographing it with a normal lens or a ﬁsheye lens, by drawing its outline, by means of chiaroscuro, or by painting it in a Cubist style—and the pictures that result from these different methods have very different features. How these features are interpreted also depends on the method by which they have been produced. For example, Kulvicki notes, we interpret pictures produced using chiaroscuro as depicting illuminated objects, although we do not interpret outline drawings as doing so, and we interpret colour photographs as depicting the colour of objects, whereas we do not interpret black and white photographs as doing so. Kulvicki argues that depictive systems are unique in competing neither syntactically nor semantically with one another. Two systems compete syntactically if they use the same features to represent features of different kinds. For example, a system which represents temperature over time by means of graphs plotting time on the x axis and temperature on the y axis competes syntactically with a system which represents monthly rainfall at a destination by plotting rainfall at that destination on the x axis and the months of the year on the y axis, because each represents something different by means of a given point on the graph. By contrast, black and white photographs do not compete syntactically with colour photographs because, although the latter is coloured and the former is not, the former does not represent colour. Because the hue of black and white photographs is syntactically irrelevant, there is no syntactic feature that the two systems share which they deploy to different representational ends. Two systems compete semantically if they use different syntactic features to represent a given property. French and English compete semantically, because French uses ‘vert’ to represent green, whereas English uses ‘green’. By contrast, where depictive systems represent the same properties, they do so in the same way. For example, both black and white and colour photographs, outline drawings, and chiaroscuro pictures represent shape in

introduction

5

the same way. Likewise, black and white photographs, colour photographs, and chiaroscuro pictures all represent variations in light and dark in the same way. Kulvicki explains this lack of syntactic or semantic competition among pictorial systems by appeal to his claim, noted above, that pictures instantiate their bare bones contents. One consequence of this is that, to have different bare bones contents, different pictures would have to have different surface features. Pictorial systems therefore cannot compete syntactically with one another. The fact that pictures instantiate their bare bones contents also precludes semantic competition, he argues, because it means that, to the extent that pictures share bare bones contents, they will share surface features. Whereas structural approaches attempt to capture what is distinctive about depiction in terms of features that pictures have as representations, the remaining three approaches attempt to do so in terms of features that they have independently of their status as representations. In addition to distinguishing pictures from other representations, these approaches seek to distinguish pictures from things that are not representations. Rather than attempting to identify features that are necessary and sufﬁcient to make a representation depictive, that is, these approaches attempt to identify features that are necessary and sufﬁcient to make anything whatever a depictive representation. In this spirit, (2) phenomenological accounts of depiction claim that it is necessary for something to elicit a perceptual experience with a certain phenomenology if it is to depict an object. Ernst Gombrich claimed that it must produce the illusion of seeing that object (Gombrich 1977); Richard Wollheim claimed that it must elicit a twofold experience, seeing-in, one fold of which is analogous to that of seeing the object face to face (Wollheim 1987); Kendall Walton claims that it must support a rich and vivid visual game of make believe featuring that object (Walton 1990); Christopher Peacocke claims that our experience of it must resemble the experience we would have of the object were we to see it (Peacocke 1987); while Malcolm Budd and Robert Hopkins claim that it must be experienced as resembling that object (Budd 1993; Hopkins 1995). (3) Recognitional accounts deny that something need elicit any particular phenomenological response in order to depict an object, but claim instead

6 catharine abell and katerina bantinaki that it must engage particular sub-personal perceptual mechanisms if it is to do so. Flint Schier and Dominic Lopes, for instance, both claim that, in order to depict an object, it is necessary for something to engage those mechanisms that are responsible for our ability to recognize that object in the ﬂesh (Lopes 1996; Schier 1986). Finally, (4) resemblance accounts claim that to depict an object, something must resemble that object. For example, John Hyman claims that it must resemble the object in respect of its occlusion shape or silhouette (Hyman 2006), and Catharine Abell claims that it must resemble the object in respect of perceptible features that jointly capture the object’s overall appearance (Abell 2009). While something’s resembling an object is not inconsistent with its bearing any of the phenomenological or perceptual relations just described to that object, such accounts remain neutral regarding both the psychological salience of the resemblances at issue, and whether the same sub-personal mechanisms are involved in recognition of something as depicting an object as are involved in recognizing that object in the ﬂesh. Although the proponents of each approach claim that it is necessary for depiction, none of these relations plausibly sufﬁces to make something a picture. This is because things that do not depict an object may nonetheless bear each of these relations to it. For example, the stains on a wall may be visually indistinguishable from a picture of a face, such that elicit experiences with the same phenomenology as those evoked by a picture of a face, engage the same sub-personal perceptual mechanisms as are involved in recognizing a face in the ﬂesh, and bear the same resemblance relations to a face as does a picture of a face. 1.2. A Further Constraint: The Standard of Correctness It thus seems that phenomenological, recognitional, and resemblance accounts must all identify some further relation that something must bear to an object in order to depict it. What prevents the stains on the wall from depicting a face is that it is purely accidental that they bear the perceptible relations described above to a face. A further relation which constrains the manner in which something must have been produced if it is to depict an object therefore looks as if it can overcome this problem. This historical constraint comprises a standard of correctness which determines whether something that meets the necessary condition for depiction

introduction

7

described above (whether that condition is to be speciﬁed in terms of phenomenology, recognitional resources, or resemblance) in fact depicts that object. There are two distinct schools of thought regarding the nature of the standard of correctness. Some philosophers hold that something must have been intended by its maker to bear the relevant phenomenological, recognitional, or resemblance relation to an object if it is to depict it (Hopkins 1998; Wollheim 1987). One objection sometimes made to this view is that, while this may be true of paintings and drawings, it does not apply to photographs, which bear causal rather than intentional relations to their objects. This has led some philosophers to claim that two distinct standards of correctness govern depiction, a causal standard for photographs, and an intentional standard for non-photographic pictures (Hopkins 1998; Wollheim 1987). Others object that even non-photographic pictures cannot be governed by an intentional standard, since picture makers can, due to technical incompetence or other factors, succeed in depicting things other than they intended to depict. Many who reason in this latter way claim that some form of causal standard governs all depiction. While non-photographic pictures are not directly causally related to their objects in the way that photographs are, philosophers such as Lopes claim that they are indirectly causally related to their objects via a series of preceding information states that have those objects as their sources (Lopes 1996). However, because the information chain linking picture and object may, but need not, incorporate agents’ mental states, it is not clear that such a standard captures the fact that pictures are artefacts. A pressing epistemological problem that faces any account of the standard of correctness is how we succeed in applying this standard and thus work out both what things are pictures and what content they have. However the standard is construed, we need to be able to determine that the constraints it imposes on pictures’ histories of production are met if we are to succeed in interpreting something as a picture. It seems unlikely that simply perceiving something would alone enable us to access either its causal history or its makers’ intentions and thus to determine that it has the history of production that is necessary for it to depict an object. Some philosophers who advocate an intentional standard of correctness argue that we are able

8 catharine abell and katerina bantinaki to identify the intentions with which pictures were produced because we possess background knowledge of particular kinds. Thus, Hopkins argues that our knowledge of the kinds of thing that exist, the kinds of thing that are generally depicted, and the various ways in which pictures are produced enable us, together with perception of a picture, to work out what it depicts (Hopkins 1998: 137–8). Others have suggested that we identify picture makers’ intentions in the same way as we identify speakers’ intentions and thus work out what their utterances mean (Abell 2009). Although relatively little has been said about how interpreters might identify a picture’s causal history, it is important to note that causal standards of correctness raise similar epistemological problems to intentional standards. 1.3. Epistemic Implications of the Standard of Correctness The nature of the standard of correctness may have important implications for the epistemic value of pictures. If photographs are governed by a causal standard of correctness, while the standard that governs non-photographic pictures is intentional, it may be possible to explain the differences in the epistemic roles that photographs and non-photographic pictures play by appeal to the differences in the standards of correctness that govern them. Thus, Hopkins argues that a photograph depicts an object or property only if that object or property was causally responsible for the way in which the photograph’s surface is marked. He argues that it does so because camera designers intend viewers to see in photographs the objects and properties that were causally responsible for the way those photographs are marked. It follows from this that photographs are necessarily accurate (Hopkins 1998: 72–3). If this is the case, it would explain why photographs provide better evidence than non-photographic pictures that their objects existed, and had the properties they are depicted as having, at the time the photographs were produced. In Chapter 3 in this volume, however, Catharine Abell denies that this is the correct explanation of the epistemic differences between photographic and non-photographic pictures. She characterizes the distinctive epistemic value of photographs as consisting in their being, in general, more reliable and richer than non-photographic pictures. Their greater reliability consists in the fact that they are more likely than non-photographic pictures to carry information about their objects in depictive form. Some feature of a picture carries the information that its object has a certain property if, had

introduction

9

that object lacked the property in question, the picture would have lacked that feature. A picture carries the information that its object has a certain property in depictive form, she argues, if it both carries the information that its object has that property, and depicts the object as having that property. The relative richness of photographs in comparison to non-photographic pictures consists in the fact that they are reliable regarding more features of their objects. That is, she claims, they carry information in depictive form about more of their objects’ properties than non-photographic pictures. She explains the limited richness of non-photographic pictures as due to the serial processes by which they are produced. The makers of nonphotographic pictures can depictively encode information about objects only bit by bit. She argues that this places practical limitations on how much information picture makers can encode in depictive form. She explains the limited reliability of non-photographic pictures as resulting from the way their makers’ intentions affect the relations they bear to their objects, and from the limitations to their technical abilities. Their intentions limit the reliability of the pictures they produce because, even though picture makers often intend accurately to depict objects, they sometimes intend to depict them inaccurately, and it is always possible that any non-photographic picture was produced by a maker who intended it to depict its object inaccurately. Moreover, even when the makers of non-photographic pictures intend to depict objects accurately, a lack of technical competence may prevent them from doing so. Photographs, she argues, are not subject to these limitations on richness and reliability. They do not suffer the same limitations on richness because they are generally produced by parallel processes. They do not suffer the same limitations on reliability because they are produced by largely mechanized processes, over which their makers have limited inﬂuence. However, Abell claims that the epistemic differences between photographs and nonphotographic pictures are purely contingent. Richness in photographs, she claims, is a technological achievement: although photographic technology has developed towards richer and richer photographs, it might not have done. Likewise, she argues, relative reliability is a purely contingent feature of photographs. Their reliability results from the fact that photographic processes exhibit a high degree of standardization with respect to the outputs they produce given certain inputs. In particular, she claims, most photographic mechanisms have been largely standardized with respect to

10 catharine abell and katerina bantinaki the accuracy of the photographs they produce. In Abell’s view, this standardization is not a consequence of photographs being governed by a causal standard of correctness. Instead, she argues, camera designers might have designed mechanisms that produce photographs that misrepresent their objects, in which case photographs would have been no more reliable—and perhaps even less reliable—than non-photographic pictures. Camera designers’ intentions impose important constraints on how the photographs produced with those mechanisms depict their objects, and it is a contingent fact that they generally intend the photographs produced with the mechanisms they design to depict their objects accurately. 1.4. Aesthetic Implications of the Standard of Correctness The existence of a standard of correctness governing depiction may also have signiﬁcant consequences for the aesthetic value of pictures. According to most phenomenological, recognitional, and resemblance accounts, the standard of correctness operates independently of the relevant phenomenological, recognitional, or resemblance relation. This opens up the possibility of the two relations yielding different verdicts regarding what a picture depicts. For example, assume that Hopkins is correct in holding both that a relation of experienced resemblance between picture and object governs depiction, and that an intentional standard of correctness governs non-photographic depiction. In some instances of depiction, there is a gap, or separation, between what a picture is experienced as resembling, and what its maker intends it to depict. For example, a stick ﬁgure drawing is experienced as resembling an emaciated human being, although its maker is likely to have intended it instead to be experienced as resembling a normally shaped human being. Hopkins attempts to solve the problem posed by mismatches between what a picture is experienced as resembling and what its maker intended it to depict by arguing that pictures have two distinct levels of content. He distinguishes between the content of seeing-in, which is determined by experienced resemblance alone, and pictorial content, which is determined by the content of seeing-in together with the picture-maker’s intentions (Hopkins 1998: 128). In Chapter 8 in this volume, John Brown explores the implications of the differences between these two levels of content for the phenomenology of our experiences of pictures. He claims that a phenomenologically distinct experience is associated with each level of content. Separation seeing-in

introduction

11

involves seeing in the picture Hopkins’s content of seeing-in. As Hopkins notes, it involves seeing in the picture something other than it actually depicts, and reﬂects the limitations of the pictorial medium (in particular, the limitations to the verisimilitude achievable within a given medium). In addition to separation seeing-in, Brown claims, we also see in pictures what they actually depict (that is, their pictorial content). Although viewers must see in a picture what it actually depicts in order to interpret it accurately, Brown claims that such seeing-in is phenomenologically less immediate than separation seeing-in, since it requires the mediation of interpretative assumptions condoned by pictorial practices. Awareness of the eccentric properties that ﬁgure in separation seeing-in is perceptually more immediate than awareness of the properties that the picture ascribes to its subject, and quite persistent, Brown argues, as it is remarkably difﬁcult to expel such properties from view. Hopkins’s account of the distinction between the content of seeing-in and pictorial content suggests that only the latter is governed by a standard of correctness. However, Brown argues that separation seeing-in is also governed by something akin to an intentional standard of correctness. It is not arbitrary or unfounded, he claims, but is instead guided and constrained by the surface properties of the picture, which are themselves intentionally produced by the picture’s maker. Rather than being an unintended consequence of the limitations of the available media, Brown argues, separation seeing-in is something that picture makers actually set out to elicit. Consequently, one can correctly or incorrectly see in a picture certain eccentric properties even if its maker did not intend to impute those properties to its subject. Brown claims that picture makers standardly intend viewers to see in their pictures objects with properties other than those depicted because awareness of such properties is of enormous aesthetic signiﬁcance. It is by getting viewers to see such properties in a picture that picture makers achieve their aesthetic and artistic aims, as opposed to their merely depictive aims. Separation seeing-in is not only an intended but also an aesthetically merited response to a picture: ‘full and nuanced appreciation of the picture as a picture requires it . . . art criticism could hardly proceed without implicitly recognizing it, and . . . much of the charm of pictures derives from such seeings-in being intended and taken up’ (Chapter 8, Section 2). Through detailed discussion of speciﬁc examples, Brown argues

12 catharine abell and katerina bantinaki persuasively that viewers who see in a picture only what it depicts (where such an experience is even possible) will overlook features essential to an accurate and adequate assessment of its aesthetic value.

2. Pictorial Experience 2.1. The Nature of Seeing-In Brown’s argument highlights the central importance of the nature of pictorial experience to the philosophy of depiction. An adequate understanding of the experience interpreters have when they see a picture as a picture is important both because of the central importance phenomenological accounts give it in determining what it is for something to depict an object, and because of its implications for the aesthetic value of pictures. One way to think of pictorial experience is as continuous with ordinary perception. On this view, in seeing a picture of a tree, one has an experience that exactly matches in phenomenology the experience of seeing the depicted tree face to face. Accordingly, Ernst Gombrich describes pictorial experience as a case of illusion (Gombrich 1977). However, this experience is not ordinarily continuous with seeing that object face to face since the perceiver is generally aware of the medium as well as the subject of the picture. This has led most philosophers to endorse the view that pictorial experience is generally twofold: it has both a conﬁgurational and a recognitional dimension to its content, related to the picture’s medium and its subject, respectively. Following Wollheim, the contributors to this volume use the term ‘seeing-in’ for this twofold phenomenological state. An adequate account of seeing-in must answer the questions of how each dimension of seeing-in should be characterized, and of how the two are related. As Robert Hopkins indicates in Chapter 6 in this volume, answers to the ﬁrst of these questions may be either divisive or unitary. Divisive accounts take each dimension of seeing-in to correspond to a distinct experience or component of experience, each with a single content. By contrast, unitary accounts treat the two dimensions of content of seeing-in as simply ‘abstractions from the complex, structured content of pictorial experience as a whole’ (Chapter 6, Section 1). According to Wollheim, seeing-in is a single perceptual act that incorporates two distinguishable but inseparable aspects of awareness: an aspect

introduction

13

which relates to the marked surface, and an aspect which relates to the object depicted (Wollheim 1987). This formulation is divisive because it takes seeing-in to combine two aspects of awareness, each representing a single object. However, Wollheim is notoriously elusive when it comes to the question of exactly how each aspect relates to its content. Lopes provides a more comprehensive divisive account of seeing-in, claiming that pictures typically elicit two spontaneous experiences, one representing their marked surfaces, and the other representing the scenes they depict (Lopes 2005). The ﬁrst experience relates to the marked surface in virtue of visually representing it, he claims, while the second relates to the depicted scene because it bears a signiﬁcant resemblance to face-to-face visual experience of that scene. Neither of these divisive accounts explains how the two distinct and, in certain respects, incompatible objects of awareness (one of which, furthermore, is actually absent from the viewer’s visual ﬁeld) merge, in the viewer’s experience, into an integrated whole. Without an answer to this question, however, the notion of seeing-in remains incomprehensible. Hopkins attempts to meet this challenge, arguing that seeing-in involves an experience whereby the marks on the pictorial surface are seen as resembling the object depicted (Hopkins 1998). This is a unitary construal of seeing-in, since it holds that there are not two objects of visual awareness but only one: the marked picture surface. Although he denies that the viewer is perceptually aware of the depicted object, Hopkins nonetheless claims that the object enters into the content of our experience because the thought of the object transforms the look of the marks in a way that we could characterize, in part, by reference to the object and its properties. In Chapter 5 in this volume, Katerina Bantinaki suggests an alternative unitary construal of seeing-in. She focuses particularly on the fact that, in seeing-in, the medium and the subject of the picture are somehow united in perception to form a single, complex object of awareness, the picture-of-X. An understanding of this uniﬁed object of awareness, is, Bantinaki claims, the key to an adequate understanding of seeing-in and, in particular, of twofoldness. She begins by noting that such unity is not unique to pictorial experience but is rather an aspect of all acts of generic perception. As perceptual psychologists have long indicated, an object of perception is ordinarily a gestalt, i.e. a unity or an organized whole. In this gestalt, distinct material elements are grouped together under an identiﬁable

14 catharine abell and katerina bantinaki generic structure: a shape or form that enables us to discriminate the object from its surroundings and (if the form is familiar) to identify it. According to Bantinaki, the unity of the object of pictorial experience is analogous to the uniﬁed character of the (gestalt) object of perception. Drawing on Aristotle’s analysis of compound substances in order to illuminate this unity, she explains that, on the one hand, the matter of the object—the materials out of which it is made and its distinct components—is uniﬁed in perception under its identiﬁed shape or form (unless, that is, the viewer focuses her attention on those materials and parts). Moreover, the form is tied to the matter in perception—it is the form of the particular piece of matter, rather than a form thought of in the abstract, that we perceive. Bantinaki suggests that a picture, as it ﬁgures in our awareness, exhibits just this kind of unity. Although it has both a medium, and a form (a shape or outline) characteristic of a real or imaginary object, these are distinct components of the object of experience only in the sense that they can be abstracted from the homogeneous whole. The two components are (ordinarily) perceptually inseparable: as one cannot (without effort) see a table without seeing, for instance, the wood that constitutes the table; so one cannot (without effort) see, for instance, the subject of a portrait without seeing the particular manner in which that subject has been construed. Consequently, she argues, any analysis of pictorial experience as involving two aspects of awareness, one of which relates to the medium and the other to the object of representation, is only an abstraction from the complex but homogeneous content of the experience, the aspects or dimensions of which are inextricably enmeshed. This account, Bantinaki argues, explains Wollheim’s claim that awareness of the medium and the subject of representation, as they ﬁgure in seeing-in, differ phenomenologically from, respectively, the experience of seeing a non-representational surface and that of seeing the represented object in the ﬂesh. In pictorial seeing, medium and subject are enmeshed. Awareness of the medium when seeing the picture-of-X is not to be compared with seeing inchoate material elements because, in the former case but not in the latter, the material elements form a meaningful whole. They are accordingly seen through (and so inseparably from) what together they make seen. Conversely, the experience of seeing an object in a picture is

introduction

15

not to be compared with seeing the object face to face (or simply imagining it) because the picture presents a medium-speciﬁc construal of that object, such that the object is (normally) seen as construed. 2.2. The Scope of Seeing-In Because they do not take awareness of the picture’s subject to be a distinct component of seeing-in, unitary theorists deny that in seeing-in the subject is represented in the same way as it is represented when seen in the ﬂesh. That is, they deny that the subject is perceptually experienced, if by ‘perceptual experience’ we mean a mode of representation typical of ordinary perception. As we have noted, many philosophers who offer accounts of seeing-in do so with the further aim of answering the metaphysical question of what it is for something to depict an object. Whether or not any of the accounts of seeing-in described above can be deployed to this further end depends on whether or not all pictorial experience is twofold. If it is not, and some pictures do not elicit seeing-in, such accounts will, at best, characterize the experiences that some pictures elicit, but will be of little help in answering questions about depiction in general. Wollheim has argued that pictures not only permit, but also require, seeing-in, so that seeing an object in a picture always differs phenomenologically from seeing that object face to face. Lopes, on the other hand, acknowledges that pictures often trigger a twofold perceptual experience, but denies that pictorial experience is, by default, twofold (Lopes 2005; see also Levinson 1998). Twofold experience, Lopes argues, is relevant to art-pictures, especially art-pictures of a certain ‘painterly’ style, but he denies that we must be aware of the marked picture surface in order to interpret or use documentary or illustrative pictures. Moreover, he notes, trompe l’œil paintings are designed with the express purpose of producing the illusion of seeing their objects by prohibiting awareness of their medium properties. He argues that the claim that all pictorial experience is twofold therefore excludes such paintings from the domain of depiction by ﬁat. Wollheim is happy with this upshot and, indeed, sees in trompe l’œil the boundaries of depiction (Wollheim 1987). But trompe l’œil paintings, Lopes notes, are pictures and a comprehensive account of depiction ought to recognize them as such.

16 catharine abell and katerina bantinaki 2.3. Inﬂection and the Puzzle of Mimesis Even if pictorial experience admits of variety and the project of deﬁning depiction phenomenologically is abandoned, the nature of pictorial experience is still of signiﬁcant aesthetic importance. Whether the aesthetic appreciation of pictures as pictures involves simultaneous or sequential awareness of design and object, it presents a puzzle, which Lopes has labelled ‘the puzzle of mimesis’ (Lopes 2005). We evaluate pictures in part for eliciting experiences as of the scenes they depict and, moreover, Lopes claims, these experiences bear signiﬁcant resemblances to face-to-face experiences of those scenes. However, he notes, we often value pictures when we would not value seeing their objects face to face. For example, we value seeing Van Gogh’s painting of a pair of old boots when we would not value seeing a pair of old boots in the ﬂesh. How can this be if the former experience resembles the latter in important respects? Lopes argues that the solution lies in the phenomenon of inﬂection. With Michael Podro (1998), he holds that the experience of many art-pictures is inﬂected: that is, features of the design are recruited into the depicted scene, so that the scene no longer looks the way it would look in any experience purported to be of that scene. Although pictures elicit experiences that resemble face-to-face experiences of their objects, the two experiences can also differ in certain respects. In particular, only the former can be inﬂected, because only the former enables us to see objects as having properties they could not be seen to have face to face. Because it explains the different value that we assign to each experience, Lopes argues, this difference solves the puzzle of mimesis. In doing so, it construes inﬂected pictorial experience as having a central signiﬁcance to pictorial aesthetics. Such signiﬁcance, however, is by no means secure. As Hopkins argues in his chapter, the signiﬁcance of inﬂection for pictorial aesthetics is contingent on the puzzle of mimesis, but the puzzle of mimesis arises only if pictorial experience is construed as divisive, such that the depicted object is represented by a separate experience, or fold of experience, from that representing the picture’s design. However, on a unitary construal of pictorial experience, it exhibits a complex, structured content from which the element representing the design and that representing the depicted object cannot be separated, except as abstractions over that content as a whole. Consequently, this experience is always different

introduction

17

phenomenologically from the experience of seeing the depicted scene in the ﬂesh, and the puzzle of mimesis never arises. Therefore, inﬂection loses the aesthetic signiﬁcance it appeared to gain by solving that puzzle. The phenomenon of inﬂection, however, may also have repercussions for the nature of seeing-in. In his chapter, Hopkins aims to show that inﬂection undermines divisive accounts of seeing-in. Inﬂection, he argues, occurs when ‘what is seen in a surface includes properties a full characterization of which needs to make reference to that surface’s design (conceived as such)’ (Chapter 6, Section 2). He argues that this phenomenon raises two challenges for the divisive theorist. Firstly, if seeing-in is understood on the divisive model, the design of the surface should ﬁgure twice over in inﬂected seeing-in (in the component of the experience that captures the design, and in the component that captures what is seen in the design). However, he notes, such doubling is not present in the phenomenology of inﬂected seeing-in. Secondly, Hopkins argues, the phenomenon of inﬂection is incompatible with the divisive theorist’s account of the scene-oriented component of seeing-in. From the perspective of a divisive account such as Lopes’s, pictorial experience includes a component that represents a picture’s design, and a component that represents the scene it depicts. If we are to understand the character and content of seeing-in as a whole, the form of representation each dimension of content involves must be speciﬁed. With regard to the design-oriented dimension, at least, it can be assumed without controversy that the representation is the same as that involved in ordinary face-to-face experience; call this Standard Visual Representation (SVR). Inﬂection aside, the divisive theorist can further assume that the scene-oriented component of the experience also involves SVR: that this component represents the depicted scene in the way that ordinary face-to-face visual experience represents its objects. However, Hopkins argues, acknowledging inﬂection puts pressure on this account of the scene-oriented component of pictorial experience. When inﬂection occurs, the depicted scene is seen as having properties that no face-to-face experience of that scene could represent (i.e. properties that need characterizing in part by reference to features of the design). This, Hopkins claims, makes it implausible that the depicted scene is represented by SVR. So either inﬂected and uninﬂected seeing-in represent the depicted scene differently (only the latter being a case of SVR), or the scene-oriented

18 catharine abell and katerina bantinaki component of seeing-in is never a case of SVR. In either case, Hopkins concludes, there is pressure on the divisive theorist to explain what type of representation the scene-oriented component of pictorial experience involves, when, or if, it does not involve SVR.

3. Pictorial Perception: Philosophical Implications 3.1. The Mechanism of Pictorial Perception The question of how the two components of pictorial experience are visually represented unveils a further concern for pictorial theory, particularly relevant, perhaps, to recognitional accounts of depiction. This is the need to understand the process of pictorial perception in a way that illuminates how the character and content of pictorial experience relates to that of ordinary perception. Since we do not generally mistake pictures for their objects, but usually recognize them as pictures, it seems unlikely that pictures engage exactly the same perceptual mechanisms as their objects. An adequate account of picture perception must therefore explain how the mechanisms it engages are like, and how they are unlike, those engaged in face-to-face object perception. In Chapter 7 in this volume, Bence Nanay proposes such an explanation, and explores its implications for pictorial experience and, in particular, for inﬂection. Nanay draws on empirical research which suggests that there are two anatomically and functionally distinct visual subsystems, the dorsal and the ventral, the former of which provides visual control for motor actions, and the latter of which is responsible for object recognition. He argues that the processes involved in interpreting a picture of an object are like those involved in face-to-face recognition of that object in that both ventrally represent the object. However, he argues, the two processes differ in that the latter dorsally represents the object while the former does not, and the former dorsally represents the picture surface while the latter does not. Moreover, he claims, features of the picture surface may be ventrally represented in perception of a picture of an object, although this is never true of face-to-face perception of that object. The nature of picture perception processes has implications for the character and content of pictorial experience and thus for how seeingin is to be understood. Nanay points towards a possible ambiguity in

introduction

19

Wollheim’s characterization of seeing-in that, if removed, could resolve the dispute concerning the scope of seeing-in. He argues that the notion of awareness that ﬁgures in Wollheim’s claim that pictorial experience involves simultaneous awareness of the medium and the object of representation is ambiguous. This claim has two possible interpretations: (a) we consciously attend both to the depicted object and to some properties of the surface, and (b) we perceptually represent both the depicted object and the surface of the picture (irrespective of whether we attend to them). These interpretations differ, crucially, on the issue of whether seeing-in requires consciousness of those features of the picture surface that are responsible for the picture’s depictive content. Nanay denies that (a) is a necessary condition for pictorial experience, but argues that (b) is plausibly construed as one, and suggests that we construe pictorial experience as twofold only in the sense that it involves the perceptual representation of both medium and object. However, this construal of seeing-in is not intended to bolster the project of deﬁning depiction phenomenologically. It suggests that there need only be one dimension to the phenomenology of pictorial experience—one relating to the object depicted—so it does not clearly distinguish pictorial experience from the face-to-face experience of objects. Instead, Nanay argues, its signiﬁcance lies in its ability to meet the challenges that, according to Hopkins, inﬂection raises for divisive accounts of seeing-in. Nanay offers a competing characterization of inﬂection as involving attention to ‘designscene properties’: relational properties of the way the depicted object emerges from a picture’s design. Both Nanay and Hopkins agree that inﬂected pictorial experience involves the perception of properties that cannot be fully characterized without reference to both the picture’s design and the depicted object. However, in response to Hopkins, Nanay contends that the design does not ﬁgure twice over in our awareness in inﬂected seeing-in, because seeing-in does not generally require consciousness of the design (the design is represented dorsally but not ventrally, so it is visually processed without entering conscious experience), while inﬂected seeing-in requires consciousness of the design only in so far as it is involved in our consciousness of the design-scene properties we attribute to the depicted scene (in this case, that is, some properties of the design are represented not only dorsally but also ventrally). Further, Nanay argues, to the extent that the depicted scene ﬁgures in inﬂected or uninﬂected

20 catharine abell and katerina bantinaki seeing-in (being represented by our ventral vision), that scene is perceptually experienced—no matter how else the overall experience may differ from ordinary perception. 3.2. The Content of Pictorial Perception Like Nanay, John Dilworth, in Chapter 4, focuses on the nature of pictorial perception. Dilworth’s primary aim is to illuminate the content of the perceptual states that give rise to pictorial experience, and thereby to solve a problem that faces any attempt to explain how these states can produce an experience with two distinct contents. As Wollheim’s seeing-in account makes clear, when we see a picture as a picture, our experience represents both the picture surface and the object it depicts. However, Dilworth notes, both contents are derived from a single distal stimulus: namely, the picture surface. An adequate account of the contents of the perceptual processes responsible for pictorial experience must explain how these two forms of content are extracted from a single stimulus. Moreover, he notes, it must do so in a manner consistent with the general nature of human perceptual abilities and representational processing structures. The explanation Dilworth offers for our ability perceptually to extract two forms of content from a single pictorial stimulus begins with the observation that ordinary perceptual processes extract two levels of content from the environment. Firstly, their low-level, aspectual content represents properties that pertain to an object under a certain aspect: that is, to the object as it appears from the viewpoint from which and under the environmental conditions in which it is seen. At this level, perceptual content is inherently ambiguous because it fails to distinguish between intrinsic properties of the object, properties that it has only relative to the viewpoint from which it is seen, and properties of the environment in which it is seen. For example, the retinal image formed when one sees a white object illuminated by blue light may be identical to that formed when one sees a blue object illuminated by white light. The retinal image represents blueness, without specifying whether that blueness is a property of the object seen or the environmental conditions under which it is seen. However, our visual experience will normally determinately represent either a white object illuminated by blue light, or a blue object illuminated by white light. Perception extracts an unambiguous, upperlevel, object-related content from lower-level, aspectual representations,

introduction

21

Dilworth claims, by means of further perceptual investigation and appeal to probabilities of certain aspect–object correlations derived from previous perceptual experience. It is by this process that perception represents the surface of a picture. Like ordinary perceptual processes, Dilworth argues, perceptual representations of depictive content exhibit a double content structure. Object-related representations of a picture surface, like retinal images, fail to distinguish between different interpretations of its depictive content. For example, an elliptical mark on a picture surface could depict either a circular object seen from side on, or an elliptical object seen from front on. Dilworth claims that perceptual representations of depictive content have a low-level, aspectual content which represents the elliptical shape but does not specify whether it is an intrinsic or viewpoint-relative property of the depicted object. From this ambiguous, aspectual content, we perceptually extract the unambiguous, object-related content of a picture, which unambiguously construes ellipticity as either an intrinsic or a viewpoint-relative property of the depicted object. Object-related depictive content differs from ordinary object-related perceptual content, however, in that it is extracted from aspectual depictive content by appeal to artistically relevant factors such as artists’ intentions and depictive conventions, rather than features of the perceptual environment. Dilworth proposes an orientational double content theory in terms of which he analyses the range of possible interpretations in depictive double content ambiguity cases. Finally, he explores the aesthetic implications of this theory, arguing that perceptual disambiguation of an aspectual content depends on features of the medium employed—whether it be painting, etching, or charcoal—and that while we may ascribe the same object-related content to pictures in different media, different aspectual content ascriptions have different expressive signiﬁcance. 3.3. The Function of Pictorial Perception The nature of the processes involved in pictorial perception may have important implications for the functions that it can serve. For example, in Chapter 2 Lopes argues that these processes are such as to enable pictures to ground demonstrative reference perceptually. Ordinary perceptual experience can ground demonstrative reference. That is, if I look at an apple and say ‘That’s an apple’, what I say is true. By contrast, if I look at a picture of an apple and say ‘That’s an apple’, it is generally thought, my statement

22 catharine abell and katerina bantinaki is false, since ‘that’ is a picture of an apple, not an apple. However, by investigating the features of ordinary perceptual experience that enable it to ground demonstrative reference and arguing that pictorial perception shares these features, Lopes seeks to show that the latter statement is in fact true. Lopes begins by outlining an argument of Mohan Matthen’s (Matthen 2005). Matthen argues that ordinary visual perception grounds demonstrative reference because it exhibits two features. Firstly, he claims, the phi phenomenon shows that it represents objects, as well as features and locations. Secondly, he argues, it represents objects deictically. That is, it represents objects in an egocentric framework, providing information about their location relative to the viewer, and it represents them nondescriptively. To see that vision represents objects non-descriptively, it is important to distinguish between the different ways in which the two visual subsystems represent objects. Descriptive vision (or ventral vision, as Nanay terms it) represents features of objects, whereas motion-guiding vision (dorsal vision) represents objects egocentrically and non-descriptively. Thus, Matthen claims, it is in virtue of the way that motion-guiding vision represents objects that perception grounds demonstrative reference. He claims that motion-guiding vision is not implicated in seeing objects in pictures. It thus appears to follow that pictures cannot perceptually ground demonstrative reference. Although he agrees with Matthen (and Nanay) that seeing-in does not involve motion-guiding vision, Lopes argues that pictures can nonetheless perceptually ground demonstrative reference to their objects. He claims that the deictic representation of an object is best construed as comprising the egocentric representation of that object, where that egocentric representation is not accomplished by attributing features to the object. Although descriptive vision represents features of objects, Lopes argues, this does not preclude it from representing them deictically. Indeed, he claims that descriptive vision must be capable of representing objects deictically, since it is able to communicate with motion-guiding vision, and the latter cannot represent objects descriptively. Thus, he claims, descriptive vision represents objects both descriptively and deictically. Lopes claims that it follows from most theories of depiction that seeing-in implicates descriptive vision. However, he acknowledges, it does not follow

introduction

23

automatically that seeing-in represents objects deictically, since descriptive vision need only represent objects deictically when it communicates with motion-guiding vision, and seeing-in does not implicate motion-guiding vision. Nevertheless, he claims, there is reason to think that seeing-in does represent objects deictically. Firstly, pictures represent their objects centrally (that is, from a viewpoint). Secondly, he notes, there is a signiﬁcant overlap between the direction they represent their objects as occupying relative to that viewpoint, and the direction ordinary visual experience of those objects represents them as occupying relative to the viewer. The upshot, Lopes claims, is that pictures support a process called ‘deep decentering’, whereby we transform central representations of objects as being to the right of the picture’s viewpoint to deictic representations of those objects as being simply to the right. Thus, Lopes concludes, depiction perceptually grounds demonstrative representation.

Conclusion The chapters in this volume span the range of issues (metaphysical, epistemological, evaluative, and phenomenological) with which the philosophy of depiction is concerned. While some of their authors share their underlying views about the nature of depiction, the phenomenology of pictorial experience, or the processes involved in picture perception, these chapters nonetheless reveal a signiﬁcant disparity of opinion on these issues. Indeed, there is nothing even approximating philosophical consensus on any of the major issues in the philosophy of depiction. Recent research has succeeded in identifying a range of constraints that adequate solutions to these issues must meet, and has thereby narrowed the range of plausible solutions to these issues, showing certain purported solutions to be inadequate. It has also given rise to a variety of new theories designed to meet these constraints. In doing so, it has drawn on research in science, art history, and other areas of philosophy. Considerable progress has also been made in exploring the implications of various prominent theories in the philosophy of depiction, both for philosophy and for other disciplines. Nevertheless, perhaps the most striking outcome of this research is the realization that the number and scope of the philosophical issues raised by depiction is far

24 catharine abell and katerina bantinaki greater than had previously been recognized. In virtue of their collective scope and individual achievement, the chapters collected here advance debate in an area that is only beginning to elicit the philosophical attention it deserves.

References Abell, Catharine (2009), ‘Canny Resemblance’, Philosophical Review, 118/2: 183–223. Budd, Malcolm (1993), ‘How Pictures Look’, in Dudley Knowles and John Skorupski (eds), Virtue and Taste (Oxford: Oxford University Press). Gombrich, Ernst (1977), Art and Illusion, 5th edn (London: Phaidon). Goodman, Nelson (1976), Languages of Art, 2nd edn (Indianapolis: Hackett). Hopkins, Robert (1995), ‘Explaining Depiction’, Philosophical Review, 104/3: 425–55. (1998), Picture, Image and Experience (Cambridge: Cambridge University Press). Hyman, John (2006), The Objective Eye: Color, Form and Reality in the Theory of Art (Chicago: Chicago University Press). Kulvicki, John (2006), On Images (Oxford: Clarendon Press). Levinson, Jerrold (1998), ‘Wollheim on Pictorial Representation’, Journal of Aesthetics and Art Criticism, 56/3: 227–33. Lopes, Dominic McIver (1996), Understanding Pictures (Oxford: Clarendon Press). (2005), Sight and Sensibility (Oxford: Oxford University Press). Matthen, Mohan (2005), Seeing, Doing and Knowing (Oxford: Oxford University Press). Peacocke, Christopher (1987), ‘Depiction’, Philosophical Review, 97/3: 383–410. Podro, Michael (1998), Depiction (Cambridge, Mass.: Harvard University Press). Schier, Flint (1986), Deeper into Pictures (Cambridge: Cambridge University Press). Walton, Kendall (1990), Mimesis as Make-Believe (Cambridge, Mass.: Harvard University Press). Wollheim, Richard (1987), Painting as an Art (Princeton: Princeton University Press).

1 Pictorial Diversity JOHN KULVICKI

Once I did a drawing of what seemed to be a cube and someone looking at it said your perspective is off your perspective is wrong and I said no the perspective is true that’s a perfectly accurate drawing of an odd shaped object. (Paul Ziff, Antiaesthetics)

There are undeniably many ways of depicting things, but it is unclear what this pictorial diversity is. What is a way of depicting something, and how many ways of depicting things are there? There are many individual pictures, of course, and many of them differ from one another. They have different surface features and they represent different things, but all of this is unsurprising. No one thinks that all the world’s pictures depict one and the same thing and no one denies that pictures depicting different things can differ quite a bit in their surface features. Pictorial diversity starts to seem interesting and confusing when one tries to ﬁx a picture’s content while varying its surface features, or vice versa. Could this picture, ostensibly one of Van Gogh’s bedroom portraits, depict instead Gauguin’s aunt? Could Fra Angelico’s Annunciation and Bosch’s Ship of Fools depict exactly the same thing? Sensible individuals, as well as most philosophers, would answer each of the above questions with a resounding ‘No!’ But it is obscure whether such denials are directed at the particular examples or at the more general claims that these questions interrogate. Most would agree An early version of this chapter was presented at the Depiction conference, organized by the editors of this volume, at the University of Manchester, May 2007. I thank the participants for helpful feedback. I am also grateful to audiences at the American Philosophical Association, Paciﬁc Division, in 2007 and the American Society for Aesthetics, Paciﬁc Division, in 2008. In the latter case, Ed Winters provided very useful comments. Finally, Catharine Abell and Katerina Bantinaki provided truly constructive criticism that much improved the chapter.

26 john kulvicki that some pictures admit of multiple interpretations. And it will become clear that many agree there are pictures with rather distinct surface features that nevertheless have exactly the same contents. This chapter seeks to explain what is at issue in discussing pictorial diversity and to introduce two tools for understanding it. The value of these tools will become clear as they are used. After Section 1 clears up some preliminary issues, Section 2 argues that, while there are various means of making pictures, viewers do not use a complementary variety of schemes for interpreting them. So, interpretative diversity is outmatched by the myriad means for making pictures. Section 3 shows that the distinct interpretative schemes employed by observers meet an interesting constraint: they compete neither syntactically nor semantically with one another. Interpretations of non-pictorial representations, like diagrams and linguistic representations, do not obey this constraint. Section 4 considers an objection to the previous section’s claims based on Ernst Gombrich’s discussion of landscape painting, and Section 5 considers a related objection based on Heinrich W¨olfﬂin’s distinction between linear and painterly styles. Section 6 discusses related worries about the depiction of space. The overall conclusion is that people have a very powerful tendency to avoid competing interpretative schemes for pictures. Sections 7 and 8 consider how one might explain this surprising pattern and suggest that, while no accounts of pictorial representation predict or preclude such a pattern, some make room for it more naturally than others. The following does not take any particular account of depiction as its starting point. There are indeﬁnitely many ways in which one could pair representations with contents. Some systematic ways of doing so are pictorial and some are not; some are diagrammatic, some linguistic, and so on. In order to avoid committing to a particular account of depiction from the outset, it will be important to focus on the practice of interpreting representations that, intuitively at least, are pictorial, and sometimes comparing these to representations that are not. Just as an account of depiction overall must begin in part with intuitions concerning which representations are pictures, so this discussion begins by focusing on the respects in which, at least intuitively, systems of representation can differ from one another and nevertheless all be pictorial. The upshot of this investigation will be an understanding of pictorial diversity that any account of depiction must respect and, if possible, explain.

pictorial diversity

27

1. Objects, Interpretations, and Means of Production It is useful for present purposes to think in terms of systems of representation. At a minimum, systems of representation couple a sense of which objects in the world count as representations with a scheme for interpreting them (Goodman 1976, Ch. 1; Kulvicki 2006, Ch. 2). For example, coloured two-dimensional surfaces are good candidates for being pictures, while chairs, trees, and skyscrapers are not. That’s not to say that it is sufﬁcient for being a pictorial representation that something is a two-dimensional surface. It’s just that being so is plausibly a necessary condition for being a picture, or even more minimally, that pictures are quite typically two-dimensional surfaces. In addition to knowing which things count as pictures, one also needs to know how to interpret them as representations. Interpretation is a rather broad category, encompassing one’s sense of a work’s expressive features, its representational content, associations and other connections between the work and others, etc. The following ignores most of this complexity to focus on representational content. This coloured surface is a good candidate for being a picture, but if it is a picture, what does it depict? Put slightly differently, if it is a picture, what does it represent qua picture? There might be many pictorial ways of interpreting coloured surfaces, or there might be very few. Observers’ practice with pictures can shed light on this issue, even absent an account of what depiction is. And while there might be an impressive diversity to the expressive features of pictures, that topic is beyond the scope of this chapter. For present purposes then, a system of representation is just the combination of some conditions on what objects count as representations with a scheme for interpreting them as representations. It will become clear that there are, intuitively at least, a number of representational interpretative schemes that are distinct from one another but nevertheless all pictorial. Alongside these systems of representation, there are systems for producing representations: ways of making objects that count as members of one system or another (Willats 2003: 129). For example, the Renaissance saw the proliferation of techniques for rendering objects in linear perspective (Elkins 1994, Ch. 4). These techniques shared a goal: the creation of marked planar surfaces that, when interpreted as pictures in linear perspective, have

28 john kulvicki certain contents. So, if you want a surface to depict a ﬂoor tiled regularly with squares and octagons from a certain point of view within a linear perspective system, these techniques will allow you to do so. There are also techniques for producing pictures in reverse perspective and in axonometric perspective (Willats 1997: 55–9, 65–9). In a similar vein, different lenses placed on a camera can produce, at least intuitively, different kinds of pictures. The right lenses can yield pictures that obey the rules of linear perspective, while ﬁsheye lenses produce something else. Like the painting techniques, these cameras serve as tools for producing certain kinds of marked surfaces. In particular, a system of production is a way of making marked surfaces that bear certain chromatic and spatial relations to their chosen objects, or at least to objects that would bear such relations to the marked surfaces were the objects to exist. It would prejudice matters to say that a system of production for, say, linear perspective pictures also determines that the result of such a process should, or would, be interpreted as a picture in linear perspective. Perhaps this is what typically happens, perhaps not. One would need to consult our practice with pictures to determine whether this is so. As the next section will show, it is far from clear that the products of many methods of production are interpreted in such a straightforward manner. At this point it should be clear that there are three dimensions along which pictorial diversity can manifest itself. There could be diversity in the kinds of objects that count as pictures, in the interpretative schemes that assign contents to those objects, and in the means one has for producing such objects. The latter two dimensions are the most interesting. Perhaps some systems of depiction only deal with coloured spheres, to the exclusion of coloured planes, but this kind of diversity is not philosophically compelling. The next section shows that, while there are many schemes for producing pictures, observers use a comparatively thin set of schemes for interpreting them.

2. Interpretative Diversity versus Productive Diversity Consider two photographs of a lemon, the only difference between them being that one was made with an ordinary lens and the other with a

pictorial diversity

29

ﬁsheye lens. One lens is a means for producing pictures within a system akin to linear perspective while the other produces ﬁsheye photos. Fisheye photos bear different, albeit systematic, spatial relations to their objects than ordinary photos do. One could imagine an interpretative scheme for the ﬁsheye photos that merely, as it were, inverts the mapping from world to plane. In that case, the regular photo interpreted within the regular scheme has the same content as the ﬁsheye photo interpreted within the ﬁsheye system: an ordinary lemon. Perhaps each of these interpretative schemes is pictorial, though reasons will be offered later for why the ﬁsheye scheme is not. While less outrageous than the Fra Angelico–Bosch example mentioned earlier, the ﬁsheye–normal case suggests, at least prima facie, that pictures with different surface features can have the same contents. (It is trivial that representations with different surface features can have the same contents.) In one sense, these pictures manifest different ways of representing, and perhaps even depicting, the lemon. They represent it as having the same qualities, but each photo has its own set of what Dominic Lopes has called ‘design–content correlations’ (1996: 153). While one could imagine interpretative schemes that give each of these photos the same contents, however, this does not ﬁt well with how viewers actually interpret the pictures. In practice, these pictures have quite different contents. The ﬁsheye photo depicts a lemon that bulges much more than the lemon depicted by the normal photo. True, both pictures were produced by pointing a camera at the very same lemon. But that does not mean the resulting photos represent the lemon as having the same qualities, given the interpretative schemes viewers deploy. The ﬁsheye photo and its normal counterpart are similar in depicting lemons, and perhaps in depicting yellow lemons, and they may very well depict the same particular lemon, but their contents are not identical overall. Each depicts the lemon as having a shape different from what the other depicts it as having. This makes sense if the regular picture and its ﬁsheye counterpart are interpreted as though they are both members of the regular system of depiction. While each photo is made by a different system of production, they are both interpreted as members of the same representational system. A rather bulgy lemon would result in a linear perspective picture just like the ﬁsheye photo in its surface details, which suggests that something akin to linear perspective is a kind of default or standard system for interpreting pictures.

30 john kulvicki Similar considerations apply to pictures in many ‘systems’. Cubist portraits seem to depict rather shattered subjects. It is also true that the picture surface of a Cubist portrait is broken up and shattered, but this has an important effect on how that picture is interpreted in so far as it seems to depict shattered things. This makes sense if Cubist portraits are interpreted with the scheme appropriate to a linear perspective system, and it makes little sense otherwise. Similarly, Picasso’s blue period paintings depict exaggeratedly emaciated people (Kulvicki 2006: 213). That’s not to say there is nothing more to the interpretation of such pictures: cf. Hopkins (1998, Ch. 6; 2003: 663). Interpretation is a rather multifaceted activity, after all. But interpretations such as these are clearly signiﬁcant for understanding these pictures. All of this suggests that a diversity of systems for making pictures—normal lenses, ﬁsheye lenses, Cubist stylizations, or what have you—does not entail a similar diversity of systems for interpreting them. There are many interpretative schemes, and a good number of them might even seem to be pictorial. But when one considers the way in which pictures are actually interpreted, it seems as though this group suffers a high rate of unemployment. The default interpretative scheme(s) might change over time, or differ from place to place. None of the foregoing suggests that linear perspective must be special in some deep sense, though it happens to be one of the interpretative schemes that viewers use. The claim is merely that interpreters use fewer schemes than one might think, given the many means for making pictures. But is it really true that the ﬁsheye photo is interpreted as depicting a bulgy lemon? It’s generally easy to identify ﬁsheye photos as such, which suggests they are interpreted differently than ordinary photos. Perhaps the ﬁsheye photo depicts an ordinary lemon after all, not a bulgy one. The properties responsible for depicting an ordinary lemon shape are just bulgier in the ﬁsheye system than the corresponding properties in ordinary photos. Similar considerations also apply to the other examples mentioned above. Richard Wollheim (2003: 144) suggested that ‘The long neck of Parmigianino’s Madonna can be seen in his picture, but it cannot be correctly seen in it, so it should be allowed only to bring about how the Madonna is perceived.’ By extension, it is possible to see a ﬁsheye photo’s subjects as bulgy, but incorrect to do so. The ﬁsheye aspect of the photo should just inform, in some sense, how the subjects are perceived, but not the shapes they are perceived as having. Perhaps

pictorial diversity

31

people use a vaster variety of interpretative schemes than the foregoing suggests. In response, it sufﬁces to ask how one identiﬁes ﬁsheye photos as such and notice that it cannot be solely because of their surface features. What makes a patterned surface seem like a ﬁsheye photo is the fact that it depicts an oddly bulging scene. This interpretation plays an essential role in determining that the photo is a ﬁsheye, but this is not the interpretation that merely inverts the ﬁsheye mapping. It is the interpretation characteristic of ordinary photos. Also, the realization that the photo was made with a ﬁsheye lens does not entirely undermine the bulgy interpretation. Try it. The bulgy interpretation endures even though one acquires beliefs about the picture’s method of production: ‘this picture of a bulgy lemon was made with a ﬁsheye lens camera pointed at an ordinary lemon’. Similarly, we know that Parmigianino is up to something interesting because he depicts a long-necked Madonna. This does not mean that what makes the painting interesting is that the Madonna is depicted as having a long neck, and it might even be that this interpretation is somehow trumped by other considerations. But it is impressively implausible to suggest that the long-necked interpretation does not play a central role in understanding the picture. Brown (Chapter 8 in this volume, Section 2 and n. 30) is sympathetic to this line of reasoning. The discussion so far has uncovered a hazardous ambiguity in the claim that a pair of pictures manifests different ways of depicting things. On the one hand, a way of depicting something can be identiﬁed with a correlation between design and content. So, an extraordinary way of depicting something is a system of picture production and interpretation that differs from the norm. This emphasizes the vehicle of representation as special, not its content: something like this depicts something ordinary like a lemon. On the other hand, a way of depicting something is just the set of properties something is depicted as having. This requires no departure from ordinary design–content rules, only a use of those rules to represent something as being somewhat extraordinary, like an extra-bulgy lemon. This emphasizes an ordinary system of interpretation coupled, perhaps, with an extraordinary method of picture production. The vehicle is treated as being rather ordinary—in that it’s part of an ordinary interpretative scheme—while its content stands out as special: this ordinary photo depicts an extraordinary thing. Pictures can be distinctive as vehicles or as bearers

32 john kulvicki of a rather distinctive content (or both, of course). Practice suggests that, by and large, it is the content and not the vehicle that is regarded as distinctive.¹ There are many interpretative schemes out there, but precious few of them are employed by interpreters. The ﬁsheye example does not exemplify all kinds of pictorial diversity, and it’s important to keep the limited claims of this section in sight. None of the foregoing suggests that interpreters use only one interpretative scheme. Indeed, interpreters use many, and this must ﬁgure in any account of pictorial diversity. For example, outline drawings typically are not interpreted as representing objects as illuminated in any particular way, while systems that employ some form of chiaroscuro are. It is wrong to interpret black and white photos as representing a black and white world; they are just silent concerning things’ colours. These systems differ with respect to the properties they characteristically represent objects as having, and are interpreted as such. So, even though there are not as many interpretative schemes employed as there are methods for making pictures, there is significant range to the way in which viewers are prepared to interpret them. In order to appreciate, rather than just enumerate, these various interpretative schemes, one needs to understand how they relate to one another. The next section suggests that amidst this variety there is an impressive regularity: these interpretative schemes do not compete with one another. This distinguishes the diversity of pictorial interpretative schemes from the range of schemes for interpreting diagrams and linguistic representations.

3. Competition What is competition? Interpretative schemes compete syntactically when they use a given syntactic feature to serve distinct semantic ends, and they compete semantically when they use distinct syntactic features to serve a given semantic end. ¹ Robert Hopkins’s (1998: 94–7) discussion of misrepresentation in caricature lines up well with this section’s claims about pictorial interpretation. We do not interpret caricatures as odd depictions of ordinary things, so much as ordinary depictions of odd things. Indeed, Hopkins’s sense of the phenomenon of pictorial diversity overall (1998, Ch. 6, esp. 147–58; 2003) ﬁts quite well with that developed herein, as will become clear. One important difference is that Hopkins distinguishes what one sees in a picture from what that picture depicts. At least sometimes, for Hopkins, these two things come apart. I have not drawn such a distinction herein because I am skeptical of its usefulness.

pictorial diversity

33

Consider some property of pictures, p, in virtue of possessing which, ceteris paribus, a picture represents P within a given interpretative scheme. Another interpretative scheme in which possession of p is ceteris paribus responsible for representing Q, which is not identical to P, competes syntactically with the ﬁrst scheme. Each competing scheme makes use of p in the service of different representational ends. This is ‘syntactic’ competition only in the sense that p is a feature of picture surfaces, not (necessarily) a feature of what these pictures depict. It might often be difﬁcult to determine just what a picture represents, ceteris paribus, in virtue of possessing some property, but some examples make this idea clear enough to be useful. The scheme envisioned by Wollheim (1980) and Nelson Goodman (1976), in which colours stand for their complements, competes syntactically with an ordinary interpretative scheme. In such an inverted system, red areas of a picture surface generally depict green things, while in the ordinary system red regions represent red things. That is to say, ordinary pictures represent red things as such by being coloured red in some region of their surfaces. In the inverted scheme, such pictures represent green. Similarly, it’s generally true that pictures representing four-sided surfaces or regions of space as such tend to do so by having four-sided regions of their surfaces marked off by lines, regions of colour, or what have you. The ﬁsheye system competes syntactically with the ordinary system, because ﬁsheye photos with rectilinear regions of their surfaces depict regions of space with concave sides. By contrast, black and white photography does not compete syntactically with its colour counterpart because in the former interpretative scheme hue is irrelevant to the interpretation of photos, and otherwise these schemes are the same. There is no syntactic feature common to both systems that serves different representational ends. So, these interpretative schemes differ, even though they do not compete. One might object that black and white photos do compete syntactically with colour photos: both of these schemes use the colours black and white differently. In one system, white represents white, while in the other it just represents a degree of lightness. But it’s not the colour white per se that plays such a role in the black and white picture scheme. In black and white photos, what matters are not colours like black and white but degrees of lightness. Many black and white photos have hues, whether they are shades of sepia, yellow, or brown, but these hues play no role in determining what such photos represent. This is important because it

34 john kulvicki reveals a sense in which indistinguishable pictures can nevertheless have rather different contents. A black and white photo of a black and white checkerboard could be indistinguishable from a colour photo of the same. But the former represents a board with alternating light and dark squares, while the latter represents a board with alternating black and white squares. These interpretative schemes do not compete syntactically, even though some members of the black and white system are indistinguishable from members of the colour system. This example is not quite as surprising as a picture of Van Gogh’s bedroom depicting Gauguin’s aunt within an alternate interpretative scheme, but it’s nevertheless an instance of intrinsically identical pictures having distinct contents depending on the interpretative scheme one employs. While syntactic competition involves distinct semantic uses of a given syntactic feature, interpretative schemes compete semantically when they represent a given property by using distinct syntactic features. Consider some property of pictures, p, in virtue of possessing which, ceteris paribus, a picture represents P within a given interpretative scheme. Another interpretative scheme in which possession of q, which is not identical to p, is ceteris paribus responsible for representing P competes semantically with the ﬁrst scheme. Each competing scheme makes use of different syntactic features in the service of a given representational end: they compete over how they represent P. The colour-inverted system competes syntactically with its ordinary counterpart because green in one system represents green, while in the other it represents red. These systems compete semantically because green is represented by one system with green while it is represented with red by the other system. Systems that compete semantically need not share any syntactic features—they just need to share contents—while systems that compete syntactically need not represent any of the same qualities. The different schemes we actually use to interpret pictures compete neither syntactically nor semantically with one another. As mentioned above, black and white photos do not depict hues, and they depict spatial features, as well as variations in light and dark, in the same manner as colour photographs do. That is to say, where black and white photos overlap semantically with colour photos—the cases in which we are prepared to say they represent the same properties, like shapes—they do so in the same way, rather than in competing ways. Similar remarks apply to the contrast

pictorial diversity

35

between chiaroscuro and line drawing. The former represents variations in light and dark while the latter does not, but they represent shapes and the like in much the same manner.² The next section will say a lot more about the representation of light and dark, because there is at least some art-historical evidence that there is semantic competition in the way light and shadow are represented. Odd interpretative schemes that are not used with any regularity compete both syntactically and semantically with the ones that are. As mentioned, in the colour complement system green represents red, while in the ordinary system green represents green (syntactic competition). And in the complement system green is represented by red, while in the ordinary system green is represented by green (semantic competition). Similar remarks apply to the possible ﬁsheye system of interpretation vis-à-vis ordinary linear perspective as well, but with respect to shapes, not colours or degrees of lightness. For example, a region with four straight sides in a linear perspective picture tends to represent a quadrilateral region of space. In the ﬁsheye system, by contrast, four sides with a given degree of curvature would represent quadrilateral regions of space. And a region of a ﬁsheye picture that has four straight sides would represent a region of space with four concave sides. It is easy to ﬁnd interpretative schemes that compete both syntactically and semantically with the ones that people do use, but it’s a very interesting fact that such competing schemes are not employed. The next two sections deepen and defend this proposal. Before moving on, it is worth noting that this pattern of non-competition does not appear in other representational kinds. Many diagrammatic interpretative schemes compete syntactically but not semantically, and conversely. There are numerous ways to diagram a given data set, for example, so semantic competition is common. Similarly, one and the same abstract diagram can be applied to many different representational ends, which amounts to syntactic competition. This seems true of the more logical or linguistic representations as well. There are many names for red, and ‘red’ could have named indeﬁnitely many things. This makes the pattern one ﬁnds with pictures all the more puzzling. Pictures are often taken to be close cousins to diagrams and related representations, but in this ² Cf. Willats (1997), who would agree about the representation of shape in each system but claim that such pictures make use of different picture primitives (lines vs regions of light and dark) constituting different ‘denotational systems’.

36 john kulvicki respect at least it seems as though diagrams are more like languages than like pictures.

4. Objection: Light and Shadow The portrait of pictorial diversity that emerges here is one in which viewers latch onto a somewhat diverse but non-competing set of interpretative schemes amidst an abundance of methods for making pictures. The distinctions between interpretative and productive systems and between competing and non-competing interpretative systems reveal an impressive regularity in practice that otherwise would have been missed. This section considers a signiﬁcant potential counter-example due to Ernst Gombrich, while the next section considers a related worry due to Heinrich W¨olfﬂin. Gombrich begins Art and Illusion (1961, pt 1) with a discussion of the ways in which light, shadow, and colour have been represented in landscape painting, especially between the seventeenth and nineteenth centuries. The features of paintings that represent such properties changed over that period, Gombrich claims. Observers with the right exposure and training readily understand pictures that differ substantially in their surface features as representations of one and the same thing, or so the story goes. The history of picture-making thus suggests that some pictures at least compete semantically, over how they represent light, colour, and shadow; and they might compete syntactically as well. Gombrich’s discussion of this issue is curiously inconclusive, however. On the one hand, he suggests that there is genuine semantic competition between, say, seventeenth-century Dutch landscapes, like Hobbema’s Watermill (1670), and their nineteenth-century English counterparts, such as Constable’s White Horse (1819) (Gombrich 1961: 58–9). Both depict scenes unevenly illuminated by the sun, and both face the challenge of representing signiﬁcant contrasts in illumination using a colour palette that lacks such differences in brightness. Each accomplishes this goal by employing different aspects of that common colour palette, which is a clear case of semantic competition. The earlier work, for example, eschewed great variation in hue as a technique for suggesting changes in the illumination of a depicted scene. Constable, by contrast, made liberal use of greens and blues in the service of similar representational ends.

pictorial diversity

37

On the other hand, Gombrich (1961: 46–9) suggests that nineteenthcentury developments in technique from Constable through Corot and Monet allowed the depiction of brighter scenes than those depicted by Hobbema, Lorrain, or Gainsborough. Picking up on a remark by Ernst Brücke, Gombrich claims that Monet let the midday sun into paintings (1961: 49). Though it is undeniable that landscape painting over the centuries has maintained a relatively consistent subject matter—landscapes—and that the surface features of such paintings have changed quite a bit, it is also true that the aspects of landscapes upon which artists focus have changed. Does one depict a landscape in the evening or at midday? In full sun, dappled, or shrouded by clouds? Since there is a sense in which the contents of landscapes seem to have changed over the centuries, the changes in landscapes’ surface features do not establish any semantic competition in interpretative practice. In addition, it is certainly possible that interpretative practices change over time, but that would not undermine this chapter’s main claim. The interesting fact about interpretative practice is that viewers at a given time (and in a given cultural milieu) make use of a diverse set of interpretative schemes that do not compete syntactically or semantically. Hobbema’s contemporaries might have found his Watermill just as bright as Constable’s contemporaries found his White Horse. In that sense, the interpretative scheme deployed in seventeenth-century Europe might compete semantically with the one used a century and a half thereafter. But that doesn’t mean anyone actually makes use of competing interpretative schemes. Constable’s work might very well have rendered Hobbema’s scenes dark to nineteenth-century eyes. Landscape painting from the seventeenth and eighteenth centuries has a tendency to seem dark, or at least quite misty, and perhaps the cost of letting the midday sun into painting was letting it set a bit on one’s forebears. Gombrich’s slippage between discussing changes in subject matter and changes in the non-semantic features that accomplish certain representational ends might result from the fact that he too used non-competing interpretative schemes, even though his historian’s eyes could see the diachronic development of interpretative practice. Overall, then, while it’s not implausible that interpretative practice has changed over time, Gombrich’s landscape examples are far from establishing that viewers deploy competing interpretative schemes, and it is an interesting question how one might go about settling this issue decisively.

38 john kulvicki

5. Objection: Linear and Painterly One ﬁnds a related discussion of painting in Heinrich W¨olfﬂin’s Principles of Art History (1932), which introduces ﬁve opposed pairs that qualify sixteenth- and seventeenth-century European pictures, respectively. Most famous is the development from linear to painterly style. Linear style emphasizes things’ edges while the painterly stresses masses. Everything depends on how far a preponderating signiﬁcance is assigned to or withdrawn from the edges, whether they must be read as lines or not . . . Therefore, what makes the difference between Dürer and Rembrandt is not less or more in the exploitation of light and shade, but the fact that in the one case the masses appear with stressed, in the other with unstressed edges. (W¨olfﬂin 1932: 19)

It was important to W¨olfﬂin that this distinction does not depend essentially on the contents of the pictures in question. It is not a matter of the representation of light and shadow, since chiaroscuro came into its own in the Cinquecento, well before the painterly style bloomed a century thereafter. Linear and painterly are thus candidates for semantically competing, and perhaps syntactically competing, systems of pictorial representation. That is, if there is no bar to which subjects each style can represent, there can be pictures in each style that differ from one another in their surface details even though they do not differ in content. W¨olfﬂin was not blind to the fact that there was a shift in subject matter over the two centuries on which he focused, however. Some subjects’ outlines lend themselves to being stressed. Others do not. He even went so far as to explicate the ‘picturesque’ (1932: 23–7) in terms that would help to explain this. Picturesque scenes lend themselves to painterly depiction. For example, ‘there is a picturesque beauty of the ruin. The rigour of the tectonic form is broken up, and while the wall crumbles and holes and ﬁssures arise, a life quickens which quivers and shimmers over the surface’ (1932: 24). The idea is that ruins, having outlines that are rather indeterminate and thus de-emphasized from the perspective of an ordinary observer, lend themselves to depiction that does not emphasize edges so much as it stresses masses. The picturesque effect on perceivers is ‘impossible for non-ruinous architecture’ (1932: 24), though, as will become clear, it is

pictorial diversity

39

possible for pictures of non-ruinous architecture. Light and shadow also play a role in the picturesque. Unevenly illuminated scenes (1932: 25), presumably even those that involve ‘non-ruinous architecture’, de-emphasize the outlines of objects because luminance boundaries compete for viewers’ attention with object boundaries. And in dimly illuminated scenes, ‘instead of a number of separate objects, we see indeterminate lighter or darker masses which ﬂow together in a common tonal movement’ (1932: 26). The dim light at dusk is picturesque; the bright light at noon is not. The picturesque is a problem if one thinks that linear and painterly can be explained independently of content. Indeed, the challenge is similar to the one faced by Gombrich. Might it be that the painterly is merely the depiction of relatively picturesque scenes? Or, to put it more carefully, might it be that the painterly is merely the depiction of scenes as being picturesque? In that case, linear and painterly depiction might be explained solely in terms of pictorial content, rather than independently of it. W¨olfﬂin seems to have been aware of this worry. His response was that there are painterly pictures of non-picturesque scenes. ‘A delicate architectural painter does not require picturesque buildings to make a painterly picture’ (1932: 26). That is to say, even non-ruinous architectural scenes can be portrayed in a painterly manner by a skilled artist. He suggests that all of this should be obvious by a mere ‘glance at art history’ (1932: 26). But there is nothing obvious about it, and even if it is true, it is unclear how this would support W¨olfﬂin’s thesis. The idea is supposed to be that a scene might not be picturesque when seen face to face, but that some pictures of it can nevertheless be painterly. This suggests to W¨olfﬂin that we cannot understand the painterly exclusively in terms of pictorial content. But different pictures of any given scene can depict it as having different features. So it might be that the painterly style amounts to a particular set of features that objects are represented as having, and that these features are distinct from those typically represented by linear pictures. This alternative explanation is not considered by W¨olfﬂin, even though much of what he says strongly suggests that it is true. W¨olfﬂin believed that painterly pictures of picturesque scenes are merely the ‘preliminary stage to the higher forms of the painterly taste’ (1932: 27), and that just about any scene can be the basis of a painterly picture once painterly technique has been fully developed. Why is the picturesque an

40 john kulvicki important ﬁrst step, however? Why are such contents what matter, at least originally, to the development of painterly style? The most obvious explanation is that what W¨olfﬂin calls ‘painterly’ consists of a certain set of features that objects are depicted as having. Techniques developed in the sixteenth century, applied to picturesque scenes, yield something close to the painterly depiction of the seventeenth century. This is because picturesque scenes have the features that painterly pictures strive to depict objects as having: indeterminate or, at least, unstressed outlines, etc. One can depict scenes not otherwise picturesque as having features characteristic of picturesque scenes, once the techniques for producing such pictures have been learned by depicting picturesque scenes. Indeed, this might be easier than one would think, because the painterly is understood in terms of lacking what linear pictures possess. The painterly does not emphasize edges or outlines, so a picture of a non-ruinous classical fac¸ade can be painterly if it says little about the edges that, if depicted in detail, would result in a more linear effect. The next section shows how a tendency to give up on depicted detail reinforces the view that pictorial interpretative schemes do not compete with one another. For now, the point is that W¨olfﬂin does little to establish the linear–painterly distinction in a way that demonstrates interpretative competition.³ Gombrich was inﬂuenced by W¨olfﬂin in particular, and by Burkhardtian art history in general, but in his famous essay ‘Norm and Form’, he criticizes W¨olfﬂin’s distinctions precisely because they seem more tied to the contents of paintings than W¨olfﬂin wanted to admit (Gombrich 1971: 93–4). The details of Gombrich’s criticism and his own counter-proposal are beside the point, but the gist of his response is important. W¨olfﬂin underestimated the complexities of pictorial representation, or understood them as uninteresting compared to non-representational, formal aspects of pictures (Gombrich 1971: 94–5). Of course pictures represent, but they are artistically interesting because of the formal aspects that distinguish them from one another. Gombrich notes that this is something of a reversal from the Renaissance, when formal and compositional aspects of pictures received almost no attention from critics and other commentators, compared to their representational aspects. The problem is that by ³ Similar remarks apply to the other four distinctions Wo¨ lfﬂin draws between sixteenth- and seventeenth-century pictures. Indeed, they are even less plausibly divorced from content than linear and painterly.

pictorial diversity

41

de-emphasizing representational aspects of pictures, W¨olfﬂin simply ignored what might be essential to understanding the distinction between linear and painterly. Meyer Shapiro voiced a similar worry: ‘It is not always clear which formal traits are independent of representation’ (1994: 81), and ‘it may be asked whether the formal schemas, like W¨olfﬂin’s, are not veiled categories of representation’ (1994: 76).

6. Space: Less Is Sometimes More Pictorial While the representation of light and shadow is a decent candidate for interpretative competition, the representation of space is not. Robert Hopkins points out that the representation of spatial features is central to depiction in a way that the representation of light, colour, and shadow is not: ‘Colour and tone are features of a picture which only sometimes play a role in dictating what it depicts . . . Shape, on the other hand, is always relevant—if anything is’ (1998: 52). Gombrich famously exempted linear perspective from the class of pictorial conventions, claiming that its special place is due merely to the fact that we cannot see around corners (1961: 250). Goodman (1976, Ch. 1) took Gombrich to task for this, but for present purposes it’s not the conventionality debates that matter. Gombrich was acutely aware of the fact that pictures are much less forgiving of spatial variation than they are of variations in other dimensions, and this seems true regardless of whether one explains the phenomenon by appeal to non-conventional features of pictures, or by appeal to habits. The ﬁsheye system’s pictures, for example, bear a different kind of spatial relation to the objects they depict than ordinary pictures do. Similar claims hold for pictures in reverse perspective and for anamorphic distortions of all sorts. It is contrary to our practice to suggest that we interpret such pictures as competing syntactically or semantically with one another. Fisheye photos seem to depict bulgy things, and anamorphic distortions seem to depict objects distorted. It’s even tempting to think that there is no spatial diversity manifest in our practice with pictures, but this would go too far. Viewers use many distinct, but non-competing, interpretative schemes, even when it comes to the representation of space, and these spatial examples are rather striking illustrations of how interpreters avoid competition.

42 john kulvicki Recall that there is a sense in which the ﬁsheye photo is interpreted as depicting a bulgy lemon. Realizing that the lemon is depicted as having an odd shape does not lead one to reinterpret the photo as though it depicts an ordinary lemon in a ﬁsheye interpretative scheme. What does plausibly happen, however, is that one discounts the bulgy interpretation in favor of something much less speciﬁc. The photo is understood to depict a lemon without depicting it as having any particular shape aside from a generally lemony one. Faced with a ﬁsheye photo, or even an odder anamorphic monstrosity, there is a temptation to discount shape as something that is represented at any level of detail. Reverse perspective constructions are similar. On the one hand they seem to depict rather oddly shaped objects, while on the other they can seem to say little about spatial detail.⁴ Yes, that Byzantine icon depicts an altar, but no, it doesn’t depict the altar as having any particular shape. At best, it depicts the altar as having a certain number of sides, sharp or rounded corners, etc. So we do not give up altogether on such pictures as representing spatial qualities; we just give up on the details when the details seem particularly odd. Our willingness to give up on depicted detail is as striking as our unwillingness to adopt odd interpretative schemes for pictures, and it is not limited to spatial aspects of pictures, even though it is most prominent there. When the colour scheme of a picture is particularly extravagant, colour is discounted as contributing systematically to representational content. There is a sense in which many Matisses seem to depict things as oddly coloured and oddly shaped, but there is an equally strong tendency to regard these paintings as saying quite little about the details of shape and colour. The shapes and colours of the painted surfaces are quite interesting, aesthetically and otherwise, but they are not interesting just because of what they manage to depict. In such cases, it’s easier to focus on surface features of a picture in their own right. It’s more difﬁcult to ignore spatial and chromatic features of C´ezanne’s paintings when interpreting them than it is for Matisse’s, ⁴ Hopkins agrees that these seem like the live options for handling such odd pictures. Using different examples, he ﬁnds ‘two possible sources of legitimate diversity. One is if the two pictures depict oblong tables while differing over the details . . . The other possibility is that, while the two pictures do not disagree on the table’s properties, one is simply indeterminate with respect to features which the other represents’ (1998: 153). He develops the latter point in Hopkins (2003). Most of Hopkins’s discussion takes place in the context of defending his account of depiction against possible counter-examples. The present focus is understanding pictorial diversity, but the next section will show that Hopkins’s account is well suited to explaining the kind of diversity articulated herein.

pictorial diversity

43

for example, so the former are more apt to be interpreted as representing things as bearing odd but interesting spatial relations to one another. This back-and-forth between content and surface has been mentioned by almost every philosopher writing about pictorial representation, but it has not been noticed that this process manifests a commitment to avoid competing interpretative schemes.⁵ To the extent that shape is discounted, competition with an ordinary system of depiction is avoided, at the price of informativeness. Such pictures have a more limited vocabulary, as it were, but they don’t speak a different language. Informativeness is a price we are willing to pay in order to avoid competition. At this point, pictorial diversity seems to have a rather interesting shape, one unique to pictures. No explanation has yet been offered for this pattern of pictorial diversity. Given that it does not show up with other kinds of representation, however, one might expect accounts of what distinguishes pictures from other representational kinds to explain the pattern. As it turns out, no existing account of pictorial representation predicts such a pattern of interpretative schemes, even though no account excludes it, either. The next two sections sketch what an explanation of this phenomenon should look like, and indicate how some accounts of pictures and the perception thereof can accommodate such an explanation. There are, in a sense, two things to explain: why competition is absent and why the non-competing diversity manifests itself as it does. As will become clear, understanding both facts requires keeping one’s eyes ﬁrmly ﬁxed on both the syntactic and semantic features of pictures.

7. Explaining Diversity’s Presence To begin, why would one expect the diversity that exists? The previous section suggests that details of a picture surface are often excused from carrying semantic weight, when their doing so would result in a picture with a rather odd content. Other interpretative schemes that viewers use are, in a sense, abstractions from linear perspective: they leave out what linear perspective leaves in. Rather than take the ﬁsheye photo to depict ⁵ Two interesting examples are in this volume (Hopkins and Brown, Chs 6 and 8). For a related discussion of these phenomena as they relate to pictorial realism, see Kulvicki (2006, Chs 11–12).

44 john kulvicki something oddly bulgy for a lemon, it is taken to say something less speciﬁc about the lemon’s shape. So what happens when the ﬁsheye photo is reinterpreted as representing something rather indeterminate? The most plausible account is that the features of the picture surface that are responsible for depicting the speciﬁc, bulgy shape are themselves ignored. One merely regards less determinate features of the picture as those that carry semantic weight, while the properties responsible for depicting the bulgy shape are relieved of their semantic burdens. So, it’s not the speciﬁc shape of the region of the lemon picture that determines what it represents so much as an abstraction from that speciﬁc shape. Such a watered-down interpretative scheme is the same as the richer one where they overlap syntactically and semantically, so these two schemes do not compete with one another. It is a striking feature of pictures that one can ignore some of their surface details and still regard the artefact as having content. Indeed, this is not true of most other kinds of representation, and this fact about pictures goes some way towards explaining the pattern of pictorial diversity that we ﬁnd here. Consider, by contrast to the picture case, that friend of yours who always exaggerates. His most recent missive describes two feet of snow with winds over ﬁfty and temperatures below zero. That’s a rather extraordinary spring storm. The charitable reader in you understands him to describe merely a large amount of snow, somewhat high winds, and fairly chilly temperatures. In effect, you abstract from the details of the content because you know those details are unlikely to be accurate. The details constitute more an aspect of your friend’s expressive style than an accurate reﬂection of the way things are. In such cases, notice, you abstract over features of the contents alone: features of the inscriptions play no role in the reinterpretation. There is no interesting abstraction over the properties of the representation, such as the letters or words inscribed, that corresponds to an abstraction over the properties of what is represented. If you abstract over those properties, the likely result is uninterpretable nonsense. This situation with the exaggerating friend contrasts sharply with pictures. It is easy to abstract over features of pictures and still have interpretable representations. For example, one can ignore the garishly coloured landscape and treat the picture as though it merely represents a landscape while remaining silent concerning the landscape’s colours. Working with the representation itself, and discounting some of its features, leads readily to

pictorial diversity

45

abstractions over its content. With your friend’s letter, the abstraction relies completely on your conception of what your friend describes. You know what storms are like in the spring and you know what your friend is like, so certain charitable reinterpretations suggest themselves. With a picture, such abstractions are always possible based on an acquaintance with the features of the picture surface. Ignoring some of them leads quite readily to an abstraction over the picture’s content. This feature of pictures deserves further discussion (see e.g. Kulvicki forthcoming; 2006, Ch. 12). For now, the point is that this aspect of pictures makes the pattern of diversity that we ﬁnd rather plausible. Pictures lend themselves straightforwardly to such reinterpretations.

8. Explaining Diversity’s Absence But why isn’t there more diversity to pictures? Why isn’t there any competition among pictorial schemes? After all, many diagrams and graphs lend themselves to abstractions over their contents in the way that pictures do (Kulvicki forthcoming), but they are also readily interpretable in ways that compete. A graph could just as well chart changes in temperature over time as changes in pressure over time, for example, while pictures do not readily admit of such reinterpretations. Why? There are two claims that seem particularly relevant to explaining this lack of pictorial diversity. The ﬁrst is uncontroversial, in part because it is fairly weak, while the second is rather more controversial. The second claim entails the ﬁrst. Even if one accepts both claims, they do not complete the story. The full story concerning pictorial diversity is not one that a philosopher is in a position to tell. First, many agree that there is a fairly signiﬁcant constraint on what can be depicted. In particular, to use Hopkins’s phrase, ‘what can be depicted can be seen’ (1998: 28).⁶ Hopkins’s remark does not just concern the semantics ⁶ Elsewhere, I suggest that it’s more accurate to characterize this constraint by claiming that visible pictures depict visible things, audible pictures depict audible things, and so on (2006: 117). It’s best to ignore the possibility of audible and other kinds of pictures in this context. Wollheim (1987: 64) cites Lessing’s Laocoon (1910) as a source for an extreme version of the view that the realm of depiction is limited by the realm of the visible. Wollheim himself disagrees (1987: 65–9) for reasons inspired by Goodman (1976, Ch. 1) and Evans (1982), but these worries do not cut against the point being made here.

46 john kulvicki of pictures, as his formulation might suggest. Pictures, as Hopkins conceives of them, are themselves visible, in the sense that the features of pictures responsible for them depicting what they do are all visible features. That’s not to say that being a picture is merely being something with the right set of visible features, since on any reasonable view of what an artefact is, the intentions of makers play some role in determining its identity. It’s just that which picture something is depends on such an artefact’s perceptible features, and these relevant features are all visible. This is a claim about pictorial syntax. The non-semantically identiﬁable features of a picture that are responsible for it depicting what it does are all visible features and, adding a semantic point to this syntactic one, ‘what can be depicted can be seen’. So there is an important relationship, on most views of depiction, between the syntactic features of pictures and the features they depict objects as having: they are all visible. This fact prevents much competition, because something whose visible features are responsible for it representing invisible features of objects, like a graph of temperature over time, is not a picture. One is thus not going to ﬁnd syntactic competition in which some visible feature is used to represent some visible quality on the one hand and some invisible quality on the other. This fact about pictures does not preclude all competition, of course. The colour-complement scheme, reverse perspective, anamorphosis, and other techniques ﬁt the visibility desideratum while at the same time competing syntactically and semantically with our ordinary interpretations of pictures. The second kind of claim upon which an explanation of diversity can build is rather controversial, but not without its supporters. Not only is it the case that visible pictures depict visible things, but pictures depict what they do partly by instantiating properties of what they depict (Kulvicki 2006) or, at least, by being experienced as resembling what they depict (Hopkins 1998). Each of these views takes to heart the idea that pictures resemble what they depict, while parrying famous objections to such a proposal (e.g. Goodman 1976, Ch. 1). To be a little more speciﬁc, such proposals suggest that in order to represent X pictorially, a representation must either share some properties with X or be experienced as resembling X in some respects. Such proposals are controversial for many reasons, and it would take the present study off topic to discuss them in great detail. The question on the table is whether and to what extent they help explain the lack of pictorial diversity in question.

pictorial diversity

47

For each of these accounts, the properties of a picture’s surface place signiﬁcant constraints on what such a picture can depict. My view is that pictures have, following up on an idea of John Haugeland’s (1991), bare bones contents.⁷ These contents are rather abstract, typically including things like indeterminate shape and colour properties. A linear perspective picture of a rectangular table seen obliquely, for example, includes a quadrilateral region of space as part of its bare bones content. Bare bones contents constrain ﬂeshed out contents in the sense that the latter include and are consistent with the former, but bare bones content does not fully determine ﬂeshed out content. Being a rectangular table at an oblique angle to the line of sight is consistent with occupying a quadrilateral region of space, but so is being a trapezoidal object at an angle perpendicular to the line of sight. Alternative ﬂeshed out interpretations of pictures are always available, while alternative bare bones interpretations are not. Pictures are interesting partly because they instantiate their bare bones contents. Pictures whose bare bones contents include a quadrilateral region of space are themselves quadrilateral in some region of their surface. This helps to explain why alternative interpretations with respect to bare bones content are unavailable and why ﬂeshed out alternatives are. In order to have different bare bones content, the picture would need to change its surface features. Also, this helps further to explain the presence of certain kinds of diversity. If one systematically ignores, by abstracting over, certain features of pictures ordinarily included in their bare bones content, the result is a less determinate constraint on ﬂeshed out content. By ignoring hue, for example, interpreters are less committed to the colours a picture represents things as having. But more to the immediate point, this fact about pictures, if it is a fact about them, also helps to explain the lack of competition among interpretative schemes. Syntactic competition involves treating a given feature of pictures, like being quadrilateral in a certain region, as being responsible for representing distinct properties in distinct interpretative schemes. If pictures instantiate their bare bones contents, however, it’s going to be difﬁcult to ﬁnd pictorial schemes of interpretation that compete syntactically. At least one of those purported competing schemes will fail to be pictorial because it will fail to ⁷ What follows is not a defense of my view, but merely a statement of it sufﬁcient to show how such an account might explain the facts about pictorial diversity. For more detail and defense, see Kulvicki (2006, esp. Chs 3–4).

48 john kulvicki meet the constraint that the picture instantiate its bare bones content. For example, the interpretative scheme that inverts the ﬁsheye mapping is not a pictorial scheme of interpretation on the current view. Fisheye photos with quadrilateral regions of their surfaces do not represent quadrilateral regions of space according to the ﬁsheye interpretative scheme. Semantic competition is similarly constrained. Two pictures alike in some respect of their bare bones contents are also alike in some respect of their surface features. Perhaps this last proposal sounds like trying to solve a substantive issue with a merely legislative decision. The ﬁsheye photos and countless anamorphic distortions have been excluded from discussion because they are not, strictly speaking, pictorial interpretative schemes. It is of interest why we do not merely invert the ﬁsheye projection, however, and the present approach avoids that issue by declaring these systems non-pictorial. But the point is that if this is the correct account of depiction, then the range of pictorial interpretative options is quite limited, and thus it is easy to explain some of the lack of pictorial diversity. This would be a merely legislative move if the account of pictorial representation upon which it depends were either unmotivated or motivated solely by a desire to explain the absence of certain kinds of diversity. While it is beyond the scope of this chapter to motivate the view of pictorial representation in question, it helps to notice that it was fashioned without the end of handling this speciﬁc problem. So, what motivation it has does not come from here. The foregoing does not fully explain the lack of pictorial competition, since there are distinct pictorial interpretative schemes, even given the constraint that pictures instantiate their bare bones contents. Reverse perspective, for example, is a pictorial system of representation, even though it works quite differently than ordinary linear perspective. Similar remarks apply to axonometric perspective and variants thereupon (Kulvicki 2006, Ch. 5). Failure to use such interpretative schemes cannot be explained in the above manner, and there is nothing else in the account that will explain it. This is the point at which facts about the ways in which we can engage perceptually and otherwise with artefacts must play a role in limiting our range, and psychologists are in a better position to identify and explain those limitations than philosophers are. Hopkins’s view is that pictures depict what they are experienced as resembling in outline shape. Outline shape is a relational, three-dimensional

pictorial diversity

49

spatial property that coloured planes can share with the three-dimensional objects they typically depict. Pictures need not share outline shape with objects to depict them, but they need to be experienced as doing so. The details about outline shape are irrelevant to present purposes, but it is easy to see how such a constraint on depiction restricts pictorial diversity. Syntactically competing schemes would require one and the same pattern on a surface to be experienced as resembling distinct outline shapes, depending on the interpretative scheme being used. There is no reason to think this is impossible, but it is highly implausible. What one experiences an object as resembling in outline shape might depend on one’s prior experience with pictures, one’s care in observing things, etc., but it hardly seems to be the kind of thing one can shift at will depending on the kinds of pictures one interacts with. Indeed, Hopkins even has an easy time explaining why reverse perspective pictures are not readily interpretable within a reverse perspective system: they systematically distort outline shape (Hopkins 1998: 156–8; 2003). Semantic competition is more plausibly permitted by the experienced resemblance view. It requires that distinct surface patterns are experienced as resembling one and the same thing. There is no reason to think this is impossible or even terribly uncommon. If there were something systematic to such experiences, it’s plausible that one could make use of semantically competing pictorial systems. It is encouraging to note from this perspective that the most plausible, though not decisive, instances of competition are semantic and not syntactic, as noted in the discussion of Gombrich and W¨olfﬂin above. The reason there is not much semantic competition, if there is any at all, is presumably that there is no such systematic bivalence to our experiences. This latter fact is in no way internal to the experienced resemblance theory of depiction, so this is another point at which an account must look beyond itself for a full explanation of pictorial diversity. Other views of depiction are less readily equipped to explain pictorial diversity as unpacked above because they place less stress on the speciﬁc relationships between syntactic features of pictures and their semantic features. Recognitional accounts of depiction, like those of Flint Schier (1986) and Lopes (1996), for example, explain pictorial content merely by appeal to the recognitional capacities pictures trigger and engender. Goodman’s symbolic account is designed to make the range of depictive

50 john kulvicki possibilities quite open-ended. Even so, the pattern of pictorial diversity explicated herein is not inconsistent with these other accounts of depiction. It’s just that whatever limits the range of pictorial diversity, or explains its presence, is more completely external to such views than it is to mine or to Hopkins’s. Proponents of such views tend to suggest that pictorial diversity is a bit vaster in scope than this chapter suggests it is. Goodman was exceedingly cavalier in discussing the scope of depiction—‘almost any picture may represent almost anything’ (1976: 38)—and Lopes suggests that the range of interpretative schemes in use is quite broad (1996: 32–5).

9. Conclusion The goal of this chapter has been to sketch the shape of pictorial diversity and suggest how the phenomenon might be explained. Most discussions of pictorial diversity start with an account of depiction and ask how that account can handle the diversity we ﬁnd. But too little time has been spent trying to articulate the phenomenon in detail, and this is likely because the focus of most discussions is the vindication of one or another account of depiction. The hope is that this chapter remedies such past offenses by introducing two conceptual tools that help to sort out the respects in which pictures do and do not vary. Focusing on the distinctions between systems for producing and interpreting pictures and between competing and non-competing interpretative schemes reveals an impressive pattern in what might otherwise seem like a rather anarchic jumble. Independently of how one wants to explain the pattern, the pattern itself distinguishes pictorial representations from other kinds of representation. And it is an interesting challenge for any view of depiction, though one not fully in the philosophers’ court, to explain the pattern that makes pictures so distinctive.

References Elkins, James (1994), The Poetics of Perspective (Ithaca, NY: Cornell University Press). Evans, Gareth (1982), The Varieties of Reference (Oxford: Clarendon Press). Gombrich, Ernst (1961), Art and Illusion, 2nd edn (Princeton: Princeton University Press).

pictorial diversity

51

(1971), Norm and Form (London: Phaidon). Goodman, Nelson (1976), Languages of Art, 2nd edn (Indianapolis: Hackett). Haugeland, John (1991), ‘Representational Genera’, in W. Ramsey, S. Stich, and D. Rumelhart (eds), Philosophy and Connectionist Theory (Hillsdale, NJ: Lawrence Erlbaum). Hopkins, Robert (1998), Picture, Image, and Experience (Cambridge: Cambridge University Press). (2003), ‘Perspective, Convention, and Compromise’, in H. Hecht, R. Schwartz, and M. Atherton (eds), Looking into Pictures (Cambridge, Mass.: MIT Press). Kulvicki, John (2006), On Images (Oxford: Clarendon Press). (forthcoming), ‘Knowing with Images: Medium and Message’, Philosophy of Science. Lessing, Gotthold (1910), Laocoon: An Essay upon the Limits of Painting and Poetry (1766), trans. E. Frothingham (Boston: Little, Brown). Lopes, Dominic McIver (1996), Understanding Pictures (Oxford: Clarendon Press). Schier, Flint (1986), Deeper into Pictures (Cambridge: Cambridge University Press). Shapiro, Meyer (1994), Theory and Philosophy of Art (New York: George Braziller). Willats, John (1997), Art and Representation (Princeton: Princeton University Press). (2003), ‘Optical Laws or Symbolic Rules? The Dual Nature of Pictorial Systems’, in H. Hecht, R. Schwartz, and M. Atherton (eds), Looking into Pictures (Cambridge, Mass.: MIT Press). W¨olfﬂin, Heinrich (1932), The Principles of Art History, trans. M. D. Hottinger (New York: Holt). Wollheim, Richard (1980), Art and its Objects, 2nd edn (Cambridge: Cambridge University Press). (1987), Painting as an Art (Princeton: Princeton University Press). (2003), ‘What Makes Representational Painting Truly Visual?’, Proceedings of the Aristotelian Society, suppl. vol., 77/1: 131–47. Ziff, Paul (1984), Antiaesthetics (Dordrecht: D. Reidel).

2 Picture This: Image-based Demonstratives DOMINIC MCIVER LOPES

Settling down after the big meal at the family reunion brings on a little nostalgia. Out come the photo albums. As the pages turn, you see familiar faces as they looked long ago. One photo shows a surprisingly sexy young woman, and you exclaim, ‘That’s Aunt Jane!’ What you say is true. The explanation is this: what you say is true in part because the picture puts you in the same kind of position with respect to Aunt Jane as the position you are in when you see her face to face. In general, (DR) Pictures perceptually ground demonstrative reference to depicted objects. This chapter makes a case for DR.

1. Pictures, Perception, and Demonstrative Reference Philosophers mostly hold that your exclamation is not on a par with ‘That’s Scarlett Johansson’, said in plain sight of her. A contrast with such views sharpens DR and indicates how to defend it. One line of thinking is pushed by Jos´e Luis Berm´udez (2000). Berm´udez assumes that a perceiver can demonstratively refer only to an object to which he stands in a certain perceptual relation. The details of the relation I am grateful to Catharine Abell, Katerina Bantinaki, Vincent Bergeron, Mohan Matthen, Bence Nanay, Dustin Stokes, Alberto Voltolini, and the Press’s anonymous referees for many helpful suggestions.

picture this: image-based demonstratives

53

do not matter now: what matters is that the relation grounds demonstrative reference—the perceiver is able to refer demonstratively to an object in part because he stands in the given perceptual relation to it. However, according to Berm´udez, looking at Aunt Jane’s photograph does not place you in the given perceptual relation with her. So that is not your aunt, that is a photograph, and ‘That’s Aunt Jane’ is false if it refers demonstratively. Since this is obvious, we seek a non-demonstrative reference, and what Berm´udez calls the ‘pictorial nature’ of the photograph swiftly brings us to see that the intended referent is the woman in the photograph. Berm´udez suggests that your exclamation contains an elliptical deﬁnite description and should be parsed as ‘The woman represented by that photograph is Aunt Jane’. This reasoning highlights two facts that need to be explained. First, we often succeed in referring when we point to images and say things like ‘That’s Aunt Jane’ or ‘That’s a mongoose’. You open yourself to correction if you point to the photograph of Aunt Jane and say ‘That’s Aunt Rose’ or if you point to a photograph of a mongoose and say ‘That’s a mink’. The correction will be ‘No, that’s Aunt Jane’ or ‘No, that’s a mongoose’. How is reference made in these cases, if not demonstratively? Descriptively, according to Berm´udez—though other answers are available (e.g. Voltolini 2009). Second, we need to explain why, in your exclamation, the appearance of demonstrative reference exerts so little strain. It is as natural as can be to point to an image and use a ‘this’ or a ‘that’ to refer to its depictum. To see this, consider three cases: you say ‘That’s Aunt Jane’ when you point to her, when you look at a picture of her, and when you pick out her number from the phone book. The second case is not just like the third case, where you ostend a phone number and say ‘That’s Aunt Jane’ (Quine 1968: 194). Even if the second and third cases both involve a reference shift, the mechanisms are different, and this difference can be expressed in any of a number of ways. Perhaps the second case is grounded in a ‘quasi-perceptual experience’ as of Aunt Jane; or perhaps, as Berm´udez proposes, the second case exploits the ‘pictorial nature’ of the photograph. Whatever we say to express this difference should explain the relative ease and automaticity of the second case, by contrast with the third case. Gareth Evans makes similar points about moving images. He writes that in watching TV, ‘ ‘‘this’’ and ‘‘that’’ pop out without the slightest

54 dominic mciver lopes sense of strain’ (Evans 1982: 147–8). The reason is that the TV image provides an information link to its depictum—albeit via an information channel with ‘circuits and time lags’ (Evans 1982: 147). But although the information link to the depictum explains why ‘this’ and ‘that’ so easily pop out, it is not enough to ground demonstrative reference. Evans proposes that statements like ‘That’s Luongo in net’ (said as you watch the hockey game on TV) exploit a concept of the information link via an image from object to perceiver, so they do not involve pure demonstrative reference. Like Berm´udez, Evans accepts that refusals to read your exclamation at face value, as referring demonstratively to Aunt Jane, are compelling so long as we are told how your exclamation does refer to Aunt Jane and why the apparent demonstrative reference to Aunt Jane exerts practically no strain. DR explains without a bit of fuss why ‘That’s Aunt Jane’ refers without strain to Aunt Jane. First, your exclamation involves a genuine demonstrative reference to Aunt Jane. Second, this demonstrative reference is grounded in the photograph’s putting you in the same kind of perceptual state as you are in when you see her face to face. So now the challenge is not to explain how the ‘pictorial nature’ of the image is sufﬁciently like ordinary perception to ease the strain caused by the apparent demonstrative. The challenge is rather to show how the ‘pictorial nature’ of the image is so like ordinary perception as to ground genuine demonstrative reference. Berm´udez and Evans take all demonstrative reference to be grounded in a perceptual relation, but DR does not require this assumption. Colin McGinn (1981: 161–3) and Marga Reimer (1991: 191–4) discuss cases where demonstrative reference seems not to rely on any perceptual link to the referent, and they go on to infer that perception is incidental to demonstrative reference. As Susanna Siegel points out, this inference implies that demonstrative reference is a uniform phenomenon, and she gives some reasons to reject this assumption (Siegel 2002: 13–14). So DR is consistent with the view that demonstrative reference must be grounded in a perceptual relation, but it is also consistent with the view that demonstrative reference is not uniform and only some forms of demonstrative reference are grounded in perception. Thus, the case for DR hinges on an account of a perceptual relation that grounds some cases of demonstrative reference. P. F. Strawson (1959: 18) wrote that demonstrative reference requires that a perceiver ‘can pick out

picture this: image-based demonstratives

55

by sight or hearing or touch, or can otherwise sensibly discriminate, the particular being referred to’. Evans complained that the ordinary concept of perceptual picking out is too vague to make Strawson’s comment much more than a starting point: it is quite undetermined by the ordinary concept what kinds of spatial circuitousness and time-lags in the information channel are consistent with the subject’s being said to perceive the object. We speak of seeing someone on the television, or hearing him on the radio. We speak of seeing someone in a mirror as well as of seeing his reﬂection—but only of seeing someone’s shadow, not of seeing him in a shadow. (Evans 1982: 144)

If this is right, then we should expect accounts of the ordinary concept of perception more to reﬂect than to settle contending views about your exclamation. If analysis of the ordinary concept of perception is no help, then we should see what science has to offer. Mohan Matthen’s Seeing, Doing, and Knowing uses ﬁndings in vision science to construct an account of perception as, in part, anchoring demonstrative reference (Matthen 2005: 271–324). Elements of this account contribute to the case for DR. Moreover, since Matthen mounts an argument that gives legs to some philosophers’ worries about DR, the best hope for DR is to ﬁnd resources for its defence within Matthen’s argument.

2. Seeing Objects Perception can ground demonstrative reference only if it represents objects. However, some philosophers hold that perception cannot represent objects (e.g. Lowe 1993; Davies 1997; Clark 2000). Matthen’s ﬁrst step is to show that perception does represent objects. Or rather, since he holds that not all perceptual modalities represent objects, he shows more speciﬁcally that vision represents objects (Matthen 2005: 271–82). The argument for this will provide materials essential when we come back to consider demonstrative reference through pictures. Vision represents features. It represents grey and black, square and round, and rough and smooth, for instance. Obviously enough, seeing grey or

56 dominic mciver lopes

Figure 2.1. Grey square and black circle

Figure 2.2. Grey circle and black square

round is not normally an experience of grey or round throughout a visual scene. Normally, part of the scene is grey and part of the scene is round. Added to that, features are normally represented together, or bound. Some grey is bound to the round; it has a determinate size and shape. These facts generate the binding problem. Vision can discriminate the scene in Figure 2.1, containing a grey square and a black disk, from the scene in Figure 2.2, containing a grey disk and a black square. The difference between the scenes cannot be represented if vision merely represents the features grey, black, round, and circular. Vision must represent more in order to bind grey to square and black to circular in Figure 2.1 and grey to circular and black to square in Figure 2.2. What more is needed? One answer is location (Clark 2000). In Figure 2.1, vision represents grey and square as at the same place and black and round as at the same place; in Figure 2.2, vision represents round and grey as at one place and black and square as at another place. So vision represents features at places. Now consider the phi phenomenon. You are looking at a screen. A light ﬂashes on and off, then a light slightly displaced from the ﬁrst light

picture this: image-based demonstratives

57

Figure 2.3. The phi phenomenon

Figure 2.4. The phi phenomenon with change in colour

ﬂashes on and off. If conditions are right, you will seem to see one light moving rather than one light ﬂashing followed by a second light ﬂashing. (The phenomenon is a function of the distance between the lights, the size of the lights, the duration of each ﬂash, and the interval between ﬂashes.) The phi phenomenon is the basis of motion perception, since it generalizes beyond the regularities of the two-light case. For example, suppose four lights are arranged in the pattern shown in Figure 2.3, with the distance between lights 2 and 3 set at twice the distance between 1 and 2 and 3 and 4. You will seem to see one light moving across the screen only if the interval between the ﬂashes of lights 2 and 3 is about twice that of the interval between 1 and 2 and 3 and 4. Finally, Nelson Goodman asked what happens if the lights are different colours, as in Figure 2.4. The answer is that one light is seen as moving and changing colour. Matthen argues that the phi phenomenon and hence the visual representation of motion require more than the representation of features at places. Let n be the interval between ﬂashes of two white lights of a given size at a given distance when the ﬂashes occur for a given duration. The duration of the ﬂashes, the size of the lights, and the distance between lights are ﬁxed, but we vary n, the length of time after light 1 switches off and before light 2 switches on. The phi phenomenon occurs when n falls below a threshold value, otherwise it looks as if light 1 switches off and light 2 switches on. Suppose that the threshold value is 200 ms. So here are two candidate operating assumptions of vision: Assumption 1. If white-here at t and white-there at t + 100 ms, then white-here belongs to the same object as white-there.

58 dominic mciver lopes Assumption 2. If white-here at t and white-there at t + 300 ms, then white-here does not belong to the same object as white-there. Accordingly, when the antecedent of Assumption 1 is satisﬁed, vision represents one object, and when the antecedent of Assumption 2 is satisﬁed, vision represents two objects. However, there is no difference in the features and locations that are represented in each antecedent. Thus, it is not possible to explain the identity of white-here and white-there in Assumption 1 in terms of features at locations. Vision must represent something more than features and locations if it represents the difference between the 100 ms interval, which triggers the phi phenomenon, and the 300 ms interval, which does not trigger the phi phenomenon. Matthen proposes that vision represents objects. The operating assumptions of vision are in fact these: Assumption 1∗ . If object 1 appears to be here at t and object 2 appears to be there at t + 100 ms, then object 1 is the same as object 2. Assumption 2∗ . If object 1 appears to be here at t and object 2 appears to be there at t + 300 ms, then object 1 is not the same as object 2. Vision therefore attributes features to moving, changing objects. As Matthen sums up the point, ‘perceptions of change and motion demand an identity that underlies change. Locations do not provide such an identity. Hence the proper representation of visual states demands reference to objects, which can continue through time’ (Matthen 2005: 282). Notice that this is a minimal conception of object representation. Of course, vision captures the difference between the case which triggers the phi phenomenon and the case which does not by representing times in addition to features and locations. That is made clear in the antecedents of Assumptions 1 and 2. However, the point is that representing features as bound at places and times sufﬁces to represent objects. Representing change and motion represents objects. This is what is captured in the antecedents of Assumptions 1∗ and 2∗ . We cannot solve the binding problem just by saying that vision represents features at locations. Adding time to location is not an alternative to Matthen’s proposal, it is Matthen’s proposal, given a minimal conception of object representation.

picture this: image-based demonstratives

59

3. Seeing and Doing and the Feeling of Presence Representing objects is not enough for vision to ground demonstrative reference. What more is needed? Matthen proposes that a representation grounds demonstrative reference if it represents objects deictically. For future reference, (DX) Vision grounds demonstrative reference by representing objects deictically. A deictic representation is one that ‘makes reference to an object relative to a perceiver’s own body without intermediary descriptive information’ (Matthen 2005: 302). Thus, there are two conditions on deixis. First, a deictic representation represents an object in an egocentric framework. Second, it represents the object non-descriptively. Matthen uses this notion of deictic representation in part to explain a phenomenological contrast between seeing an item in a picture and seeing it face to face, and the explanation of this contrast adds to DX. Seeing Aunt Jane in a picture differs phenomenologically from seeing her face to face. For one thing, the latter comes with a feeling of presence that is obviously absent in looking at her portrait. In general, seeing things in pictures lacks this feeling of presence (though there are exceptions, like trompe l’œil images). Beyond this, little can be said about the phenomenological character of the feeling—about what the feeling is like. Matthen (2005: 304) calls it ‘inarticulate’, though he does say something about what the feeling is a feeling of. It is a feeling of things being ready to hand, of being perceptually coupled with them (Matthen 2005: 315–16). You do not feel as if you can reach out and touch your aunt when you see her in the photograph, as you would were you to see her in the ﬂesh. Perhaps this is not saying much, but not much more needs to be said. Matthen’s project is not descriptive phenomenology. His project is to explain why face-to-face seeing comes with this feeling of presence that seeing-in lacks. The explanation makes ingenious use of the two-pathway model of vision. Vision integrates two systems, which

60 dominic mciver lopes can be called ‘descriptive vision’ and ‘motion-guiding vision’. The systems recruit anatomically distinct structures, one running along ventral pathways from the occipital lobe to the inferior temporal lobe, and the other along dorsal pathways from the occipital lobe to the parietal lobe. For this reason, the two systems are often called ‘ventral’ and ‘dorsal’ vision (and this is the terminology Bence Nanay uses in Chapter 7 in this volume). However, it will be important to characterize them functionally rather than anatomically. Descriptive vision represents features of objects, such as shape. As a result, it feeds long-term memory and has epistemic signiﬁcance. Motion-guiding vision subserves short-term motor control: it orients the body to objects of attention and guides the limbs in acting on objects. To do this, it represents objects in an egocentric spatial framework. Since it is not descriptive, it does not feed long-term memory and it has no epistemic signiﬁcance. Descriptive and motion-guiding vision are doubly dissociable. An impairment of one system may spare the other (Milner and Goodale 1995; Weiskrantz 1997). A person with impaired descriptive vision may be able to reach for and pick up an object with the correct grip although she cannot describe its shape. A person with optic ataxia cannot properly manipulate an object that he can describe quite well. For example, he might be able to describe the distance and orientation of a slot and yet prove unable to put a letter through the slot without trial and error. The dissociation is also seen in unimpaired subjects. For instance, motionguiding vision is immune to size-contrast illusions that affect descriptive vision. In the Titchener (or Ebbinghaus) illusion, the size of the centre disk seems to vary as a function of the size of the disks surrounding it (Figure 2.5). Melvyn Goodale and his collaborators constructed a threedimensional version of the illusion using disks like poker chips (Aglioti et al. 1995; see also Ellis et al. 1999). As expected, the centre disk appeared smaller when surrounded by large disks than when surrounded by small disks. Nevertheless, subjects who were asked to pick up the centre disk scaled their grip to its actual size. Matthen proposes that motion-guiding vision is immune to the sizecontrast illusions because it only represents features of objects that are needed to guide action: it only represents target points. Consider a seabird diving for a ﬁsh (Matthen 2005: 301–3). The ﬁsh is moving fast but the bird must grip it with its beak. To succeed in catching the ﬁsh, the bird

picture this: image-based demonstratives

61

Figure 2.5. Titchener–Ebbinghaus illusion

must ﬁrst target two grip points, such that connecting the top beak with one point and the bottom beak with the other point allows it to carry off the ﬁsh (Figure 2.6). There is initially an interaction with descriptive vision, which uses shape information to determine the target points. It might use the information that striped ﬁsh are tasty and round ﬁsh are bony to choose a particular ﬁsh and it might then use the shape of that ﬁsh to decide what precise points (at the tick marks on Figure 2.6) on the ﬁsh to target. But once the target points have been set, the bird need only home its upper beak on one target point and its lower beak on the other, making rapid adjustments as its prey moves. Motion-guiding vision performs this task. Further updates from descriptive vision about the ﬁsh’s shape are not required: all the bird needs to secure its dinner is timely information that allows it to reorient itself continuously to the target points as they move and as the bird gets closer. If this distinction between motion-guiding vision and descriptive vision is correct, then only motion-guiding vision is deictic. It represents target points for action in an egocentric spatial framework without the intermediary descriptive information that is vulnerable to size-contrast illusions. With this distinction and the notion of deixis in hand, Matthen proposes that deixis is responsible for the feeling of presence. We are aware of the outputs of descriptive vision through conscious experience of shapes, sizes,

62 dominic mciver lopes

Figure 2.6. Two ﬁsh

colours, and the like; and we are aware of the outputs of motion-guiding vision through the ‘inarticulate’ feeling that things are ready to hand, that we are perceptually coupled with them for action. Moreover, motion-guiding vision is not implicated in seeing objects in pictures, though it does represent the surfaces of pictures. Suppose that you are about to reach for a poker chip on a table, gripping its near and far sides (as indicated by the tick marks on Figure 2.7). Your grip at these edges will be the same diameter as a grip targeting its left and right sides. Now consider a chip that is depicted as lying on a table. If motion-guiding vision set you up to reach for the depicted chip, it would use the same grip diameter for the near and far sides as for the left and right sides. In fact, motion-guiding vision sets a much smaller grip on the near and far sides of the depicted chip than on its left and right sides. The grip it sets for the near and far sides of the depicted chip is the same as the grip it sets for the top and bottom edges of the ellipse on the picture surface which depicts the chip. Motion-guiding vision does not guide your reach into the depicted space. It stops, as it were, at the picture surface. (In most cases anyway. One exception is being taken in by a trompe l’œil picture. To be taken in by such a picture, you must fail to see it as

picture this: image-based demonstratives

63

Figure 2.7. Poker chip on a table

a marked, two-dimensional picture surface: seeing the surface as a surface blocks any illusionistic experience of the depicted scene and hence any chance of being ‘taken in’ (Lopes 2005: 29–34). Thus, when taken in, one would expect motion-guiding vision not to set a grip diameter for the near and far sides of a trompe l’œil chip that matches the diameter of the grip it would set for the top and bottom edges of the corresponding ellipse on the picture surface. Instead, it should set the same grip on the near and far sides of the depicted chip as on its left and right sides.) To sum up, only motion-guiding vision is deictic, deictic representation comes with the feeling of presence, and seeing in pictures does not engage motion-guiding vision. These claims explain why seeing in pictures lacks the feeling of presence: (M1) (M2) (M3) (M4) (M5)

Only motion-guiding vision is deictic. Seeing O in P does not implicate motion-guiding vision. So seeing O in P is not deictic. Deixis is responsible for the feeling of presence. So seeing O in P lacks the feeling of presence.

64 dominic mciver lopes Remember, though, that the goal was to say what is needed, in addition to representing objects, in order for perception to ground demonstrative reference. According to Matthen, perception grounds demonstrative reference to objects by representing them deictically: (DX) Vision grounds demonstrative reference by representing objects deictically. We are now in a position to build on DX. Deictic representation is representation in an egocentric framework without intervening descriptive information. Motion-guiding vision is deictic, but not descriptive vision. Visually based demonstrative reference is grounded in motion-guiding vision.

4. Deixis in Motion-Guiding Vision Although Matthen never draws the inference explicitly, DX conjoined with M1 and M2 implies that (M6) Seeing O in P does not ground demonstrative reference to O. The empirical evidence suggests that motion-guiding vision is not involved in seeing in pictures. So if only motion-guiding vision is deictic, then seeing-in is not deictic. This explains why we do not feel as if we are in the presence of depicted objects—why we do not feel ready to reach out and act upon them. It also means that we cannot refer to them demonstratively. The explanation of why seeing O in P does not come with a feeling of presence also implies, given DX, that it does not ground demonstrative reference to O. But since M6 follows directly from M1, M2, and DX, making a case for DR means rejecting one or more of these premises. M2 is on solid ground and DX is plausible. That leaves M1.

5. Deixis in Descriptive Vision Matthen explains that seeing in pictures lacks the feeling of presence because deixis is responsible for the feeling of presence. Seeing face to face is deictic; seeing-in is not. However, there is another explanation of why seeing-in

picture this: image-based demonstratives

65

lacks the feeling of presence: it does not implicate motion-guiding vision. If this explanation is correct, M4 is not needed, and if M4 is not needed, then there is room to doubt M1. Seeing things in pictures normally lacks the feeling of presence that is part of seeing things face to face (or in trompe l’œil). According to Matthen, we have a feeling of O’s presence because O is represented deictically, and O is represented deictically because motion-guiding vision represents O. An equally good explanation is this. We have a feeling of O’s presence because motion-guiding vision represents O. Our experiences of depicted objects normally lack this feeling of presence because motion-guiding vision does not normally represent depicta—it stops at the picture surface. This explanation ﬁts our intuitions about the character of presence. Matthen describes presence as a feeling of things being ready to hand, of being perceptually coupled with them. Of course, you do not feel as if you can reach out and touch your aunt when you see her in the photograph, as you would were you to see her in the ﬂesh. The reason is that seeing her in the picture does not implicate motion-guiding vision. If M4 is not needed to explain why seeing in pictures lacks a feeling of presence, then there is room to contemplate rejecting M1, breaking the monopoly of motion-guiding vision over deixis. A word about ‘deixis’. As we saw, Matthen uses the term quite speciﬁcally: a deictic representation is one that represents objects in an egocentric framework ‘without intermediary descriptive information’ (Matthen 2005: 302). In the context of M1, the ﬁnal clause of this deﬁnition is a little puzzling. The claim that only motion-guiding vision is deictic is tautological if deictic representation excludes descriptive vision by deﬁnition. Since M1 is not intended as a tautology, the deﬁnition must be read another way. Here is one reading: R is a deictic representation of O just in case (1) R represents O egocentrically and (2) R does not accomplish (1) in virtue of attributing features to O. This reading is consistent with Matthen’s account of the seabird. As a matter of fact, the story of the seabird shows what is wrong with M1. An explanation of how motion-guiding vision works non-descriptively requires that descriptive vision represent objects deictically. The story tells of a seabird who is diving for ﬁsh. Its motion-guiding vision system targets two good grip points on the chosen ﬁsh, enabling the bird to catch its prey with its beak (see Figure 2.6). In order to do this, descriptive vision must

66 dominic mciver lopes indicate for motion-guiding vision which points to target. Only once these points are indicated can motion-guiding vision take over and complete the task at hand. But how does descriptive vision indicate the points for motion-guiding vision? There are two options. One is that motion-guiding vision processes descriptive information and descriptive vision indicates the points by describing features of the ﬁsh. It ‘says’ something like this: ‘See the striped ﬁsh? Go for the spot just forward of the dorsal ﬁn and the spot just below the gills’. (The point is the same if the vocabulary is restricted to shapes and colours.) The problem is that this kind of descriptive information is not available to motion-guiding vision. If it were, motion-guiding vision would yield size-contrast effects, and it does not yield these effects. The alternative is that descriptive vision speaks the language of motionguiding vision and represents the target points in that language. That is, it hands over to motion-guiding vision a set of coordinates in an egocentric spatial framework. It targets those spots: there and there, without using descriptive information about shapes and colours and other features whose representation yields size-contrast effects. So target assignment requires that descriptive vision represent objects deictically. (This is why it is crucial for present purposes to describe the two systems functionally rather than anatomically, as ‘dorsal’ and ‘ventral’ streams. It may be that the ventral stream does not process egocentric target information but descriptive vision integrates data from both streams so that it is deictic.) The claim that both systems represent objects deictically is consistent with the evidence that they are dissociable. What distinguishes the systems is not that one is deictic and the other is not. Rather, only one represents features in a way that makes it susceptible to size-contrast effects; and only one is responsible for precision grip. An optic atactic can describe one poker chip as ‘to my right’ and the other as ‘to my left’ but nevertheless have difﬁculty when told to pick up either one. In the Titchener illusion, descriptive vision underestimates the size of the solid angle subtended by the centre circle when surrounded by larger circles and overestimates the size of the solid angle subtended by the centre circle when surrounded by smaller circles. Motion-guiding vision does not make this error: it equips us with accurate, precision grip. Descriptive vision does not guide precision grip, but reduced accuracy does not imply a difference in spatial framework.

picture this: image-based demonstratives

67

Breaking the monopoly of motion-guiding vision over deixis also makes sense of seeing things in the distance. Motion-guiding vision does not stretch to distant objects. I have no precision grip for Crown Mountain, which I see across Burrard Inlet from my study window. Still, I can point to it (and very poorly gauge its distance) and I see it as lying to the left of Grouse Mountain. Nevertheless, descriptive vision places Crown Mountain in my action space—it primes me to set out in a certain direction to reach it or to point my binoculars in a certain direction to see it ‘up close’. So descriptive vision non-descriptively represents objects in an egocentric spatial framework: it is deictic. Replacing M1 with the claim that descriptive and motion-guiding vision are deictic breaks the argument for M6.

6. Deixis in Seeing-In If the argument for M6 fails, then a case can be made for DR, the thesis that seeing an object in a picture grounds demonstrative reference to the object. The case goes like this: (D1) Seeing O in P implicates descriptive vision. (D2) Descriptive vision represents objects deictically. (D3) So seeing O in P represents O deictically. Given DX, which takes perception to ground demonstrative reference by representing objects deictically, it follows that (D4) Seeing O in P grounds demonstrative reference to O. This reasoning vindicates DR, if it is sound. As long as it is given a weak reading, D1 follows from most theories of depiction. On a strong reading, seeing O in P involves some of the very same processes of descriptive vision as are engaged in seeing O face to face. According to recognition theories of depiction, if P depicts O then P engages the same visual processes as are involved in seeing O (Schier 1986; Lopes 1996). As a result, recognition theories endorse the strong reading of D1. Most other theories of depiction imply, at the least, a weaker reading of D1. For example, according to some theories, seeing O in P is explained as seeing a resemblance between O and P (Budd 1993; Hopkins 1998). Seeing a resemblance between O and P involves descriptive vision, for descriptive

68 dominic mciver lopes vision represents shapes, sizes, colours, and other such features of the visual scene, and it is also responsible for representing resemblances among these. So most leading contemporary theories of depiction imply D1, though perhaps not all do. An unclear case is Kendall Walton’s (1990) proposal that seeing O in P is explained by one’s imagining of one’s seeing P that it is a case of seeing O. This theory implies D1 only if imagining seeing involves descriptive vision. Only Goodman’s (1976) theory of depiction clearly fails to imply the weak reading of D1, though some have argued that it is consistent with it (Lopes 2000). In the circumstances, D1 looks like a safe bet. The more likely target is D2, the claim that descriptive vision represents objects deictically. The objection is not that D2 is out-and-out false, but rather that it must be qualiﬁed in a way that breaks the step to D3. After all, the account of the seabird shows only that descriptive vision sometimes represents objects deictically. In particular, it represents objects deictically when it must pass on information to motion-guiding vision. But seeing-in does not pass on information about depicta to motion-guiding vision, since seeing-in does not implicate motion-guiding vision. So maybe descriptive vision does not represent objects deictically when it is implicated in seeing them in pictures. If that is right, the step to D3 is invalid. This objection ﬁts Matthen’s discussion of pictures. Perceivers normally ‘cross-index descriptive information with information in egocentric coordinates’ (Matthen 2005: 315), but this does not happen when they see scenes in pictures (trompe l’œil aside). He writes, Suppose you are looking at a picture of two men shaking hands. Where are they? As far as you can tell by seeing in the picture, the question has no answer. The space within the picture is connected, as any good three-dimensional space should be: since the two men are shaking hands, they are spatially connected to one another in quite determinate ways. But their spatial relations are visually represented in object-centred, not viewer-centred coordinates. (Matthen 2005: 315)

As this passage suggests, the space within the picture is not part of the picture viewer’s space. In Matthen’s slogan, the picture lacks a ‘here’ (Matthen 2005: 316). The account given in Section 2 of how vision represents objects explains what it is for depicted objects to be represented in a space within the

picture this: image-based demonstratives

69

picture—not as part of the space of the picture surface. Vision represents features as bound together. According to the location theory of binding, features are bound at locations; but this does not explain the phi phenomenon, in which vision binds features to (potentially) moving and changing objects. Matthen proposes that vision represents objects just in case it represents features at different places at different times as belonging to the same object. On either theory, vision represents depicted objects in a space which is not the space of the picture surface, because not all features are bound with features located at the picture surface. In particular, vision does not bind every surface shape, colour, and texture with a depicted shape, colour, or texture. A striking example is Picasso’s 1906 portrait of Fernande Olivier (Figure 2.8), where the vertical lines to the viewer’s upper left are not bound to the sitter’s facial features (they are not part of her ‘do’). Indeed, they have no depictive function. What about surface features that do have a depictive function? The famous landmark in Seurat’s painting of 1889 (Figure 2.9) is depicted by dots of paint, but as long as the viewer is close enough to discriminate the dots, vision does not bind them to the tower’s structure. Just as the Picasso evokes an experience like seeing a woman through a scratched glass, the Seurat evokes an experience like seeing an object through a pattern glass window. So depicted features need not be bound to the surface features in virtue of which they are represented. In these examples, features of the picture surface are bound together and features of the depicted object are bound together; and some features of one are bound to features of the other—but not all. Assuming that vision represents distinct objects just in case it binds features at different places at a given time, it follows that vision represents depicted objects in a space distinct from that of the picture surface. It does not follow just from this that vision represents depicted space in a non-egocentric—allocentric or object-oriented—framework. Independent evidence is needed, and the picture of the two men shaking hands does seem to indicate that depicta are not represented egocentrically. Might another example indicate the contrary? The following example takes advantage of the full account of object perception as the representation of features bound as objects moving through space and time. After all, the phi phenomenon gives us the motion in motion pictures.

70 dominic mciver lopes

Figure 2.8. Pablo Picasso, Portrait of Fernande Olivier (1906)

picture this: image-based demonstratives

Figure 2.9. Georges Seurat, Eiffel Tower (1889)

71

72 dominic mciver lopes Table 2.1. Three Shots of a Rolling Ball

Static Shot Pan with the Ball Dolly with the Ball

Distance

Direction

Constant Changes Constant

Changes Constant Constant

Figure 2.10. Three shots of a rolling ball. Left: Static Shot; middle: Pan with the Ball; right: Dolly with the Ball

Suppose that you are directing Rolling Ball, a movie whose script reads simply: ‘A ball rolls across the grass’. Here are three shots for you to choose from. A very low-budget version is Static Shot. You mount the camera on a tripod, run the ﬁlm, and kick the ball across a playing ﬁeld. With a bigger budget you can choose Pan with the Ball. You hire an assistant to kick the ball and you pan the camera to keep the ball in the centre of its viewﬁnder. The high-budget version is Dolly with the Ball. You mount the camera on a dolly and track down the playing ﬁeld alongside the ball. All three movies depict the ball as moving relative to the playing ﬁeld, but they show it in ways that look quite different (see Table 2.1). In Static Shot, the ball travels across the screen, from one side to the other. In Pan with the Ball, the ball stays at centre screen but starts out far, comes in close, and rolls out far again. In Dolly with the Ball, the ball stays at a constant distance and direction in the ﬁeld of view. The difference between these three shots can be represented either egocentrically or allocentrically. Egocentric and allocentric representations of the same scene may carry the very same information, though they carry it in different ways, or as Matthen (2005: 313) puts it, in different ‘forms’. The three drawings in Figure 2.10 represent the trajectory of the ball relative to the camera in the three versions of Rolling Ball allocentrically, yet each carries the same information as its movie counterpart about the apparent distance and direction of the ball from the camera. Indeed, the same is true

picture this: image-based demonstratives

73

of Table 2.1! The question is whether the three movie versions represent the distance and direction of the ball from the camera in allocentric or egocentric form. The objection that D2 should not apply to seeing in pictures hinges on an answer to this question.

7. Varieties of Central Representation The question calls for a closer look at deixis, and especially egocentric representation, but also at the idea that vision sometimes represents objects egocentrically without relying on intermediary descriptive information. A compelling account of the three versions of Rolling Ball is that they differ in how they represent the ball as seen from a point of view. Static Shot looks as the scene would look if you saw it without moving. Pan with the Ball looks as the scene would look if you saw it by turning your head. Dolly with the Ball looks as the scene would look if you saw it while running alongside it. None of them look like the three drawings in Figure 2.10—not because the drawings are so sketchy, but rather because they do not represent the ball as seen from the same point of view as their cinematic counterparts. What point of view is that? The point of view of the camera is not the same as the point of view of the movie’s spectator. By extension, the representation of the ball as seen from the camera’s point of view is not the same as the representation of the ball as seen from the point of view of a spectator watching the movie in the cinema. To make this distinction a little more precise, it helps to see that both kinds of representation are species of the same genus. Call this genus ‘central representation’. Central representation is perspectival representation. It represents objects in a scene from a viewpoint, an origin, or a ‘here’, as Matthen puts it; and it represents them as lying in various directions and at various distances from that viewpoint. Normal vision is central in this way. It represents objects in your environment from a viewpoint (your location), as lying in various directions and at various distances from that viewpoint. Likewise, Rolling Ball represents the ball from a viewpoint (the camera’s location), as lying in various directions and at various distances from that viewpoint. The difference

74 dominic mciver lopes between these varieties of central representation has to do with their viewpoints. In ordinary vision, the viewpoint is the perceiver’s location. In Rolling Ball, it is the camera’s location—that viewpoint is the ‘here’ of the movie. Matthen (2005: 302) restricts egocentric representation to the former, which is ‘relative to a perceiver’s own body’. Call the latter species of central representation ‘cameracentric representation’. Egocentric and cameracentric representation are distinct varieties of central representation. The complement of central representation is allocentric representation, which represents objects in relation to each other, but not from a viewpoint. Neither egocentric nor cameracentric representations are allocentric. These distinctions might seem to show why the intuitive account of Rolling Ball lends no support to applying D2 to seeing in pictures. The intuitive account only shows that Rolling Ball represents the ball’s movement cameracentrically. So the movie does not represent the ball’s movement egocentrically—that is, it does not represent the ball’s movement in relation to the movie’s spectator. Since deictic representation is egocentric, seeing the ball in the movie is not deictic. One strategy in reply to this argument denies the assumption that deictic representation is egocentric and asserts instead that it is central. However, it is better to grant the assumption and focus instead on a reply that picks up on features of depiction. The claim that deixis requires only central representation does not vindicate the application of D2 to seeing-in unless seeing-in represents objects centrally without the beneﬁt of intermediary descriptive information. Since the viewpoint of the camera is not the same as the viewpoint of the spectator in the cinema, the worry is that we see objects in pictures centrally only with the aid of intermediary information about the location of depicta in relation to the camera. An idea borrowed from John Campbell (2002) helps to express the worry. Here is Rob’s Story: ‘Rob stood facing the summit, a path winding up the ridge to his right. He took a deep breath and then turned up the path.’ This story represents Rob’s environment centrally. If asked where the summit is at the end of the story, you can easily ﬁgure out that it is to Rob’s left. Maybe you use a bit of reasoning: the summit is ahead of Rob, if x is ahead of y and y turns right, then x is to the left of y, and Rob turns right, so the summit is to the left of Rob. Whether or

picture this: image-based demonstratives

75

not you actually reason like this, you do use concepts like ‘to the left’ that are the same as the concepts used in this reasoning. Campbell calls them relational. The concept ‘to the left’ is a concept of a relation with (at least) two places, one of them occupied by Rob (the relation could be represented by ‘Lsr’). Such concepts contrast with what Campbell calls monadic concepts, which we deploy in experience. Rob sees the summit is ahead, then he sees the summit is to the left. This concept ‘to the left’ does not represent a relation with a place occupied by Rob (it is represented by ‘Ls’). The worry is that the viewer’s experience of seeing the ball in Rolling Ball represents the ball’s location centrally and also relationally. It represents the ball from the viewpoint of the camera (hence centrally), in such terms as ‘to the left of the camera’. In deictic representation, the locations of objects relative to the viewpoint are represented monadically. Deixis is immediate in the sense that it is monadic. It does not allow a place for intermediary descriptive information about the locations of viewpoints to come in to represent the locations of objects relative to those viewpoints. We now know what to look for in a case for stretching D2 to seeing in pictures. Grant that deictic representation is egocentric. Also grant that the intuitive account is correct and Rolling Ball represents its protagonist’s movement cameracentrically. An additional assumption is needed for the inference that seeing the ball in the movie is not deictic—namely, that the movie does not represent the ball egocentrically if it represents the ball cameracentrically. Why not allow that there are hybrid species of central representation, that the movies represents the ball cameracentrically and egocentrically? Put another way, why not monadically as well as relationally? Viewpoints individuate central representations. Two central representations may represent the very same objects and yet they are distinct if they represent those objects from different viewpoints—that is, if they represent the objects as lying in different directions from their viewpoints. Thus, the viewpoint is implicit in the directions the object is represented as lying in. Call the spatial content of a central representation that determines its viewpoint its viewpoint-determining content. Central representations of the same objects from different viewpoints must differ in their viewpoint-determining content.

76 dominic mciver lopes What Rolling Ball represents cameracentrically differs from what your egocentric experience of the depicted scene represents in so far as the viewpoint-determining content of the movie does not match that of your visual experience. However, there is a rich and systematic overlap in the two viewpoint-determining contents—that is, between the represented directions from the camera to depicted objects and directions the objects appear to lie in from the picture viewer in normal viewing conditions. The ball is represented as to the cameracentric right of the tree just in case the ball looks to you to be to the right of the tree, and the ball is represented as moving to the upper left relative to the camera just in case it looks to you to be moving to the upper left. Relative distance also determines viewpoint, and there is an overlap in the representation of relative distances. The ball is represented as closer than the goal-post to the camera just in case it looks to you to be closer than the goal-post. The overlap is not complete. If you view the movie from the far right side of the theatre, then a cameracentric direction to the ball—say, two o’clock—will not match the egocentric direction to the ball—say, eleven o’clock. When you view Rolling Ball from straight ahead, there will often be some discrepancy in visual angles subtended by objects, so that, for example, an object that is represented as subtending a visual angle of twenty degrees cameracentrically may seem to subtend a visual angle of ﬁfteen degrees egocentrically. So the viewpoint-determining contents of the movie and the experience of seeing in the movie do not match precisely (except when the viewer occupies exactly the right spot). The claim is merely that the viewpoint-determining contents do overlap. Pictures exploit this systematic overlap in viewpoints to facilitate what Campbell calls ‘deep decentering’ (in a discussion of memory demonstratives). In deep decentring, a central relational representation is transformed into a central monadic representation. Suppose that Pan with the Ball represents the ball as moving to the right of the camera. Your experience of seeing the ball in the shot also represents it as moving to the right. This content could be relational (Rbc) or it could be monadic (Rb). We need not use relational concepts to represent the ball’s motion: we need not represent the ball as moving to the right from the viewpoint of the camera. We simply see it as moving to the right. Contrast this case with an avant-garde version of Rolling Ball —call it Sideways—where the camera is rolled ninety degrees, so that viewers see

picture this: image-based demonstratives

77

the ball moving from the top to the bottom of the screen. Sideways subverts the systematic overlap between cameracentric and egocentric viewpoints. Either an effort of imagination is needed to decentre into the cameracentric viewpoint or else we think of the ball as moving right relative to the camera, and not, as it appears, moving downwards relative to us. Putting the point another way, there’s a difference between our experiences of seeing the ball in Static Shot, Pan with the Ball, and Dolly with the Ball, on one hand, and Sideways, on the other hand. All four are cameracentric. The difference is that in only the ﬁrst three is there a rich and systematic overlap between egocentric and cameracentric directions, which facilitates decentring and hence a monadic representation of the location of objects in space. The experience of seeing in a picture represents objects egocentrically and monadically, so it represents them deictically.

8. Seeing and Seeing-In The claim that seeing objects in pictures is deictic does not imply that seeing-in is just the same as seeing O face to face. Some differences are worth noting. First, the fact that seeing-in exploits overlaps in central spatial content between cameracentric and egocentric representations does not imply that vision represents the picture and the depicted scene in the very same space. Vision does not represent Aunt Jane as in your living room when the photograph of her is in your living room. In part this is because the overlap between egocentric experience and cameracentric representation is not normally complete. Even when the overlap is complete (when you occupy the right spot in front of the picture), the spaces are not identical if the depicted space is not represented by motionguiding vision. Only in trompe l’œil does motion-guiding vision represent depicted space, but trompe l’œil spaces are experienced as part of ecological space. In addition, seeing O in P involves seeing P, and one sees O in P by seeing P. Seeing O in P is seeing indirectly. However, indirect seeing is not obviously incompatible with deixis. According to Paul Snowdon (1992: 56), for example, S directly perceives O just in case S stands, in

78 dominic mciver lopes virtue of S’s perceptual experience, in such a relation to O that S can make the true, non-dependent demonstrative judgement ‘That is O’. Yet Snowdon describes your contact with Aunt Jane as ‘fairly indirect’. It follows that your exclamation ‘That’s Aunt Jane’ involves a dependent rather than a non-dependent demonstrative reference (Snowdon 1992: 59). This does not mean that as you look at the picture you must think ‘That is a photo of Aunt Jane’. On the contrary, you might lose all awareness of the photograph. But unless you would acknowledge under questioning the truth of ‘That is a photo of Aunt Jane’, then you are mistaking the photo for your aunt. Vision grounds a dependent demonstrative reference to O just in case you can count as demonstratively referring to O only if you are prepared to acknowledge under questioning that O bears a certain relation to P. By this deﬁnition, seeing-in grounds dependent demonstrative reference (except in trompe l’œil). Whereas direct perception grounds non-dependent demonstrative reference, seeing-in is indirect and grounds dependent demonstrative reference. Defending Snowdon’s analysis of direct perception is a task for another occasion, but it does illustrate one way to capture an obvious difference between seeing-in and seeing face to face. And it makes no concession to views like those of Evans and Berm´udez. They hold that your exclamation does not involve a pure demonstrative reference to Aunt Jane—descriptive elements are mixed in. Snowdon’s dependent demonstrative judgements are pure: your being prepared to acknowledge under questioning that you are looking at a photo of Aunt Jane is not part of the meaning of your exclamation. Before wrapping up, it should be noted that nothing has been said against the arguments of Evans and Berm´udez. They must be assessed against what we know about how vision works. Matthen’s Seeing, Doing, and Knowing begins with vision science. If we begin with Evans and Berm´udez, we will end up reading Matthen; and if we begin with Matthen, then a good next step might be to look at the arguments of Evans and Berm´udez. Assuming that deictic visual representation grounds at least some cases of demonstrative reference, a case can be made that seeing O in P grounds demonstrative reference to O. Given that you see Aunt Jane in the photograph, your so seeing her grounds the demonstrative reference in your exclamation. Therefore, what you say is true.

picture this: image-based demonstratives

79

References Aglioti, Salvatore, Joseph F. X. DeSousa, and Melvyn A. Goodale (1995), ‘SizeContrast Illusions Deceive the Eye But Not the Hand’, Current Biology, 5/5: 679–85. Bermu´ dez, Jos´e Luis (2000), ‘Naturalized Sense Data’, Philosophy and Phenomenological Research, 61/2: 353–74. Blaser, Erik, Zenon Pylyshyn, and Alex Holcombe (2000), ‘Tracking an Object Through Feature Space’, Nature, 408/6809: 196–9. Budd, Malcolm (1993), ‘How Pictures Look’, in Dudley Knowles and John Skorupski (eds), Virtue and Taste (Oxford: Blackwell). Campbell, John (2002), Reference and Consciousness (Oxford: Oxford University Press). Clark, Austen (2000), A Theory of Sentience (Oxford: Oxford University Press). Davies, Martin (1997), ‘Externalism and Experience’, in Ned Block, Owen Flanagan, and Güven Güzeldere (eds), The Nature of Consciousness (Cambridge, Mass.: MIT Press). Ellis, Robert R., J. Randall Flanagan, and Susan J. Lederman (1999), ‘The Inﬂuence of Visual Illusions on Grasp Position’, Experimental Brain Research, 125/22: 109–14. Evans, Gareth (1982), The Varieties of Reference (Oxford: Oxford University Press). Gombrich, E. H. (1961), Art and Illusion (Princeton: Princeton University Press). Goodman, Nelson (1976), Languages of Art, 2nd edn (Indianapolis: Hackett). Hopkins, Robert (1998), Picture, Image, and Experience (Cambridge: Cambridge University Press). Lopes, Dominic McIver (1996), Understanding Pictures (Oxford: Oxford University Press). (2000), ‘From Languages of Art to Art in Mind’, Journal of Aesthetics and Art Criticism, 58/3: 227–31. (2005), Sight and Sensibility: Evaluating Pictures (Oxford: Oxford University Press). Lowe, E. J. (1993), ‘Self, Reference, and Self-Reference’, Philosophy, 68/263: 15–33. McGinn, Colin (1981), ‘The Mechanism of Reference’, Synthese, 49/2: 156–86. Matthen, Mohan (2005), Seeing, Doing, and Knowing (Oxford: Oxford University Press). Milner, David A., and Melvyn A. Goodale (1995), The Visual Brain in Action (Oxford: Oxford University Press).

80 dominic mciver lopes Quine, W. V. O. (1968), ‘Ontological Relativity’, Journal of Philosophy, 65/7: 185–212. Reimer, Marga (1991), ‘Demonstratives, Demonstrations, Demonstrata’, Philosophical Studies, 63/2: 187–202. Schier, Flint (1986), Deeper into Pictures (Cambridge: Cambridge University Press). Siegel, Susanna (2002), ‘The Role of Perception in Demonstrative Reference’, Philosophers’ Imprint, 2/1: 1–21. Snowdon, Paul (1992), ‘How to Interpret ‘‘Direct Perception’’ ’, in Tim Crane (ed.), The Contents of Experience (Cambridge: Cambridge University Press). Strawson, P. F. (1959), Individuals (London: Methuen). Voltolini, Alberto (2009), ‘How Demonstrative Pictorial Reference Grounds Contextualism’, Paciﬁc Philosophical Quarterly, 90/4: 402–18. Walton, Kendall (1990), Mimesis as Make-Believe (Cambridge, Mass.: Harvard University Press). Weiskrantz, Lawrence (1997), Consciousness Lost and Found: A Neuropsychological Exploration (Oxford: Oxford University Press).

3 The Epistemic Value of Photographs CATHARINE ABELL

Photographs play epistemic roles that most other pictures cannot. Firstly, they can provide compelling evidence that the things they depict existed at the time they were taken. We might doubt that the unfamiliar animals drawn by zoologists of the past ever existed, but be convinced by photographs of Tasmanian tigers that this now extinct marsupial once populated its namesake state. Likewise, a photograph of a politician lunching with a known criminal may convince us of her corruption, when a drawing of the same scene would not. Secondly, photographs play a unique investigative role, enabling us to identify features of their objects that are easily overlooked. Police detectives consult photographs, rather than paintings, of crime scenes, looking for clues that will help them solve cases. Similarly, consumers pore over photographs of potential purchases on eBay, looking for ﬂaws that a hand-drawn picture would not reveal, whether unintentionally or through deliberate omission. This does not mean that all photographs are alike in respect of epistemic value. While many photographs are suited to the epistemic roles described above, others—for example, those that are badly focused—are not. Token photographs may lack the epistemic signiﬁcance that photographs generally possess, and token non-photographic pictures might have such epistemic

Thanks to Katerina Bantinaki and to the audience at Mimesis, Metaphysics, and Make-Believe, a conference in honour of Kendall Walton, University of Leeds, 2007, particularly my commentator, Aaron Meskin, for helpful feedback on earlier versions of this chapter.

82 catharine abell signiﬁcance, although non-photographic pictures generally do not. For example, a painting by Vermeer may better enable us to identify readily overlooked features of the scene it depicts than would a photograph of that scene. Photographs have a distinctive epistemic value because the epistemic properties they generally possess differ from those that non-photographic pictures generally possess. The contrast between the epistemic properties of photographic and non-photographic pictures obtains between photographs, considered in general, and non-photographic pictures, considered in general. Nor does it follow that photographs generally have greater epistemic value than non-photographic pictures generally do. Their epistemic signiﬁcance simply differs. Non-photographic pictures can play epistemic roles that photographs generally cannot. For example, botanical drawings may make salient features of plants that would not be prominent in photographs of them, but information about which is of particular epistemic value to us. The scope of non-photographic pictures also lends them an epistemic signiﬁcance that photographs do not have. Whereas photographs can depict only particulars, non-photographic pictures can depict, in addition, objects of a certain type that are not existent particulars of the type at issue. Unlike photographs, they can thus be used to depict the prototypical features of a type in isolation from the accidental features of particular instances of that type. What nonphotographic pictures can teach us may be just as valuable as what photographs can teach us, but each teaches in virtue of different epistemic capacities. My aim, in this chapter, is to explain why photographs are generally able to play the epistemic roles described above, and why non-photographic pictures generally are not. This requires accounts, ﬁrst, of the difference between the two sorts of picture and, second, of the form of epistemic value (let us call it EV) that a picture must have in order to play those roles. These I offer, respectively, in Sections 1 and 2. In Section 3, I examine three proposed explanations of why photographs generally have EV and non-photographic pictures generally lack it, and argue that none is adequate. In Section 4, I explain why nonphotographic pictures typically have little EV. In Section 5, I explain why the EV of photographs typically exceeds that of non-photographic pictures.

the epistemic value of photographs

83

1. Photographic versus Non-Photographic Pictures For the purposes of this chapter, we can understand a picture as something that depicts. Depiction is a distinctive form of representation, characteristic of ﬁgurative paintings, drawings, and photographs. Some philosophers claim that pictures depict their objects in virtue of looking like, or visibly resembling, them (Hopkins 1998). Others deny that depiction can be explained in terms of resemblance (Kulvicki 2006; Lopes 1996). Nevertheless, none denies that pictures generally seem, to those viewers who successfully interpret them, to resemble the things they depict (Kulvicki 2006: 81; Lopes 1996: 4). Although it is beyond the scope of this chapter to provide an account of depiction,¹ we can therefore understand a picture’s depictive content as that content a grasp of which is generally accompanied by the visual perception of apparent resemblances. In specifying a picture’s depictive content, we must be careful to distinguish between its internal and its external object. Its internal object is what it depicts, whereas its external object, if it has one, is the independently existing object whose properties are causally responsible for the picture’s surface being marked in the way it is. A black and white photograph of you has a person of indeterminate colour as its internal object, but a person of perfectly determinate colour as its external object. What distinguishes photographic and non-photographic pictures is not the form of representation each employs, but the way in which each is produced. Photographs are produced by largely mechanical means, whereas non-photographic pictures are not. Moreover, photographic mechanisms operate such that all photographs have external objects, and the design features of any photograph (those features in virtue of which it depicts its object) depend causally on the features of its external object, such that the former depend counterfactually on the latter. When I photograph a mountain, for example, the mountain causes my photograph to have the design features it does, such that, had the mountain been different, my photograph would likewise have differed. By contrast, not all nonphotographic pictures have external objects (there need be no existent particular which they depict) and, while the features of some of those that ¹ I provide such an account in Abell (2009).

84 catharine abell do are counterfactually dependent on features of their external objects, not all are. I might seek to draw a mountain accurately such that the features of the drawing I produce depend counterfactually on those of the mountain. However, faced with the same mountain, I might instead employ a schema or formula for the depiction of a mountain, in which case the features of my drawing need not depend counterfactually on those of the mountain. One might likewise sever the counterfactual dependence of a photograph’s features on features of its external object—by over-painting it, for example. However, this would render the photographic mechanism redundant. Consequently, the resultant picture would not be genuinely photographic. Photographers’ intentions play a signiﬁcant role in determining what their photographs depict. Their intentions determine where they point their cameras, what kind of cameras they choose, and how they set their variable parameters. They can also affect depictive content after the photographic mechanism has been employed. By using the darkroom technique of dodging and burning, for example, photographers can realize their intention to represent certain aspects of the photographed scene as darker, and others as lighter, than they would otherwise have been depicted. Likewise, by cropping, they can select which aspects of the photographed scene are depicted.² However, photographers’ intentions can sever the counterfactual dependence of features of the photograph on features of its external object only by overriding the effect of the photographic mechanism, and producing a picture that is not purely photographic.

2. An Account of EV What features enable a picture both to provide compelling evidence that an object existed and had certain properties at the time the picture was produced, and to reveal features of an object that are easily overlooked? We need a clear grasp of EV, the form of epistemic value that enables a picture to play these roles, before attempting to explain why photographs and non-photographic pictures play different epistemic roles. ² This understanding of what it is to be a photograph is thus broader than Scruton’s notion of the ideal photograph (Scruton 1983).

the epistemic value of photographs

85

A picture provides compelling evidence of an object’s existence, and of its possessing certain features, only if it is produced by a reliable process. That is, the process linking picture and object must be such that we are likely to form true beliefs about objects on the basis of pictures produced by that process. This requires two things. Firstly, the pictures produced by this process must be likely to carry information about their external objects. Secondly, they must be likely to do so in depictive form. For the purposes of this chapter, we can understand information carrying, with Jonathan Cohen and Aaron Meskin, as a counterfactual relationship between independent variables, such that x’s being F carries information about y’s being G if and only if the counterfactual conditional (if y were not G, then x would not have been F) is true (Cohen and Meskin 2006: 3). A picture’s having a certain feature carries the information that an object has a certain property if and only if, had the object not had that property, the picture would have lacked that feature. Things may carry information in virtue of a variety of different features. A tree trunk may carry information about the tree’s age in virtue of the number of rings in its bark. A history book may carry information about a war in virtue of the words written on its pages. A picture-production process is reliable only if the pictures it produces are likely to carry information about their external objects in depictive form. A picture carries the information that an object has a certain property in depictive form if it both carries that information and depicts the object as possessing that property. Pictures may carry information about objects without doing so in depictive form. For example, a photographic negative may carry the information that an object had a certain colour because, had the object’s colour differed, the colours of the negative would also have differed. Nevertheless, the negative does not depict the object as having the colour it had: adequate knowledge of the photographic process involved may enable one to work out, from the negative’s colour, what colour its external object was, but doing so does not make the negative appear to resemble that object in respect of colour. It is difﬁcult to work out, from a colour negative, the colour of the object about which it carries information. By contrast, it is easy to discern such information from a picture which carries it in depictive form. The information pictures carry in depictive form is easier to grasp than that which they carry in other forms. Consequently, we are more likely to form

86 catharine abell true beliefs about objects on the basis of the former information than the latter. It is an interesting question why this is so. One possible explanation is that the interpretation of depictive content engages our visual recognitional abilities (Lopes 1996; Schier 1986), and thus inherits the reliability of our perceptual processes. Nevertheless, detailed discussion of this issue must await another occasion. The upshot of these two conditions is that a process will be more reliable the more likely it is to produce pictures whose depictive content is accurate: whose internal objects have only features possessed by their external objects. Consequently, it is tempting to analyse the reliability of a picture-production process as the probability, given any picture produced by that process, that its object was as it is depicted as being at the time it was produced. However, doubts that there is any non-epistemic account of probability able to accommodate probabilities of inverse conditional properties make this approach seem unpromising (Cohen and Meskin 2006: 4). Fortunately, an alternative, modal analysis is possible. If a picture depictively encodes information about its object, then, according to David Lewis’s semantics for counterfactuals (Lewis 1973), its depictive content will differ in the closest possible world in which its object differs. The likelihood that a process will produce pictures that depictively encode information about their external objects can be construed in terms of the distance between worlds in which the depictive content of pictures produced by that process is accurate, and worlds in which it is not.³ Accordingly: A given picture-production process⁴ is more reliable the greater the distance, for any non-actual picture produced by that process, between the nearest possible world to the actual world in which the picture exists and its external object was as it is depicted as being at the time the picture was produced and the closest possible world to that world in which the picture exists and its external object was not as it is depicted as being at the time the picture was produced. This construal of reliability assumes that the actual world is closer to the former than to the latter world.⁵ ³ In what follows, I construe the likelihood of this outcome in the way speciﬁed, not as synonymous for its probability. ⁴ The level of generality at which any picture-production process is to be individuated depends on how it is conceived by those who designed it (where it is mechanized), or those who use it (where it is not). ⁵ As Cohen and Meskin point out, Lewis construes distance relations between worlds as interestdependent, which would make counterfactuals depend tacitly on the doxastic (Cohen and Meskin

the epistemic value of photographs

87

We cannot understand EV solely in terms of the reliability of pictureproduction processes. Even if it is the result of a reliable process, a picture will not reveal features of an object that are easily overlooked unless it carries a reasonable amount of information about that object. To play such a role, a picture must also be rich, where richness is a measure of the amount of depictively encoded information a picture carries. The more such information it carries, the richer it is. How much information a picture carries is a measure of the number of features of its external object about which it carries information in depictive form. How ﬁnely the features of its object can be individuated is determined by the relations of counterfactual dependence that obtain between picture and object. If a difference in the shape of a single hair on a woman’s head would result in a corresponding difference in the shape a portrait of that woman depicts the hair as having, the shape of the hair counts as a feature of the woman about which the picture depictively encodes information. However, if she would have to have lacked hair altogether in order for the portrait to depict her hair differently, her having hair counts as a feature about which the picture depictively encodes information, but neither the shape nor the existence of the individual hair counts as such. A picture may be richer—and the process by which it is made more reliable—regarding some types of features than others. For example, black and white photographic processes may be very likely to produce pictures that depictively encode information about the shape properties of their external objects, and may produce pictures that carry a lot of information about such properties, but are generally both unlikely to produce pictures that depictively encode information about the colour properties of their external objects, and incapable of producing pictures that carry much information about colour properties. We can understand a process’s reliability relative to features of a certain type as follows: A given picture-production process is more reliable, relative to features of type F, the greater the distance, for any non-actual picture produced by that process which depicts F features, between the nearest possible 2006: 10). However, like Cohen and Meskin, I take the distance relations between worlds to be mind-independent.

88 catharine abell world to the actual world in which the picture exists and its external object had the F features it is depicted as having at the time the picture was produced, and the closest possible world to that world in which the picture exists and its external object did not have the F features it is depicted as having at the time the picture was produced. Again, it is assumed that the actual world is closer to the former than to the latter world. A picture is richer, relative to features of type F, the more information about F features it carries. We thus need to distinguish between a picture’s EV simpliciter, and its EV relative to features of a certain type. A picture’s EV simpliciter is a measure of its richness and of the reliability of the process by which it was produced, while its EV relative to features of a certain type is a measure of its richness relative to features of that type and the reliability of its production process relative to features of that type. This account construes EV as independent of individual viewers’ particular interests and attributions. A picture could have a high EV simpliciter despite the information it conveys being less important to some viewers than the information conveyed by a picture with a lower EV simpliciter, and thus of less overall epistemic value to them. As I noted earlier, EV is not the only form of epistemic value a picture may have. Nevertheless, given that information about features of a certain type is important to viewers, one picture may have greater EV than another relative to features of the type at issue. I have not speciﬁed how richness and reliability combine to determine EV. Rather, I construe them as independent dimensions, each of which may contribute to EV in isolation from the other. This reﬂects the fact that different purposes impose differing demands for reliability and for richness. Someone consulting a picture to determine whether a galloping horse ever has all its hooves in the air simultaneously seeks reliability above richness. The detective who consults pictures of a crime scene looking for information that will enable her to solve the crime seeks both richness and reliability. A fashion stylist who consults pictures of the Paris catwalks looking for inspiration seeks richness over reliability. Individual viewers may make erroneous attributions of EV. A viewer who mistakenly takes a photorealist painting to have been produced by a photographic process will attribute to it greater EV than it possesses.

the epistemic value of photographs

89

Moreover, even when their attributions of EV are accurate, viewers will derive inappropriate beliefs from pictures if they misidentify their external objects on the basis of their internal objects. Suppose the photographs apparently showing the ﬁrst moon landing were in fact taken in a Hollywood studio. While our beliefs about photographic processes may justify an attribution of high EV to such photographs, it would be wrong to construe them as providing good evidence that a moon landing occurred, since their external object is not a moon landing. Let us now consider three different explanations of the epistemic differences between photographic and non-photographic pictures. Their adequacy will depend on whether they explain it as a difference in EV.

3. Existing Explanations 3.1. Photographs as Transparent According to Kendall Walton, photographs are aids to vision, not mere representations (Walton 1984: 263). Just as we see a scene when we examine it through a telescope or regard its reﬂection in a mirror, so too, he claims, we literally see a scene when we look at a photograph of it. Photographs are transparent: we see through them to their objects. Like mirrors, they are transparent although their surfaces are visible when we see through them, and although they afford only indirect visual access to things. Unlike mirrors, photographs enable us to see spatially and temporally distant scenes. Seeing through photographs is a unique form of perception that enables us to see things that could not otherwise be seen. Contrarily, Walton argues, most non-photographic pictures are opaque: when we look at them, we see depictive representations, not the scenes they depict. He identiﬁes three conditions that, he claims, are individually necessary for transparency. Firstly, to see one object through another, the object through which it is seen must be counterfactually dependent on the object seen through it. Secondly, this relation of counterfactual dependence must be belief-independent. Walton asks us to suppose that an explorer goes into the jungle and comes back with either photographs or sketches of a dinosaur. Both the photographs and the sketches may convince us that there is a dinosaur in the jungle. However, he argues, ‘The important difference is that, in the case of the sketches, we rely on the picture maker’s

90 catharine abell belief that there is a dinosaur in a way in which we don’t in the case of the photographs’ (Walton 1984: 263). If the explorer did not believe there were a dinosaur in the jungle, Walton thinks, his sketches wouldn’t depict a dinosaur, whereas his photographs would depict a dinosaur no matter what he believed. Finally, Walton claims, transparency requires the preservation of the real relations of similarity and dissimilarity between objects. Pears are similar to apples and dissimilar to bears. Visual experiences of pears reﬂect this fact: perceivers may mistake pears for apples, but not for bears. Both photographic and non-photographic pictures, and our visual experiences of them, likewise preserve the similarity relations among objects: we may mistake a picture of a pear for a picture of an apple, but would not mistake it for a picture of a bear. However, neither written descriptions nor our experiences of them preserve real similarity relations between the objects they represent. ‘Pear’ is more like ‘bear’ than ‘apple’, and this is reﬂected in the mistakes we are likely to make about a description’s content. Even though a mechanically generated description might exhibit belief-independent counterfactual dependence on its object, therefore, it is not transparent. Walton’s account suggests an explanation of photographs as having EV because photographic processes have the reliability of perceptual processes and the richness of perceptual experiences. However, while many visual experiences may be the rich products of reliable processes, not all are. The visual experiences of the myopic and the colour-blind are not. The transparency of photographs therefore does not ensure that they have EV, since photographic processes may be among those visual processes that are unreliable. To explain the EV of photographs, we need an explanation of why photographic processes typically count among reliable visual processes, and why photographs are akin to rich visual experiences.⁶ Walton suggests that photographic processes are reliable in virtue of ‘the fact that our photographic equipment and procedures happen to be standardised in certain respects’ (Walton 1984: 273). However, there are many respects in which photographic equipment and procedures are not standardized, but differ considerably. For example, while most ⁶ A general evaluation of Walton’s claim that photographs are transparent is beyond the scope of the present chapter, which is concerned solely with its epistemic implications. For criticisms of this claim, see Cohen and Meskin (2004), Currie (1991), and Dretske (1984).

the epistemic value of photographs

91

cameras utilize visible light, infra-red cameras do not, and both camera types may have either analog or digital mechanisms. Moreover, the mere standardization of such equipment and procedures would achieve only uniformity, not reliability. More needs to be said about the respects of standardization at issue and how they ensure reliability if this explanation is to succeed. One could relinquish the claim that photographs are transparent and simply insist that their epistemic value is explained by their meeting Walton’s three conditions. On his account, photographs differ from those non-photographic pictures that depend counterfactually on their objects only in their belief-independence. However, many non-photographic pictures are also belief-independent, but lack the EV of photographs. Picture makers may, and often do, depict objects by employing schemata, or formulae, for the depiction of objects of particular types. In such cases, which schema they employ depends on what they believe about the object they seek to depict. However, many picture makers instead depict objects by attending carefully to their visible features and allowing those features to guide the way in which they mark the picture surface. As Dominic Lopes notes, ‘In drawing, the eye and the hand work together, perhaps bypassing the mind, or rather that portion of the mind that deals in concepts and beliefs’ (Lopes 1996: 186). There is empirical evidence to support this claim, which suggests that the visual information that guides our motor actions may differ from that carried by conscious visual experience. For example, subjects presented with disks arranged to produce the Tichener or Ebbinghaus illusion have conscious visual experiences that misrepresent the relative sizes of the disks. They experience disks of the same size as differing in size, and disks of different sizes as being of the same size. Nevertheless, when subjects reach for those disks, the aperture produced between ﬁnger and thumb is perfectly suited to the actual sizes of the disks, rather than the sizes they are represented as having in subjects’ conscious visual experiences (Milner and Goodale 1995: 168).⁷ Since it is the information carried by conscious visual experience, rather than that used for visually based motor control, that is made available for conceptual mobilization, this evidence suggests that the ⁷ For philosophical discussion of these results, see Clark (2001) and Jacob and Jeannerod (2003). In their chapters in this volume, Lopes and Nanay discuss further the implications this experiment has for the philosophical understanding of depiction (see Lopes, Ch. 2, Sect. 3; Nanay, Ch. 7, Sect. 7).

92 catharine abell information that guides picture makers’ actions could indeed bypass their cognitive centres. Walton acknowledges that some non-photographic pictures, such as tracings and ‘doodles done automatically, while the doodler’s mind is on other things’, may be belief-independent (Walton 1984: 267). They too, he claims, are probably transparent. While some such pictures may have EV, however, not all do. An automatic doodle of a crime scene would lack the EV of a photograph of that scene. The belief-independence of photographs does not sufﬁce to explain EV.⁸ 3.2. Photographs as Necessarily Accurate Robert Hopkins has argued that photographs are necessarily accurate (Hopkins 1998).⁹ For something to depict an object, he argues, it must not only have the right appearance, but also the appropriate history of production. This history comprises a standard of correctness, which determines whether something that appears to depict an object actually does so. Hopkins argues that the standard of correctness for non-photographic pictures is intentional. Something is a non-photographic picture of an object only if its maker intended it to represent that object. Even if it is visually indistinguishable from a picture of a lion, the blotch made by accidentally spilling a bottle of ink is not a picture, since it was not produced with the requisite intention. However, he claims, the standard of correctness that governs photographs cannot depend on photographers’ intentions, since photographs that are taken accidentally depict scenes that photographers did not intend them to depict. Instead, he argues, this standard depends on the intentions of those who design the mechanisms with which photographs are taken. Since camera designers cannot always control how these mechanisms are used, their intentions are, of necessity, quite general. Hopkins argues that ‘the relevant intention is better put thus. The camera’s designer intended that we see in the photographs whatever is causally responsible for those surfaces being marked as they are’ (Hopkins 1998: 72). In his view, therefore, the ⁸ One might think the real difference between photographic and non-photographic pictures lies in the latter’s dependence on, and the former’s independence from, their maker’s intentional states in general, not their beliefs in particular. However, as I will argue in Sect. 5, some photographs depend on developers’ visual states. ⁹ See also Scruton (1983) and Currie (1999: 288).

the epistemic value of photographs

93

standard of correctness for photographs comprises both an intentional and a causal element. The ink blotch is not a photograph, since it is not the product of a mechanism designed with the relevant intentions, and no lion is causally responsible for its appearance. Since it follows from his characterization of camera designers’ intentions that they intend the photographs produced with the mechanisms they design to depict their external objects accurately, Hopkins concludes that photographs necessarily depict their objects accurately (Hopkins 1998: 72–3). However, the nature of camera designers’ intentions is an empirical matter, not one that can be determined a priori. Like makers of nonphotographic pictures, camera designers may intend that the photographs produced with the mechanisms they design misrepresent their external objects. For example, the designer of a camera whose lens systematically distorts the shape of the things seen through it is best understood as intending that the photographs produced with his camera depict their external objects as having shapes that differ in various systematic ways from the shapes they actually possess. Moreover, even if camera designers’ intentions were as Hopkins claims, this would not ensure the accuracy of the photographs produced using their cameras, since development processes are beyond camera designers’ control. To the extent that techniques such as dodging and burning are available, which enable the depictive content of photographs to be altered during the development process, photographers can produce photographs that misrepresent their objects, irrespective of camera designers’ intentions. Photographs are no less able to misrepresent their objects than other pictures. There is a sense in which photographs are necessarily accurate, but the form of representation at issue is causal (indexical, in Peirce’s terminology), rather than depictive. Just as the number of rings in the trunk of a tree may accurately represent the tree’s age, in virtue of the causal relation between its age and the rings, so too photographs accurately represent the objects that caused them. When photographs depict their objects accurately, their depictive and causal contents coincide. However, the two kinds of content can come apart. Even if photographs were necessarily accurate qua pictures, Hopkins’s claims do not yield an adequate explanation of their epistemic signiﬁcance. His claims would explain their reliability: there would be no possible world in which a photograph exists but its object was not as the photograph

94 catharine abell depicts it at the time the photograph was taken. However, they would not explain their richness. Badly focused photographs have relatively little EV, not because they are produced by unreliable processes, but because they are not rich. Despite their accuracy, a detective would clearly prefer to consult a properly focused photograph of a crime scene. 3.3. Photographs as Spatially Agnostic Informants Cohen and Meskin argue that photographs are epistemically valuable because they are spatially agnostic informants: they carry information about their objects’ visually accessible properties other than their egocentric spatial locations, without carrying information about their egocentric spatial locations (that is, about their objects’ spatial locations relative to oneself) (Cohen and Meskin 2004: 204). Contrarily, visual experiences can carry information about the former properties of their objects only if they also carry information about the latter. On Cohen and Meskin’s view, photographs are epistemically valuable because they can provide information about their objects’ visually accessible properties, in conditions in which information about those objects’ egocentric spatial location is unavailable (Cohen and Meskin 2004: 205). Many non-photographic pictures are also spatially agnostic informants. Cohen and Meskin construe the epistemic differences between photographs and such pictures as merely apparent, not real. Viewers take photographs, but not most other pictures, to be spatially agnostic informants, they argue, because the type photograph is salient to viewers and because they believe tokens of that type typically to be spatially agnostic informants, whereas this is not true of most non-photographic picture types. A picture type is salient if viewers typically categorize tokens of the type in question as belonging to that type (Cohen and Meskin 2004: 205). The type photograph is salient because viewers typically categorize token photographs as belonging to that type. Contrarily, the type accurate landscape drawing is not salient, as viewers will typically categorize tokens of that type as belonging, not to it, but to the more general type landscape drawing. Whereas viewers believe that tokens of the type photograph typically carry information about visually accessible properties, they do not believe the same of tokens of the type landscape drawing (Cohen and Meskin 2004: 205). Viewers do not take tokens of most non-photographic picture types to be spatially agnostic informants because the types at issue are either not salient to them, or

the epistemic value of photographs

95

they don’t believe that tokens of those types are typically spatially agnostic informants. There are two reasons why Cohen and Meskin’s account does not sufﬁce to explain our attributions of epistemic value. Firstly, like the type photograph, the type photographic negative is salient. Moreover, while some viewers may not believe that photographic negatives typically carry information about visually accessible properties, others—those with insight into the nature of information—believe that they do. Nevertheless, such viewers will take the photographs produced from such negatives to differ from the negatives in respect of EV. A lawyer would not use negatives, in lieu of the photographs produced from them, to provide a jury with evidence about an object’s colour, even if she knew the members of the jury believed the negatives to carry information about the object’s colour, since it would be much more difﬁcult for them to derive such information from the negatives than from the photographs. Cohen and Meskin’s account fails to accommodate the relevance of the form in which information is carried. Secondly, we may attribute different EV to pictures of two different, salient types, both of which we know to be spatially agnostic informants. For example, we attribute different epistemic value to court drawings than to photographs. If it were permissible to take photographs in the courtroom, most newspapers and television news programmes would show photographs instead of court drawings, precisely because of their distinctive epistemic properties. Like the type photograph, the type court drawing is salient to viewers. The distinctive content of such drawings, and the contexts in which they are usually presented—on television news programmes and in newspaper reports—mean that viewers typically categorize such pictures as belonging to the type court drawing rather than to some more general type. Moreover, viewers also usually believe that court drawings typically carry information about their objects’ visually accessible properties. We attribute different epistemic capacities to court drawings and photographs because we take the latter to be both richer and more reliable sources of information about their objects. Because Cohen and Meskin’s account does not mention the amount of information pictures carry about their objects’ visually accessible properties, it does not capture the epistemic signiﬁcance of richness. Neither does it capture the epistemic signiﬁcance of reliability: it appeals to our beliefs about the information different

96 catharine abell types of picture typically carry, but not our beliefs about the likelihood that pictures of the relevant types will carry information. It therefore fails to capture the epistemic differences between photographs and court drawings. None of the accounts considered here succeeds in explaining why photographic and non-photographic pictures differ—or appear to differ—with respect to EV. In the remaining sections, I propose an alternative explanation of this difference, which I construe as an actual difference in the EV pictures of the two types actually possess.

4. The EV of Non-Photographic Pictures Non-photographic picture-making processes typically lack the reliability required for their pictures to play the epistemic roles to which photographs are suited, because their makers’ intentions affect the relations they bear to their objects. Picture makers’ intentions are variable. Sometimes they intend to depict objects accurately, while, at others, they intend to misrepresent them. The fact that someone who uses any non-photographic process might have done so with the intention of misrepresenting an object severely limits the reliability of that process. Generally speaking, those nonphotographic pictures that are the product of intentions to misrepresent objects will misrepresent their external objects. However, it is not true that non-photographic pictures that are products of intentions accurately to represent objects will generally depict them accurately. Picture makers’ ability to realize their intentions is constrained both by the limitations to their technical abilities and by the reliability of their visual processes. Even in cases in which they intend to depict something accurately, the former constraint in particular signiﬁcantly diminishes the likelihood that they will succeed in doing so. Picture makers’ intentions affect the reliability of some non-photographic processes more than others. The processes involved in court drawing are more reliable than many other non-photographic processes, for example. A variety of factors—professional norms, the threat of unemployment—make the likelihood that such processes are exploited for the purpose of misrepresenting objects much lower than the likelihood that many other non-photographic processes were used for such a purpose. However, the

the epistemic value of photographs

97

fact that their makers might have intended that they misrepresent their objects affects their reliability nonetheless. The possible world in which a picture produced by a non-photographic process depicts its external object accurately is therefore never very much closer to the actual world than one in which it misrepresents that object. Furthermore, even when non-photographic pictures do carry information about their external objects, the serial process by which they are made limits their richness. Picture makers cannot mark a picture surface so as to achieve simultaneous depictive encoding of information about the various different features of an object. Instead, they must encode information little by little. Consequently, the more information a picture maker seeks to encode depictively, the longer the picture-making process takes. This places practical limits on how much information picture makers can encode in depictive form. Moreover, many of the scenes picture makers seek to depict change during the time taken to produce rich non-photographic pictures of them. This further limits how much depictively encoded information most non-photographic pictures can carry. Some picture makers (Vermeer, for example) possess the tenacity required to pursue richness despite the time required to achieve it. Nevertheless, it is no accident that many of those non-photographic pictures that rival photographs in richness are still lifes or are copied from photographs.

5. The EV of Photographs Most photographic processes are parallel, rather than serial: they enable the simultaneous depictive encoding of information about the various different features of objects. Most cameras simultaneously record light reﬂected from the various different parts of the scene photographed. While some photographic processes are serial—such as those involved in astronomical photography—this does not affect the richness of the resultant photographs, in so far as the things they are used to photograph remain unchanged during the time taken to photograph them. Moreover, the automation of such serial photographic processes means that there are fewer practical limitations on their employment than on that of serial non-photographic picture-production processes, which require their makers’ participation throughout.

98 catharine abell Photographs are therefore generally immune to the factors that limit the richness of non-photographic pictures. However, this alone does not explain why they are typically richer than the latter. Poorly focused photographs may be the result of parallel processes, but are less rich than many non-photographic pictures. Photographic richness is a technological achievement. The richness of the photographs that could be produced using early cameras was no greater than that of many paintings. Later cameras could be used to produce much richer photographs, and the advent of colour ﬁlm further enhanced their capacity for richness, by enabling photographs to encode information depictively about more kinds of features of their objects. Contemporary colour ﬁlm has enhanced the capacity for photographic richness further still, by enabling photographs to encode more information depictively about the colours of their objects. The fact that a photograph was produced using photographic technology that enables a high degree of richness is no guarantee that it is rich. There are certain conditions that need to be met in order for such technology to yield rich photographs. For example, many cameras will produce them only if there is a certain level of ambient light in the photographed scene, or if the ﬁlm is exposed for a suitable length of time. Nevertheless, photographs are typically considerably richer than non-photographic pictures because most photographs are produced using photographic technology that ensures a high degree of richness under the conditions in which they are taken. The reason why both photographic technology has developed in the direction of increasing richness and the conditions required for richness are usually met is precisely that, in general, we value richness. We may not value it enough to pursue it at the expense of the other activities we could pursue in the time required to produce a rich non-photographic picture, but we value it enough to want it when the time required to achieve it does not prohibit us from pursuing other activities we value. The reliability of photographic processes results from the standardization, not of the processes themselves, but of the functions they perform. This standardization is enabled by the fact that photographic processes are largely mechanical. Although there is a variety of different photographic mechanisms, both analog and digital, the mechanisms involved in the production of most photographs are alike in respect of the outputs they produce, given certain inputs. In particular, given a certain scene, most such

the epistemic value of photographs

99

mechanisms will yield an accurate depictive representation of that scene. This is a purely contingent fact about photographic mechanisms. Contra Hopkins, camera designers might have intended that the photographs produced with the mechanisms they designed misrepresent their external objects, and thus have designed mechanisms that function to produce inaccurate pictures. The reason why they did not—in the most part, at least—may be that they designed them for consumers who, they assumed, generally value depictive accuracy over inaccuracy. There are two respects in which this standardization is incomplete. Firstly, it is incomplete to the extent that photographic mechanisms do not function to produce accurate photographs. As we have seen, some photographic mechanisms, such as those that incorporate distorting lenses, do not produce accurate photographs. Moreover, many mechanized processes are reliable relative to most, but not all, feature types. For example, many produce photographs that, while otherwise accurate, misrepresent the determinate colours of the scenes they depict. Even those mechanisms that generally produce photographs that are accurate in all respects may have the potential to produce photographs that misrepresent the scenes they depict. This may be because, while the mechanisms at issue produce accurate photographs given most inputs, there are some inputs—for example, scenes of very low absolute illumination—given which they produce inaccurate photographs. Secondly, it is incomplete to the extent that photographic processes are not mechanized. Whereas photographs taken with instamatic cameras and developed in commercial labs are produced by wholly mechanized processes, others are not. To the extent that photographic processes are not mechanized, their outcome is subject to the inﬂuence of the same factors as affect the reliability of non-photographic processes. For example, non-mechanized development procedures may depend, not just on photographers’ intentions, but also on their visual states. Photographers sometimes include colour charts in the scenes they photograph so that, when they come to develop their photographs, they can check the colours the photographs depict the charts as having against the colours of such a chart (Snyder and Allen 1975: 162). This process will only ensure colour accuracy if photographers both intend to depict objects’ colours accurately, and they are not colour-blind. Photographic processes are reliable only to the extent that they function to produce accurate pictures. The easier it is to produce an inaccurate

100 catharine abell photograph using a given photographic process, the smaller the distance will be between the nearest possible world in which a picture produced by that process depicts its object accurately and the nearest possible world in which it depicts it inaccurately (assuming the latter world to be further from the actual world), and the less reliable that process will be. Nevertheless, the extent to which photographic processes standardly function to produce accurate pictures makes them considerably more reliable than non-photographic processes. The great majority of photographic mechanisms either function to produce such pictures, or do so under most conditions, or function to produce photographs that are accurate in most respects. Moreover, although the processes by which many photographs are produced are not wholly mechanized, and although photographers’ intentions need be no less variable than those of other picture makers, two factors make it considerably harder for photographers to realize their intentions to misrepresent objects than for the makers of non-photographic pictures to do so. Firstly, unlike the latter, they can affect depictive content only if they preserve the counterfactual dependence of features of photographs on features of their external objects. This makes them dependent on the existence of techniques such as dodging and burning, or the use of colour ﬁlters, that enable them to inﬂuence the causal relation between picture and object without breaking it. Secondly, the extent to which photographic processes are governed by standardly functioning mechanisms signiﬁcantly restricts the range of available techniques, and thus the capacity for photographic misrepresentation. For example, the use of distorting lenses is one of few techniques that enable the photographic misrepresentation of shape. However, the fact that the process of recording light reﬂected from scenes photographed is largely governed by standardly functioning mechanisms means that this technique is not readily available to photographers. What of photographic manipulation? There has been a dramatic rise, in recent years, in the frequency with which photographs are intentionally manipulated to misrepresent the scenes they depict. This is a direct result of the increasing prominence of digital photographic equipment; the greater accessibility of the computer technology required to manipulate digital photographs; and the spread of the skills required to use this technology. Digital photographic processes, it seems, can readily be exploited to produce photographs that misrepresent their objects.

the epistemic value of photographs

101

However, it is important to distinguish two forms of digital manipulation. The development of techniques which enable the systematic alteration of the shape or colour a digital photograph depicts its object as having, while preserving the counterfactual dependence of features of the photograph on features of its external object, has made photographic misrepresentation much easier than it previously was, and thus reduced the reliability of digital photographic processes and the EV of the photographs produced by them. By contrast, while techniques which enable the pixel-by-pixel retouching of photographs produce pictures which misrepresent objects’ shapes and colours, they override the effect of the photographic mechanism by severing the counterfactual dependence that features of those pictures would otherwise bear to features of the objects photographed, and thus produce pictures that are not purely photographic. Such techniques do not affect the reliability of photographic processes, because they make them redundant. Consequently, although they can be used to undermine the EV of particular photographs by transforming them into non-photographic pictures, they do not diminish the EV that photographic pictures generally possess. Nevertheless, such techniques affect the EV we are likely to attribute to the pictures they are used to produce. The seamlessness of the manipulation enabled by such techniques makes us likely to mistake such pictures for photographs, and thus to attribute to them the EV that photographs generally possess. While digital manipulation of the ﬁrst kind is still sufﬁciently rare for the EV typical of photographs to exceed that typical of non-photographic pictures by a comfortable margin, the epistemic future of photographs is by no means assured. As Barbara Savedoff notes, advances in digital photographic processes may one day eradicate the epistemic differences between photographic and non-photographic pictures (Savedoff 1997: 212).

Conclusion On two of the explanations I have considered, the epistemic differences between photographic and non-photographic pictures result from essential differences between the two types of picture. Walton sees the difference between the two as consisting in the fact that photographs are aids to vision,

102 catharine abell while non-photographic pictures are mere representations. Hopkins sees it as consisting in the fact that they are governed by different standards of correctness. By contrast, while they take photographs to differ epistemically from those non-photographic pictures that do not carry information about objects, Cohen and Meskin take certain of the epistemic differences attributed to pictures of the two types to be merely apparent. I have steered a somewhat different course, arguing that, while photographs in general do differ epistemically from non-photographic pictures in general, this difference is due to wholly contingent factors, namely advances in photographic technology and the standardization of the function of photographic processes.

References Abell, Catharine (2009), ‘Canny Resemblance’, Philosophical Review, 118/2: 183–223. Clark, Andy (2001), ‘Visual Experience and Motor Action: Are the Bonds Too Tight?’, Philosophical Review, 110/4: 495–519. Cohen, Jonathan and Aaron Meskin (2004), ‘On the Epistemic Value of Photographs’, Journal of Aesthetics and Art Criticism, 62/2: 197–210. (2006), ‘An Objective Counterfactual Theory of Information’, Australasian Journal of Philosophy, 84/3: 333–52. Currie, Gregory (1991), ‘Photography, Painting and Perception’, Journal of Aesthetics and Art Criticism, 49/1: 23–9. (1999), ‘Visible Traces: Documentary and the Contents of Photographs’, Journal of Aesthetics and Art Criticism, 57/3: 285–97. Dretske, Fred (1984), ‘Abstract of Comments: Seeing Through Pictures’, Nous, 18: 73–4. Hopkins, Robert (1998), Picture, Image and Experience (Cambridge: Cambridge University Press). Jacob, Pierre and Marc Jeannerod (2003), Ways of Seeing: The Scope and Limits of Visual Cognition (Oxford: Oxford University Press). Kulvicki, John (2006), On Images (Oxford: Oxford University Press). Lewis, David (1973), Counterfactuals (Oxford: Blackwell). Lopes, Dominic McIver (1996), Understanding Pictures (Oxford: Clarendon Press). Milner, A. D. and M. A. Goodale (1995), The Visual Brain in Action (Oxford: Oxford University Press). Savedoff, Barbara (1997), ‘Escaping Reality: Digital Imagery and the Resources of Photography’, Journal of Aesthetics and Art Criticism, 55/2: 201–14.

the epistemic value of photographs

103

Schier, Flint (1986), Deeper into Pictures (Cambridge: Cambridge University Press). Scruton, Roger (1983), ‘Photography and Representation’, in Scruton, The Aesthetic Understanding (London: Methuen). Snyder, Joel and Neil Walsh Allen (1975), ‘Photography, Vision and Representation’, Critical Inquiry, 2/1: 143–69. Walton, Kendall (1984), ‘Transparent Pictures: On the Nature of Photographic Realism’, Critical Inquiry, 11/2: 246–77.

4 Depictive Seeing and Double Content JOHN DILWORTH

This chapter provides an overview of a comprehensive double content (DC) approach to the depictive issues of seeing-in, as discussed by other contributors to this volume such as Robert Hopkins, Dominic Lopes, Bence Nanay, and John Brown. A potential advantage of the DC approach to pictorial depiction is that it has previously been systematically worked out in two much broader contexts. First, my 2005 book The Double Content of Art argues that all of the arts—and not just depictive arts such as painting or photography—involve similar double content structures. And second, the DC account has also been extended to perceptual content generally in several recent articles (Dilworth 2005b,c). These latter articles securely embed the DC account in a cognitive science framework, which—among other things—enables the account to avoid the potential dangers of an over-reliance on everyday assumptions concerning experiences of pictures. A useful point of entry for subsequent discussion is provided by Richard Wollheim’s inﬂuential concept of ‘seeing-in’ (Wollheim 1968, 1987). In particular, Wollheim’s distinction between the conﬁgurational, or design, features of a picture and its recognitional features is taken as a point of departure for such accounts. Any adequate account of depiction must explain how two distinct elements—namely, the purely physical design features of the surface of a picture, and the subject matter that is seen in the picture—are related to each other in our experience of pictures as pictures. Consequently, Lopes, Hopkins, and Nanay follow Wollheim in assuming that, in so far as pictorial experience has two kinds of content, they are best My thanks to the editors, Catharine Abell and Katerina Bantinaki, for their very helpful comments on earlier versions of this chapter.

depictive seeing and double content

105

distinguished as design-related, conﬁgurational content on the one hand, and subject-related, recognitional content on the other hand. However, from a cognitive science point of view this is a confused—or at least superﬁcial or inadequately analysed—content-based contrast. This is because the purely physical design or conﬁgurational features of the surface of a picture must provide the physical basis for any kinds of content whatsoever that are associated with the picture. If, for example, the relevant two kinds of content are kinds of perceptual content associated with viewings of the picture, then both kinds must be derivable from viewings of the same single physical design on the surface of the picture. Thus, there is a sense in which all pictorial content must be conﬁgurational content, since all of it must be derived from the same visible design. This is not to deny that there are two saliently different kinds of content that are thus derivable, but it is to insist that their differences cannot be explained purely in terms of their common derivation from a single design. An informal exposition of some legitimate content differences will now be provided, after which a more systematic account will be presented in subsequent sections. A typical picture, such as Rembrandt’s pen drawing of pastor Jan Cornelisz Sylvius, as discussed by Hopkins (Fig. 6.1), in some respects resembles an actual person seen face to face, while in other respects—such as in the directions of its inked lines—it instead resembles the design conﬁguration on the surface of the drawing. Presumably some such contrast underlies Wollheim’s intuitive conﬁgurational versus recognitional content distinction, as further developed by Hopkins et al. However, the comparison is superﬁcial or preliminary only, because it is those very inked lines themselves that also must provide the respects in which the picture resembles a person seen face to face, since strictly all that is visible in the picture is those inked lines. But, it could reasonably be asked, how could two different kinds of perceptual content be derived from the very same visible areas of a picture? Would they not visually compete with each other, so that, for instance, once all of the conﬁgurational, inked-lines content of the pen drawing has been accounted for, there would be no areas of the picture left over to represent the person who is portrayed by the drawing? In reply to such concerns, clearly the visible ink lines have to do double duty—both providing content or information about the features of these speciﬁc drawn lines that show

106 john dilworth Rembrandt’s pen-drawing style, and providing information about the person depicted in the picture as well. I have argued elsewhere (Dilworth 2005a,e) that in cognitive science terms there is really only one possible solution to this potential problem of informational conﬂict—namely, that the two kinds of content or information must be hierarchically arranged, in two distinct levels, with one level encoding or representing the other. Also, arguably the most plausible hierarchical arrangement is one in which lower-level, design-related content—providing information about the conﬁguration of lines etc. on the picture surface—is used to encode or represent a higher level of subject-related content. This is the structure adopted by the current DC account of depiction. So, on this DC account, viewers of a picture see the depicted subject in the picture by seeing the manner in which the lower-level conﬁgurational content of the picture represents that higher-level subject. The next section will give a preliminary, intuitive account of how these two levels of content could explain seeing-in, and will preview further details of the DC theory.

1. Double Content and Perceptual Ambiguity Issues It is important to situate the DC account in a broader context of human perceptual abilities and representational processing structures. It will be shown that closely related double content structures must be involved in normal perception of objects that are not pictures, so that the postulated DC structure of depictive perception is plausible, in that it is no more than a natural extension of more generic perceptual abilities. But in addition, it will be important to be able to theoretically account for a signiﬁcant ambiguity problem that is unavoidably associated with the suggested DC approach—both with respect to pictures, and for perception generally. The problem is that any object having a two-level DC structure is inevitably potentially ambiguous between various distinct DC interpretations, because by hypothesis at most one of the levels is directly or basically perceived—the upper-tier represented level must be inferred or interpreted from the data provided by the lower level. But in fact the situation is even worse than this preliminary description suggests, because there are potential conﬂicts between the two levels of content as well, so that, strictly speaking, neither level of content is fully determined by the

depictive seeing and double content

107

initial data as derived from the surface of a picture or from worldly objects generally. A legitimate interpretation would assign appropriate determinate content to each level, but the basic problem is that there would always be more than one possible way of doing this for each level. Hence, representational ambiguity ensues: potentially there is a range of distinct, at least minimally legitimate compatible double content interpretations of the retinal conﬁguration, so that there is no one unique double content structure associated with a concrete perceptual state S. For example, retinal information indicating that blue light was visually received might show that object X was blue and illuminated by ambient neutral white light, or it might instead show that object X was white but illuminated by ambient blue light. (There is also an indeﬁnite range of equally compatible, though less extreme, aspect and object data combinations for colour factors.) A similar point holds for some concrete shape pattern on the retina, such as an elliptical shape. That shape might have been caused by light emanating from an actually elliptically shaped object X directly facing the perceiver, or from a circular object X that was viewed from an oblique angle, and so on. Hence, any concrete retinal shape properties are just as ambiguous with respect to double content interpretations of them as are concrete colour properties of the retinal image. But, since arguably all low-level information provided by retinal images is either shape or colour information, it follows that all of the basic information derivable from concrete retinal states is potentially multiply ambiguous with respect to compatible double content interpretations of those states (Dilworth 2005c).

2. Basic Similarities in the Perceptual Content of Pictures and Non-Pictures It might initially be assumed that perception of an external representation, such as a picture, is fundamentally unlike perception of an ordinary real object, such as a cow, which does not represent anything. However, we must be careful not to confuse the legitimate differences in representational status of the relevant objects themselves—pictures versus non-pictures—with differences in the perceptual content of a person who is viewing each kind of item. There are at least three fundamental kinds of

108 john dilworth similarities in perceptual content that apply to both classes of object, which will now be brieﬂy summarized. As a preliminary, note that some sort of distinction of two kinds of perceptual content is inevitable in the case of pictures, because of the differences between their surface designs and their depicted subject. One way to express this point is that, for any given subject S, such as a scene or a person, there could be many different pictures of it, each having the same subject S but with a different design content D1, D2, . . . , Dn. However, there is an analogous distinction between perception of an object O, and perception of various aspects A1, A2, . . . , An of that object O, as seen from various distinct perceptual viewpoints. To invoke a G. E. Moore-style argument, just as I can know that I see a hand when I look at one of my hands, so also can I know that in so doing, I also see some particular aspect of that hand—such as the palm of my hand, or the back of the hand. But since my perceptual content always includes both the hand and some aspect of the hand, and since the two must be distinct—since there is only one hand, but many aspects of it—there must be two distinct kinds of perceptual content involved in one’s perception of the hand, just as there must be in one’s perception of a picture and its subject. The relevant concept of an aspect of an object may be generalized to include any broadly aspectual or contextual conditions under which objects are perceived, including lighting, distance, atmospheric conditions such as rain or mist, and partial occlusion by other objects. This generalization is legitimate because any such factors can affect our retinal images of objects, which provide the proximal source of all our visual information about them. So we may distinguish the aspect or aspectual content of retinal visual images from their speciﬁcally object-related intrinsic content. Arguably, aspect and intrinsic content together provide a generic two-level DC structure of perceptual content, which is closely analogous to the two-level DC structure of design and subject content in the case of pictures. The second fundamental similarity in perceptual content for pictures and non-pictures is provided by the mediating role of retinal images themselves. Indeed, it is arguable that such retinal images are themselves closely analogous to external pictures, such as projected screen images, in that, physically speaking, a retinal image literally is a projected image focused on the retina by the eyeball. In addition, the double perceptual content of external pictures must always be perceptually mediated via the

depictive seeing and double content

109

retinal images of viewers of the pictures, so that inevitably there must be an intimate relation between the contents of perception of external pictures and of non-pictures. Or, in more detail, when a picture is perceived, the image of the picture that is formed on the retina of the perceiver will approximately duplicate the main qualitative features of the surface design of the picture itself. Consequently, a perceptual interpretation of the design of the picture would involve—as applied to the closely similar retinal image—the same standard double content extraction procedures as would be used to extract content from any retinal image of worldly objects. In other words, since all visual perceptual interpretation is mediated by retinal images, it should not be surprising that those retinal images that represent external pictures are internally processed in similar structural ways to other more miscellaneous retinal images. As for the third fundamental similarity in perceptual content for pictures and non-pictures, it is that generic perceptual content is just as fundamentally ambiguous as is pictorial content—as will be further demonstrated in the next section.

3. More on Double Content and Representational Ambiguity in Perception and Depiction This section provides a more detailed theoretical overview of issues concerning generic perceptual double content and its inevitable ambiguities. These generic perceptual issues will also be closely related to pictorial double content issues (Dilworth 2005a,b,c). Normal perception of an actual object X involves a particular viewpoint of the perceiver with respect to X, so that different aspects Y of object X will be seen from differing viewpoints. Perception of such an aspect Y provides information both about object X itself and about various concrete aspectual factors Y1 , Y2 , . . . , such as intervening objects, lighting and atmospheric conditions, and so on, that together make up the totality of what can be perceived from a given perceptual viewpoint during observation of object X. A primary task of normal perception is to extract from such viewpoint-dependent aspects correct information about object X itself, free of the potentially distracting aspectual factors. An early stage of this perceptual extraction involves the processing of sensory information derived

110 john dilworth from a concrete sensory input state S, such as a concrete pattern of retinal excitation caused by light rays emanating from a given viewpoint-dependent aspect of X. Hence, in symbolic form, the initial or low-level informational content Z derivable from such a concrete sensory state S thoroughly intermixes data Y —about speciﬁcally aspectual worldly factors Y—with data X about X itself. Hence, in order to extract useful information about object X itself, the perceptual system must somehow transform or decode the low-level data Z = Y (X ) into upper-level data X that is speciﬁcally about object X itself. (In typical perceptual cases, the extracted aspect content factors Y , which provide information about worldly aspectual factors Y, would be discarded as being irrelevant to the main perceptual task.) Consequently, perceptual data has a two-level or double content structure in which low-level, intermixed information Y (X ) of a raw, broadly aspectual kind, has to be decoded to extract the upper-level objectual information about object X. However, there is a fundamental epistemic problem associated with this task, namely that a concrete sensory input state S, such as a concrete pattern of retinal excitation as discussed above, provides no guaranteed or automatic procedure for separating out the genuinely X-related information X from more miscellaneous information Y that is about the aspect Y and its aspectual factors Y1 , Y2, . . . , rather than speciﬁcally about X. The perceptual state S potentially contains information both about object X and about the ambient, aspectual conditions under which X is being perceived. But since two distinct content factors Y and X have to be extracted from the single retinal state, arguably there is a range of distinct possible twofactor interpretations, each of which is compatible with the same concrete retinal state. The general point that perceptual data is fundamentally ambiguous or indeterminate in various respects is widely accepted in vision science (see, for instance, Purves and Lotto 2003 for further examples). However, the speciﬁc double content structures summarized here, and defended in more detail elsewhere as noted, provide a novel, DC-based theoretical approach for articulating what are, arguably, the two most salient kinds of content involved in such perceptual cases of informational ambiguity.

depictive seeing and double content

111

Turning now to external representation cases such as pictures, arguably a closely analogous structure, along with similar possibilities of representational ambiguity, holds in their case as well. Just as a concrete perceptual state S has a low-level, broadly aspectual content of form Z = Y (X ), so also does a concrete external representation R have a low-level, broadly design or conﬁgurational content of the same form Z = Y (X )—which content Z is the raw or unprocessed design information derivable from the visible surface of the relevant picture. In addition, in both perceptual and external representation cases, an upper-level content X must be extracted by some means—where X is content that is about an object X in generic perceptual cases, and about a subject matter X of the picture in a speciﬁc pictorial case. As to the status of the decoded lower-level aspectual or design content factors Y in both generic object and picture perception cases, this information may simply be discarded in utilitarian cases, such as everyday perception of objects or vacation snapshots, in which all that is desired is accurate information about the actual properties of the objects that were seen or photographed. But, arguably, artistic pictures gain much of their signiﬁcance—including their typical inﬂective differences from face-to-face seeing of an actual object X—from perception of such lower-level design factors Y . As for the issue of how potential double content ambiguities could be resolved in generic perceptual cases, various views are possible. Purves and Lotto (2003) argue that broadly historical statistical factors, such as probabilities of previous correlations between a given kind of low-level retinal data pattern and actual facts about the objects being perceived, are sufﬁcient to resolve the ambiguities. I would add that in many cases of normal perception of non-representational objects, the potential ambiguities could be resolved by more comprehensive current perceptual observation. Potential ambiguities in allocation of colour information to either aspectual or object factors could be reduced by examining the ambient illumination directly. Or potential shape ambiguities could be reduced or eliminated by viewing an object from a range of viewpoints. By contrast, in typical pictorial cases, artistic intentions help to resolve ambiguities. Here is an example that involves artistic intentions and the nature of artistic media (Dilworth 2005a). Presumably, Rembrandt intended

112 john dilworth viewers of his pen drawing (as discussed previously) to see it as depicting a person. On such a legitimate interpretation, arguably only some of the pen lines could be concurrently noticed as such by a viewer, because noticing all of them as pen lines—i.e. as lower-level content only—would be perceptually inconsistent with concurrently noticing Rembrandt’s intended depicted subject at all. Or, to put the matter another way, though all of the pen-line-derived data must be initially processed as lower-level conﬁgurational content, in a legitimate interpretation some of it should ultimately be interpreted as transparently representing the subject—i.e. as upper-level content only—while other elements may be semi-transparent—i.e. providing content elements at both levels—while still others might provide low-level content only that provides no subject information (such as the frame or signature). Semi-transparent cases, involving both levels of content, could be illustrated as follows. For these cases too there could be both appropriate and inappropriate interpretations. For example, Rembrandt’s drawing could be incorrectly interpreted as indeed depicting a person, but nevertheless one who has strange pen-like markings all over his face. This possible interpretation is semi-transparent (involving both levels of content) because the pen markings are seen as being on the depicted person’s face—rather than just as being meaningless lower-level pen strokes—but it is nevertheless wrong because it would betray a fundamental misunderstanding of the artistic medium of pen drawing. Or, as Ernst Gombrich remarks in a similar case, a drawing of Tivoli should not be interpreted as depicting Tivoli as being bounded by wiry lines (Gombrich 1962: 78). A DC-based theory capable of explaining the basic cognitive factors involved in such ambiguity cases will be provided in the following sections. To sum up the points raised in this section, the two-level perceptual content associated with viewing a picture is potentially ambiguous between various distinct two-factor content interpretations of its design. The potential ambiguities would need to be resolved via constraints provided by artistically relevant factors, such as the artist’s intentions, or relevant media conventions, such as those applying to drawings rather than paintings. In addition, more generic kinds of perception of worldly objects involve structurally similar kinds of perceptual double content—except that, in their case, the standards for appropriateness of double content interpretation are provided by normal standards for veridical perception of objects and their properties.

depictive seeing and double content

113

4. An Orientational Double Content Theory: Basic Concepts Because of the importance of double content ambiguities for the understanding of both perception in general and depiction in particular, it is appropriate to develop a theoretical structure that is speciﬁcally geared to the analysis and articulation of the full range of relevant factors in such double content ambiguity cases. The following theoretical structure was initially developed (Dilworth 2005a, Chs 8–10, 12) to explain pictorial ambiguities that become salient in cases of changes in spatial orientation of a picture, and hence it is appropriately described as an orientational theory of double content ambiguities. As a preliminary, since a given picture—such as Rembrandt’s pen drawing—could be interpreted in more than one distinct double content way, we shall describe each of those ways as characterizing a distinct pictorial content (PC) PC1, PC2, . . . of the concrete design provided by the picture. The orientational theory to be provided shows how these various possible distinct double content interpretations PC1, PC2, . . . of the picture relate both to each other and to the basic raw or uninterpreted information that can be perceptually extracted from the picture surface via the concrete retinal conﬁgurations of a perceiver of the picture. To begin, here is a basic example that shows a need to introduce orientational concepts. A picture is capable of being interpreted in at least two different pictorial ways, involving distinct pictorial contents PC1 and PC2, as follows. If a picture of an upright house is spatially inverted, an ambiguity becomes apparent: the inverted picture can either be interpreted as an inverted picture of an upright house (PC1), or as an upright picture of an inverted house (PC2 ). The relevant pictorial contents PC1 and PC2 must be distinct, because each involves a distinct subject matter—namely, an upright house in the ﬁrst case and an inverted house in the second case. But the two distinct pictorial contents PC1 and PC2 are associated with only a single concrete picture. Hence, it becomes necessary to distinguish at least three different kinds of items and three corresponding relevant orientational factors—respectively, those applying to the concrete picture, to each distinct pictorial content

114 john dilworth PC1 and PC2, and to the distinct subjects S1 and S2 of each pictorial content. An adequate orientational theory must articulate all of the relevant orientational factors involved in such cases. It is immediately clear from the example that at least two distinctive orientational factors are involved: namely, the spatial orientation of the concrete picture P on the one hand, and the differing spatial orientations of the subject matters S1 and S2 of each pictorial content PC1 and PC2 on the other hand. But arguably a third kind of orientational factor, which concerns the spatial orientation of the distinct contents PC1 and PC2 themselves, must also be involved. This is so for the following reason. The orientation of the subject matter of a pictorial content (PC) cannot be determined independently of knowing the orientation of the PC itself. For example, if a picture P depicts an inverted house, then it must be possible to see that the depicted house is inverted when the relevant PC is viewed in its normal upright orientation. But this is only possible if the PC itself has an orientation that is independent of the orientation of its subject matter S. And ﬁnally, arguably a fourth, broadly environmental orientational factor E is required as well, so that the orientation of other items relative to environment E can be described. The most basic orientational concept in the current DC theory is that of a ﬁeld. A ﬁeld (to be discussed further in the following section) is an abstract structure of different speciﬁc qualities of some generic, homogeneous kind, such as some interrelated spatial positions, or some interrelated colours, or shapes. In the case of a spatial ﬁeld, a ﬁeld is a three-dimensional structure of abstract spatial points, the coordinates of which could be used to describe or index corresponding positions in some region of threedimensional space. Each of the four orientational factors described above would involve, among other things, its own distinct spatial ﬁeld, so that four implementations or tokens of the same abstract spatial ﬁeld type would be required to fully explain the orientational structure of the relevant example. Second, a concept of ﬁeld orientation needs to be deﬁned. This concept concerns the way in which two distinct copies of the same ﬁeld type relate to each other. For example, if a concrete pictorial token T has had a spatial ﬁeld F(T) assigned to it, then the ﬁeld orientation of token T with respect to the environmental ﬁeld E is deﬁned in terms of the correspondence or alignment between the differing coordinates of ﬁeld E and ﬁeld F(T).

depictive seeing and double content

115

Third, a concept of intrinsic orientation is also required. Whereas the concept of ﬁeld orientation (or alignment) is a relational concept specifying the relations of two distinct though type-identical ﬁelds, the concept of intrinsic orientation instead presupposes that there is some functionally deﬁned standard that speciﬁes a unique internal positional structure for elements of the ﬁeld (in some particular application of the ﬁeld structure). In particular, a top distinguished element or layer must be speciﬁable, which thereby counts as an intrinsic top of the relevant intrinsically oriented ﬁeld. In contrast, a ﬁeld top of an item X is whichever side or region of X is aligned with the intrinsic top of the relevant environmental ﬁeld E, whether or not X has an intrinsic top—and, if it does, whether or not its current ﬁeld orientation with respect to E results in its intrinsic top being aligned with its ﬁeld top. For example, a spatial environmental ﬁeld E is normally regarded as having an intrinsic orientation in which elements of the ﬁeld that are more distant from the earth’s surface are taken to be higher than elements that are closer to the earth, so that the intrinsic top of the ﬁeld would be the layer of elements that are at the highest altitude (most distant from the earth). Or, as another example, a car has an intrinsic orientation, with its intrinsic top being the top of the roof, because that is the functionally based orientation in which cars are normally driven. However, by contrast, miscellaneous items of hardware, such as small hand tools do not have an intrinsic orientation because they lack any unique, functionally speciﬁed orientation in which they are normally used. The concept of intrinsic orientation is particularly important in analysing pictures. Both PCs PC1 , PC2, . . . and their corresponding subject matters S1 , S2 , . . . possess an intrinsic spatial orientation, and in the analysis to be given, concrete pictures P do so as well, as part of the analysis, even though such pictures need not independently have an intrinsic orientation in their own right, simply as physical objects. Also, it will turn out that when the orientational theoretical framework is generalized—to include colour, shape, etc. ﬁelds, as well as spatial ﬁelds—pictorial contents, subject matters, and pictures will continue to possess an intrinsic orientation and an intrinsic top element with respect to their relevant ﬁelds. Fourth in this list of basic orientational concepts is a concept of uprightness, along with a correlative concept of inversion. An item X is upright just in case it has an intrinsic top that is currently aligned with the top of an

116 john dilworth intrinsically oriented environmental ﬁeld E—i.e., in the spatial case, if X’s intrinsic top is currently spatially above any other parts of X, relative to the environmental ﬁeld. Correspondingly, an item X is inverted in spatial cases if its intrinsic top is currently spatially below any other parts of X. It will also be convenient to use a more inclusive concept of inversion for any case in which item X is not upright. Finally, a ﬁfth basic orientational concept comes in the form of a principle: the oriented subject matter invariance principle, or OSMI principle (Dilworth 2005a, Chs 9, 12). As a simple example, we normally assume that if a concrete picture P of an upright house is spatially inverted, so that the ﬁeld orientation of picture P relative to environmental ﬁeld E changes by 180 degrees, nevertheless the ﬁeld orientation of the subject matter S—which is deﬁned relative to the intrinsic top of the relevant pictorial content PC1 —does not change during the rotation, in that the picture remains one that is interpreted as being a picture of an upright house, rather than its changing into a picture of an inverted house.

5. Applications of Orientational Concepts The orientational concepts just deﬁned will now be applied both to concrete retinal tokens Tr, and to aspect Y (X ) and object X contents. A parallel application to concrete pictures P, double pictorial contents PC = Y (X ), and subject matters S will also be described. The orientational concepts provide ways in which to keep track of changing relationships between these three kinds of items, plus of their relations to a fourth item—an environmental, intrinsically oriented ﬁeld E. However, these structured orientational relations are not just a perspicuous way of describing the relevant relations. In addition, they are intended to be a useful tool in articulating a broadly truth-functional or correctness-based view of depiction—which closely associates depictive content with a correctnessbased view of perceptual content as well. Thus, on the present view, pictures can correctly or incorrectly represent their subjects, just as perception can correctly or incorrectly represent worldly objects. Consequently, the current orientational DC framework must fully support a clear distinction between correctly and incorrectly depicted subjects. It does so by identifying orientational uprightness of subject in some

depictive seeing and double content

117

respect—such as colour, shape, or spatial orientation—with correctness of subject in that respect, and generic inversion in the relevant respect with incorrectness of subject. Similarly, in the case of object-related perceptual content X , uprightness of its relevant ﬁeld is identiﬁed with correctness of object-related content. Next, here are some further details on the structure of a ﬁeld. Because of the emphasis on correctness of subject matter, the elements of the relevant ﬁeld are deﬁned in the following way. First, there is an intrinsic top element, consisting of the relevant kind of correct subject matter—such as a particular correct spatial orientation, or a correct colour, or a correct shape, and so on. As a result, all four copies of the relevant ﬁeld have the same intrinsic top and intrinsic orientation. Environmental ﬁeld E provides a ﬁxed reference standard, relative to which the other three intrinsically oriented ﬁelds may have differing ﬁeld orientations. Second, the rest of the elements in the ﬁeld are deﬁned in a structured relational way that corresponds to (what could be called) relevant psychophysical alternatives for the kind of property being considered. For example, with spatial orientation, the relevant alternatives are other spatial orientations to which a picture could be rotated. In the case of colours, because of our psychophysical makeup, the other colour elements in the ﬁeld would roughly correspond to some standard ‘colour wheel’ structure for the colours. In the case of perceptual shapes, the actual shape of the subject matter would be the intrinsic top of the ﬁeld, with other retinal shapes produced by actual viewing of such an object from various angles making up the rest of the ﬁeld elements. In depictive cases, stylistic and medium-related considerations also serve to structure other characteristic kinds of ﬁeld, in ways to be discussed further in Section 6 below. For example, a medium such as pencil drawing would use ﬁelds in which top elements were fully realistic renderings of a subject, with increasingly graphic elements such as linear shading and cross-hatching becoming more prominent in the less realistic, more inverse elements of a ﬁeld. As an initial application of the orientational framework, consider a colour change case. During stormy weather, daylight that is normally a neutral white colour tends to take on a bluish tinge—as evidenced by any number of undesirably bluish photographs taken under such conditions. As perceivers with sophisticated perceptual constancy mechanisms, we are

118 john dilworth usually able to compensate for such cases of abnormal ambient lighting. Hence, normally, a white sheet of paper continues to look white to perceivers of it, whatever the weather conditions, or time of day, may be. But colour information extracted directly from a perceiver’s concrete retinal image of the paper in such cases would still contain an unusually high amount of bluish light. How are we able to correctly perceive the inherent colour of the paper as being white under such conditions? The orientational framework in this case would involve a colour ﬁeld with a white intrinsic top, since white is the correct colour of the paper. As a preliminary, recall that low-level aspectual double perceptual content is of form Y (X ), involving aspectual factors Y and objectual factors X , and that this double content information has to be extracted from a concrete retinal image token Tr. Also, since the paper X is indeed white, and it is correctly perceived as white by normal perceivers, the paper-related perceptual content X of such a perceiver must be upright relative to the relevant environmental ﬁeld E. Now in the orientational model, the colour that the paper is perceived to have is the colour that is the current ﬁeld top (relative to ﬁeld E) in the object-related content X . Since this is a case of correct perception of colour, the ﬁeld top of X is also identical with the white intrinsic top of the colour ﬁeld, so that the paper is perceived to be white. But nevertheless, the colour-related properties of the concrete retinal image token Tr include a predominance of bluish-light-related properties, so that consequently token Tr’s copy of the relevant colour ﬁeld is not upright, but instead it is generically inverted with respect to reference ﬁeld E. Hence, there is a potential compatibility problem: how can the actual white colour of the paper be made consistent with the physical bluish properties of retinal token Tr, which consequently makes token Tr’s own copy of the colour ﬁeld non-upright relative to ﬁeld E? Clearly the answer is that perceivers can correctly compensate for the bluish ambient light conditions by adopting a full double content perceptual interpretation Y (X ) of the retinal image token Tr, in which the aspectual content Y (X ), and its included aspectual colour factors Y , are non-upright, or inverted, in the orientational model, to the same degree as token Tr itself . Such a double content interpretation is fully compatible with the nonupright orientation of token Tr, in that the predominance of bluish light properties in token Tr is fully explained as being entirely due to aspectual

depictive seeing and double content

119

ambient light factors, rather than its being due to the inherent colour of the paper. An alternative interpretation of the retinal image token Tr will now be considered. To make the alternative interpretation plausible, and to link it to a case of depiction, suppose that a colour photograph Tp was taken of the original piece of white paper X from the same position, and at roughly the same time, as the original perceptual observation of X that produced the bluish retinal image Tr. Then a subsequent viewing of that photographic token Tp could produce a retinal image Tr2 that was qualitatively very similar to the initial retinal image Tr of the actual sheet of paper. However, suppose that the lighting conditions have now changed to neutral white light, so that, in viewing the photograph Tp, there is no automatic perceptual compensation by the viewer for the bluish tinge of retinal image Tr2. Instead, the photograph is perceived by its viewer as depicting a bluish sheet of paper, rather than a white sheet of paper. How is this difference from the original perception of the paper as being white to be explained? To begin, since the sheet of paper X depicted by the photograph—i.e. the subject matter of the photograph—is in fact white as before, the appropriate colour ﬁeld is also the same as before, with the same white intrinsic top. However, the depictive token depicts the subject matter—the paper—as being bluish in colour rather than white. Consequently, in the orientational model the colour ﬁeld for the subject matter is no longer upright relative to the environmental reference ﬁeld E. Instead, its ﬁeld top is the bluish colour that the photograph is perceived as depicting with respect to its subject matter, rather than the actual white colour of the intrinsic top of the subject matter ﬁeld. Hence, the subject matter ﬁeld must be non-upright, or broadly inverted, relative to the environmental ﬁeld E, which indicates in the model that the colour of the subject matter is incorrectly depicted by the photograph. It still remains to be explained how this broadly inverted subject matter interpretation can be rendered compatible with the full current double content interpretation Y (X ) of each of the two tokens—namely, the bluish retinal token Tr2, and the bluish concrete depictive token Tp itself. But perhaps it is clear enough that the broadly inverted subject matter, with its incorrect bluish ﬁeld top, fully explains by itself how the bluish properties of retinal token Tr2 and depictive token Tp have been processed. Clearly, it has been perceptually assumed that the full informational content

120 john dilworth of the bluish qualities of concrete tokens Tr2 and Tp is exhausted by the bluish qualities of the relevant subject matter, namely the paper. Consequently, in the orientational model the pictorial content PC itself—i.e. the full, low-level aspectual content Y (X )—would normally be being interpreted as being upright in its ﬁeld, i.e. with a white ﬁeld top. Hence, the full aspectual content Y (X ) that constitutes the pictorial content PC is assumed to be a neutral white colour, and, consequently, assumed to have no impact on the relation between the bluish qualities of tokens Tr2 and Tp and the perceived bluish subject matter of picture P. Or, in terms of the speciﬁcally perceptual content of token Tr2 itself, the aspectual content Y is assumed to involve nothing more than neutral ambient white light that could have no effect on the colour content of the perceived object—which ambient light conditions, as it happens, were in force during viewing of the photograph itself. Consequently, these viewing conditions help to explain why the photographic picture is misperceived as depicting a bluish sheet of paper. As for notation, the following simpliﬁed version is appropriate for pictures: upright picture content (UPC), inverted picture content (IPC); and upright subject matter (US), inverted subject matter (IS). Or, for perceptual content in general, including the special case of perception of a picture, we may distinguish an upright aspect (UA), inverted aspect (IA); and upright object (UO), inverted object (IO). Then the case of double content ambiguity given above for retinal tokens Tr is one in which a given kind of bluish retinal image can be given an orientational double content interpretation either as an inverted aspect–upright object (IA–UO) case, or as an upright aspect–inverted object (UA–IO) case. As for a more detailed analysis involving the relevant concrete tokens as well, upright token (UT) and inverted token (IT) factors are also required. In the example given, the correct colour of the paper was white, but both the retinal token Tr and the picture P were bluish. Consequently, these were both inverted token (IT) cases, or, more fully, IT–IA–UO and IT–UA–IO cases.

6. The Artistic Functions of Design Content In this section the context will be broadened to brieﬂy consider characteristic ways in which artists can exploit orientational factors in their pictures

depictive seeing and double content

121

to produce results that normally could not be seen in face-to-face seeing. These factors will serve to sharply distinguish artistic cases of depiction from more generic perceptual cases. Such phenomena, as previously discussed by Podro (1998) and Lopes (2005) for instance, can readily be explained in terms of Wollheim’s two-factor concept of ‘seeing-in’, so it is important that the current rival DC view should also be able to explain the phenomena, in order to effectively compete with Wollheim’s inﬂuential account. To begin, I have argued elsewhere that perceptually based double content phenomena in the arts are not conﬁned to depictive cases of visual artworks, but instead they are endemic in all forms of art (Dilworth 2005a). But such a synoptic approach to the arts in general thereby takes on the burden of explaining all facets of artistic meaning, not just those that are particularly salient in a given art form. In particular, it is not enough for a double-content-based approach to the arts to explain just standard, purely representational kinds of content or subject matter. Instead, the approach must also explain factors in artistic meaning specifically associated with each particular art form and with artists themselves, including factors such as the idiosyncratic expressive means used by a particular artist within a given medium, more general stylistic considerations, artistic intentions and attitudes, and so on. My own attempt to explain both kinds of meaning—of both artistic and representational kinds—closely associates each kind with a particular kind of content in the two-level structure. Standard representational content is explained in terms of upper-level subject-matter content, while broadly artistic kinds of content, involving expressive artistic styles, attitudes, and intentions, are explained in terms of lower-level aspectual content. This section will concentrate on showing how characteristic aspectual factors are recruited by visual artists for their own expressive purposes. Fortunately, the orientational framework established in the previous sections is particularly useful in articulating speciﬁcally artistic factors in the content of depictive and other artworks. A leading idea is that an art medium, such as the medium of painting or etching, is closely associated with its own, medium-speciﬁc orientational ﬁelds. These ﬁelds have fully realistic intrinsic tops, such as particular shapes, textures, or colours, but what makes them artistically appropriate is that their non-top elements are mediumspeciﬁc. For example, an etching uses groups of purely linear elements to depict subject matter. So the subtle shading on a woman’s face would

122 john dilworth be depicted using combinations of dense cross-hatched lines for darker shadings, along with looser patterns of lines to depict lighter shadings. Nevertheless, as previously discussed, the etched linear design on the surface of a print is not intended to be a realistic picture of a woman’s face as one that is covered with a network of ﬁne lines. The lines are part of the medium, or the depictive means of etching, not part of the subject matter that it depicts. In orientational terms, correct perception of what an etching depicts is of form IP–IPC–US. A correct interpretation of the wiry linear elements in the concrete design understands them as expressing aspectual depictive content of form Y (X ), with this aspectual content being inverted because its ﬁeld top is etching-speciﬁc linear elements rather than the realistic facial contours that are the intrinsic top of the relevant ﬁeld or ﬁelds. Hence, just as we can perceptually compensate for an oblique view of a shape by interpreting it as an IA–UO case, so also can we perceptually compensate for etching-speciﬁc aspectual content in a picture by a structurally similar IPC–US interpretation; whereas, by contrast, a naive incorrect interpretation of an etching as depicting a woman with wiry lines on her face would be using a dual, UPC–IS orientational interpretation instead. Artists may also use their own inimitable stylistic ﬁelds for speciﬁcally expressive purposes. For example, many of the Swiss sculptor Giacometti’s wiry, vertically elongated sculptures may initially look, to the uninitiated, like three-dimensional pictures, having the outline shapes of very thin or spindly people. Arguably, however, they are instead expressive sculptural pictures of ordinary-shaped people. Their apparent distortions in shape require an IP–IPC–US interpretation in which the prominent aspectual divergences in shape from realistic US shapes express the drained and marginalized lives of ordinary individuals in modern societies. By contrast, an incorrect IP–IPC–IS interpretation would instead interpret such works as inexpressive, pointless realistic sculptures of spindly people (Dilworth 2008). Colour can also be used for aspectual expressive purposes, such as in the case of Andy Warhol’s various serigraphs of Marilyn Monroe. One of these has a strong red colour in the area depicting her face (Dilworth 2005a: 258). The uninitiated might think that this incorrectly depicts Marilyn as having a red face rather than her actual pink face—a strange kind of deliberate error by Warhol, of orientational form UPC–IS or IPC–IS. However, much

depictive seeing and double content

123

more artistically appropriate is a dual IPC–US interpretation in which the red colour has the aspectually expressive function of allowing us to see Marilyn, with her ordinary pink face, through the garish, red-light-district glow of hyperactive media publicity. Indeed, the whole series of Marilyn serigraphs, with their various colours, only makes any artistic sense if thus interpreted in IPC–US expressive colour terms. To brieﬂy summarize these points, the current account has a ready explanation of such cases in terms of inverted aspectual expressive content (IPC). Since this aspectual content is inverted, it is not realistic and hence does not resemble the object seen face to face. But, at the same time, it is not seen as being visually inconsistent with the correct subject matter content either, since it is perceived as expressive pictorial (IPC–US) rather than as incorrect subject matter content (UPC–IS). This seeing of the red-faced Marilyn picture is not like seeing Marilyn face to face, but nor is it like incorrectly seeing Marilyn as having a red face. Instead, the potent alchemy of aspectual inversion enables us to see Marilyn as viewed via Warhol’s idiosyncratic expressive vision.

7. Other Conceptions of Pictorial Content The current double content (DC) theory of depiction distinguishes aspect representation and content from object representation and content. It postulates that this speciﬁc representational or content-based distinction is applicable both to perceptual content generally, and to depictive content in particular. However, there are various other extant distinctions of kinds of representation or content in the literature. This section surveys some of these, with an emphasis on distinctions that are particularly relevant to depiction and to the current collection. In order to avoid any invidious comparisons, after a brief discussion of Wollheim’s seeing-in view, the other views are presented alphabetically by the names of their proponents. The comments are brief because their main purpose is to distinguish other content distinctions from those of the current DC view. 7.1. Wollheim Recall that Wollheim (1968, 1987) introduced the two-factor seeing-in view, which is taken as basic by the other participants, according to which

124 john dilworth seeing of both the physical design of a picture and its content typically occurs during seeing-in. Though the DC view denies Wollheim’s claim that the physical design as such of a depictive token is seen during depictive seeing, it is possible to sympathetically reinterpret his view so as to render it consistent with the DC view, using as a basis other views of his concerning artistic media that license such a reinterpretation (Dilworth 2005a, Ch. 6). Nevertheless, for present purposes it is simpler to regard Wollheim’s views of content as directly entailing a theory of depiction that is incompatible with the DC view. 7.2. Brown Brown’s pioneering investigations in Chapter 8 in this volume, in addition to being of signiﬁcant value in their own right, could be regarded as providing further evidence for the current DC theory. Brown is able to successfully show, via further development of Hopkins’s concept of separation, that standard conceptions of seeing-in are overly narrow and simplistic, in that they neglect various kinds of possibilities of separation between what can be seen in a picture and its ofﬁcial subject. From the perspective of the DC theory, arguably such additional cognitive possibilities are only to be expected, for reasons such as the following. If the DC orientational theory is correct, speciﬁc artistic media involve socially constructed, artistically relevant ﬁelds, which are arrived at in the following way. Groups of artists experiment with speciﬁc media, such as the medium of pen and ink drawing, and, in collaboration with informed audiences, discover how pen and ink marks may be used, both to depict people and places, and to express an artist’s own attitudes and ideas about those subjects. However, what this means is that the relevant cognitive ﬁelds are social constructions—the products of whole cultures over many years—whose depictive functioning inevitably outruns the speciﬁc intentions of any one artist as applied to any one speciﬁc drawing. Consequently, precisely because the social functioning of a ﬁeld thus outruns individual intentions, that artist is unable to fully communicate to the viewer exactly how each penned mark is intended by him to characterize his ofﬁcial intended subject in the current particular picture. Nevertheless, according to the DC theory, the lower-level pen marks are inevitably capable of representing some higher-level subject matter in some way. But since not all of this higher-level content can be attributed to

depictive seeing and double content

125

the ofﬁcial subject (because of the intentional gap between social meanings of ﬁelds and individual artistic intentions), the DC theory predicts that there would, in general, be surplus higher-level represented content that would piggyback on content associated with the ofﬁcially intended subject. But this widespread, separated surplus higher-level content or meaning is exactly what Brown’s insightful investigations have uncovered. Also, presumably artists have learned to control (via prior observations of their own earlier pictures) these additional kinds of higher-level content for further artistic purposes of their own. 7.3. Hopkins Robert Hopkins (1998, Ch. 6) suggests that in some cases what can be seen in a picture is separate from what the picture depicts. An example he gives (1998: 128) is of a photograph of a dog. If the photograph is overexposed, what is seen in it might be a greenish dog, because the overexposure has resulted in an overall greenish tint to the picture—even though what the photograph depicts is a dog of no particular colour. Hopkins also argues that in part the separation occurs because there are standards of correctness for what a picture depicts, but not for what can be seen in the picture. Arguably this general distinction—of what can be seen in a picture from what it depicts—is a legitimate one for the following reason (though it is not one that is explictly discussed by Hopkins). Wollheim (1987: 67–71) has argued that any general account of representation, including of iconic kinds such as depictive representation, must acknowledge that it is possible to represent objects that are not particular actual objects—such as a painting of a lake that depicts no particular actual lake. In such a case, one could see the lake in the painting, even though there is no actual lake depicted by the painting, and hence no standards of correctness for what it depicts. I have suggested elsewhere that in order to explain such cases, a general distinction is needed between cases of external representation—in which some actual object is represented—versus cases of internal representation, in which some object is represented, whether or not the representation also externally represents an actual object (Dilworth 2005a, Ch. 11). As applied to Hopkins’s greenish dog, a greenish dog can be seen in the picture because the concrete depictive token does indeed internally represent a greenish dog, even though the depictive token does not externally represent a greenish dog. Perhaps there are other ways to defend Hopkins’s distinction,

126 john dilworth including his own explanatory framework, but the main point is that his distinction is indeed a legitimate one that is distinct from, and hence does not conﬂict with, the DC distinction. As for Hopkins’s current contribution in this volume (Chapter 6), he makes a good case that unitary, rather than divisive, accounts of seeing-in have much to recommend them. To that extent, his analyses support both his own unitary theory, and the current unitary DC theory as well. 7.4. Kulvicki John Kulvicki (2006) distinguishes, roughly following Haugeland (1991), the bare bones from the ﬂeshed out content of an image, and a similar distinction plays a more minor role in his current contribution (Chapter 1 in this volume). However, there are two signiﬁcant problems in attempting to compare his two kinds of content with my aspectual and subject matter kinds of content. First, on my account it will typically be possible to ﬁnd more than one depictive double content interpretation of a given depictive token. But those alternative interpretations would not typically differ in level of speciﬁcity or richness as do Kulvicki’s two kinds of content. Second, Kulvicki’s interesting claim in the current volume—that there is limited diversity and competition in viewers’ interpretive schemes, which fails to match the diversity in picture production schemes—can be directly questioned in various ways such as the following. First, since depictive tokens always permit dual or ambiguous interpretations, those interpretations do semantically compete with each other. And second, unusual production schemes, such as those of ﬁsheye photography, do produce legitimate pictures and depictive interpretations (pace Kulvicki) in that they involve specialized shape ﬁelds that deﬁne the relevant art medium. Arguably a bulgy ﬁsheye picture of a lemon is just as much a picture as is a wiry etching of the same lemon, once their double content structure is properly understood.

References Dilworth, John (2005a), The Double Content of Art (New York: Prometheus Books). (2005b), ‘The Double Content of Perception’, Synthese, 146: 224–43. (2005c), ‘The Twofold Orientational Structure of Perception’, Philosophical Psychology, 18: 187–203.

depictive seeing and double content

127

(2005d), ‘The Perception of Representational Content’, British Journal of Aesthetics, 45: 388–411. (2005e), ‘A Double Content Theory of Artistic Representation’, Journal of Aesthetics and Art Criticism, 63: 249–60. (2007), ‘In Support of Content Theories of Art’, Australasian Journal of Philosophy, 85: 19–39. (2008), ‘The Abstractness of Artworks and its Implications for Aesthetics’, Journal of Aesthetics and Art Criticism, 66: 341–53. Gombrich, Ernst (1962), Art and Illusion, 2nd edn (London: Phaidon Press). Haugeland, John (1991), ‘Representational Genera’, in W. Ramsey, S. Stich, and D. Rumelhart (eds), Philosophy and Connectionist Theory (Hillsdale, NJ: Lawrence Erlbaum). Hopkins, Robert (1998), Picture, Image and Experience (Cambridge: Cambridge University Press). Kulvicki, John (2006), On Images (Oxford: Clarendon Press). Lopes, Dominic McIver (2005), Sight and Sensibility (Oxford: Clarendon Press). Podro, Michael (1998), Depiction (Cambridge, Mass.: Harvard University Press). Purves, Dale and R. Beau Lotto (2003), Why We See What We Do (Sunderland, Mass.: Sinauer Associates). Wollheim, Richard (1968), Art and its Objects (Cambridge: Cambridge University Press). (1987), Painting as an Art (London: Thames & Hudson).

5 Picture Perception as Twofold Experience KATERINA BANTINAKI

The notion of twofoldness has been one of Richard Wollheim’s important contributions to the study of depiction. By means of this notion Wollheim aimed to highlight the fact that in seeing a picture the viewer can be visually aware of both the object that is being depicted and the medium in a single perceptual act. In seeing Lucian Freud’s portrait of Queen Elizabeth II, for instance, the viewer can be visually aware of the subject of the portrait and, at the same time, of the materials and the technique that the artist used to make up the portrayed subject. ‘Twofoldness’ names the phenomenal character of the perceptual experience that incorporates these two aspects, an experience that Wollheim calls seeing-in, and which he takes to be partly deﬁnitive of pictorial representation. The success of Wollheim’s theory rests on the intelligibility of twofoldness, which in turn seems to call for a comprehensive account of the character of, and relation between, the alleged aspects of visual awareness. Wollheim, thinking that an inquiry into the phenomenology of perceptual experience would not be fruitful, did not take on such a project. Consequently, it has been thought, he failed to provide us with a comprehensive theory of pictorial representation. My aim in this chapter is to show that we can make good sense of the notion of twofoldness, and so of Wollheim’s conception of seeing-in, even without further phenomenological investigation. To this end, I will draw on Aristotle’s doctrine of the unity of matter and form in compound substances as this is expounded mainly in his Metaphysics. I will then explain how Aristotle’s account of this unity may give us a way to think of the complex content of a twofold act of perception, in a way that allows

picture perception as twofold experience

129

us to account for properties that Wollheim attributed to pictorial seeing, and further, to resolve objections raised to the idea that pictorial seeing is twofold.

1. Seeing-In The starting point of Wollheim’s theory of pictorial representation was the fact that seeing an object in a picture is not like seeing the object face to face: he acknowledged that there is a difference in the phenomenology of the two experiences. The difference, he thought, consists in the fact that the physical properties of the picture—the surface, and the marks on that surface—are perceived along with what is depicted therein. So when, for instance, we see a picture of a table, not only do we know that what we see is a marked canvas rather than a table; but also we never lose track of the canvas, of the color pigments and lines. It is this complexity that for Wollheim differentiates pictorial perception of an object from ordinary perception of that object. That pictorial perception has this complexity, he argues, is evidenced by the fact that, when appreciating a picture, we can ‘marvel endlessly at the way in which line or brushstroke or expanse of color is exploited to render effects or establish analogies that can only be identiﬁed representationally’ (1996: 216). The capacity for visual experiences that have this sort of complexity is for Wollheim prior to depiction, both logically and historically:¹ it is the capacity for seeing-in. As manifestations of this faculty Wollheim cites hallucinations and dreams, thus implicating (visual) imagination. Seeing-in, however, differs from ordinary exercises of imagination in that in this case ‘visions of things not present to the senses come about through looking at things present’ (1998: 218). But not any sort of object can generate seeing-in. Wollheim explains that although commonly triggered by pictures, seeing-in further occurs in our encounter with surfaces that are not, and are not believed to be, representational. However, it is only triggered by surfaces of a certain kind: for seeing-in to occur, the surface in the viewer’s visual ﬁeld has to be adequately differentiated—think, for instance, of crumbling walls, rocks, or frosted glass: the heterogeneity of ¹ On the logical and historical priority of seeing-in, see Wollheim (1998: 47–8).

130 katerina bantinaki such surfaces seems to feed into the viewer’s imagination, triggering a certain visual experience, not any more, or not just, of the surface but also of something ‘in’ the surface. Wollheim named the distinctive character of seeing-in ‘twofoldness’, to highlight the fact that a single perceptual act incorporates two distinguishable but inseparable aspects of awareness.² The two aspects of awareness are speciﬁed as (a) the conﬁgurational aspect, which relates to the differentiated surface, and (b) the recognitional aspect, which relates to whatever is recognized in the differentiated surface. When it comes to pictures, the two aspects of seeing-in relate to the medium and the object that is being depicted respectively. With regard to the conﬁgurational aspect, Wollheim (1996: 212–13) explains that, in seeing a picture, the viewer should be aware ‘of the sustaining features of representation—features of x that permit me to see it as y’, of the ‘pigmented features of the canvas’, of an ‘unrestricted range of features of [the artist’s] panel’. The recognitional aspect of the experience, on the other hand, relates to what one sees in the picture, and accordingly to what the picture is recognized to be a picture of .³ Wollheim takes seeing-in to be the mode of seeing appropriate to pictures. However, he acknowledges that it is not necessary that the viewer undergoes a twofold experience for as long as she engages with a picture. It is possible that, according to her interests, the viewer shifts her attention towards the depicted object or the medium; when the shift of attention is such that the viewer focuses exclusively on either of these aspects of awareness, the experience ceases to be a seeing-in as it is not twofold (1996: 47). There is, in this case, a further outcome: the picture is not experienced as a picture any more as, in Wollheim’s scheme, seeing-in is partly deﬁnitive of pictorial representation. Accidental formations, such as ² This is the case according to the second version of the seeing-in theory (Wollheim 1998). According to the ﬁrst version (Wollheim 1996), seeing-in involves two different but simultaneous experiences. ³ Wollheim imposes a standard of correctness on pictorial seeing, so that not just any visual experience in response to the picture determines what the picture represents; rather, only experiences that correspond to the artist’s intention do so. The standard of correctness has been a controversial aspect of Wollheim’s account of depiction. As it does not directly impinge on the nature of seeing-in itself, however, it will be latent in my discussion: I shall take the recognitional aspect of awareness to have as its proper object whatever the viewer can see in a picture, given her cognitive background. Further, according to Wollheim, to see something in a picture is to ‘discern something standing out in front of, or (in certain cases) receding behind something else’ (Wollheim 1996: 46). However, in what follows, I will remain neutral regarding the truth of this claim, assuming only that seeing-in involves at least seeing something in the marked surface.

picture perception as twofold experience

131

cloud formations, permit, but they do not require, twofold seeing. Pictorial representation, in contrast, requires twofoldness. Wollheim (1996: 213) argues that ‘If I look at a representation as a representation it is not just permitted but required of me that I attend simultaneously to object and medium’; and further (1996: 216) that ‘In painting . . . twofoldness must be a normative constraint upon anyone who tries to appreciate works of [this art].’ When the viewer’s focus is exclusively on either the medium or the object that is being depicted, the object of sight does not ﬁgure in her visual awareness as a picture. Wollheim has put forth a wholly perceptual account of pictorial representation. He deﬁnes pictorial representation in terms of a perceptual experience, seeing-in, so that an object is not a picture unless it can trigger seeing-in; and further concedes that it is the content of that experience that reveals what is represented in a picture. The success of Wollheim’s account thus relies on the intelligibility of seeing-in. But seeing-in has proven to be a troublesome notion. In what follows I will consider the main objection to Wollheim’s account of pictorial experience and explain how it can be met. It needs to be made clear from the outset, however, that my aim is to improve our understanding of Wollheim’s conception of pictorial experience, not to defend his account of pictorial representation, so in what follows I will remain neutral on the question whether the notion of seeing-in can ground a successful deﬁnition of pictorial representation.

2. Objection: Making Sense of Twofoldness The most important objection to Wollheim’s theory is that it has not been sufﬁciently well explained to qualify as a complete theory of depiction (see, for instance, Budd 1992). An initial worry is that the two aspects of seeing-in seem to be incompatible. As Wollheim accepts, ‘it is true that each aspect of the single experience is capable of being described as analogous to a separate experience. It can be described as though it were a case of simply looking at a wall or a case of seeing a boy face-to-face’ (1998: 46). If we take this analogy at face value, then we can assume that the contribution of the conﬁgurational aspect of awareness to the overall content of the experience of seeing-in is that one is in the presence of a

132 katerina bantinaki (ﬂat) marked surface, while the contribution of the recognitional aspect is that one is in the presence of a three-dimensional object of some kind. Thus, one and the same object, for instance the picture of a tree, will ﬁgure in the viewer’s awareness as both a ﬂat and a three-dimensional object: as a disposition of pigments and as a tree. In this case, simultaneous awareness of the medium and the depicted object entails that the picture is simultaneously perceived under two different, and incompatible, descriptions—as a present marked surface and as an absent 3D object of some kind. This, however, is an impossible experience. However, this objection does not undermine the seeing-in hypothesis—only a certain understanding of it. The objection relies on the assumption that each aspect of seeing-in is (in terms of its overall character) identical to a face-to-face experience of the object relevant to the aspect. But, according to Wollheim (1998: 46), this is not how we should understand the two aspects of seeing-in: confusion will ensue if we ask how experientially like or unlike each aspect is to the analogous face-to-face experience after which it can be described. As it applies to pictures, for instance, the complexity of seeing-in is, supposedly, such that the phenomenology of each aspect of the experience is incommensurate to the phenomenology of seeing either a non-representational surface or the actual object depicted in a picture. If the way that a picture surface ﬁgures in our awareness when we see something in it is incommensurate to the way that a non-representational surface ﬁgures in our awareness, then we cannot assume that the content of the former kind of awareness (and thus its contribution to the overall content of seeing-in) is analogous to the content of the latter. The same would be true of the recognitional aspect of awareness. But with this qualiﬁcation a different challenge emerges. If we are not to model our understanding of each aspect of seeing-in on the face-to-face experience after which it can be described, then we seem to have no resources to understand what each aspect is like in terms of its character and content, and so what seeing-in is like. According to Budd (1992), for instance, the fact that the phenomenology of each aspect of awareness is incommensurate to the phenomenology of the simple face-to-face experience after which it can be described prevents us from drawing conclusions about the phenomenological character and the content of the experience of seeingin, in which case the experience is rendered incomprehensible. Since

picture perception as twofold experience

133

Wollheim denies that an inquiry into the phenomenology of seeing-in would be fruitful, and thus does not analyse the experience further, he has not provided us with a comprehensive account of seeing-in. We are left with only our own experience of seeing pictures to help us understand a theory developed as an account of this kind of experience. We may be able, however, to resist this conclusion, even without further phenomenological investigation. But ﬁrst we need to be clear on where the problem lies, i.e. which aspect of Wollheim’s account of pictorial experience needs to be clariﬁed. Wollheim claims that pictorial seeing is twofold, i.e. that it has two aspects to its content, one relating to the medium and one relating to the subject of representation. To accept this claim we need to be able to understand what such a twofold perceptual experience is like in terms of its character and content: such an understanding requires that we are able to specify both what is the contribution of each aspect to the overall content of the experience and how the two aspects merge in a single homogeneous experience (so we need to be able to understand what is the complex content of the experience as a whole). Wollheim’s incommensurability claim speciﬁed above provides only a negative guide to such understanding: we are not to model our understanding of each aspect of seeing-in on the face-to-face experience after which it can be described (i.e. the visual experience of a non-representational marked surface and the visual experience of the actual object that is being depicted), and so we cannot assume that seeing-in is just a compound of the analogous face-to-face experiences. The gist of the incommensurability claim is thus that we cannot draw on the relevant simple face-to-face experiences in order to illuminate the experience of seeing-in: such simple experiences lack the complexity that seeing-in is supposed to have as a whole, a complexity that is, presumably, reﬂected in its component aspects. In accordance with the incommensurability claim, seeing-in does not represent its object, for instance the picture of a table, as a marked surface plus a table: in the object of awareness of seeing-in, the medium and the object of representation merge in a way that precludes such a description. I believe that Wollheim is on the right track with regard to both the characterization of pictorial seeing and the incommensurability claim. However, since he refused to provide a further positive characterization of seeing-in—one that would specify the precise content of each aspect of

134 katerina bantinaki the experience (in a way that would clearly set it apart from the analogous face-to-face experience) and the way that the two aspects merge in a uniform single but complex perceptual act—he exposed his theory to the incomprehensibility objection that Budd raises. To meet this objection our aim should therefore be to provide a positive characterization of seeing-in that will allow us to understand what this complex experience is like in terms of its character and content.⁴ To do so, I believe we need to divert our attention from pictorial perception to generic perception and the compound character of its objects. This is certainly not an explanation that Wollheim developed, but I will pursue it in order to improve the intelligibility of Wollheim’s conception of seeing-in. Consider, for instance, a wooden table: the object has a particular material constitution (it is made of wood) and it has a particular form or shape in terms of which it is identiﬁed as a table; the two aspects of the object—the table as matter, the table as identiﬁable form—are perceptually inseparable (or, at least, they cannot be perceived separately without effort). The wooden table, as it ﬁgures in our awareness, is a compound of matter and form: these form a single, complex but homogeneous, object of awareness and they are distinct aspects of the object (or, better, distinct aspects of awareness) only in the sense that they can be abstracted from the complex content of the experience as a whole. The unity of matter and form in concrete objects of perception is a central theme in Aristotle’s account of natural objects and artefacts. In the following paragraphs I brieﬂy consider Aristotle’s conception of such unity and then explain how it can help us form an intelligible conception of the twofoldness of pictorial perception.

3. The Unity of a Compound Substance Aristotle classiﬁes as substances all independent things, things on which everything else (qualities, quantities, relations) somehow depends. A human being, a dog, a tree, a bed, are all examples of individual substances. Every concrete (as opposed to abstract) substance is thought to be a single but ⁴ For an alternative explanation to the one that I suggest in this chapter, see Hopkins (1998). For Wollheim’s reply to this alternative explanation, see Wollheim (2003).

picture perception as twofold experience

135

complex entity, in the sense that each is a compound of matter and form. The matter is what constitutes the substance, the material out of which it was made: for example, the bronze of a statue and the wood of a bed. Form is the functional organization of an object in terms of which it satisﬁes a deﬁnition; the form of a table is, then, exactly that, being a table, having a structure (serviceable to a function) identiﬁable as table-ness. The matter and the form of a concrete particular are not physically distinct components but are distinct at least because they can be thought of separately, i.e. they can be abstracted from the whole, for instance in analysis or description (see, e.g. Scaltsas 2001: 116). At least matter, however, can be comprised of physically distinct components: for example, the matter of a house involves bricks, planks, and roof tiles.⁵ If the concrete particular involves distinct—physical and abstract—components, how does it become a uniﬁed whole? How does the house become one for us, instead of a collection of parts and properties? According to Aristotle, it is the presence of form in the concrete object—as the essence of the object and as its functional organization—that actively uniﬁes all the different components and makes them one determinate object (Aristotle 1924: 7. 17). The form is that in terms of which the different parts become a uniﬁed determinate something. Under the purposeful form the component parts lose their distinctness: they lose their boundaries. Being organic parts of a whole (and for as long as they are thought of as parts of a whole), they are identity-dependent on that whole: on what the whole is and what the whole does. For instance, because the functional organization that makes a house a house (e.g. being a roofed, enclosed, habitable space) is present in a given collection of materials, this collection of materials is not, and is not perceived as, a juxtaposition of so many bricks and planks; it is, rather, a unity, a concrete house, and the different materials are organic parts of that whole. On the whole, the bricks ⁵ It should be noted here that Aristotle speaks of matter at two different levels: on the one hand there is prime matter, which is abstract, and on the other hand there is concrete matter, which can itself be given a basic hylomorphic explanation to the extent that it has chemical properties but also physical properties like mass and shape (although not yet the meaningful shape of the concrete whole of which it is the matter). In the analysis of concrete objects as compounds of matter and form, Aristotle uses both senses of the term ‘matter’, and certainly the latter since he often talks of the material parts of an object being united under the meaningful form (e.g. bricks and planks being united under the form of the house). It is on this latter use that I will focus since this is most relevant to Wollheim’s conception of a medium (a medium which can be described also as a design, a design that itself has parts and shape, although not yet the meaningful shape of the pictorial representation of which it is a design).

136 katerina bantinaki are not just bricks but parts of the wall of the house, and the ﬂoorboards are not just planks but parts of the ﬂoor of the house. The loss of boundaries, and the ensuing identity-dependence, applies not only to material parts but also to form (Scaltsas 2001: 125). Although form as a universal is not tied to any particular set of materials, the concrete form is so tied: the type ‘house’ is not any particular set of bricks or planks, but this house is these bricks and these planks, organized in this way. The concrete house is existentially and qualitatively dependent on or tied to its matter, so that a defect in the bricks is a defect in the house and the destruction of the bricks entails a destruction of the house. Since the components of a concrete particular are identity-dependent on the (functional and material) character of the whole, physical or abstract separation from that whole (as when one focuses one’s attention on the bricks of the house, and thinks of them as bricks rather than as parts of the wall of the house) entails reidentiﬁcation of the entity thus separated. What results from abstraction (the pure matter or the pure form) is an entity that is not present—in that pure state—in the concrete thing, given that the concrete thing is a uniﬁed or homogenized whole. According to the homonymy principle suggested by Aristotle, separation from a substantial whole entails reidentiﬁcation of the emerging components: separated from the whole, the components are independent of what the whole is or of what the whole does, so they have to be redeﬁned (for instance, when attention is focused on the bricks of a house and these are considered as just bricks rather than as parts of the wall of a house, their identity changes to the extent that they are not considered in terms of the essence and the function of the concrete compound object, i.e. the house). Consequently, what a concrete object shares with other objects (its matter, its form) does not exist in that pure state in the object. Different objects may share the same matter, but in each object the matter is inseparable from the whole that it constitutes: in each object the ‘wood’, for instance, has taken on a certain shape and function: it is necessarily informed—it is, for instance, a (wooden) bed, a (wooden) table, a (wooden) chair. The wood can be separated from the whole only by abstraction; it cannot be physically separated from the whole and it cannot be perceived separately from the whole: you cannot see the wood without seeing the bed that it constitutes—without seeing its functional character. To be sure, the wooden bed is a piece of wood, but from the moment the wood took

picture perception as twofold experience

137

shape it is not just wood: the wood necessarily constitutes (and ordinarily it is seen as constituting) a bed. The case for form is the same as it is for matter. If we think of a concrete particular, for instance the wooden bed, just in terms of its functional structure or its identity, we shall have to abstract the form from the uniﬁed whole. The emergent entity, the abstract form (‘a bed’), is not an item that exists in the whole: in the concrete object the form is necessarily enmeshed with matter. The actual form (the concrete token of the abstract type) is existentially and qualitatively dependent on the underlying matter. The actual form is not—it is never—‘a bed’; it is always a wooden bed, or an iron bed, etc., depending on the constitution of the whole, and that constitution determines the qualitative character of the object. Thus, the concrete particular is en-mattered form (or en-formed matter), and in that whole, matter and form are enmeshed; such unity determines how the object ﬁgures in our perceptual awareness: what we see the object as being and how we see it as being—what its qualities are. Aristotle’s account of the unity of compound substances can be summarized as follows: the form, as functional organization and as the essence, unites the different (physical and abstract) parts of a concrete thing into a meaningful whole, a whole that has a particular identity and a particular function. In that whole, the different material elements lose their distinctness and become organic parts of the whole. Equally, the form is tied (existentially and qualitatively) to the constituting matter. Consequently, abstraction from that whole entails reidentiﬁcation for both matter and form: abstracted, the components become different entities, as they are independent of what the whole is and of what the whole does. The unity of matter and form in compound substances can give us, I believe, a way to think of twofoldness.

4. Gestalt Aristotle’s account of the compound character of concrete substances aims to capture their intrinsic nature. Our concern with twofoldness, however, relates to the character of the visual experience to be had in response to a particular subset of such entities, pictures, and not to their intrinsic nature. But Aristotle’s account of the unity of matter and form in concrete

138 katerina bantinaki substances may give us a way to think of the unity of a twofold act of perception: in this case, the focus is not on the way that the object is in itself but in the way that the object appears as being. My claim is that objects of perception, as such, exhibit (in our experience) the kind of unity that Aristotle ascribes to compound substances, and it is such unity that gives us an understanding of twofoldness, in particular of the complex content of a twofold act of perception. That objects of perception have a compound character—that visual experience represents its objects as compounds of matter and form—has long been acknowledged by perceptual psychologists. As Rudolph Arnheim notes, ‘[perceptual] identiﬁcation presupposes an identiﬁable pattern. One cannot recognize something as a thing known, expected or to be reacted to unless it is discriminated by its sharply deﬁned character’ (1969: 29). What Arnheim highlights is the fact that—for reasons that need not concern us here—we tend to perceive organized patterns rather than inchoate material elements. An object of perception is ordinarily a gestalt, i.e. an organized whole: distinct material elements are grouped together under an identiﬁable generic structure—a shape or form that enables us to discriminate the object from its surroundings. The phenomenon of perceptual gestalt is intriguing and certainly pivotal to depiction; Wollheim acknowledges this fact when he traces the ﬁrst important step in the creation of painting in the recognition of a gestalt: ‘As the agent continues to mark the canvas . . . he notices that some of these marks, because of how they have landed on the surface, coalesce. Perceptually they form wholes or units. Groups of them are seen as one’ (1998: 20). How is this phenomenon relevant to the explanation of seeing-in, and, in particular, of twofoldness? Its relevance is the uniﬁed, homogeneous character of the (gestalt) object of perception: on the one hand, the matter of the object—the materials out of which it is made and its distinct components—are uniﬁed in perception under the identiﬁed shape or form (unless, that is, the viewer focuses her attention on those materials and parts), and on the other hand, the form is tied to the matter in perception—it is the form of the particular piece of matter (rather than a form thought of in the abstract) so it is qualitatively dependent on that matter. A picture as it ﬁgures in our awareness can be similarly explained as such a compound object—a compound of medium properties and meaningful form, i.e. a form characteristic of an entity in the world—and if we can give an account

picture perception as twofold experience

139

of the unity of these aspects of awareness, then we might gain a better grasp of the complex content of seeing-in. Aristotle’s account is useful to us in serving as a model for the explanation of this unity.

5. The Unity of a Picture What does Aristotle’s account of the unity of concrete substances suggest for pictures as objects of awareness? Let’s take as an example Freud’s portrait of Queen Elizabeth II; and let’s think of matter as corresponding to the medium (in particular, the marks on the pictorial surface), and of form as the functional organization of the marks on the pictorial surface, in terms of which the whole qualiﬁes as a portrait of Elizabeth II. In Poetics Aristotle indicates that pictorial form is a distinctive shape or outline, and so a distinctive overall appearance.⁶ Pictorial form corresponds to the subject of the picture (or, in Wollheim’s terms, the object of representation) in the following manner: if the material elements on a picture’s surface are organized so that they are seen as forming a shape or outline (and thus the overall appearance) characteristic of an entity X, then—at least in standard cases—the subject of the picture is X. Aristotle’s account can perhaps help us elucidate the twofold but uniﬁed character of the object of pictorial perception, the picture-of-X. Much like an Aristotelian compound substance, the picture appears to us as a composite of medium properties and meaningful form—the form characteristic of X. But as in compound substances, the components do not seem to us to coexist as an aggregate; rather, they are experienced as comprising a uniﬁed homogeneous whole. How does this unity come about? In this case, too, the principle of unity is form. This is clearly exempliﬁed in the case of physically distinct components. Although the pictorial surface is marked with different elements, the elements have been intentionally arranged or organized in a certain way; having been so arranged, in Wollheim’s terms, perceptually they seem to make a whole or unit—they seem to make up a particular shape: a shape which is a characteristic identiﬁable form, in terms of which the object of sight qualiﬁes as a ⁶ Aristotle (1996: 1. 47a , 6. 50b ). The term ‘overall’ here is to exclude aspects of ﬁne-grained appearance relevant to constitution.

140 katerina bantinaki determinate this, a picture-of-X (in the case, that is, that the viewer recognizes the overall shape as the form characteristic of X). The identiﬁable form unites in perception the different material elements on the surface in an organic whole (thus, the object of pictorial seeing is a gestalt): under its organizing inﬂuence, the material elements lose their distinctness, their boundaries, and become identity-dependent on the meaningful whole. We can draw a parallel with the example of a particular house: under the inﬂuence of the form-of-a-house the bricks and the planks are not just (and do not ﬁgure in our awareness as just) bricks and planks, rather they are parts of the wall and the ﬂoor, they have become organic parts of the whole that they constitute and are, to that extent, redeﬁned. In the case of the portrait of Elizabeth II, under the inﬂuence of the organizing form (the form of the Queen) the material elements lose their distinctness in perception. For instance, the red pigment at the bottom half of the canvas and the brown patch on the upper left function now as organic parts of the whole; they ﬁgure in our awareness not as just red and brown patches, but as the lips of the Queen’s portrayed face and as a shadow on her portrayed forehead. Although the distinct material parts of the design may have physical properties like shape or colour, under the inﬂuence of the identiﬁable form such properties assume a certain meaning for us: they are seen as meaningful, rather than just as physical, properties. At the same time, pictorial form is tied in perception to the material elements on the picture’s surface: it is those elements that constitute the form of the individual that is seen in the picture. As an effect, the qualitative character of the object of sight (how the object looks) depends on the constitutive elements, on the medium as a whole—on what materials have been used, on what surface, and how these have been manipulated and organized by the artist in order to construe the intended subject. To that extent the object of sight, the picture, manifests the medium: the medium is seen through what it makes seen. The picture, then, as an object of awareness, has the sort of unity that Aristotle attributes to concrete substances: although it is carried forth by a medium and has a form (a shape or outline) characteristic of a real or imaginary object, these are distinct components of our experience of the object only in the sense that they can be abstracted from the complex but homogeneous content of the experience as a whole. The material elements are perceptually inseparable from the form that the elements make up:

picture perception as twofold experience

141

as one cannot (without effort) see a table without seeing the wood that constitutes the table, so one cannot (without effort) see, for instance, the Queen construed in Freud’s painting without seeing the particular manner in which she has been construed. Consequently, the analysis of the content of our experience of a picture into a medium and a subject is a sort of division of the experience: and, in parallel to Aristotle’s claim about compound substances, it results in abstract entities that are not, in that pure state, part of our experience of the object. In the entity that ﬁgures in our awareness when we see a picture (with understanding), medium and subject are ordinarily enmeshed: the medium (speciﬁcally the marks on the pictorial surface) is informed by the subject—in particular, by the form characteristic of the relevant real-world or imaginary entity—and the subject is being constituted by the medium, so that it is qualitatively dependent on the medium.

6. On Twofoldness According to the suggested account, what an appropriate viewer sees when she looks at a surface which is suitably marked is the picture-of-X, which is a single but compound object of awareness. The compound character of the object of awareness may help us understand Wollheim’s conception of twofoldness, and so meet Budd’s objection to this theory; in particular, it can help us understand Wollheim’s claim that (a) in seeing a picture we are visually aware of both the medium and the object of representation; and that (b) being thus aware of the medium and the object of representation is different phenomenologically from seeing either a non-representational marked surface or the actual object represented. With regard to claim (a), since the object of awareness is a mediumspeciﬁc construal of a real-world (or imaginary) object, then both the object that is being construed and thus depicted, and the medium, as this is evidenced in or manifested by the construal, are parts of the viewer’s awareness—they are part of the complex content of her experience of the picture. The fact that, in seeing a picture of an object, the viewer is visually aware of the object that is being construed and of the medium, in contrast to what Budd claims, allows us to draw some conclusions about the content of

142 katerina bantinaki the viewer’s experience: we can rightly assume that aspects of the object’s appearance—minimally the (two-dimensional) shape that is characteristic of this entity from the point of view implicit in the picture—as well as aspects of the medium, are both part of the content of the viewer’s experience. In the case of Freud’s portrait of Elizabeth II, for instance, we can assume that a viewer who sees the Queen-in-the-painting, sees a face that is characteristic of the Queen from the implicit point of view, and sees that the Queen-in-the-painting is old, austere, wears a tiara, has a ‘textured’ grayish look, and so on. It is true, however, that having said only that much about the viewer’s experience, we haven’t completely speciﬁed the content of this experience, which thus remains partly obscure: the content of the twofold act of pictorial perception cannot be entirely speciﬁed by a speciﬁcation of its aspects, if our understanding of each aspect is based on an analogy with the simple face-to-face experience from which it derives. As mentioned earlier, the reason is that the two aspects of the experience—thought of as uniﬁed aspects of the twofold experience rather than as abstract components of it—are phenomenologically incommensurate to the (simple face-to-face) experiences from which they derive. Wollheim highlights this fact by means of a non-pictorial example: ‘Sometimes we experience a pain in the knee. This is a complex experience, but it is not to be understood by seeing how one part of it compares with having a pain, but nowhere in particular, and how the other part compares with being aware of one’s knee and where it is’ (2001: 20). Following the analogy with Aristotelian compound substances, we can understand this claim as follows: the experience of having a pain in the knee is a complex but homogeneous experience; having a pain and being aware of one’s knee are distinct components of the experience only in the sense, and to the extent, that they can be abstracted from the homogeneous whole in a description of the experience. In the concrete experience the two components are enmeshed: as in the case of a concrete substance, for instance a wooden table, one cannot (without effort) see the table without seeing the wood that constitutes the table; so in the case of a pain in the knee one cannot (without effort) feel the pain without feeling one’s knee as being in pain. In this case, a description of the experience that appeals to its components in an abstract manner fails to capture the phenomenology and content of the concrete experience where the two components are

picture perception as twofold experience

143

enmeshed (an experience where the one component, so to speak, infuses the other). To return to the case of pictorial perception, we can explain Wollheim’s claim (b) in a parallel manner: in the object of sight, i.e. the picture-ofX, medium and subject are enmeshed. Awareness of the medium when seeing the picture-of-X is not to be compared with seeing inchoate material elements, because in the former case but not the latter the material elements form a meaningful whole; they are accordingly seen through (and so inseparably from) what together they make seen. Conversely, the experience of seeing an object in a picture is not to be compared with seeing the object face to face (or simply imagining it): the picture presents a medium-speciﬁc construal of that object, so the object is (normally) seen as construed; for instance, in seeing Freud’s painting we don’t see the real Queen, rather we see the Queen-in-the-painting, and the Queen-in-thepainting, as described above, is old, austere, wears a tiara, has a ‘textured’ grayish look, and so on. Now, this fact can account for two phenomena relevant to seeing-in discussed in this volume, inﬂection and separation. According to Robert Hopkins, when inﬂection occurs ‘what is seen in a surface includes properties a full characterization of which needs to make reference to that surface’s design (conceived as such)’ (Chapter 6, Sect. 2). Whereas separation, as John Brown explains, is an experience ‘representing a ﬁgure or scene which in some way and to some degree is distinct from, because incompatible with, the picture’s proper subject’ (Chapter 8, Sect. 2). Both phenomena, I take it, are manifestations of the unity of medium and subject in pictorial perception. It was noted above that, as an effect of this unity, the material elements on the picture’s surface are seen as forming a meaningful whole: inﬂected properties are parts of that whole (as Hopkins notes, they lie in pictorial space), but they are properties that simply resist a description in subject terms (no meaning can be attached to these properties). Separation seeing-in, on the other hand, relates to the fact that, as was noted above, the subject is (ordinarily) seen as construed by the medium—its qualitative character, how the object looks, depends on what materials have been used and how these have been handled by the artist. In separation seeing-in the material elements on the picture’s surface are seen as forming a meaningful whole and, in contrast to inﬂection, they can be given a description in subject terms, but they do not all form subject

144 katerina bantinaki properties that an informed viewer would ascribe to the intended subject according to the norms of pictorial practice. Now, regarding the scope of seeing-in, one may ask: is it necessary that the experience to be had in response to a picture always has the complex content that I have ascribed to it—does the object of sight appear always as a unity of medium properties and meaningful form—or rather, in some cases, can the one be seen separately from the other? My analysis is consistent with, and thus allows us to make sense of, the fact pointed out by Wollheim that in seeing a picture we can focus our attention on either of its aspects: we can focus exclusively on the medium (the materials or the technique that has been used); or we can focus exclusively on the depicted object (for instance, in Freud’s painting of Elizabeth II, we can contemplate the real Queen and facts about her that possibly justify the properties that the painting ascribes to her). But notice that such a focus of attention would be a form of abstraction: so to focus on the medium is to see or think of the medium in separation from the form that it has been used to bring forth—it is, for instance, to see a line as a line, rather than as forming the curve of a picture-body. It was noted with regard to Aristotelian compound substances that any division of the uniﬁed whole generates distinct entities, entities that are independent of what the whole is, or what the whole does, so that they have to be reidentiﬁed. This qualiﬁcation seems to correspond to Wollheim’s claim that when ‘preference for one aspect of the experience gets carried to the point where the other aspect evaporates . . . twofoldness is lost, and then seeing-in succumbs to an altogether different kind of experience’ (1998: 47). The experience changes because the object of awareness changes (in cases of focused attention that Wollheim describes); the object of awareness is no longer a uniﬁed whole, a medium-speciﬁc construal of the subject, but it is a (pure) medium or a (pure) subject. When the object that ﬁgures in our awareness is the picture-of-X, the two components are enmeshed: an experience where the one component is seen or thought of in separation from the other is an experience where the object of awareness is different—different in either its identity or its qualitative character—thus, it is a different experience. But now, if we understand seeing-in as an experience that has the complex content that I have assigned to it, then there would seem to occur a divide in the pictorial domain between pictures that can and pictures that cannot give rise to such an experience, especially trompe l’œil

picture perception as twofold experience

145

pictures that seem to preclude awareness of their medium. Wollheim, who aimed to provide a deﬁnition of depiction in terms of seeing-in, saw this as an unhappy upshot that we are forced to accept, thus leaving trompe l’œil out of the domain of depiction. But this, it has been argued, is a counterintuitive assumption (Levinson 2001: 30 and n. 5). According to Dominic Lopes (1996: 50), for instance, ‘it simply begs the question to assert that twofoldness explains the distinctive phenomenology of our experiences of pictures seen as pictures, if what we mean by seeing pictures as pictures is seeing them in a way that excludes trompe l’œil’. Now, my aim in this chapter has been to offer an explanation of twofoldness—of the unity of a twofold act of perception—leaving aside the question whether seeing-in (understood as a twofold act of perception) can ground a deﬁnition of depiction. Even from this standpoint, however, the question may still arise whether the experience to be had in response to trompe l’œil is ever, or in any sense, twofold in the way speciﬁed here. In relation to this question it is worth noting a fact that is rarely acknowledged in discussions of trompe l’œil: as John Hyman (2006: 132) notes, the belief that trompe l’œil pictures are designed to produce an illusion and sustain it for as long as the viewer sees the painting, is an exaggeration which distorts the aim and the effect of trompe l’œil painting. The play element would be lost and the enjoyment of skill and virtuosity, which trompe l’œil cultivates and caters to, would be frustrated if it were true. That is why, as Ruskin remarks, trompe l’œil invariably ‘has some means of proving at the same time that it is an illusion’.

If this is the case, as I believe it is, then Wollheim unduly excluded trompe l’œil works from the domain of depiction and, with the same sleight of hand, his critics mischaracterize the total experience to be had in response to a trompe l’œil work: once the artefactual character of what is seen is acknowledged by the viewer, once the real-life appearance is thematized as lifelikeness, the technique of the trompe l’œil picture comes to the fore and the medium is thereby evident in the viewer’s experience, which now has the complex character of seeing-in. However, even while the viewer is not aware of the medium, is there any sense in which her experience of the trompe l’œil picture is twofold, so that the trompe l’œil objection to Wollheim’s theory could be met? In Chapter 7 in this volume, Bence Nanay argues that pictorial experience is by necessity twofold in the minimal sense that both the surface and the

146 katerina bantinaki object of the picture are represented in the viewer’s experience (with or without awareness). The experience to be had in response to trompe l’œil pictures is regarded as twofold in that minimal sense—in which case it seems that there is after all a speciﬁcation of twofoldness that can support the project of deﬁnition for the whole spectrum of depiction. Be that as it may, I doubt whether this is a speciﬁcation of twofoldness that would protect Wollheim’s theory from the trompe l’œil objection: through the notion of twofoldness Wollheim aimed to specify the distinctive phenomenology of pictorial seeing; even if the type of twofoldness that Nanay speciﬁes may apply to the seeing appropriate to trompe l’œil pictures, it is not a type of twofoldness that shows up in the phenomenology or the content of the experience but, rather, it concerns the process of pictorial seeing. If in seeing a trompe l’œil picture—at least initially—we are not aware of the medium (even if the medium is visually processed), then the medium is not part of the complex content of the experience, which therefore lacks the distinctive character that Wollheim aimed to attribute to pictorial seeing: the objection that Lopes raises to Wollheim’s account thus persists, and for a theorist interested to meet this objection her best chance is to acknowledge with Hyman that trompe l’œil pictures can give rise to twofold seeing-in, although they may not support such an experience for the duration of our engagement with the picture.

7. The Limits of the Explanation So far we have managed to specify, at least partly, the content of seeing-in, and we have managed to explain why this twofold experience cannot be rightly speciﬁed by means of an analogy with the simple face-to-face experience of its component aspects. One may still object, however, that a comprehensive description of the phenomenology and content of the experience of seeing-in has not yet been provided: if we cannot pursue the above analogy in order to illuminate the character and content of seeing-in, then we are left to reﬂect on our own experience of seeing a picture in order to understand the twofold character of this experience—in particular, the manner in which the two aspects of the experience merge. One may ask, however, whether we are in any better position with regard to ordinary perception. Can we give a full description of the phenomenology

picture perception as twofold experience

147

and content of the experience of seeing, for instance, a concrete house, an object that—as it ﬁgures in our awareness—is a compound of distinct material elements and form? I take it that the answer is no: in that case, too, we need to appeal to our own experience of seeing such a compound object in order to understand how the relevant components merge in the perceptual experience. In both cases, I believe, a complete description of the phenomenology and content of our experience would require something that we do not have so far, and which we would need to have in order to explain further how ordinary perception of X differs from pictorial perception of X: we should be in a position (in which we are not, or at least, not yet) to offer a comprehensive explanation of the way in which meaningful form and, especially, material constitution ﬁgure in the uniﬁed, homogeneous, perceptual awareness of real objects and of pictorial construals of these objects. In that case, the limits to the comprehensiveness of seeing-in are the limits to the comprehensiveness of ordinary perception. If, at the moment, we are happy to accept those limits in the latter case, then we should be happy to accept the same limits in the former case. So are we in a position to meet Budd’s objection that the experience of seeing-in is incomprehensible, and further that we cannot describe its phenomenology and its content? The objection is, at least partly, met: the contrast with the unity of concrete compound substances allows us to make sense of the twofoldness of the experience; the speciﬁcation of its aspects and of their unity allows us to draw at least some conclusions on its content; and although it has been admitted that we cannot give a complete description of the phenomenology of the experience, it has been explained (i) that we are in no better position with regard to ordinary perception, (ii) why such a description cannot be given, and (iii) where we should turn to provide it. I thus contend that we can understand Wollheim’s conception of the twofold character of pictorial perception if we draw a parallel to perhaps less obscure cases of non-pictorial perception: cases where a piece of matter (or distinct material elements) is recognized as a whole with a certain identity, an identity tied to a particular form. In all such cases the perceiver is aware not just of a piece of matter and not abstractly of a form: she is aware of the particular piece of matter as having this form. Pictorial perception is twofold in just this sense. If we can make good sense of the twofoldness of

148 katerina bantinaki ordinary perception, then we can make good sense of the twofoldness of pictorial perception. Although I have not engaged here with the project of deﬁnition of depiction through the notion of twofoldness, where this notion is meant to capture the distinctive character of seeing-in, it may seem nevertheless that my analysis runs counter to this project. I have tried to illuminate the twofoldness of pictorial seeing by drawing on the twofoldness of ordinary perception—the compound character of its (gestalt) objects. This might be seen to cast doubt on Wollheim’s project of deﬁning pictorial representation (partly) through the phenomenon of twofoldness, as here twofoldness is taken to be a feature of ordinary perception, i.e. it is assumed that its objects ﬁgure in the perceiver’s awareness as compounds of matter and form. The project of deﬁnition, however, can still be supported, not by the notion of twofoldness itself, but by the difference in the objects of awareness in a twofold experience (especially the difference in the matterrelated dimension of the experience) in the case that one is seeing a real object and in the case that one is seeing a picture of the object. It is such difference that grounds the distinctive complex content of pictorial seeing. One may want to retort, however, that the difference that as a matter of fact exists between the relevant experiences is not a difference that is reﬂected always in the content of the experience—that ﬁgures in our awareness: we may be aware of seeing a picture while looking, for instance, at a photo of a wooden table, but the medium, one could argue, is not part of the content of the experience—it is only the wooden table, as such, that is part of that content. I doubt, however, whether this is an accurate description of our experience of the photo. Perhaps trompe l’œil aside, seeing a wooden table in the ﬂesh is bound to differ from seeing a wooden table in a photo, even if only in minute detail: there will be textural differences, or differences in luminosity, or differences in spatial awareness (as Brown argues in Chapter 8, Sect. 5, in this volume, ‘pictorial space is perceived as shallower than accords with the subject’). The complex content of the experience of the wooden table thus differs from the complex content of the experience of a photo of a wooden table, and the differences are a direct manifestation of the viewer’s awareness of the constitution of the relevant entities: the organizing inﬂuence of perception acts on different material components (a piece of wood in one case and, for example, an arrangement of colours in the other) and

picture perception as twofold experience

149

the difference is manifest in the viewer’s experience, even if only in its ﬁne-grained content. In Chapter 6 (Sect. 6) in this volume, Hopkins raises the question whether the representation of the subject involved in seeingin is Standard Visual Representation—i.e. whether pictorial perception represents that subject in the way that ordinary seeing represents its objects. Following my analysis, the answer to this question should be negative, to the extent that, trompe l’œil aside, in seeing-in the medium is manifest in the experience of the subject (it is seen as constituting the form of the subject). The interesting question now is: can we specify further how seeing-in relates to ordinary perception—in particular, how seeing X in a picture relates to seeing X face to face? Wollheim thought that an inquiry into the phenomenology of pictorial perception would not be fruitful, so he did not provide us with an explanation that would help us answer this question. Contrary to Wollheim, however, we could elucidate the phenomenology of visual perception in a way that would help us mark the differences between pictorial and ordinary visual perception of an object. The object of pictorial perception, I have argued, is a medium-speciﬁc construal of a certain subject. Such medium-speciﬁc construals present us with aspects of the appearance of the depicted entities. What these aspects are, and how close the correspondence is between the real (or imaginary) entity and its pictorial construal depends on the particular use of the medium, on the constitution of the pictorial entity. To indicate the differences between pictorial and ordinary perception of an object one would have to offer a comprehensive explanation of the way in which meaningful form and, especially, material constitution ﬁgure in perceptual awareness both of real objects and of pictorial construals of these objects. Until this kind of (phenomenological) explanation has been offered, the seeing-in theory, admittedly, will not be complete.

References Aristotle (1924), Metaphysics, trans. W. D. Ross (Oxford: Clarendon Press), . (1996), Poetics, trans. M. Heath (London: Penguin). Arnheim, Rudolph (1969), Visual Thinking (Berkeley: University of California Press). Budd, Malcolm (1992), ‘On Looking at a Picture’, in J. Hopkins and A. Savile (eds), Psychoanalysis, Mind and Art (Oxford: Oxford University Press).

150 katerina bantinaki Hopkins, Robert (1998), Picture, Image and Experience (Cambridge: Cambridge University). Hyman, John (2000), ‘Pictorial Art and Visual Experience’, British Journal of Aesthetics, 40/1: 21–45. (2006), The Objective Eye: Color, Form, and Reality in the Theory of Art (Chicago: University of Chicago Press). Lamarque, Peter (2000), ‘Objects of Interpretation’, in J. Margolis and T. Rockmore (eds), The Philosophy of Criticism (Oxford: Blackwell). Levinson, Jerry (2001), ‘Wollheim on Pictorial Representation’, in R. van Gerwen (ed.), Richard Wollheim on the Art of Painting (Cambridge: Cambridge University Press). Lopes, Dominique McIver (1996), Understanding Pictures (Oxford: Oxford University Press). Scaltsas, Theodore (2001), ‘Substantial Holism’, in T. Scaltsas, D. Charles, and M. L. Gill (eds), Unity, Identity, and Explanation in Aristotle’s Metaphysics (Oxford: Oxford University Press). Thomasson, Amy (2005), ‘Ingarden and the Ontology of Cultural Objects’, in A. Chrudzimski (ed.), Existence, Culture, Persons: The Ontology of Roman Ingarden (Frankfurt: Ontos). Wollheim, Richard (1996), ‘Seeing-As, Seeing-In, and Pictorial Representation’, in Wollheim, Art and its Objects (Cambridge: Cambridge University Press). (1998), Painting as an Art (London: Thames & Hudson). (2001), ‘On Pictorial Representation’, in R. van Gerwen (ed.), Richard Wollheim on the Art of Painting (Cambridge: Cambridge University Press). (2003), ‘What Makes Representational Painting Truly Visual?’, Proceedings of the Aristotelian Society, suppl. vol., 77/1: 131–47.

6 Inﬂected Pictorial Experience Its Treatment and Signiﬁcance ROBERT HOPKINS

According to some, our experience of pictures is sometimes ‘inﬂected’ by our awareness of properties of the picture’s surface (Podro 1991, 1998; Lopes 2005). As a result, what we see in these pictures are things that could not be seen face to face. Moreover, our experience of these pictures exhibits a phenomenology that differs from that of any possible experience of the pictures’ objects in the ﬂesh. All this is supposed to be of central importance for pictorial aesthetics. It helps us understand why we value pictures (why we are interested in looking at them) when we would not value seeing their objects directly. Thanks to inﬂection, pictures offer us a world we could not encounter in any other way and forms of experience that cannot be enjoyed outside the pictorial realm. Inﬂection thus promises to be of some signiﬁcance. But what exactly is the phenomenon supposed to be, and does it really occur? Once clear on those matters, we can ask whether it is really as important as claimed.

1. What Is Pictorial Experience? First a preliminary. The notion of inﬂection is dependent on that of pictorial experience. It is pictorial experience, or seeing-in as I shall also call it, that is, under the right circumstances, inﬂected. But what is seeing-in? Seeing-in is a special experience that pictures offer us. It is in experiencing pictures in this way that we grasp what they depict. Consider, for instance, I am grateful to Katalin Farkas, Bence Nanay, and the editors for illuminating discussion.

152 robert hopkins Rembrandt’s pen and ink sketch of the pastor Jan Cornelisz Sylvius (Figure 6.1). If I go to the British Museum and look at this drawing, I see ink marks on a piece of yellowed paper. But I also (in some sense) see a man, holding his left hand outwards, as if engaged in conversation. My experience of the picture thus has two dimensions to its content. It represents what is before me, a marked surface; and it represents something else, a man with certain features. When I see one thing as a picture of something else, my experience has this double content. This is how I know that a picture is before me, and how I know what the picture’s own content is, what it depicts. Unless my experience represented a surface as before me, I would take myself to be confronted, not with a picture, but with a gesturing man. And unless my experience in some way represented a man, I would take what is before me to be merely a marked surface, and not a picture of something else. It is hard to say more about seeing-in without controversy. There are various detailed accounts of what the experience amounts to. However, we need to proceed for as long as possible without choosing between them. Later (in Sections 4–6) it will emerge that disagreements over the nature of pictorial experience have consequences for the signiﬁcance of inﬂection. Since our job is in part to establish what that signiﬁcance might be, we cannot afford to beg any questions by rejecting accounts of seeingin without argument. Still, we can usefully lay the ground for that later discussion by identifying the key issues on which accounts of pictorial experience disagree. As noted, seeing-in standardly has two dimensions to its content. The key disagreements concern how it comes to exhibit these two dimensions, the kind of content each dimension involves, and whether both dimensions are necessary. On the ﬁrst issue, some hold that seeing-in is a combination of two experiences, one representing the surface, the other representing the depicted object. Seeing-in thus has a double content because it is made up of two component experiences, each of which has a single content. That combination might be sequential, each experience following the other (Gombrich 1961); or simultaneous, both occurring at the same time (Wollheim 1968; Lopes 2005). Others suggest that the two dimensions are embodied, not in distinct component experiences, but in distinct ‘aspects’ or ‘folds’ of a single experience (Wollheim 1987). Still others reject the idea that seeing-in has any components corresponding to the two dimensions of content (Walton

inflected pictorial experience

153

Figure 6.1. Rembrandt, Jan Cornelisz. Sylvius (c.1646), sketch for the posthumous etching

154 robert hopkins 1990; Hopkins 1998). Those two dimensions are present in seeing-in, but it does not decompose into discrete elements to which the dimensions can be assigned. Rather, the two are abstractions from the complex, structured content of pictorial experience as a whole. Turning to the second issue, we need to say something about the way in which the two dimensions of the experience’s content represent their respective objects, the marked surface and the depicted scene. The marked surface is represented as actually before one. It is represented in seeing-in in the same way that vision in general represents its objects. Put another way, seeing-in really is a kind of seeing, and what is seen is the marked surface. But what of the scene-oriented dimension of content? We might say the same there: the depicted scene is represented in whatever way ordinary seeing represents its objects, i.e. it is given as really before me. And we might say this even though the depicted scene is not in fact before me (so there is at most the illusion of seeing it), and, in the context of pictorial experience as a whole, I know that it is not (so there is not even illusion, if that involves being misled). However, this is not the only option. We might, to take just one alternative, say that the scene visible in the surface is represented in something like the way involved in visualizing an item. These ﬁrst two issues are the deepest on which accounts of seeing-in disagree, and we will have to return to them below (Sections 5–6). The third, in contrast, we can set aside. Must seeing-in exhibit both dimensions of content? What of trompe l’œil, in which a picture is mistaken for what it depicts? In this case we seem to see, not a picture of that object, but the object itself. If so, our experience hardly involves the representation of a marked surface. If this counts as pictorial experience, that experience need not involve more than the scene-oriented dimension of content. Opinions divide on whether experience of trompe l’œil counts as seeingin (Gombrich 1961; Wollheim 1987; Lopes 2005). We can avoid that issue. Our topic is inﬂected pictorial experience, and nobody thinks that experience of trompe l’œil is inﬂected. If one is not aware of the marked surface before one, features of that surface cannot ‘inﬂect’ one’s experience of the picture. Of course, when we are fooled by such a picture, there is a sense in which we are aware of the marks that compose it: if those marks weren’t in front of us, we wouldn’t seem to see whatever it is the picture depicts. But this is not the sort of awareness we have of most pictures. In most cases, even as we see things in the marked surface, we see the marks as

inflected pictorial experience

155

marks. Our experience of the picture represents a marked surface as before us. This is what is missing in trompe l’œil, and yet what inﬂection requires. In what follows, then, we can ignore trompe l’œil. Indeed, we can focus our attention further. Dominic Lopes (2005: 25) usefully distinguishes two sorts of pictorial feature. A picture’s design comprises those features in virtue of which it supports seeing-in. If, for instance, we see a horse in a drawing in virtue of the way the pencil marks are shaped, then the shape of the marks is an aspect of the picture’s design. Not every feature of a surface plays this role. The grain of the paper, for instance, may make no difference to what is seen in it. As Lopes notes, we may, while seeing things in surfaces, be aware of their design—we may see certain features of the surface as the features responsible for our seeing the scene in it. The central cases of inﬂection are those in which this awareness plays a key role in transforming other aspects of pictorial experience. In what follows, then, we need only consider seeing-in that includes awareness of design. Given this restriction, we can cut through the various controversies surrounding seeing-in to the following common ground. In the cases that concern us, we can distinguish within pictorial experience the picture and its design, as that experience reveals them to us; and the depicted object or scene with its properties, again as revealed in the experience. More concisely, pictorial experience itself seems to distinguish the picture from what is seen in it. The natural expectation, given what we have said so far, is that these two items will need characterizing in distinct terms. Of course, what is seen in the picture depends on the design. No doubt sometimes the properties of the one will match those of the other—as when a scarlet robe is visible in a part of painted canvas that is itself scarlet. But such dependence and overlap do not undermine the key expectation, that in principle the experienced design and the thing seen in it can be characterized without reference to each other. It is precisely this expectation that inﬂection challenges.

2. What Is Inﬂection? So to the ﬁrst of the questions it is our goal to answer. Proponents of inﬂection offer various characterizations of the phenomenon. What follows

156 robert hopkins is a series of statements that might capture what they have in mind. By examining them in the light of the role inﬂection is intended to play, we will be able to sift the central formulations from the more peripheral. The simplest place to start is with (1) Sometimes, our experience of pictures is inﬂected by awareness of properties of the picture’s design. This might be a useful slogan, but only as a gloss on other claims. For what does ‘inﬂected’ mean? All it clearly suggests is some causal inﬂuence, of awareness of design on pictorial experience as a whole. The claim that seeing-in involves such inﬂuence is a truism. For that is what distinguishes seeing-in from mere fantasy or hallucination. You only see a man in Rembrandt’s sketch because you see the dark lines composing the ﬁgure. Suppose you were aware of the lines and aware of the ﬁgure, but there was no causal connection between your awareness of the two. Then you would be seeing the design while merely visualizing the ﬁgure, or seeing the former while hallucinating the latter, or some such. What you could not be doing is seeing the man in the lines. Everyone should accept this basic truth about seeing-in, quite independently of the issue of inﬂection. Since all (1) clearly asserts is this uncontroversial thought, taken alone it does not clearly describe a phenomenon more speciﬁc than seeing-in itself. Perhaps the intended meaning can be revealed by supplementing (1) with other claims. Consider (2) Sometimes, the phenomenology of seeing something in a surface differs from that of seeing that object face to face (Lopes 2005: 128). If the Rembrandt is an instance of inﬂection, (2) claims that what it is like to see a gesturing man in it differs from what it is like to see such a man in the ﬂesh. This is certainly not truistic. Below (Section 5) it will emerge that not everyone should accept (2) in the sense intended. However, we are not yet in a position to explain what is controversial about it. For the moment, we can continue our attempt to understand what inﬂection is supposed to be by turning to another claim, one supposed to explain why (2) is true: (3) Sometimes, what is seen in a surface is an object with properties that it could not be seen to have face to face. (Podro 1991: 173; 1998: 28; Lopes 2005: 40, 128–9)

inflected pictorial experience

157

What exactly does (3) claim? One reading should be set aside. It is obvious that in some pictures we see objects with properties they do not in fact have (think of the smooth complexions visible in airbrushed portraits of ageing celebrities). It is also obvious that in some pictures we see objects with properties they could not have (think of Escher’s impossible staircases). Since one can see face to face only what exists to be seen, the states of affairs visible in these pictures cannot (in the Escher case, could not) be seen in the ﬂesh. Should (3) be read in such a way that these pictures count as examples of inﬂection? I think not. Despite their distinctive subject matter, our experience of these pictures is ordinary seeing-in. There is no interesting sense in which it is ‘inﬂected’. Further, if inﬂection does reduce to seeing non-actual or impossible states of affairs in pictures, its aesthetic signiﬁcance is no greater than that of the depiction of such things. Its proponents clearly have a more important phenomenon in mind. How, then, should (3) be read? The previous reading turned on how the world is. If it doesn’t contain objects like those depicted, then it is not possible to see such objects in the ﬂesh. A better reading takes (3) instead to concern the ways that, in experience, things might seem to be. (3) concerns, not the accuracy, but the content of face-to-face seeing. What it should claim is that inﬂected seeing-in offers us objects with properties that face-to-face experience could not even seem to put before us: (3∗ ) Sometimes, what is seen in the surface is an object with properties that no experience purporting to be of that object seen face to face could represent. This is progress. The celebrity portrait and Escher no longer count as cases of inﬂection. For, while as things are we could not have veridical face-to-face experiences of such things, we could certainly seem to see them.¹ Those cases thus no longer threaten to render inﬂection mundane: cases of inﬂection will be more exotic than these pictures. However, (3∗ ) does raise a question. What are the properties it describes? What sort of property can be seen in a surface, but cannot even seem to be seen in any experience (that purports to be) of the relevant object face to face? Here is the answer that ﬁts best with the sorts of things proponents of inﬂection say, and the sorts of things they want to use the notion to do. ¹ Indeed, Escher’s pictures of endless staircases were inspired by three-dimensional models presenting paradoxical appearances similar to those of the objects in those pictures (Penrose and Penrose 1958).

158 robert hopkins It is that the relevant properties have to be described by reference to the picture’s design. In inﬂected pictorial experience, we see in the surface properties that need characterizing, in part, by reference to properties of the surface itself. In particular, they need characterizing by reference to the very properties of the surface that sustain seeing-in, i.e. the surface’s design. It helps to crystallize the proposal in some deﬁnitions. Let’s call the properties it postulates as seen in surfaces inﬂected properties. We deﬁne them as follows: Inﬂected properties =df those properties visible in a surface, a full characterization of which needs to make reference to that surface’s design (conceived as such). Then the phenomenon of inﬂection is that what is seen in a surface sometimes includes inﬂected properties. More explicitly: (4) Sometimes, what is seen in a surface includes properties a full characterization of which needs to make reference to that surface’s design (conceived as such). This proposal goes beyond the truistic. As I noted at the end of the last section, the natural expectation is that, while design might determine what is visible in a surface, the two can be described independently. The idea here is that this is not always so. When inﬂection occurs, features of the surface’s design must be cited in characterizing the properties visible in it. Thus, awareness of the design not only causes one to see something in it; it is awareness of features that partly constitute the world seen in. That world ends up possessing features that in some way involve the design itself. To anticipate the example explored in the next section, what is seen in the Rembrandt’s ink-marked surface is not just a hand, but a hand itself composed of ink strokes. This way of giving substantial content to the idea of inﬂection also squares with other things its proponents say. They talk of features of design being ‘recruited’ to the scene visible in the picture (Podro 1998: 13, 26). They say that the scene itself somehow straddles the boundary between the marked surface before the viewer and the other world conveyed by those marks (Podro 1998: 17, 28). Or they say that, as well as seeing the scene in the design, we see the design in the scene (Podro 1991: 172). The proposal offers a way to cash out these metaphors.

inflected pictorial experience

159

The proposal also makes sense of (3∗ ). If the properties of which it speaks are inﬂected properties, it will indeed be impossible for them to be represented in face-to-face experience. Inﬂected properties need characterizing, in part, by reference to a surface’s design. In these cases, therefore, what is seen in the surface itself makes reference to the fact that a surface is before the viewer, bearing a certain design in virtue of which a scene is visible in it. Thus, no experience could represent such properties while purporting to be a case of seeing the object face to face. The role of design in characterizing the properties of the object guarantees that it is not the sort of thing that can seem to be seen in the ﬂesh.² A ﬁnal advantage of the proposal is that, spelled out carefully, it enables us to distinguish inﬂection from two other phenomena. One is overlap. Sometimes the properties visible in a surface and the properties of its design match. To reuse our earlier example, a scarlet robe may be seen in a patch of paint that is itself scarlet. In such cases, the scene visible in the surface needs characterizing by reference to properties that, in fact, are also properties of the design. The idea here is more demanding. Inﬂection occurs when properties visible in the surface need characterizing by reference to properties of the design conceived as such. Mere overlap does not count. The other phenomenon is nested seeing-in. Sometimes what we see in a picture is itself a picture in which things are seen. For instance, in the picture before us we see a room, with a picture on the wall; and in that inner picture we see a country dance. In this case too, what is seen in the design needs characterizing by reference to a design, conceived as such. For a design, with something visible in it, is precisely what we see in this picture. The phenomenon of nested pictorial experience is both interesting in itself and in some respects similar to inﬂection. Nonetheless, the two are distinct. We can keep them apart provided we get the current proposal just right. In inﬂection, but not nested seeing-in, what is seen in the design ² In Hopkins (1998: 134; cf. 28–30) I make the unqualiﬁed claim that only what can be represented in face-to-face visual experience can be seen in a surface. If inﬂection is deﬁned as I am suggesting and if, as I go on to argue, it does occur, that claim is false. However, it is close to the truth. Nothing here casts doubt on my thought that properties only given to the other senses, or properties not given to the senses at all, cannot be seen in pictures. Moreover, the exceptions to the unqualiﬁed claim themselves essentially involve the apparatus of seeing-in, and in particular design. Thus, inﬂection extends the range of what can be seen-in beyond that of what can (seem to) be seen face to face, but only by elaborating on a more basic phenomenon, uninﬂected seeing-in, which does ﬁt my unqualiﬁed claim.

160 robert hopkins before us needs characterizing, in part, by reference to that very design, conceived as such. (In nested seeing-in, what is seen in the design before us needs characterizing by reference, not to that design, but to a design visible in that design.) Thus, I suggest, (4) offers the deﬁnitive formulation of the notion of inﬂection. From it (3∗ ) follows, since inﬂected properties are indeed properties that no experience purporting to be of the relevant object faceto-face could represent. If (2) (in the sense the proponent of inﬂection intends) is true at all, it is so because (3∗ ) is, and so because (4) is. And (4) offers the non-truistic content of (1).³

3. Is Any Pictorial Experience Inﬂected? A proper understanding of inﬂection in hand, we can ask whether the phenomenon ever obtains. However, to address this question we need a criterion for what is seen in a surface. After all, on our understanding inﬂection precisely involves intermingling of properties of the surface’s design with properties seen in it. To test whether inﬂection occurs, it is thus not enough to be able to identify which experiences count as seeing-in, and to identify which properties those experiences represent. We also need a way to sort the properties represented as belonging to the surface from those seen in it. Moreover, since inﬂected properties straddle that divide, we need the criterion for what is seen-in to be open to the possibility that some features might ﬁgure on both sides. Unfortunately, despite all the argument about how to characterize seeing-in, there has been very little discussion of how to identify what is ³ In Ch. 7 in this volume, Bence Nanay offers a different account of inﬂection. For him, inﬂected pictorial experience is in essence awareness of emergence, of how the design enables us to see the depicted object in it. As Nanay notes (Sect. 5), the experience he describes ﬁts a key element in my deﬁnition of inﬂection: in being aware of how scene emerges from design, we are indeed aware of features of the picture that require characterizing, in part, by reference to design. (Nanay calls these features ‘design-scene properties’.) However, these are not inﬂected properties in my sense. For inﬂected properties are properties seen in the surface. It is very odd to think that what I see in a surface includes how the surface enables me to see that thing in it. That is indeed something of which seeing-in can make me aware—and so emergence is something we can appreciate. But my awareness of emergence is given by seeing-in as a whole, not the scene-oriented dimension of its content. Emergence, while important to the aesthetics of pictures (as acknowledged below), is not to be confused with inﬂection.

inflected pictorial experience

161

seen in a given surface (though see Wollheim 2001: 23–4; Hopkins n.d.) Nonetheless, given our restriction to cases in which the subject is aware of design, there are two obvious candidates. Since one criterion only functions to supplement the other, we can legitimately exploit both. The ﬁrst criterion appeals to elements in the content of one’s pictorial experience that do not appear as properties of what is before one. Suppose my experience presents me with a trapezoidal block of colour, and yet also in some way represents what lies in that direction as square (Walton 1990: 54–5). If it represents what is before me as a trapezium, but does not represent a square as before me, then the trapezium is seen, and the square is seen in it. (I see in the trapezium a square tilted away from me.) The second criterion steps in when the ﬁrst fails to deliver a result. It appeals to the way experience groups the properties and objects it represents. Suppose that considered in isolation squareness has as much claim to be represented before me as the property of being a trapezium. (So the ﬁrst criterion lets us down.) It may, nonetheless, be that one of the two is represented as belonging with some further property clearly not given as before me. (Perhaps squareness goes with being tilted, but nothing before me looks tilted.) Provided only one of the pair is so represented, we can sort them by ﬁrst identifying what they are grouped with, and then applying the ﬁrst criterion to the properties so grouped. Consider, then, one of Podro’s most convincing examples of inﬂection, the now familiar Rembrandt sketch.⁴ (I reproduce the ﬁnished engraving too (Figure 6.2), since it helps one to appreciate what is going on in the sketch.) As Podro notes, Rembrandt captures Sylvius’ hand gesture particularly well. The upward thrust of the hand is clearly visible. But to see it one must see the ink strokes which depict the hand as themselves driving upwards. Indeed, the hand itself seems to be both body part and rising splash of ink. The movement of the ink is grouped with the movement of the hand (our second criterion) and the movement of the hand is not seen as before one (our ﬁrst).⁵ Thus, what is seen in this picture is a hand ⁴ Podro discusses the example in his (1998: 16–17), although he does not make any of the points I go on to. ⁵ We might think that the movement of the ink ﬁts the ﬁrst criterion too, since I do not see moving ink as before me. However, this depends on how we understand that property. Should we take it literally, in which case it is a property the ink is only imagined to have (Podro 1998: 8)? Or should we take it as a metaphorical description of some property the ink really possesses, and which is seen as before me?

162 robert hopkins

Figure 6.2. Rembrandt, Jan Cornelisz. Sylvius (1646), posthumous etching

inflected pictorial experience

163

composed of rising ink. Since what is seen-in needs characterizing in part by reference to properties of the picture’s design, prima facie what is seen in the sketch is an inﬂected property. We could resist this conclusion only by dividing the elements seen, placing the upward-turned hand with the scene visible in the marks, the inky splash with the picture’s design. The claim would then be that, while we see the hand in the inky splash, the two are not more closely related than that. But a further feature of the picture blocks this move. To grasp the full effect of the gesture’s upward thrust, one needs also to see it as contrasting with the adjacent downward cascade of Sylvius’ robe. Again, the fall of the cloth seems shot through with the movement of the ink that represents it. Now try seeing the two sets of ink marks, the downward driving ones that represent the robe and the rising ones representing the hand, as merely features of the design. To do so dampens the upward thrust of the hand. The strokes composing it now rise in the same plane as that in which the strokes composing the robe descend. The movements of the two are in tension, and work to cancel each other out. A proper visual appreciation of the upward thrust of the hand requires us to avoid this tension. That can only be achieved by seeing the hand strokes in a very particular way. They pick up on the downward thrust of the robe strokes, as if they were the same ﬂuid rebounding off a resistant surface. But they do not ﬁght against the motion of the robe, by ﬂowing back in the direction from which that motion came. Rather, they continue its force by ﬂowing not only upwards but outwards, towards the viewer. Thus, fully to see the upward-turned hand in the picture, one must see the movement of the ink strokes as occurring not on the marked surface, but in the space visible therein. The properties of the hand cannot be fully characterized without reference to this movement seen in the surface. But since it is movement of ink strokes, the hand cannot be fully characterized without reference to design. At least one of the properties visible in this picture is inﬂected. There is thus at least one example of inﬂection. If the reader is persuaded of this, she will probably be ready to accept that there are others. The proponents of inﬂection sometimes make much bolder claims. Lopes says that whenever pictorial experience includes awareness of design, the latter inﬂects the former (2005: 128–9). I won’t attempt to settle whether that stronger claim is true.

164 robert hopkins

4. Does Inﬂection Matter? We know what inﬂection is, and we know it occurs at least sometimes. It is time to ask what its signiﬁcance might be. It is supposed to be important to the aesthetics of pictures. I consider three accounts of how it might be. At the heart of our account of inﬂection lies (3∗ ), the idea that we sometimes see in pictures things of a kind that could not be seen face to face; and (4), the idea that the difference lies in inﬂected properties: properties that, although seen in pictures, require characterizing by reference to the design in which they are seen. The ﬁrst account of inﬂection’s signiﬁcance locates it in these same claims. The contents of seeing-in outstrip those of seeing face to face, since only seeing-in represents inﬂected properties. Seeing-in is central to our appreciation of representational pictures, it is the experience through which we engage with whatever they have to offer aesthetically. There is more to seeing-in than what is seen in the surface—it also involves awareness of the surface itself, thus allowing us to appreciate such features as the way the brush has been handled, the delicate colours that have been used, the vigour with which the whole has been organized, and so on. But a key part of our experience of pictures is nonetheless a matter of what objects and properties are visible in them. Any expansion in what can be seen-in thus bears directly on what there is to appreciate in looking at pictures. Since the phenomenon of inﬂection constitutes just such an expansion, it is of signiﬁcance for pictorial aesthetics. While plausible, this only establishes so much. It does not obviously give inﬂection the central importance its proponents take it to have. Yes, what we see in pictures is an important element in what there is to appreciate in them. Why, though, is it so important that what can be seen in them differs from what we can (seem to) see face to face? And why does it matter that inﬂected properties in particular can ﬁll the former role, but not the latter? Absent answers to these questions, the ﬁrst account threatens to leave inﬂection not so much holding the key to pictorial aesthetics, as an interesting curiosity. In search of some more substantial signiﬁcance for inﬂection, we might turn to the idea of emergence, of how what is seen in a surface is sustained by that surface’s design. The challenge in making representational pictures is precisely to tap and to develop this phenomenon. A skilled artist is in

inflected pictorial experience

165

part one who can elicit a rich world visible in the surface from a relatively limited design—someone able to convey a great deal with scant resources. Appreciating pictures is in part appreciating how this has been achieved (Wollheim 1987). Indeed, this is central to appreciating them as pictures, since it is only in pictures that those resources (design) are used to produce that effect (seeing-in). The second account of inﬂection’s importance takes it to lie in the way it allows us to appreciate emergence. After all, inﬂection precisely straddles the divide between design and the world visible in it. The idea is that in doing so it offers us the opportunity better to appreciate how the one emerges from the other.⁶ The problem with this proposal is that, with or without inﬂection, the subject already has all she needs to appreciate emergence. Remember we are restricting our attention to cases in which seeing-in involves awareness of design. So the subject already both sees something in the surface and sees some of the surface’s properties as responsible for that thing being seen therein. What more does she need to appreciate how the one sustains the other? Of course, it is possible to see design and to see something in it without seeing how the design enables one to see that in it. But it is not clear that inﬂection ﬁlls this gap. When pictorial experience is inﬂected, in the surface one sees properties that in some way involve the picture’s design. Why should that phenomenon not coexist with failure to appreciate how the design supports seeing the scene in it? Indeed, why think that inﬂection even increases one’s chances of gaining that appreciation, let alone guarantees it? On this proposal, then, inﬂection is at best one possible route to a beneﬁt that might equally be gained by other means. If inﬂection is to be clearly more than a curiosity (as on the ﬁrst account) or an optional means to a worthwhile end (as on the second), a certain assumption has to be made. This is that the task of an aesthetics of pictures is in key part to explain why we value looking at them in ways we do not value seeing objects—perhaps the very same objects as those depicted—in the ﬂesh. That immediately gives central importance to any contrasts between seeing-in and ordinary seeing. Dominic Lopes uses this assumption to frame what he calls the ‘puzzle of mimesis’, a puzzle he then uses the notion of inﬂection to solve. This provides the third account of ⁶ Above (n. 3) I discussed Nanay’s idea that inﬂected experience simply is experience of emergence. Here the suggestion is rather that, though the two experiences are distinct, the former opens the way to the latter.

166 robert hopkins inﬂection’s importance that I consider. Since it is easily the best developed, it merits longer discussion than the other two. Indeed, in one way or another, the rest of the chapter will be devoted to assessing Lopes’s claims. Lopes’s puzzle stems from two claims: Pictorial Evaluation Thesis. In part, to evaluate a picture as a picture is to evaluate it as eliciting experiences of the picture itself and as of the scene it depicts. (2005: 4) Mimesis Thesis. Pictures typically elicit experiences as of the scenes they depict, which experiences resemble, in important respects, face-to-face experiences of the same scenes. (2005: 12)

The puzzle is to reconcile these claims with the fact that we do indeed often evaluate pictures in ways in which we would not evaluate seeing their objects face to face (Lopes 2005: 20–1). For instance, we value Van Gogh’s painting of a pair of old boots, even though we would not value seeing such a pair of boots in the ﬂesh (Schier 1993). But how can that be if, as the pictorial evaluation thesis claims, we evaluate pictures in part for offering us experiences as of what they depict, and, as the mimesis thesis claims, those experiences are signiﬁcantly like the experiences of seeing those scenes in the ﬂesh? Of course, the evaluation thesis leaves open that we value pictures for other reasons too—for presenting us with marvellously coloured and skilfully marked surfaces, for instance. We might hope to locate their value over and above that of their objects in features such as this. But, as Lopes notes (2005: 22–3), this won’t do. The value of the Van Gogh does not reduce to that of looking at old boots plus that of looking at a canvas that is skilfully marked, marvellously coloured, and so forth. Even qua presenting us with boots, it seems valuable as the boots themselves would not be. The solution, says Lopes, lies in inﬂection (2005: 192). (For discussion of the other elements Lopes invokes, and defence of the claim that inﬂection must play the key role, see Hopkins 2008.) Although the experiences elicited by pictures, as of what they depict, do indeed signiﬁcantly resemble experiences of those objects in the ﬂesh, nonetheless the two also differ signiﬁcantly. For our experience of pictures can be inﬂected. When it is, it involves experience as of objects that has a different phenomenology from any experience of those objects face to face. This is what (2) claims. And this holds because we see in those pictures those objects with properties

inflected pictorial experience

167

they could not be seen to have in the ﬂesh. This is (3∗ ), which, I argued, is best understood in terms of (4). We can see inﬂected properties in pictures, and those properties cannot be represented in (what purports to be) experience face to face. No surprise, then, that we value the two sorts of visual experience differently. They have different natures, and the worlds they offer us do not match more than in part. Inﬂection thus allows us to pinpoint the key differences between the two experiences, and to make sense in those terms of their differing value. If the puzzle of mimesis is genuine, it presents a serious challenge for pictorial aesthetics. Inﬂection solves the puzzle, at least for inﬂected seeingin (and remember Lopes thinks that almost all relevant seeing-in is inﬂected). Thus, if the puzzle is genuine, inﬂection is of considerable signiﬁcance aesthetically. The problem with this defence of the phenomenon’s interest is that it is contingent on Lopes’s views about pictorial experience itself. It is hard to make sense of either the puzzle or inﬂection’s solution to it unless one characterizes seeing-in as Lopes does. The next section explains why.

5. Unitary versus Divisive Accounts of Pictorial Experience Lopes’s way of framing the puzzle reﬂects his account of seeing-in. He takes pictorial experience, at least in the cases that interest us, to involve two experiences that occur simultaneously. One is experience of the picture’s design, the other experience of the scene visible in that design (2005, Ch. 1). Above I suggested that any account of pictorial experience should accept that, in the cases that concern us, there are two dimensions to its content. One dimension captures the design, the other the scene visible in it. What Lopes adds to this commonplace is the idea that each dimension can be identiﬁed with the content of a distinct experience. Pictorial experience has those two dimensions because it is composed of two experiences, one representing the design, the other the scene. This view of pictorial experience underpins the puzzle, and its solution by appeal to inﬂection, in several ways. It is exploited by the two claims which frame the puzzle. When the Pictorial Evaluation Thesis speaks of pictures’ eliciting ‘experiences of the picture itself and as of the scene it depicts’, the experiences in question are the two just described. It

168 robert hopkins is the second, scene-oriented experience that the Mimesis Thesis claims resembles in signiﬁcant respects seeing the scene face to face. It is only that scene-oriented experience, experience ‘as of’ the depicted object, that Lopes calls ‘seeing-in’. (I, in contrast, throughout use the term to refer to pictorial experience per se, and thus to the whole of which, according to Lopes, that scene-oriented experience is merely a part.) This terminology, and the account of pictorial experience it presupposes, also dictates how we should interpret Lopes’s solution to the puzzle. When he asserts (2), that inﬂected seeing-in has a phenomenology different from that of seeing the relevant object face to face (2005: 128), he really means the following: (2∗ ) Sometimes, the phenomenology of one of the two experiences composing pictorial experience, that ‘as of’ the picture’s object, differs from the phenomenology of seeing that object face to face. Given how the puzzle and its solution are shot through with Lopes’s views about pictorial experience, we need to consider whether they make sense from the perspective of other accounts of seeing-in. Here’s one alternative. Seeing-in is experienced resemblance, in some respect. (Let’s not worry here about what that respect might be. Answers can be found in Peacocke 1987, Budd 1992, or Hopkins 1998.) To see something in a design is to experience the design as resembling that thing in that respect. In this account, seeing-in is a single experience with a complex content. That content is complex in that it has a particular structure: this resembles that in such-and-such respect. The two dimensions of the content of pictorial experience emerge from this structure. The design-oriented dimension reﬂects what is seen as resembling something else. The sceneoriented dimension reﬂects what the design is experienced as resembling. But there is no sense in which these two dimensions reside in distinct experiences, or anything like them. It is not, for instance, as if experiencing resemblance involves seeing one thing while visualizing another. One might, I suppose, try to understand it as having that structure, but those who offer experienced resemblance accounts of seeing-in have not. From the perspective of this view, neither the puzzle of mimesis nor the solution to it in terms of inﬂection makes sense. A key element in the puzzle, the Mimesis Thesis, goes missing. That talks of experience ‘as of’ the depicted scene, one of the two experiences supposed to constitute

inflected pictorial experience

169

seeing-in. Since the resemblance view rejects the idea of such components, it is forced to revise the Thesis. The simplest revision would be this: Revised Mimesis Thesis. Pictures typically elicit experiences of seeing in them the scenes they depict, which experiences resemble, in important respects, face-to-face experiences of the same scenes. But the truth of this is far from obvious. The resemblance view claims that seeing-in is an experience of likeness. It in no way follows that seeing-in is itself like any other experience. Moreover, given the complex structure the resemblance view ascribes to seeing-in, there are at least some prominent differences between it and seeing face to face. So the view is unlikely to accept the Mimesis Thesis, even in the Revised form. But without the Mimesis Thesis, the puzzle does not arise. And the solution fares worse. At least, it does in so far as it appeals to (2) or (2∗ ). For while (2) is true, in the experienced resemblance view, it is so in every case of seeing-in, and not merely those cases that involve inﬂection. Every instance of seeing-in differs phenomenologically from seeing the relevant object in the ﬂesh, for only the former is an experience of something else as resembling that object. Of course, the proponent of inﬂection meant (2) to capture a truth that holds for only some pictorial experience. Hence what Lopes really meant by (2) was (2∗ ). But (2∗ ) is certainly not true from the perspective of experienced resemblance. Since it concerns a non-existent experience, i.e. one of the two experiences that together are supposed to supply the components of seeing-in, its truth value is that of other statements involving empty referring terms. Thus, Lopes’s claims for the signiﬁcance of inﬂection depend on his wider view of seeing-in. Certainly, if he held the experienced resemblance view, he could not make the claims for inﬂection that he does. Does this point hold more generally? The experienced resemblance view is not the only account of seeing-in to eschew anything like the structure Lopes sees it as having. Kendall Walton’s view (Walton 1990) is, I think, best read as similarly antipathetic to that idea. But equally, it is not clear Lopes needs precisely the view above if the puzzle and his solution are to ﬂy. Suppose, for instance, he adopted something like Wollheim’s later view (Wollheim 1987), in which pictorial experience involves, not two component experiences, but two ‘folds’ or ‘aspects’ of a single experience. Wollheim takes each of those folds to be analogous, albeit in unspeciﬁable

170 robert hopkins ways, to other experiences. The fold in which the design is given is analogous to seeing the picture without seeing anything in it. That in which the scene is given is analogous to seeing that scene face to face (1987, Ch. 2, sect. b). Prima facie, we could rephrase the Mimesis and Pictorial Evaluation Theses in these terms. We could say that the scene-directed fold of pictorial experience is signiﬁcantly like seeing the scene in the ﬂesh, and an important element in what is valuable about seeing pictures. That looks enough for the puzzle to run. So ascribing to inﬂection the signiﬁcance Lopes claims for it does not depend on the details of his account. What, then, is required of an account of seeing-in if there is to be a puzzle of mimesis for inﬂection to solve, and a satisfying solution in terms of it? The crux, I suggest, lies in whether the distinction between the two dimensions of content in pictorial experience is taken to correspond to any further divide in its nature. Lopes and Wollheim think it does. To the distinction between dimensions, there corresponds a further distinction with psychological reality—be it between component experiences, or component ‘folds’ of a single experience. The experienced resemblance theorists and Walton think that it does not. For them, the two dimensions are abstractions from an experience that, however else it divides, does not divide further along those lines. Call these two sorts of account of seeing-in ‘divisive’ and ‘unitary’ respectively. The thought is that if unitary accounts are correct, inﬂection cannot have the signiﬁcance Lopes claims. That signiﬁcance requires one’s account to be divisive—though it is a further question whether every divisive account will allow for it.⁷ Unitary accounts are perfectly able to acknowledge the phenomenon of inﬂection, and to ascribe some signiﬁcance to it. However, they will not be able to give it the signiﬁcance Lopes does. They are stuck attributing to it the limited signiﬁcance described in the ﬁrst and second accounts explored in Section 4—at least until they ﬁnd something better to say. ⁷ While the later Wollheim (1987) could make sense of the puzzle of mimesis, he could not exploit Lopes’s solution. His account of seeing-in prevents him from making sense of inﬂection. The problem lies with the ‘fold’ of seeing-in which presents us with the depicted object. The only thing Wollheim thinks can informatively be said about that fold is that it is ‘somewhat analogous’ to the experience of seeing that object face to face. Inﬂection, I have argued, amounts to our seeing in pictures objects with inﬂected properties, and inﬂected properties cannot ﬁgure in (the content of) face-to-face seeing. Thus, inﬂected seeing-in would, for Wollheim, involve a ‘fold’ for which there is no analogous experience, and thus about which nothing could informatively be said. Wollheim was in general pessimistic about how much can be said about seeing-in, but even he would presumably baulk at the idea that about the key element in some seeing-in we can in effect say nothing at all.

inflected pictorial experience

171

But are unitary accounts correct? The proponent of inﬂection needs to defend a divisive theory of seeing-in if inﬂection is really to matter. Provided that can be done, inﬂection’s place in aesthetics is secure. So far I’ve said nothing in favour of either sort of account. Nor do I intend to settle the issue here. My goal is more limited. All I want to argue is that the phenomenon of inﬂection itself exerts some pressure to adopt a unitary account.

6. Accommodating Inﬂection I offer two arguments to show that divisive accounts of seeing-in struggle to accommodate inﬂection. Thus, while divisive accounts may be able to offer inﬂection a signiﬁcance it otherwise lacks, the phenomenon itself suggests those accounts are false. First, if seeing-in is understood on the divisive model, the design of the surface should ﬁgure twice over in inﬂected seeing-in. Divisive accounts hold that to each dimension of the content of seeing-in, there corresponds some psychologically real component of seeing-in itself. Those components might be experiences combining to form a complex whole, or merely ‘folds’ in a single experience. Either way, one component should capture the design-oriented dimension of content, the other the dimension oriented towards the scene visible in that design. Inﬂection just is the phenomenon of our seeing in the surface properties that need characterizing in part by reference to design. Thus, certain aspects of the design must, in inﬂected experience, be represented both by the component that captures the design itself and by the component that captures what is seen in it. Those aspects of design should thus ﬁgure twice in the experience as a whole. But this is not borne out by the phenomenology of inﬂected seeing-in. When I see the inky hand in Rembrandt’s sketch of Sylvius, the upward ﬂowing ink strokes do not ﬁgure twice over. Of course, I can concentrate on them as marks, or concentrate on them as somehow forming the hand. So there is a duality here. But that duality is in the object experienced as before me, the picture. As far as my experience itself goes, the only division is between the two dimensions of content that can be abstracted from it, one representing the design, the other the scene. And the only doubling is that the ink strokes ﬁgure in both those dimensions of content. No

172 robert hopkins further division within that experience is marked, and in particular not one between components of the whole, such that the ink strokes appear once in one and again in another. Second, it is hard to see how divisive accounts of seeing-in can allow for inﬂection without losing their grip on one of the two components they postulate, that embodying the scene-oriented dimension of the experience. Any divisive account treats seeing-in that is not inﬂected, in the cases to which we are restricting attention (Section 1), as having the following structure: (i) A component that represents the picture, including its design. (ii) A component that represents the scene. The result of these two occurring together is pictorial experience: (iii) An experience that represents the picture, including its design, and represents the scene as visible in that picture. In Section 1 I raised two issues that any account of seeing-in should address. One is the issue that separates divisive from unitary accounts, that of how seeing-in comes to have two dimensions to its content. But there is also the other issue to consider, what form of representation each dimension of content involves. There are various candidates: representation as found in vision generally, that found in the beliefs to which experience gives rise, that found in visualizing, and so on. To which of these will divisive accounts appeal? What forms of representation are in play in the components of seeing-in (i) and (ii) that such accounts postulate? In (i) the representation is the same as that involved in ordinary face-to-face experience. The picture, with its design, is given as before one. But what about (ii)? Divisive accounts must answer if they are to offer a theoretical understanding of seeing-in as a whole, (iii). After all, pictorial experience as a whole in some sense presents us, not just with a surface bearing a design, but with the scene visible in that surface. In what sense does it make that scene available to us? That is surely the central question facing any account of seeing-in. Since divisive accounts treat (iii) as simply the combination of (i) and (ii), the answer they give will turn on what they say about the representation in (ii).

inflected pictorial experience

173

Inﬂection aside, divisive accounts can try to handle (ii) in the same way as (i). The second component represents the scene in the way that ordinary, face-to-face visual experience represents its objects. (Call this Standard Visual Representation.) Of course, we know not to take that component at face value—the scene visible in a picture is not before us and, trompe-l’œil aside, we don’t take it to be.⁸ But that’s compatible with reading (ii) as proposed. For the component is merely part of a larger whole, (iii), and it is the nature of the whole, not just that of the component, that guides belief. Since that whole includes the representation of a design, (i), as well as a scene, (ii), somehow sense is made of the whole in such a way that we don’t take the represented scene to be before us. Gombrich’s divisive account (1961) apparently took this line on (ii).⁹ Lopes’s terminology suggests he is also attracted by it. He describes (ii) by talking of the component—for him a distinct experience—being ‘as of’ the scene. It’s common to use such talk when one wants to cancel any implication of veridicality in the phrase ‘experience of . . .’. Normally, that is all such talk does. It does not suggest that the experience so characterized differs from veridical perception in more fundamental ways, and in particular there is no suggestion that they differ in the form of representation involved.¹⁰ ⁸ In fact, trompe l’œil provides another reason for adopting the current line. As I noted in Sect. 1, a third issue concerning seeing-in is whether our experience of trompe l’œil counts. Anyone who thinks it does has some reason to adopt a divisive account of seeing-in, along with the claim that (ii) involves Standard Visual Representation. Since in trompe l’œil we do take the depicted object to be before us, it is natural to suppose that our experience of those pictures involves the same form of representation as seeing face to face. The simplest way to make sense of the idea that trompe l’œil experience is seeing-in is to suppose that it is normal seeing-in with component (i) missing. But if (ii) constitutes all of trompe l’œil, and the latter involves Standard Visual Representation, so must (ii). This neat package of views may be one thing drawing folk to divisive accounts. It certainly seems to motivate Gombrich and Lopes. But one moral of the argument to follow is that exploiting this package is not consistent with advocating inﬂection. ⁹ At least, that is the normal way to read his talk of ‘illusion’. For an alternative, see Bantinaki (2007). ¹⁰ What of the later Wollheim? Did he too hold that the representation in (ii) is Standard Visual Representation? The issue is delicate. In refusing to explain in what ways the relevant ‘fold’ of seeing-in is analogous to seeing the scene face to face, Wollheim in effect ducked the question. His long-standing opposition to Gombrich’s account might tempt one nonetheless to read him as taking the opposite line here, and answering ‘no’. However, in his last work (2003) Wollheim claimed that seeing-in involves ‘experience of’ the scene visible in the picture, and took this to be something that experienced resemblance views cannot accommodate. A natural reading of this combination of claims is as endorsing ‘yes’.

174 robert hopkins However, acknowledging inﬂection puts pressure on this account of the scene-directed component. Inﬂection amounts to our seeing inﬂected properties in the surfaces before us. Thus, the three elements for inﬂected pictorial experience look like this: (i) A component that represents the picture, including its design. (iv) A component that represents the scene, along with certain inﬂected properties. And the combined inﬂected experience: (v) An experience that represents the picture, including its design, and represents the scene, including certain inﬂected properties, as visible in that picture. Now, as argued above (Section 2), inﬂected properties cannot be represented in experience that purports to be ordinary visual experience of objects. Ordinary experience presents objects as before one, i.e. as seen face to face. Inﬂected properties are properties that, without reducing to design properties (consider, for example, the property of being an upturned hand), nonetheless need characterizing, in part, by reference to features of design, conceived as such (e.g. by reference to inky strokes). The experiences, or folds,¹¹ (iv) in which we are presented with inﬂected properties thus cannot even seem to be cases of seeing face to face. And that at least raises the question whether the representation in (iv) can be Standard Visual Representation. Of course, we must not assume that Standard Visual Representation is limited to face-to-face experience. That would be question-begging, since the divisive accounts under consideration precisely suggest that it also ﬁgures in seeing-in, in (ii) and in (iv). But once we acknowledge that any experience of inﬂected properties could not even purport to be a case of face-to-face seeing, the question does arise whether the two can possibly involve the same form of representation. Anything bearing inﬂected properties is not just an unusual sort of entity, but one that somehow combines aspects drawn from very different orders of reality: the world of design and the world of scenes visible in design. Can Standard Visual Representation offer us that? Until we have a satisfactory answer, the divisive account is threatened either with treating (iv) as involving a ¹¹ From now on I omit this qualiﬁcation.

inflected pictorial experience

175

different form of representation from (ii), or with retreating from the idea that (ii) involves Standard Visual Representation. Acknowledging inﬂection thus puts pressure on the understanding divisive accounts can offer of the second component in seeing-in. Why be sceptical about Standard Visual Representation’s ability to represent inﬂected properties? The problem is that those properties draw on different levels of reality, design, and scene, as the objects vision puts before us do not. But a divisive account might reply that Standard Visual Representation does sometimes cross levels in just this way.¹² Think, for instance, of seeing an object through a distorting lens, such as a very uneven piece of glass. Here one’s experience represents something that combines different orders: that of the world seen through the lens, and that of the lens through which it is seen. Doesn’t this suggest that different levels are something Standard Visual Representation is perfectly capable of capturing? If it can do so here, why not in (iv)? In a nutshell, the answer is that there are two ways to understand cases such as that of the distorting lens, and that neither serves the divisive theorist’s needs. Taken one way, the analogy leaves him unable to distinguish inﬂection from nested seeing-in. Taken the other, it leaves him unable to distinguish it from overlap. In presenting this dilemma, let’s begin with the two ways. When I see something through a distorting lens, my experience might take one of two forms. It might factor out the contribution of the lens and the object behind it, so that I see the thing for what it is, and see it as distorted by the intervening pane. Or it might not factor out the two, so that all I see is a strange sort of thing—the fact that it is an ordinary item seen through a distorting pane explaining why I see what I do, but not being part of how I see things to be. These two ways the case might go are common to any experience involving two levels: either the two levels are distinguished in that experience, or they are not. The question, then, is which way of taking the case offers the best model for (iv). Suppose we treat (iv) the ‘factored’ way. It involves the Standard Visual Representation of two levels, distinguished as such. In effect, this is to treat (iv) as very similar to (iii), the combined experience that is ordinary seeing-in. There, too, the levels of design and scene are present, with ¹² I am grateful to Berys Gaut and the editors for suggesting this reply.

176 robert hopkins experience presenting them as distinct. Of course, there is a difference: (iv) is just one component of pictorial experience as a whole, whereas (iii) is such a whole. The whole involving (iv), i.e. (v), will thus be very different from ordinary seeing-in, (iii). There’s no danger that inﬂected seeing-in will collapse into the ordinary kind. What is under threat, however, is our ability to distinguish inﬂected seeing-in from the nested variety. Nested seeing-in, remember, involves seeing in a picture something that is itself a design in which things are seen. So in such cases we have the following elements, according to a divisive account: (i) A component that represents the picture, including its design. (vi) A component that represents the scene, including a design, with some further scene visible in it. And the combined pictorial experience: (vii) An experience that represents the picture, including its design, and represents the scene as visible in that picture, which scene itself includes a design with some further scene visible in it. How will (vi) differ from (iv)? Our current assumption is that (iv) factors out the level of design from that of scene. (vi) does just that, in representing a design (the one visible in the picture) in which a scene is visible. If (iv) factors design and scene this way, it simply is (vi), and inﬂection is indistinguishable from nesting. (After all, the ﬁrst component, (i), is common to the two cases, and the experience as a whole is just the combination of that component with the second.) But if (iv) does not factor design and scene as (vi) does, how does it do so? Unless the divisive theorist can answer, the factored reading of (iv) must be abandoned.¹³ ¹³ Above we kept nesting and inﬂection apart by appeal to the idea that only in the latter is it necessary to characterize what is seen in the design before us by reference to that very design. Will that help here? The representation of a design before us is the job of the ﬁrst component of seeing-in, (i). Since (i) is a component in (v) and (vii), appeal to how the scene relates to the design before us might allow the divisive theorist to distinguish the latter pair after all. That would be to ﬁnd some difference between inﬂected and nested pictorial experience. But surely he won’t consider himself to have identiﬁed the core difference until he can distinguish the scenes they show us, i.e. (iv) from (vi). It is the search for that difference that is our present concern. Appeal to the design before us, and so to (i), could only help with that if the second components of inﬂected and nested seeing-in, respectively, can themselves differ in ways depending on the wider context in which they occur. That is to suppose that pictorial experience as a whole can transform its parts. But that is to abandon a major attraction of divisive views: the promise to account for features of seeing-in as a whole in terms of features of its putative parts.

inflected pictorial experience

177

That might anyway seem the best thing to do. After all, inﬂection as a whole, and thus the element in seeing-in which is, supposedly, the core of the phenomenon, (iv), does indeed involve a mingling of design and scene as ordinary seeing-in does not. While the latter distinguishes sharply between the design and what is seen in it, the whole point of inﬂection was to challenge that sharp divide. So it may seem obvious that divisive accounts should treat (iv) the unfactored way. Just like the experience of seeing the object through the distorting pane without seeing it as such, (iv) involves Standard Visual Representation of things at different levels, but does not represent those levels as distinct. However, if we construe inﬂected seeing-in this way, we then struggle to distinguish it from another familiar phenomenon, overlap. In overlap, a divisive account will say, we have the following elements: (viii) A component that represents the picture, including its design, and represents some part of the design as having some property F. (ix) A component that represents the scene, and represents something in that scene as F. And the combined pictorial experience: (x) An experience that represents the picture, including that part of the design that is F, and represents the F-object in the scene as visible in that F-part of the picture. How do (v) and its component (iv) differ from (x) and its component (ix)? Both pairs involve awareness of design. Both involve awareness of scene properties that can be characterized by reference to properties of design. Earlier (Section 2) we distinguished inﬂection from overlap by framing our definition of inﬂected properties carefully. They are properties seen in surfaces that need characterizing by reference to features of that surface’s design, conceived as such. The properties in (ix) and (x) don’t ﬁt this bill. They can be characterized that way—for example, such-and-such a property of the scene can be captured by saying it’s the same property as is exhibited by a given bit of the design. But they need not be so captured: we could instead simply talk of F-ness (or whatever). So all will be well provided the properties in (iv) and (v) have to be characterized by reference to design—provided, that is, they really are inﬂected properties, as we deﬁned them. The problem is that treating (iv) the unfactored way makes it hard to see how this can be.

178 robert hopkins Ex hypothesi, (iv) does not represent the two levels that are somehow bound up in the relevant properties, design and scene, in such a way as to distinguish them. Why, then, should the properties it represents need characterizing by reference to design? Consider the analogy with the distorting lens. If my experience fails to distinguish the distortion of the lens from the contribution made by the object seen through it, why think we can only characterize how that experience represents things as being by reference to a distorting lens? We can so characterize it: my experience represents things as being the way they look when seen through such a lens. But we need not. We could say instead that it represents an object with strangely shaped limbs, perhaps ones the boundaries of which are not all equally deﬁnite, and so forth. The situation is the same with (iv), treated the unfactored way. Perhaps the properties it mentions can be characterized by reference to design; but there is no necessity for this. Since the experience itself doesn’t mark out design as distinct from scene, it’s quite unclear why anything about the properties as it represents them needs describing by reference to design. We may struggle, of course, to describe those properties. But that problem confronts everyone: the sorts of properties we see in pictures such as the Rembrandt are hard to describe. It’s also true that we may ﬁnd that our best attempts to describe those properties mingle talk of very different kinds of things: hands and ink, for instance. But none of this shows that, in describing those properties, we have to appeal to properties of design conceived as such. To talk of ink is not yet to talk of ink composing a design, something in which things might be seen. Thus, the divisive account takes the unfactored route only to be left with a reading of (iv) on which that no longer clearly differs from (ix) and, as a result, (v) does not differ from (x). Inﬂection has collapsed into overlap. Thus, divisive accounts face an unappealing choice. They must say something about the representation in (iv), on pain of taking inﬂected seeing-in to fall outside their account of seeing-in in general. If inﬂected seeing-in is not to be a radically different state of mind from ordinary seeing-in, they had better say that the representation involved is the same as that involved (according to them) in (ii), Standard Visual Representation. Given that what is represented in (ii) crosses levels in a distinctive way, it’s hard, but not perhaps impossible, to see how that could be. What is not possible, it seems, is to make sense of that idea in such a way as to avoid collapsing inﬂection either into nesting or into overlap. This naturally leaves one

inflected pictorial experience

179

wondering why we should accept that the representation in (iv) is Standard Visual Representation. Apart from the fact that this is all divisive accounts seem able to offer us as an account of (iv), there seems little reason to adopt this view.¹⁴ At the least, then, I would like to be shown how a divisive account of seeing-in can continue to answer the questions it cannot duck while acknowledging inﬂection. This, along with the earlier argument from doubling up, makes me wonder whether it is wise for any divisive theorist to endorse the idea of inﬂected seeing-in. Of course, there may also be problems with unitary accounts of pictorial experience. I don’t deny that. What I have yet to see is that inﬂection provides any arguments against them. My more general conclusion is rather cautious. Inﬂection only clearly makes sense against the background of unitary accounts of seeing-in. But, given such accounts, the only signiﬁcance for aesthetics that can clearly be made out for it is rather limited. Inﬂection does occur, and I’m grateful to its proponents for bringing that to our attention. But in seeking to give it a central role in our appreciation of pictures, they are in danger of combining views that do not sit comfortably together.

References Bantinaki, Katerina (2007), ‘Pictorial Perception as Illusion’, British Journal of Aesthetics, 47/2: 268–79. Budd, Malcolm (1992), ‘On Looking at a Picture’, in J. Hopkins and A. Savile (eds), Psychoanalysis, Mind and Art (Oxford: Blackwell). Gombrich, Ernst (1961), Art and Illusion (Princeton: Princeton University Press). Hopkins, Robert (n.d.), ‘Sculpture and Visual Perspective’. (1998), Picture, Image and Experience (Cambridge: Cambridge University Press). (2008), ‘Reasons for Looking: Lopes on the Value of Pictures’, Philosophy and Phenomenological Research, 77/2: 563–73. Lopes, Dominic McIver (2005), Sight and Sensibility (Oxford: Clarendon Press). Peacocke, Christopher (1987), ‘Depiction’, Philosophical Review, 96/1: 383–410. ¹⁴ I’ve presented the difﬁculty here as confronting the idea that the second component in the various forms of seeing-in involves Standard Visual Representation. In fact, it is more general. It faces any account that both analyses seeing-in into components and a whole formed from them, as divisive accounts do; and then takes just one form of representation to be involved in every component of every kind of seeing-in. There are, as it were, only so many relations between design and scene to go round, and not enough, it would seem, to cover the varieties of pictorial experience before us.

180 robert hopkins Penrose, R. S. and L. Penrose (1958), ‘Impossible Objects: A Special Type of Visual Illusion’, British Journal of Psychology, 49: 31–3. Podro, Michael (1991), ‘Depiction and the Golden Calf’, in N. Bryson, M. Ann Holly, and K. Moxey (eds), Visual Theory: Painting and Interpretation (Cambridge: Polity Press). (1998), Depiction (Cambridge, Mass.: Harvard University Press). (2001), ‘The Artistry of Depiction’, in R. van Gerwen (ed.), Richard Wollheim on the Art of Painting: Art as Representation and Expression (Cambridge: Cambridge University Press). Schier, Flint (1993), ‘Van Gogh’s Boots: The Claims of Representation’, in D. Knowles and J. Skorupski (eds), Virtue and Taste (Oxford: Blackwell). Walton, Kendall (1990), Mimesis as Make-Believe (Cambridge, Mass.: Harvard University Press). Wollheim, Richard (1968), Art and its Objects (Cambridge: Cambridge University Press). (1987), Painting as an Art (London: Thames & Hudson). (2001), ‘On Pictorial Representation’, in R. van Gerwen (ed.), Richard Wollheim on the Art of Painting: Art as Representation and Expression (Cambridge: Cambridge University Press). (2003), ‘What Makes Representational Painting Truly Visual?’, Proceedings of the Aristotelian Society, suppl. vol., 77/1: 131–47.

7 Inflected and Uninflected Experience of Pictures BENCE NANAY

1. Introduction: Inﬂected Versus Uninﬂected Seeing-In One big question about pictures is how we perceive them: what perceptual state we are in when we see a depicted object in the picture surface. Following Richard Wollheim, I will call the experience of seeing something in a picture seeing-in (Wollheim 1980, 1987, 1998). The question about the nature of seeing-in is crucial enough, but it has also been suggested that seeing-in may play an important role in understanding what pictures are: pictures are those objects in the face of which ‘suitably informed’ perceivers are supposed to have a seeing-in experience (Wollheim 1987, 1998). I will remain neutral about whether we can use seeing-in to deﬁne what pictures are. But my concern in this chapter is not just the nature of seeingin, but mainly the distinction between two different kinds of picture perception: inﬂected and uninﬂected. It has been suggested that sometimes our experience of pictures is inﬂected. As Dominic Lopes says: ‘Features of the design may inﬂect illustrative content, so that the scene is experienced as having properties it could only be seen to have in pictures’ (Lopes 2005: 123–4). I’m grateful for Dominic Lopes’s, Robert Hopkins’s, Catharine Abell’s, and Katerina Bantinaki’s detailed comments. I’m also grateful to the Inﬂection reading group at the University of British Columbia, Summer 2007. I presented a version of this chapter at the 2009 University of Cincinnati Philosophy Symposium and I am grateful for the feedback I received there, especially from Gregory Currie, John Kulvicki, No¨el Carroll, Jesse Prinz, Cynthia Freeland, Christopher Gauker, Tom Polger, and Jenefer Robinson.

182 bence nanay An important clariﬁcation is in order: what Lopes means by design is ‘those visible surface properties in virtue of which a picture depicts what it does’ (Lopes 2005: 25). I will call these properties design properties in what follows. Lopes is clear that ‘design may inﬂect illustrative content’. In other words, it does not always do so. Some of our experiences of pictures are inﬂected, some others are not (see also footnote 4 below). When I watch a football game on TV, my experience tends not to be inﬂected. When I am admiring the composition of a C´ezanne landscape, it tends to be inﬂected. And here is the main question I would like to address in this chapter: what is the difference between these two kinds of experience of pictures, inﬂected and uninﬂected? Thus, we have two questions: (a) what is seeing-in? and (b) what is inﬂected seeing-in? It is difﬁcult to answer (b) without having a clear answer to (a), and what we say about (b) may put constraints on the way we can answer (a). I will attempt to answer the two questions at the same time. We have seen that we may be able to use the answer to (a) in order to tell what pictures are. How about (b)? Why is it relevant? It has been suggested that the answer to (b) may be the key to understanding not depiction but another big topic of aesthetics: the aesthetic appreciation of pictures. The suggestion, in a very sketchy form, is that the aesthetic appreciation of pictures in some way depends on our inﬂected seeing-in; maybe inﬂected seeing-in is even necessary for the aesthetic appreciation of pictures (see the discussion in Lopes 2005, Ch. 3, and Hopkins, Chapter 6 in this volume). Further, it is also taken to be a crucial, and maybe even necessary, feature of ‘advanced picture making abilities’ (Lopes 2005: 28). A couple of possible misunderstandings need to be dispelled. First, whether our seeing-in is (supposed to be) inﬂected does not have any direct consequence with regard to whether the picture is good or bad. We can have inﬂected seeing-in experiences even when we face really bad pictures. And we can have uninﬂected seeing-in experiences when we are looking at masterpieces. But is a ‘suitably informed spectator’ supposed to have an inﬂected pictorial experience when she is looking at a masterpiece in order to appreciate it as such? We should not rush to a positive answer to this question. Inﬂected seeing-in may allow us to focus on features of the picture that we may ﬁnd valuable. But there may be other dimensions

inflected and uninflected experience of pictures

183

of value that do not require inﬂected seeing-in. There is no obvious and straightforward connection between inﬂected seeing-in and the value that we assign to pictures. Second, inﬂected seeing-in is not to be confused with the widely discussed phenomenon of aesthetic experience (whatever that means; see Carroll 2005 for an excellent summary). Whatever we mean by the concept of ‘aesthetic experience’, it should be applicable to experiences other than those of pictures. And it is not clear how the notion of inﬂected seeing-in would apply in those cases.

2. Inﬂected Seeing-In So far I have not said much about what inﬂected seeing-in is supposed to be. The most important feature of inﬂected seeing-in seems to be that when we have experiences of this kind, the perceived object is experienced differently from the way it would be experienced face to face. As Lopes says, ‘Design seeing transforms the content of seeing-in so that it no longer matches the content of seeing the scene face-to-face. Design is ‘‘recruited’’ into the depicted scene so that the scene no longer looks the way it would when seen face-to-face’ (Lopes 2005: 40; see also Podro 1998: 13). And again: ‘Except in trompe-l’œil, seeing O in P is quite different phenomenally from seeing O face to face; O is seen to have properties in P that it is not seen to have with the naked eye’ (Lopes 2005: 128–9). And here we encounter a problem. One may suggest that in the case of any instance of seeing-in, inﬂected or not, it is true that ‘the content of seeing-in no longer matches the content of seeing the scene face-toface’. Seeing Jessica Alba face to face and seeing her in a photograph are two very different experiences. Seeing a shark swimming towards me on TV is, again, very different from seeing it swimming towards me face to face. But this is not the difference the champions of inﬂection have in mind. Michael Podro says that in the case of inﬂected seeing-in, the picture’s design is ‘recruited’ into the depicted scene and this is why the scene does not look the way it would when seen face to face (Podro 1998: 13, 26). In the Jessica Alba example and in the shark example our experience is different from the way it would be if we saw the depicted scenes face to

184 bence nanay face, but it is not different because the picture’s design is ‘recruited’ into the depicted scene. So inﬂected seeing-in, as Podro understands it, is not merely a kind of seeing-in where our experience is different from the way it would be if we experienced the depicted object face to face. It is a kind of seeing-in where our experience is different because the picture’s design is ‘recruited’ into the depicted scene. It is not clear, however, what it is supposed to mean that the picture’s design is ‘recruited’ into the depicted scene. Podro’s other formulations of inﬂected seeing-in are equally metaphorical: he says that when our seeing-in is inﬂected, then besides seeing the scene in the design, we also see the design in the scene (Podro 1991: 172). He also says that inﬂected seeing-in straddles the boundary between the marked surface and the depicted object (Podro 1998: 17, 28). If we want to understand the difference between inﬂected and uninﬂected seeing-in, we need to make sense of these metaphors. Robert Hopkins goes through a couple of possible interpretations of inﬂected seeing-in and settles for the following: ‘Sometimes, what is seen in a surface includes properties a full characterization of which needs to make reference to that surface’s design (conceived as such)’ (Chapter 6 in this volume, p. 158). If we accept this characterization of inﬂected seeing-in, then some, but not all, of our experiences of seeing-in are inﬂected. Seeing a football game on TV is (supposedly) not inﬂected as what is seen (the football game) does not (presumably) include properties a full characterization of which needs to make reference to the picture’s design, conceived as such. And admiring the composition of a C´ezanne landscape may indeed be an inﬂected experience: what I see in the painting may include properties a full characterization of which needs to make reference to the painting’s design. In other words, the difference between inﬂected and uninﬂected seeingin is that we attribute properties of a certain kind (properties that cannot be fully characterized without reference to the picture’s design) to the object of our experience in the former but not in the latter case. This will be our preliminary deﬁnition of inﬂected seeing-in for now. Some further complications, clariﬁcations, and modiﬁcations will follow as we go along. For now, it is sufﬁcient to point out that what this distinction amounts to depends on what we take seeing-in to be, inﬂected or uninﬂected.

inflected and uninflected experience of pictures

185

I pretended in this section that we have a clear understanding of what seeing-in amounts to and we are drawing a distinction, within this neatly deﬁned category, between inﬂected and uninﬂected instances of seeing-in. In fact, it is not at all clear what seeing-in amounts to. So if we want to see clearly with regard to the nature of inﬂected seeing-in, we may need to take a step back and ask what seeing-in is, inﬂected or not.

3. Seeing-In and Twofoldness Wollheim, who introduced the concept, said that seeing-in is a twofold experience: ‘The spectator is, and remains, visually aware not only of what is represented but also of the surface qualities of the representation’ (Wollheim 1980: 214–15). This feature of our experience of pictures is called ‘twofoldness’ and, in some form or other, many philosophical accounts of depiction endorsed it as a necessary feature of our experience of pictures (Walton 1990: 300–1; 1991: 423; 2002: 33; Hopkins 1998: 15–17; see also Lopes 2005, Ch. 1, and Kulvicki 2006: 172–3 for moderately critical overviews).¹ In spite of its widespread use of this notion, neither Wollheim nor other philosophical accounts of pictorial perception say much about what is supposed to be meant by the twofoldness of experience. Wollheim talks about simultaneous awareness of surface and scene (Wollheim 1998: 221; 1987: 46). But the notion of awareness he uses is ambiguous and as a result Wollheim’s notion of twofoldness itself is also ambiguous (Nanay 2005). And those who used, or criticized, the notion of twofoldness did not get rid of this ambiguity. Here are two possible interpretations of twofoldness (both of which we have good reasons to attribute to Wollheim): (i) We consciously attend both to the depicted object and to some properties of the surface.² ¹ Ernst Gombrich’s account of our experience of pictures is inconsistent with the idea of twofoldness. As he said: ‘is it possible to ‘see’ both the plane surface and the battle horse at the same time? If we have been right so far, the demand is for the impossible. To understand the battle horse is for a moment to disregard the plane surface. We cannot have it both ways . . .’ (Gombrich 1961: 279). ² One important consideration in favour of (i) is the following quote: ‘The seeing appropriate to representations permits simultaneous attention to what is represented and to the representation’

186 bence nanay (ii) We perceptually represent both the depicted object and some properties of the picture surface (while we may or may not attend to them).³ We have many reasons to reject the view that (i) is a necessary feature of seeing-in, but none of these reasons should persuade us to reject (ii) as a necessary feature of seeing-in. I will go through some of the inﬂuential criticisms of the concept of twofoldness and of the claim that twofoldness is necessary for seeing-in and point out that although they are all valid arguments if we take twofoldness to mean (i), they lose their appeal if we interpret twofoldness as (ii). First, if twofoldness implies the conscious attention to surface and scene, then this rules out the possibility of perceiving things in pictures unconsciously. But it seems that we are capable of perceiving things in pictures unconsciously. We can perceive objects in pictures even if we are not conscious of either the surface or the depicted object, as the widely discussed phenomenon of subliminal priming shows (Strahan et al. 2002; Eimer and Schlaghecken 2003; Greenwald et al. 1996). If we watch a ﬁlm and, unbeknownst to us, images of a can of a certain beverage are ﬂashed for less than 100 milliseconds, we tend to develop a desire to drink that kind of beverage. We are not conscious of the depicted object (we see it for less than 100 milliseconds), but the fact that we desire the beverage (the depicted object) seems to indicate that we did perceive it in the picture in spite of the fact that we were not aware of it. Perception can be conscious and unconscious and so can perceiving something in a picture. Thus, it seems that perceiving something in a picture does not need to be a conscious experience. So while we have good reason to reject (i) as a necessary condition for seeing-in, these considerations do not count against the claim that seeing-in entails (ii), the simultaneous visual representation of the picture surface as well as of the depicted object. Second, here is Jerrold Levinson’s attack on the idea that twofoldness is a necessary feature of seeing-in (Lopes 1996: 37–51 makes a similar point): Plausibly not all seeing-in or registering of pictorial content is aesthetic in character, or even informed by the awareness of pictures as pictures; for instance, that directed (Wollheim 1980: 213). But it is not clear whether seeing-in only needs to ‘permit’ simultaneous attention or it is constituted by it. ³ The main consideration in favour of attributing this notion of twofoldness to Wollheim is his argument in favour of the necessity of twofoldness for seeing-in from perceptual constancy (Wollheim 1980: 215–16). I will analyse this argument at the end of this section (see also Nanay 2005).

inflected and uninflected experience of pictures

187

to or had in connection with postcards, passport photos, magazine illustrations, comic strips, television shows, or movies. Thus, any view that builds aesthetic character, or even awareness of pictures as pictures, directly into seeing-in would seem to have something amiss. (Levinson 1998: 228–9)

And even more explicitly: If you see a woman in a picture in virtue of visually processing a pattern of marks, then of course in some sense you thereby perceive the medium in which those marks inhere or consist. But it is far from clear that when you see the woman in the picture you must in some measure be attending to, taking notice of, or consciously focusing on the picture’s surface or patterning as such. (Levinson 1998: 229)

Thus, if Wollheim held that (i) is a necessary feature of seeing-in, he was wrong. I agree. But Levinson’s considerations are silent about whether (ii), the other interpretation of twofoldness, could be considered to be a necessary feature of seeing-in. Lopes gives a thorough analysis of the notion of twofoldness (Lopes 2005, Ch. 1). He concludes that although twofoldness is an important feature of some of our experiences of some pictures, it is not a necessary feature of seeing-in in general. I will attempt to point out that Lopes’s arguments presuppose that the notion of twofoldness is to be interpreted as (i). While they are very persuasive that (i) is not necessary for seeing-in, they say nothing about whether (ii) could be considered to be necessary for seeing-in. Lopes introduces the notion of ‘design seeing’, a concept we shall use later, as ‘seeing design features as responsible for seeing-in’ (Lopes 2005: 28). He says: When we look at a picture, we normally see in it the scene it depicts, but we may also see its design as a design. Of course, there is a sense in which we always see the picture’s design when we see things in it, for we always see a scene in a picture by seeing the picture face to face. It is only in virtue of seeing the conﬁguration of marks on its surface, and being sensitive to visible changes in them, that we see anything at all in the picture. However, seeing a pictorial design face to face does not entail seeing the design as a design—it does not entail design-seeing. (Lopes 2005: 28)

188 bence nanay Thus, we always see the surface of the picture when we see something in it, but we may or may not see its design as design. This sounds like a rejection of the claim that (i) is a necessary feature of seeing-in and an endorsement of the claim that (ii) is. But Lopes restricts the notion of twofoldness to (i) and hence denies that twofoldness (that is (i)) is a necessary feature of seeing-in. He says: ‘According to Wollheim, seeing-in is one component of an experience with two aspects: a simultaneous or ‘twofold’ experience of design and depicted scene (Wollheim calls the complete, twofold experience ‘seeing-in’). It is impossible to see a scene in a picture without also seeing the picture’s design as a design’ (Lopes 2005: 33). Thus, the claim Lopes attributes to Wollheim is that twofoldness, that is (i), is necessary for seeing-in. And he then rightly rejects this view, pointing out that one of Wollheim’s main arguments for the claim that twofoldness is necessary for seeing-in, the argument from perceptual constancy (Wollheim 1980: 215–16), ‘requires that the visual system access information about the surface of pictures, but it does not follow that the surface is experienced. We may see the design without seeing it as a design, if design information is used by vision to correct for viewing position without entering conscious experience’ (Lopes 2005: 35). The way I read Lopes’s point is that the perceptual constancy of picture perception does not support twofoldness as (i), but it does support twofoldness as (ii). He comes close to a similar formulation: ‘An explanation of constancy requires at most that constant seeing-in be accompanied by surface seeing, not design seeing’ (Lopes 2005: 36). Another argument Lopes gives against the claim that (i) is necessary for seeing-in is the following. He introduces the notion of pseudo-twofoldness and argues that some of our seeing-in experiences are pseudo-twofold, hence, by deﬁnition not twofold (Lopes 2005: 40–2). His example is the famous picture of the Dalmatian where we can see the Dalmatian’s contour only if we see the dog in the picture. If we do not see the Dalmatian in the picture, we do not see the contour: we only see ‘a patchwork of unrelated blobs’ (Lopes 2005: 42). What this is supposed to show is that when we see the Dalmatian in the picture, we do not see the picture’s design as a design: there is no design seeing going on. Why? Because ‘the design features . . . must be visible independently of seeing anything in the picture’ (Lopes 2005: 41) and we cannot see the shape of the

inflected and uninflected experience of pictures

189

Dalmatian independently of seeing the dog in the picture. Be that as it may: seeing the Dalmatian in the picture may or may not exclude design seeing, i.e. seeing the design as a design. But it deﬁnitely does not exclude that seeing it in the picture entails representing the picture’s design: it does not exclude that our experience of seeing-in is a twofold one in sense (ii). Lopes concludes from these observations that seeing-in is a diverse phenomenon: sometimes it entails design seeing (seeing the picture’s design as a design), sometimes it doesn’t. Sometimes it entails surface seeing (seeing the picture’s design, but not seeing it as a design), and maybe sometimes (in the case of the experience of trompe l’œil paintings) it doesn’t. Importantly for our purposes, Lopes holds that if seeing-in does entail design seeing, then seeing-in is inﬂected. If it does not, it is uninﬂected. This is a coherent account of inﬂected and uninﬂected seeing-in and I do think that it is on the right track. Inﬂected seeing-in is deﬁned with the help of the concept of design seeing: seeing-in is inﬂected if it ‘doubles with design seeing’. But it is not at all clear how the concept of design seeing could be cashed out. The two characterizations Lopes gives are not extremely helpful. He says that design seeing is ‘seeing design features as responsible for seeing-in’ (Lopes 2005: 28) and that design seeing is seeing ‘the picture’s design as a design’ (Lopes 2005: 33; see also 2005: 28). In order to make sense of these suggestions, I will turn to a Wollheiminspired way of looking at the distinction between inﬂected and uninﬂected seeing-in in the next section, which I take to be consistent with Lopes’s distinction (although they use the concept of twofoldness, and of seeing-in, differently). In this section, we have found that we have a number of reasons to reject the idea that (i) is necessary for seeing-in. But none of the arguments that have been given against the importance of twofoldness show that (ii) is not necessary for seeing-in. I will assume in what follows that (ii) is necessary for seeing-in. This is not such a wild assumption: many of the critics of the notion of twofoldness and of Wollheim’s views in general would also be happy with this claim (Lopes 2005, Ch. 1; and maybe even Hopkins, Chapter 6 in this volume). The reason why I spent so much time on the distinction between (i) and (ii) is not merely to rule out a false interpretation of the concept

190 bence nanay of twofoldness. I hope to show in the next section that Wollheim’s two concepts of twofoldness may help us to understand the difference between inﬂected and uninﬂected seeing-in better. The view I will outline is not really Wollheim’s; I believe that a very charitable interpretation of his writings could yield this view, but I will not argue for this claim. That is why I call this account a Wollheimian (rather than Wollheim’s) account of inﬂected seeing-in.

4. A Wollheimian Account of Inﬂection Here is the view in a nutshell that could be attributed to Wollheim: while twofoldness, interpreted as (ii), is a necessary feature of seeing-in, twofoldness, interpreted as (i), is an important, and maybe even necessary, feature of (something like) inﬂected seeing-in. If we interpret Wollheim as having this view, then we can conclude that much of the confusion around the notion of twofoldness stems from the fact that he held both claims, often not distinguishing between the two. This chapter is not a piece of Wollheim scholarship, but it should be pointed out brieﬂy that some of Wollheim’s arguments in favour of the importance of twofoldness are in fact arguments for the claim that (ii) is necessary for seeing-in and some others are arguments for the claim that (i) is necessary for (something like) inﬂected seeing-in. One of Wollheim’s original arguments in favour of his twofoldness claim is that we would not be able to appreciate the great examples of pictorial representations unless we were consciously attending to both the design of the picture and the depicted object at the same time (Wollheim 1980: 214–16). As he writes: ‘in Titian, in Vermeer, in Manet we are led to marvel endlessly at the way in which line or brushstroke or expanse of colour is exploited to render effects or establish analogies that can only be identiﬁed representationally’ (Wollheim 1980: 216). Now, ‘marvelling endlessly’ at a Vermeer, according to Wollheim, entails something like inﬂected seeing-in: we attribute properties to the experienced object that can only be fully described with reference to the ‘lines or brushstrokes or expanses of colour’. And he claims that this experience (again, not seeing-in per se, but (something like) inﬂected seeing-in) is a twofold experience in sense (i). This claim is

inflected and uninflected experience of pictures

191

underlined by the analogy Wollheim draws between experiencing pictures and poetry: A comparison that suggests itself is with the difﬁculties that would have lain in store for us in our appreciation of poetry if it had been beyond our powers to have simultaneous awareness of the sound and the meaning of words. In painting and poetry twofoldness must be a normative constraint upon anyone who tries to appreciate works of those arts. (Wollheim 1980: 216)

Thus, Wollheim’s argument seems to suggest that twofoldness, in the sense of (i), that is, simultaneous conscious attention to the surface and the depicted object, is a necessary condition for the aesthetic appreciation of pictures and maybe for inﬂected seeing-in (and not for seeing-in per se). Contrast this argument with Wollheim’s other important argument concerning twofoldness, the one from the perceptual constancy of pictures we considered above. The argument from perceptual constancy aims to establish something about the nature of seeing-in per se, not about the aesthetic appreciation of pictures: it is true of all seeingin that the depicted object is not distorted if we look at a picture from an oblique angle. And the concept of twofoldness he appeals to in Wollheim (1980: 215–16) is closer to (ii) than to (i): the simultaneous representation of surface and scene. But what is important for our purposes here is that Wollheim seems to use twofoldness in sense (ii) to argue about seeing-in per se and he uses twofoldness in sense (i) to argue about inﬂected seeing-in and the aesthetic appreciation of pictures. It is time to summarize what Wollheim teaches us about the intricate connection between twofoldness, seeing-in, and inﬂected seeing-in. Seeing-in necessarily entails twofoldness in the sense of (ii): the simultaneous representation of surface and scene. But inﬂected seeing-in necessarily entails twofoldness in sense (i): simultaneous conscious attention to surface and scene. Maybe it is not overly charitable to attribute this view to Wollheim. The fact that he uses two concepts of twofoldness, which play very different roles, without acknowledging the difference, has muddied some waters, but it is nonetheless a coherent account of both seeing-in and inﬂected seeing-in.

192 bence nanay More recent discussion on inﬂection describes Wollheim as holding that every instance of seeing-in is inﬂected. Lopes, for example, claims that: Wollheim holds that we always see a picture’s design at the same time as we see in it the scene it depicts: the one interpenetrates the other in a single experience. Design seeing transforms the content of seeing-in so that it no longer matches the content of seeing the scene face to face. Design is ‘recruited’ into the depicted scene so that the scene no longer looks the way it would when seen face to face. (Lopes 2005: 40)⁴

Notice that the ﬁrst sentence just says that seeing-in entails seeing a picture’s design. But the second sentence says it entails design seeing, which, as we have seen above and Lopes himself meticulously pointed out, is something much stronger than seeing the design: it implies seeing the design as a design. If Wollheim in fact held that seeing-in entails design seeing, that is, conscious attention to design properties, then he seems to be confusing inﬂected seeing-in and seeing-in per se. In this case, it is unclear what logical space there remains for uninﬂected seeing-in, just as Levinson suggests in the quotation above. But if we interpret Wollheim as having two concepts of twofoldness and, to use Lopes’s terminology, if he takes design seeing to be necessary for inﬂected seeing-in, and at the same time takes seeing the surface to be necessary for inﬂection per se, then there is no prima facie problem with his account. It is worth noting that if we interpret Wollheim the way I have suggested, then one of the allegedly most problematic consequences of his account will appear much less problematic. Wollheim famously thought that our experience of trompe l’œil paintings is not twofold, which, in his wider theory, also implies that trompe l’œil paintings are not pictures. Many have found this conclusion very counterintuitive. But if seeing-in entails twofoldness in sense (ii), that is, the simultaneous representation ⁴ Lopes’s point may apply in the case of Michael Polanyi’s observations that could be taken to be an early introduction of the idea of (something like) inﬂection. Polanyi writes: ‘The arts do not exhibit things that could be really there and yet are not there; they exhibit things of a kind that cannot exist either in nature or in human affairs’ (Polanyi 1970: 234). It seems that Polanyi is committed to saying that all seeing-in is inﬂected. What I tried to show is that there is a way of interpreting Wollheim in such a way that he is not committed to this claim.

inflected and uninflected experience of pictures

193

of surface and scene, then it is not at all clear that our experience of trompe l’œil paintings is not a twofold experience in this sense. When we are deceived by a trompe l’œil painting, we are not aware of the surface: that’s why, for a moment, we are fooled into thinking that we are perceiving the depicted object face to face. But this does not exclude that we do represent the surface properties, without being aware of them. The perception of trompe l’œil paintings is a good example for uninﬂected seeing-in.⁵ Of course, Wollheim repeatedly admitted that it is a (maybe not too desirable) consequence of his account of seeing-in that our experience of trompe l’œil paintings is not seeing-in (see Wollheim 1998 and Feagin 1998 for a good summary; see also Levinson 1998: 228–9). And this may make one wonder how charitable my interpretation is. Well, it may be a bit too charitable. But, again, this chapter is not a piece of Wollheim scholarship. If he didn’t in fact hold this view, it is still a view that is coherent, does not face any prima facie problems, and gives a straightforward answer to both questions this chapter set out to answer: what seeing-in is and what inﬂected seeing-in is (and, in addition, it does so without being committed to the allegedly problematic consequences of his views regarding trompe l’œil). I do not want to endorse the Wollheimian account I outlined here. But I do think that it gives us some important considerations for giving a general account of inﬂected seeing-in. According to the Wollheimian account, inﬂected seeing-in entails simultaneous attention to design and scene. I think that’s almost right. I will argue that we are consciously attending⁶ to one property only, but this property is relational: it cannot be fully characterized without reference to both the picture’s design and the depicted object. For those who ﬁnd the Wollheimian account appealing, it needs to be pointed out that attending to this one relational property, depending on how we think of attention, could be taken to imply twofold simultaneous attention to design and scene. ⁵ The perception of trompe l’œil paintings is, of course, very different from other examples of uninﬂected seeing-in, like watching television or looking at magazine illustrations. In the latter cases, we are not fooled into thinking that we are looking at the depicted object face to face (not even for a moment). ⁶ I have been and I will be using the concepts of ‘attention’ and ‘conscious attention’ interchangeably. This does not mean that I take sides in the grand debate about whether attention is necessary and/or sufﬁcient for consciousness. On this thorny issue, see Campbell (2002) and Prinz (forthcoming).

194 bence nanay

5. A General Account of Inﬂected Seeing-In In this section, I will depart from the Wollheimian considerations and formulate the difference between inﬂected and uninﬂected seeing-in. The hope is that we can ﬁnd a general account of inﬂected seeing-in that would capture the intuitions behind the way Lopes and Wollheim draw the distinction between inﬂected and uninﬂected seeing-in. First, what is in common between inﬂected and uninﬂected seeing-in? Both inﬂected and uninﬂected seeing-in are twofold experiences: if we see an apple in a picture, we represent the picture’s design and the depicted apple simultaneously. Second, what is the difference between inﬂected and uninﬂected seeingin? In the case of inﬂected seeing-in, we consciously attend to a relational property that cannot be fully characterized without reference to both the picture’s design and the depicted object. In the case of uninﬂected seeingin, this is not necessarily the case. To put it simply, the difference between inﬂected and uninﬂected seeing-in is a matter of attention. And attention, as the famous ‘inattentional blindness’ experiments show, can dramatically change what we experience (Simmons and Chabris 1999; Mack and Rock 1998; Rensink n.d.) I said that when our seeing-in experience is inﬂected, we attend to a relational property that cannot be fully characterized without reference to both the picture’s design and the depicted object. I will call these properties that we attend to in the case of inﬂected seeing-in ‘design-scene properties’. Not all relational properties that cannot be fully characterized without reference to both the picture’s design and the depicted object are ‘design-scene properties’.⁷ Further, there are many ways of expressing a ‘design-scene property’: it could be referred to as the property of how features of the picture’s design give rise to or undergird the experience of the depicted object; or of how the depicted object emerges from the design, etc. What is important is that when seeing-in is inﬂected, we are consciously attending to ‘design-scene properties’. ⁷ The property of ‘being seen in this surface by me right now’, for example, although it probably cannot be characterized fully without reference to the picture’s design and to the depicted object, is not a ‘design-scene property’.

inflected and uninflected experience of pictures

195

That the properties we are attending to when seeing-in is inﬂected are relational is also reﬂected in many of the intuitions behind the concept of inﬂection: when we have an inﬂected seeing-in experience, we experience how the depicted object emerges from the design: from the marks on the surface. As Lopes says, we see the depicting design ‘undergirding’ the depicted scene: ‘seeing a picture as a picture amounts to seeing its undergirding—to seeing, as it were, the process of depiction and not merely its product’ (Lopes 2005: 39). Design seeing, which is necessary for inﬂected pictorial experiences, ‘amounts to seeing design features as responsible for seeing-in’ (Lopes 2005: 28). As Hopkins summarizes the view (which he himself disagrees with): ‘[An] account of inﬂection’s importance takes it to lie in the way it allows us to appreciate emergence. After all, inﬂection precisely straddles the divide between design and the world visible in it. The idea is that, in doing so, it offers us the opportunity better to appreciate how the one emerges from the other’ (Chapter 6 in this volume, p. 165). It is hard to see how these properties we see in surfaces would be capable of all this if they were not relational: if they did not make reference to both the picture’s design and the depicted object. When seeing-in is uninﬂected, we do represent some design properties perceptually. But this does not entail attending to ‘design-scene properties’. But when seeing-in is inﬂected, we do attend to ‘design-scene properties’: we consciously attribute the property of emerging from the picture’s design to the depicted object. This is the difference between inﬂected and uninﬂected seeing-in. How does this account relate to Hopkins’s characterization of inﬂected seeing-in? As we have seen, Hopkins said that, in the case of inﬂected seeing-in, ‘what is seen in a surface includes properties a full characterization of which needs to make reference to that surface’s design (conceived as such)’ (Chapter 6 in this volume, p. 158). My suggestion keeps the spirit of this suggestion: in the case of inﬂected, but not in uninﬂected, seeing-in, we consciously attribute a property to the depicted object, a property that can’t be fully characterized without reference to the picture’s design—just as Hopkins says. I said in the last section that this account could be thought to be consistent with Lopes’s views on inﬂection. Lopes holds that seeing-in is inﬂected if it is doubled with design seeing. Design seeing, in turn,

196 bence nanay is ‘seeing design features as responsible for seeing-in’ (Lopes 2005: 28). The big question is what it means to see design features as responsible for seeing-in. But as design properties are ‘those visible surface properties in virtue of which a picture depicts what it does’ (Lopes 2005: 25), it sounds plausible to claim that ‘seeing design features as responsible for seeing-in’ implies consciously attending to some relational properties that cannot be fully characterized without reference to the picture’s design (that are responsible for seeing-in) and to the depicted object (what the design features are seen as being responsible for). In other words, ‘seeing design features as responsible for seeing-in’ could be plausibly taken to imply conscious attention to ‘design-scene properties’. We have made some progress: it seems that the Wollheim-inspired account of inﬂected seeing-in and Lopes’s views converge and provide us with a way of understanding the distinction between inﬂected and uninﬂected seeing-in. Of course, the account I presented here is very sketchy: maybe I should call it a general framework for analysing inﬂected and uninﬂected seeing-in rather than an actual account. I will attempt to ﬁll in the details of this general framework in Section 7. But before I do so, I need to address some important worries about both this sketchy account and the very idea of inﬂected pictorial experience.

6. Troubles with Inﬂected Seeing-In In the case of inﬂected, but not uninﬂected, seeing-in, as we have seen, ‘what is seen in a surface includes properties a full characterization of which needs to make reference to that surface’s design (conceived as such)’ (Hopkins, Chapter 6 in this volume, p. 158). If our seeing-in is inﬂected, we attribute properties of this kind to the object we experience. If it is not, we don’t. This may sound like a neat and clear-cut distinction between inﬂected and uninﬂected seeing-in, but what it amounts to depends on what we take the object of our experience (what we attribute these properties to) to be. More speciﬁcally, it depends on one’s account of seeing-in per se. Hopkins argues that the notion of inﬂection is difﬁcult to combine with what he calls the ‘divisive’ accounts of seeing-in, that is, those accounts

inflected and uninflected experience of pictures

197

that take twofoldness to be a necessary feature of seeing-in.⁸ If he is right, then the picture I outlined here must be wrong. I think Hopkins’s arguments are not conclusive, but they do put some important constraints on possible accounts of seeing-in (and of inﬂected seeing-in). I will outline an account of seeing-in and inﬂected seeing-in in Section 7 that satisﬁes these constraints. He has two complaints. First, he argues that if seeing-in entails twofoldness, then in the case of inﬂected seeing-in the picture’s design ﬁgures twice in our overall experience. It follows from the twofoldness claim that in the case of any seeing-in, inﬂected or uninﬂected, we must represent the picture’s design. But, in the case of inﬂected seeing-in, we represent the depicted object as having properties that cannot be fully characterized without reference to the picture’s design. Thus, design ﬁgures in both ‘folds’ of the twofold experience. But, Hopkins concludes, this does not reﬂect the phenomenology of seeing things in pictures. The ﬁrst thing to notice is that appealing to the phenomenology of seeing things in pictures may not be conclusive in this case. Take uninﬂected seeing-in ﬁrst. As the representation of the picture’s design is not necessarily conscious, design may not show up in the phenomenology at all. And in the case of inﬂected seeing-in, design may show up only in the characterization of some properties (the ‘design-scene properties’) we attribute to the depicted object. Hopkins’s argument seems to presuppose that seeing-in is twofold in sense (i). But I argued that we have little reason to think so, whereas we do have some reason to think that it is twofold in sense (ii). But let us proceed more slowly. When we see an apple in a picture, our overall experience attributes a number of properties to this apple. This overall experience represents some properties of the surface and it also represents some properties of the apple. Take uninﬂected seeing-in ﬁrst. I represent (typically without attending to them) some design properties. I also represent some properties of the apple. I attribute some design properties to the surface and, at least normally, do so without attending to them (and, at least sometimes, I do so unconsciously). I attribute some non-design properties to the apple, such as being spherical, etc. As I represent design properties without ⁸ Examples for ‘divisive’ accounts of seeing-in include Wollheim’s, Lopes’s, and Podro’s account as well as mine. I have also argued that Kendall Walton’s views on seeing-in are also divisive (Nanay 2004).

198 bence nanay attending to them, these properties may not even show up in my experience at all. Now let us turn to inﬂected seeing-in. If I experience this picture in an inﬂected manner, I still represent (again, without attending to them) the design properties of the picture. Similarly, I still represent the depicted object: I attribute non-design properties to the depicted apple. But, and this is the difference between inﬂected and uninﬂected seeing-in, I also attend to some ‘design-scene properties’: properties that cannot be fully characterized without reference to the picture’s design. Does this make design properties ﬁgure in my experience twice over? I don’t think so. It is only in attending to these ‘design-scene properties’ that design features show up in our experience. Our representation of design properties that makes seeing-in possible does not show up in our experience, as we are not attending to them. In other words, design shows up only in the ‘depicted object’ fold of our twofold experience, not in the ‘surface’ fold. But Hopkins has a second worry about the prospects of combining the idea of inﬂected seeing-in with what he calls the ‘divisive’ account of seeing-in. He raises the following question. If seeing-in is a twofold experience, then we need to be able to tell how we represent the two entities this experience represents: the design and the depicted object. We presumably see the design, but how do we represent the depicted object? Do we also see it? As he puts it, do we represent it by means of a ‘standard visual representation’ (as opposed to a belief or mental imagery)? He assumes that the answer should be yes. But then inﬂected seeing-in poses a problem. As he says: Ordinary experience presents objects as before one, i.e. as seen face to face. Inﬂected properties are properties that, without reducing to design properties (consider, for example, the property of being an upturned hand), nonetheless need characterizing, in part, by reference to features of design, conceived as such (e.g. by reference to inky strokes). The experiences . . . in which we are presented with inﬂected properties thus cannot even seem to be cases of seeing face to face. (Chapter 6 in this volume, p. 174)

Hopkins is careful to put this point as a challenge: every account of seeingin needs to be able to tell how seeing-in represents the depicted object and every account of inﬂected seeing-in needs to be able to tell how inﬂected seeing-in represents the depicted object. And it is not clear how those who

inflected and uninflected experience of pictures

199

want to combine a divisive account of seeing-in with acknowledging the possibility of inﬂected seeing-in could answer this question. I will try to meet this challenge in the next section.

7. Filling in the Details of the General Account of Inﬂected Seeing-In Hopkins is right. Any account of seeing-in must be able to tell how this experience represents the picture’s surface and how it represents the depicted object. And any account of inﬂected seeing-in must be able to tell how our inﬂected experience represents these two entities and how these representations are different from the way we represent surface and scene in the case of uninﬂected seeing-in. It is not enough to say that we see both the picture’s surface and the depicted object, because in this case it remains unclear how seeing the design in the case of uninﬂected seeing-in is different from doing so in the case of inﬂected seeing-in. I will outline an account of seeing-in and inﬂected seeing-in that does respond to Hopkins’s questions and that allows for the difference between inﬂected and uninﬂected seeing-in. It is important to make it clear that the view I will outline is one possible way of ﬁlling in the details of the general account I outlined in Section 5. If this speciﬁc view failed, this would not make the more general points I made about the difference between inﬂected and uninﬂected seeing-in above obsolete. In this section I will give one way of ﬁlling in the details of the general outline of the account of inﬂected seeing-in. But there may be other ways of doing so. As an account of inﬂected seeing-in presupposes an account of seeing-in per se, ﬁlling in the details of the general outline of inﬂected seeing-in mainly means giving a precise theory of seeing-in per se and then plugging the general account of inﬂected seeing-in into this theory. Hence, I will spend a lot of time giving a theory of seeing-in per se: of what is in common between inﬂected and uninﬂected seeingin and then turn to the differences between inﬂected and uninﬂected seeing-in. Humans (and other mammals) have two visual subsystems that use different regions of our central nervous system, the ventral and dorsal streams. These two streams can be differentiated anatomically and functionally. I

200 bence nanay will stick to the latter way of drawing this distinction. To put it very simply, the ventral stream is responsible for identiﬁcation and recognition, whereas the function of the dorsal stream is the visual control of our motor actions. In normal circumstances, these two systems co-function, but if one of them is removed or malfunctioning, the other can still function relatively well (Milner and Goodale 1995; Goodale and Milner 2004). If the dorsal stream is malfunctioning, the agent can recognize the objects in front of her, but she is incapable of manipulating them or even localizing them in her egocentric space (especially if the perceived object is outside the agent’s fovea). This happens if a patient is suffering optic ataxia. If the ventral stream is malfunctioning, the agent can perform actions with objects in front of her relatively well, but she is incapable of even guessing what these objects are. This happens in the case of visual agnosia. The philosophical implications of this physiological distinction are not at all clear. Some have argued that ventral visual processing is conscious, whereas dorsal is unconscious (Milner and Goodale 1995; Goodale and Milner 2004), but this view has been criticized both on empirical and on conceptual grounds (Dehaene et al. 1999; Jeannerod 1997; Jacob and Jeannerod 2003). It has also been suggested that dorsal processing gives rise to non-conceptual content, whereas ventral processing gives rise to conceptual content (Clark 2001). I do not need to take sides in either of these questions. All I assume here is that these two streams of visual processing are functionally different. The dorsal subsystem feeds into our perceptually guided actions and provides visual control for our motor actions; thus, it represents the perceived object egocentrically (see Nanay 2008 for details about what is meant by egocentricity). The ventral subsystem, however, feeds into our epistemic apparatus and helps us to recognize objects. This way of drawing the distinction between the two visual subsystems seems to be the common denominator between various interpretations of the anatomical ﬁndings: a similar distinction is made by Jeannerod (1997), Jacob and Jeannerod (2003), Goodale and Milner (2004), and Matthen (2005). Matthen somewhat idiosyncratically labels the two as ‘motionguiding vision’ (dorsal) and ‘descriptive vision’. Lopes (Chapter 2 in this volume) follows this terminology. I will use the dorsal–ventral terminology that is commonly used by psychologists, neuroscientists, and philosophers of mind.

inflected and uninflected experience of pictures

201

In healthy humans the way the dorsal and the ventral stream works can come apart in some circumstances, as in the case of the three-dimensional Ebbinghaus illusion. The two-dimensional Ebbinghaus illusion is a simple optical illusion (see Figure 2.5). A circle that is surrounded by smaller circles looks larger than a circle of the same size that is surrounded by larger circles. The three-dimensional Ebbinghaus illusion reproduces this illusion in space: a poker-chip surrounded by smaller poker-chips appears to be larger than a poker-chip surrounded by larger ones. The surprising ﬁnding is that although our judgement of the comparative size of these two chips is wrong as we judge the ﬁrst chip to be larger than the second one, if we are asked to pick up one of the chips, our grip size is not inﬂuenced by the illusion (Aglioti et al. 1995; some worries about the experimental conditions are expressed by Gillam 1998 as well as Franz et al. 2003).⁹ The usual way of explaining this ﬁnding is that our dorsal stream is not fooled by the illusion but our ventral stream is. The same results can be reproduced in the case of other optical illusions. In the Müller–Lyer illusion, while we (mistakenly) see the two lines as having different lengths, our eye and pointing movements represent them (correctly) as being the same (Goodale and Humphrey 1998; Gentilucci et al. 1996; Daprati and Gentilucci 1997; Bruno 2001).¹⁰ Similarly, in the case of the ‘Kanizsa compression illusion’ and the ‘hollow-face illusion’, our perceptual experience is fooled but our action is not (Bruno and Bernardis 2002 and Kr´oliczak et al. 2006, respectively). Thus, it does happen under exceptional circumstances that our ventral visual subsystem attributes a different property to an object from the one the dorsal subsystem does. My claim is that our visual system functions in a somewhat similar manner when we perceive pictures. I will argue that the dorsal and the ventral visual subsystems attribute different properties to the perceived object whenever we see objects in pictures. The ventral subsystem attributes properties to the depicted scene whereas the dorsal subsystem attributes design properties to the surface of the pictures. Or, to put it very simply, it is constitutive of our experience of seeing things in pictures that the depicted scene is represented ⁹ Some philosophical implications of these results are discussed in Abell (Ch. 3 in this volume) and Lopes (Ch. 2 in this volume). ¹⁰ These results have been questioned as they are difﬁcult to parse with the phenomenon that we also experience the Müller–Lyer illusion in the haptic sense modality (Heller et al. 2002, 2005; Suzuki and Arashida 1992).

202 bence nanay by our ventral vision, whereas the surface of the picture, together with its design properties, is represented by our dorsal vision. The claim that the ventral subsystem represents properties of the scene, whereas the dorsal one represents properties of the surface, entails the following four claims: (a) (b) (c) (d)

The depicted object is represented by ventral perception. The depicted object is not represented by dorsal perception. The surface is represented by dorsal perception. The surface is not necessarily represented by ventral perception.

I give detailed arguments for these four claims in Nanay (2008). My concern here is not seeing-in per se, but inﬂected seeing-in. I argued in Section 5 that seeing-in is inﬂected if we are attending to what I called ‘design-scene properties’: relational properties that cannot be fully characterized without reference to both the picture’s design and the depicted object. The question we need to ask is how these properties are being represented. The response, in short, is that they are represented ventrally. The difference between inﬂected and uninﬂected seeing-in is then the following: in the case of uninﬂected seeing-in, the depicted object is represented ventrally while the design properties of the surface are represented dorsally but not ventrally. In contrast, in the case of inﬂected seeing-in, the depicted object is represented ventrally and one of the properties it is represented (ventrally) as having is a ‘design-scene property’. As this property is relational and it cannot be fully characterized without reference to both the picture’s design and the depicted object, the fact that it is ventrally attributed to the depicted object means that at least some features of the picture’s design must be represented ventrally (given that the ‘design-scene property’ is represented ventrally). In other words, while in the case of uninﬂected seeing-in the design features are only represented dorsally, in the case of inﬂected seeing-in at least some of them are represented both dorsally and ventrally. This distinction between inﬂected and uninﬂected seeing-in is a version of the more general account I outlined in Section 5. As the ventral subsystem is what is taken to be the locus of attention, the fact that inﬂected seeing-in represents ‘design-scene properties’ ventrally could be interpreted as a version of the claim that in the case of inﬂected (but not uninﬂected) seeing-in, we attend to ‘design-scene properties’.

inflected and uninflected experience of pictures

203

It could also be thought to be consistent with Lopes’s view on the difference between inﬂected and uninﬂected seeing-in. Lopes gives the following characterization of inﬂected seeing-in: ‘Features of the design may inﬂect illustrative content, so that the scene is experienced as having properties it could only be seen to have in pictures’ (Lopes 2005: 123–4). If it is true that in the case of inﬂected seeing-in we ventrally represent the depicted object as having ‘design-scene properties’, then the scene is experienced as having properties it could only be seen to have in pictures: as having ‘design-scene properties’. In the case of uninﬂected seeing-in, it is not experienced as having these properties (although it is dorsally represented as having design properties). Now we are in the position to answer Hopkins’s second objection about the way we represent the depicted object in the case of inﬂected and uninﬂected seeing-in. In both cases we represent it ventrally. In the case of inﬂected, but not uninﬂected, seeing-in we also attribute some ‘design-scene properties’ to it ventrally. Does this count as ‘standard visual representation’? I’m not sure. If ‘standard visual representation’ means seeing face to face, then it doesn’t, as seeing face to face entails the ventral and dorsal representation of one and the same object. I argued that this is not the way we represent the depicted object (whether or not our seeing-in is inﬂected). We only represent it ventrally. But does this way of representing the depicted object count as perceiving it? Yes, it does. Ventral perception is perception all the same. Hopkins, rightly, pointed out that any account of inﬂected seeing-in (and of seeing-in in general) needs to specify how we represent the depicted object and the surface. I tried to meet this challenge in this section. One last worry needs to be addressed. One may worry about whether this account of inﬂected seeing-in can handle Hopkins’s ‘doubling of the design features’ objection. In the case of inﬂected seeing-in, we represent the surface both ventrally and dorsally. Does this mean that the design ﬁgures in our seeing-in experience ‘twice over’? I don’t think so. Each time we perceive something face to face, we represent it both ventrally and dorsally and it would be odd to suggest that this object ‘ﬁgures twice over’ in our experience. Finally, I need to emphasize that the speciﬁc account of seeing-in and inﬂected seeing-in I outlined in this section is just one way of ﬁlling in the details of the general account I gave in Section 5. If the speciﬁcs of my

204 bence nanay dorsal–ventral account of seeing-in are questionable, this does not cast any doubt over the general account of inﬂected seeing-in as conscious attention to ‘design-scene properties’.

References Aglioti, S., J. F. X. DeSouza, and M. A. Goodale (1995), ‘Size-Contrast Illusions Deceive the Eye But Not the Hand’, Current Biology, 5: 679–85. Bruno, Nicola (2001), ‘When Does Action Resist Visual Illusions?’, Trends in Cognitive Sciences, 5: 385–8. and Paolo Bernardis (2002), ‘Dissociating Perception and Action in Kanizsa’s Compression Illusion’, Psychonomic Bulletin and Review, 9: 723–30. Campbell, John (2002), Reference and Consciousness (Oxford: Oxford University Press). Carroll, No¨el (1995), ‘Mimesis as Make-Believe’, Philosophical Quarterly, 45: 93–9. (2005), ‘Aesthetic Experience: A Question of Content’, in Matthew Kieran (ed.), Contemporary Debates in Aesthetics and the Philosophy of Art (Oxford: Blackwell). Clark, Andy (2001), ‘Visual Experience and Motor Action: Are the Bonds Too Tight?’, Philosophical Review, 110: 495–519. Cutting, J. E. (1987), ‘Rigidity in Cinema Seen from the Front Row, Side Aisle’, Journal of Experimental Psychology: Human Perception and Performance, 13: 323–34. Daprati, E. and M. Gentilucci (1997), ‘Grasping an Illusion’, Neuropsychologia, 35: 1577–82. Dehaene, S., L. Naccache, G. Le Clec’h, E. Koechlin, M. Mueller, G. DehaeneLambertz, P. F. van de Moortele, and D. Le Bihan (1999), ‘Imaging Unconscious Priming’, Nature, 395: 597–600. Eimer, Martin and Friederike Schlaghecken (2003), ‘Response Facilitation and Inhibition in Subliminal Priming’, Biological Psychology, 64: 7–26. Feagin, Susan L. (1998), ‘Presentation and Representation’, Journal of Aesthetics and Art Criticism, 56: 234–40. Franz, V. H., H. H. Bülthoff, and M. Fahle (2003), ‘Grasp Effects of the Ebbinghaus Illusion: Obstacle Avoidance Is Not the Explanation’, Experiential Brain Research, 149: 470–7. Gentilucci, M., S. Cheiffe, E. Daprati, M. C. Saetti, and I. Toni (1996), ‘Visual Illusion and Action’, Neuropsychologia, 34: 369–76. Gillam, Barbara (1998), ‘Illusions at Century’s End’, in Julian Hochberg (ed.), Perception and Cognition at Century’s End (San Diego: Academic Press).

inflected and uninflected experience of pictures

205

Goldstein, E. B. (1987), ‘Spatial Layout, Orientation Relative to the Observer, and Perceived Projection in Pictures Viewed at an Angle’, Journal of Experimental Psychology Human Perceptual Performance, 13: 256–66. (2001), ‘Pictorial Perception and Art’, in Goldstein (ed.), Blackwell Handbook of Perception (Oxford: Blackwell). Gombrich, Ernst (1961), Art and Illusion, 2nd edn (Princeton: Princeton University Press). Goodale, Melvyn A. and Keith G. Humphrey (1998), ‘The Objects of Action and Perception’, Cognition, 67: 181–207. and A. D. Milner (2004), Sights Unseen (Oxford: Oxford University Press). Greenwald, Anthony, G. Sean C. Draine, and Richard L. Abrams (1996), ‘Three Cognitive Markers of Unconscious Semantic Activation’, Science, 273: 1699–1702. Haffenden, A. and M. A. Goodale (1998), ‘The Effect of Pictorial Illusion on Prehension and Perception’, Journal of Cognitive Neuroscience, 10: 122–36. Heller, Morton A., Deneen D. Brackett, Kathy Wilson, Keiko Yoneyama, Amanda Boyer, and Heather Steffen (2002), ‘The Haptic Müller–Lyer Illusion in Sighted and Blind People’, Perception, 31: 1263–74. Melissa McCarthy, Jennifer Schultz, Jayme Greene, Melissa Shanley, Ashley Clark, Samantha Skoczylas, and Jamie Prociuk (2005), ‘The Inﬂuence of Exploration Mode, Orientation, and Conﬁguration on the Haptic Müller–Lyer Illusion’, Perception, 34: 1475–1500. Hopkins, Robert (1998), Picture, Image and Experience: A Philosophical Inquiry (Cambridge: Cambridge University Press). Jacob, Pierre and Mark Jeannerod (2003), Ways of Seeing: The Scope and Limits of Visual Cognition (Oxford: Oxford University Press). Jeannerod, M. (1997), The Cognitive Neuroscience of Action (Oxford: Blackwell). Kr´oliczak, Grzegorz, Priscilla Heard, Melvyn A. Goodale, and Richard L. Gregory (2006), ‘Dissociation of Perception and Action Unmasked by the Hollow-Face Illusion’, Brain Research, 1080: 9–16. Kulvicki, John (2006), On Images: Their Structure and Content (Oxford: Oxford University Press). Levinson, Jerrold (1998), ‘Wollheim on Pictorial Representation’, Journal of Aesthetics and Art Criticism, 56: 227–33. Lopes, Dominic McIver (1996), Understanding Pictures (Oxford: Oxford University Press). (2003), ‘The Aesthetics of Photographic Transparency’, Mind, 112: 433–48. (2005), Sight and Sensibility (Oxford: Oxford University Press). Mack, A. and I. Rock (1998), Inattentional Blindness (Cambridge, Mass.: MIT Press).

206 bence nanay Matthen, Mohan (2005), Seeing, Doing and Knowing: A Philosophical Theory of Sense Perception (Oxford: Oxford University Press). Milner, A. D. and M. A. Goodale (1995), The Visual Brain in Action (Oxford: Oxford University Press). Nanay, Bence (2004), ‘Taking Twofoldness Seriously: Walton on Imagination and Depiction’, Journal of Aesthetics and Art Criticism, 62: 285–9. (2005), ‘Is Twofoldness Necessary for Representational Seeing?’, British Journal of Aesthetics, 45: 263–72. (2008), ‘Picture Perception and the Two Visual Subsystems’, in B. C. Love, K. McRae, and V. M. Sloutsky (eds), Proceedings of the 30th Annual Conference of the Cognitive Science Society (CogSci 2008) (Mahwah, NJ: Lawrence Erlbaum). Newall, Michael (2003), ‘A Restriction of Pictures and Some Consequences for a Theory of Depiction’, Journal of Aesthetics and Art Criticism, 61: 381–94. Podro, Michael (1991), ‘Depiction and the Golden Calf’, in N. Bryson, M. Ann Holly, and K. Moxey (eds), Visual Theory: Painting and Interpretation (New York: HarperCollins). (1998), Depiction (Cambridge, Mass.: Harvard University Press). Polanyi, Michael (1970), ‘What Is a Painting?’, British Journal of Aesthetics, 10: 225–36. Prinz, Jesse (forthcoming), ‘How Do Perceptual States Become Conscious?’, in Bence Nanay (ed.), Perceiving the World: New Essays on Philosophy of Perception (New York: Oxford University Press). Rensink, Ronald A. (n.d.), ‘Robust Inattentional Blindness’. Sedgwick, H. A. and A. L. Nicholls (1993), ‘Cross Talk Between the Picture Surface and the Pictorial Scene: Effects on Perceived Shape’, Perception, 22 (suppl.), 109. Simmons, Daniel J. and Christopher F. Chabris (1999), ‘Gorillas in our Midst: Sustained Inattentional Blindness for Dynamic Events’, Perception, 28: 1059–74. Strahan, Erin J., Steven J. Spencer, and Mark P. Zanna (2002), ‘Subliminal Priming and Persuasion: Striking While the Iron Is Hot’, Journal of Experimental Social Psychology, 38: 556–68. Suzuki, K. and R. Arashida (1992), ‘Geometrical Haptic Illusions Revisited: Haptic Illusions Compared with Visual Illusions’, Perception and Psychophysics, 52: 329–35. Vishton, P. M. and J. E. Cutting (1995), ‘Veridical Size Perception for Action: Reaching vs. Estimation’, Investigative Ophthalmology and Visual Science, 36 (suppl.), 358. Walton, Kendall L. (1990), Mimesis and Make-Believe: On the Foundations of the Representational Arts (Cambridge, Mass.: Harvard University Press).

inflected and uninflected experience of pictures

207

Walton, Kendall L. (1991), Reply to Reviewers, Philosophy and Phenomenological Research, 51: 423–7. (2002), ‘Depiction, Perception, and Imagination’, Journal of Aesthetics and Art Criticism, 60: 27–35. Wollheim, Richard (1980), ‘Seeing-As, Seeing-In, and Pictorial Representation’, in Wollheim, Art and its Objects, 2nd edn (Cambridge: Cambridge University Press). (1987), Painting as an Art (Princeton: Princeton University Press). (1998), ‘On Pictorial Representation’, Journal of Aesthetics and Art Criticism, 56/3: 217–26.

8 Seeing Things in Pictures JOHN H. BROWN

A half-century of theorizing about pictorial seeing has concentrated mainly on the central experience of seeing the subject in a picture. This consists variably of seeing an individual, identiﬁed or unidentiﬁed, real or ﬁctive, in a picture, and seeing that individual as being of a given type or as having given properties.¹ The latter, the depicted properties, are in general the most problematic. There is honourable dispute, as well as honourable uncertainty, about what properties, determinate or indeterminate, a subject is depicted as having in particular cases and in general. Can we see ongoing movement as opposed to arrested poses in depicted scenes? Can we see states of mind, subjective visions, as opposed to real or ﬁctive realities? Can we see the past or the future in the subject or is that merely inferred from what we see? How far beyond the shown, or ‘on-stage’, portion of a depicted scene can we see the subject as extending, and with what speciﬁcity? How far in front of the picture plane can we see the pictorial space as extending? Can we see depicted persons addressing ourselves, the viewers? Can we see a pictorial subject as referring to its being a pictorial subject? The scale of such problems varies considerably depending on the kind of picture. They rise to near global proportions in cases of conspicuously textured or abstract modes of depiction, an ascent that begins with Impressionism and Post-Impressionism and climaxes in Cubism and other avant-garde manners, where the depiction is often bafﬂingly cryptic. The questions about pictorial seeing do not stop with those about subjects and their properties. For, as has long been recognized, we can also see things in pictures that do not belong to the proper subject. Richard Wollheim put it this way: ‘With any representational picture there is likely ¹ Many add that pictures may also depict kinds (Goodman 1968: 21 ff.; Wollheim 1987: 69; 1998: 223; Lopes 1996: 152). For the contrary view, see Walton (2002: 27 ff.).

seeing things in pictures

209

to be more than one thing that can be seen-in it: there is more than one experience of seeing-in that it can cause’ (2001: 26). This raises the question, to what extent, if at all, seeing other things in a picture than the subject or other properties than those depicted as belonging to a subject is relevant to the understanding and appreciation of the picture. I think the relevance of such seeing is substantial and acutely underappreciated, and it is my purpose in this chapter to focus attention upon this neglected but essential aspect of pictorial seeing. I believe that all the main accounts of pictorial seeing (e.g. those of Wollheim, Walton, Lopes, Budd, Hopkins) fail to attend adequately to the ubiquity and signiﬁcance, for advanced pictorial seeing, of seeing other things in pictures than what I will call the authorized pictorial subject. As preliminaries, let me lay down some parameters. First, I will deal only with what I have just called advanced pictorial seeing. By this I mean savvy, resourceful, and patient viewing under favourable conditions by a person with an extensive background in pictorial discernment, detection and interpretation of visual art, and aesthetic appreciation of the world at large. Second, I am primarily interested in aesthetically appreciative pictorial seeing, not mere recognition of the pictorial subject. Third, I deal only with handmade pictures, since photographs raise issues that cannot be well dealt with within the limits of this chapter. Finally, I will restrict the inquiry to pictures that are stylistically uniﬁed and feature fairly robust pictorial depth. My procedure is as follows. First, I present in brief a particularly explicit discussion of seeing unauthorized things in pictures, that by Robert Hopkins (1998), and outline how, on his theory, such seeings-in are properly dealt with. In Section 2, I set forth my reasons for thinking that this solution is inadequate and the phenomenon itself is more signiﬁcant than he or others have thought. I also introduce Hopkins’s technical term ‘separation’. In Section 3, I argue that the phenomenon, in one of its two main forms, also crops up in fairly naturalistic full-colour paintings, not just in the special cases to which Hopkins refers; and that it plays a role in any advanced pictorial seeing of works within this extremely wide range. Section 4 considers and rejects two ways of evading the problem posed by the phenomenon. Section 5 introduces the second form of unauthorized seeing-in, here called spatial separation, which is the direct product of the reduction of dimensions from three to two in pictures generally, and

210 john h. brown argues that this too is aesthetically signiﬁcant. The remaining two sections develop the constructive role I believe separation seeing-in should play in advanced appreciation of pictures and the consequences that accepting this role would have for received theories of depiction.

1. Separation Seeing-In Hopkins (1998: 128 ff.) gives a name to unauthorized seeings-in, characterizing them as ‘separation’ phenomena because what one sees in the picture is to some extent separated from what one recognizes as the subject. I adopt and extend this technical usage (without his authorization) to include ‘separation seeing-in’, ‘separation subjects’, and ‘separation interpretations’. He cites cases where such seeing-in seems not just possible but inevitable or at least highly likely. One of his examples is a pencil sketch by Alfred Stevens in which the nude torso of a youth is rendered in a loose style (Figure 8.1). Here is Hopkins’s account of what a viewer might see in the picture as she responds to the graphic peculiarities of the rendering and sees in them what I call a separation subject with the following features: a man of an outline shape as determinate as that she sees the drawing as having.² That man is semi-transparent, especially at the top of his leg and headpiece. His nose and mouth blur indistinctly into the rest of his face. Across his upper arm and midriff run parallel curving strands of something. The little ﬁnger of his left hand seems malformed. (1998: 139)

Perhaps not all of these descriptions will survive a scrupulous review. Certainly they fall short of precision. But they come close enough to ² Note how extreme this is, if we take ‘as determinate as she sees the drawing as having’ strictly. It appears to imply that every surface mark seen by the viewer is seen in terms of a property of the anomalous ‘man’ seen in the drawing. A more reasonable necessary condition for separation is that some marks that are not subject-indicating are seen as if they were subject-indicating. There are obviously degrees of separation. The Ghirlandaio example discussed below is clearly a mild case. Perhaps Hopkins would not consider it one at all, since another strand in his view is that separation occurs when the resemblance between the design and the outline shape ascribed to the subject is insufﬁcient to support the viewer’s seeing the subject in the design, as in a stick ﬁgure that depicts a human. In my view this (negative) resemblance condition does not obtain in the case of the Ghirlandaio. There is enough resemblance to support the intended soft-skin subject when the work is viewed so as to make that look come forward. See Sect. 3 below.

seeing things in pictures

211

Figure 8.1. Alfred Stevens, Study of a Nude Figure

make Hopkins’s general point. The imagined viewer’s way of seeing is not without visual foundation. The free-ﬂoating lines comprising the headpiece suggest neither hair or a turban nor anything else we can identify. The interior of the right upper arm can easily be penetrated by a probing eye, justifying the ascription of semi-transparency. The left hand is seriously misshapen and joined to a dark V-shaped something or other, more solid than a shadow but without any other assignable physical nature. And these references are the tip of a substantial iceberg even when one excludes anything deriving simply from the limitations of the medium, as, for instance, colour, which is left indeterminate by the complete absence of the entire range of colour properties in the medium. Seeing a colourless

212 john h. brown youth in Stevens’s drawing would thus be excluded from what I call the separation subject. But that still leaves the viewer with a visual impression of a highly non-standard subject. Hopkins endorses a resemblance theory of depiction, according to which the viewer experiences the outline shape of the motifs in the picture as resembling the outline shape of the presumed subject (from a certain point of view). Thus, the overall outline shape of the motif in the Stevens is experienced as resembling the overall outline shape of a youth. But the interior details are very far from complying with this requirement. The upper right arm and the headpiece of the youth are (or can be) experienced as resembling strangely semi-transparent structures and so pose a problem for the theory. The solution is given by the viewer’s thought about the subject being controlled by practical knowledge of pictorial practice. This knowledge consists of three broad kinds: (1) ‘knowledge of what sort of items the world contains’, e.g. ‘that things are very rarely limited in colour to differing shades of grey . . . that dogs are not green, and that people are not composed of very thin cylinders’; (2) ‘knowledge of the sort of items [that are] in general depicted’, namely ‘items exhibiting something like the features real things enjoy’; (3) knowledge that depiction interpretations are conditioned by recognition of the limitations of verisimilitude achievable within a given medium and in rivalry with competing artistic aims. For instance, an artist may employ rapid, free-hand strokes that necessarily sacriﬁce accuracy or determinacy for the sake of verve (1998: 138, 144). This understanding of depiction practice, Hopkins says, will correctly convince the viewer that the artist did not intend her to take an anomalous manlike being to be depicted, but rather a (roughly) normal man who is indeterminately depicted. She will regard her seeing-in experience as ‘eccentric’³ in those respects (1998: 139). Accounts by others suggest variations of this general line. Wollheim, for instance, privileges the artist’s intended form of seeing-in, providing that the intention is fulﬁlled by the picture (2001: 26–7). Walton (1990, Ch. 4) offers a far more complex ³ What Hopkins says about the ‘marriage’ option (1998: 127–8) implies that in some cases knowledge of the depiction practice causes a person’s seeing-in experience to become as indeterminate as the corresponding aspect of the subject. Of the instant case he comments, ‘Stevens’ picture . . . seems to me to exploit Separation and Marriage in equal measure’ (1998: 139).

seeing things in pictures

213

picture of the many devices artists use to determine the depiction subject, which for him comes to the picture prescribing a set of (serious, artrelevant) games of make-believe.⁴ These responses serve to protect the depictive subject against misconstruction, but they do not provide a reason for dismissing separation phenomena as mere nuisances. The prevalence of such phenomena suggests that they have a more constructive role to play. There must be something basic about depiction that accounts for that prevalence, which has no equivalent in literary description.

2. The Role of Separation Seeing-In in Pictorial Seeing To obviate misunderstandings let several things be noted. First, although Hopkins never discusses the question whether separation seeing-in inhibits seeing the proper subject in the picture, there is no reason to doubt, so far as I can see, that we can maintain a lively awareness of both in any well-conducted course of pictorial seeing. In this and other respects our engagement with pictures seems both in fact and by right to consist of many concurrent and sequential strands in mixed modes, some more visual, some more ideational. This theme will be developed in due course. A second point, equally vital, is that comparable compatibility holds between separation seeing-in and seeing the surface design as such. The marks that excite separation seeing-in are also seen as marks on the surface. The seeing-in differs from the seeing-on but there is no psychological bar to experiencing the surface and attending to its particulars within a single episode of pictorial viewing. The same point applies mutatis mutandis to Wollheim’s twofoldness account, since separation seeing-in merely generates an alternative to the standard recognitional fold. It also applies to Dominic Lopes’s recognition model, since multiple sets of recognitional capacities are activated during any sustained and perspicuous viewing. A third point, doubtless more controversial, is that justiﬁed separation seeing-in is not an absurdity, even when it is not conﬂated with justiﬁed seeing of pictorial equivocation, falsiﬁcation, or error in a picture. Some ⁴ These games involve not only what is depicted (what one makes-believe seeing) but also what the picture non-depictively represents (what is properly attributed to the subject but not seen).

214 john h. brown of what Hopkins imagines his viewer to see in the Stevens drawing is, I maintain, justiﬁed. By that I mean that such seeing-in ﬁnds adequate support by identiﬁable cues in the picture and that its response to these cues is sufﬁciently in tune with established depiction practices to play a signiﬁcant role in the overall depiction project. When this condition is met, the viewer is entitled to entertain the thought of a separation subject. The artist has depicted the authorized subject by way of depicting a separation subject. The justiﬁcation for referring to what is seen in the picture as a subject depends on the properties in question being experienced as standing in pictorial space. Hopkins’s account leaves the impression that the separation phenomenon is relatively limited, arising mainly with free sketches, etchings, abstract representations such as in Cubism, and the like.⁵ I think this is far too narrow an extension to do justice to the effect that design features have on what is constructively seen in the works. I hope to make a case for separation phenomena being afforded by any picture that possesses substantial pictorial or optical depth (and perhaps even more pervasively than that). All such pictures, viewed correctly, elicit experiences representing a ﬁgure or scene which in some way and to some degree is distinct from, because incompatible with, the picture’s proper subject. I also hope the examples to be discussed will edge the reader towards agreement that it is not contrary to the expectation or intention of the artist that we see in the picture these ‘eccentric’ appearances; that to the contrary it is something that artists typically tolerate, desire, and exploit; that full and nuanced appreciation of the picture as a picture requires it, that art criticism could hardly proceed without implicitly recognizing it, and that much of the charm of pictures derives from such seeings-in being intended and taken up. These propositions can be boiled down to the ubiquity claim and the aesthetic signiﬁcance claim.

3. The Ubiquity and Aesthetic Signiﬁcance of Separation Seeing-In The Stevens example elicits separation seeing-in primarily by its manner of rendering, especially by the pencil strokes used. Since this comes under the ⁵ Cf. Hopkins (2003b: 663) on Picasso’s Les Demoiselles d’Avignon and Portrait of Gertrude Stein.

seeing things in pictures

215

heading of facture⁶ in art lingo, I shall speak of it as facture-separation. Let us consider how facture contributes to the ubiquity and artistic signiﬁcance of separation seeing-in. Supporting the ubiquity thesis requires varying the cases, so consider a ﬁnely textured full-colour painting such as Ghirlandaio’s Portrait of Giovanna Tornabuoni⁷ (1488). For all its ﬁdelity of colour and texture to the intended subject, which I take to be the sitter idealized, the ﬂesh is presented as somewhat wooden and the hair as artiﬁcially, almost metallically, stiff. Vasari complained of such things in the paintings of this period.⁸ Clearly the subject is not to be taken as wooden and stiff in these particulars and hence they must be separated from the set ascribed to Giovanna. Similar subtle features elsewhere in the painting also display separation but these two will sufﬁce for present purposes. From some theories of depiction one easily gains the impression that the viewer is supposed to overlook or set aside the discrepancies just mentioned, to see Giovanna’s soft, unstreaked ﬂesh right through the brushwork, her silky hair through the two un-silky-stranded buns. But I submit that any such ‘seeing through’, if possible at all, is either counterartistic (naive or negligent) or else impure. By impure I mean that the ‘eccentric’ ostensible hardness never is expelled from the experience but persists like a veil or double-image, not just on the surface of the panel either, but there, in pictorial space. You can visualize the subject fullﬂeshed, etc. Or you can imagine that your seeing the parts of the surface is seeing only properties the subject is depicted as having. But there is a more perspicuous, sharply focused experience of the work that involves the deviant properties too. The two experiences overlap. But of the two the separation seeing-in can claim more visual immediacy since it includes more of the visible facture. The authorized seeing-in has to override some of the potentially depictive marks, e.g. the semi-transparent effect in the Stevens. ⁶ Any marking or other handling of the surface with depictive or decorative intent counts as ‘facture’ in standard artistic terminology, whatever implements and materials are used. I am particularly interested in facture as depictive. Facture contrasts with what was once called ‘invention’, namely the arranging of characters and decor in a depicted scene. Both are design features. ⁷ Domenico Ghirlandaio, Portrait of Giovanna Tornabuoni (1488), tempera on panel, 77 × 49 cm, Museo Thyssen-Bornemisza, Madrid. The image is available in high resolution on the following website: . ⁸ Vasari (1965: 251): ‘the hard, dry, harsh style that art had acquired through the excessive study’. Vasari’s list of these artists includes Domenico Ghirlandaio.

216 john h. brown With regard to the aesthetic signiﬁcance claim, it seems equally certain that appreciation of Ghirlandaio’s distinctive style depends on not only seeing the surface facture as such but also seeing the impact of this in pictorial space. And whatever Vasari may think, some part of our enjoyment of Ghirlandaio’s image depends upon seeing the hardness of the texture, even though we also see the picture as depicting Giovanna as she is meant to be, soft-skinned and silken-haired. The hardness is part of his style, and that style is one of the aspects of his artistry that viewers know how to enjoy. As Vasari goes on to note, it was left to painters of the next period to achieve full mastery of the effect of ﬂesh, as in Leonardo’s Mona Lisa, in whose throat, he says, the pulse seems to beat, and that of hair, as in Correggio’s pictures, where ‘it appears soft and downy, with each golden strand ﬁnely distinguished and coloured, so that the result is more beautiful than in real life’ (Vasari 1965: 266, 253). When we turn our searching attention to them, however, we see that achieving these effects has consequences, for now some part of the sfumato of shadows extends like a light haze over everything, the like of which we never encounter face to face. Better than life, perhaps, but in its own way subtly, if endearingly, anomalous, not properly ascribed to Correggio’s subject.⁹ In stressing the importance of separation seeing in the Ghirlandaio I do not mean to suggest that there is anything illicit in subordinating one’s visual awareness of the slightly wooden look of Giovanna’s ﬂesh. To the contrary, I believe that at some point in one’s negotiation with the painting it is required to give the lady the beneﬁt of the doubt concerning her skin and imagine it to be as silky sleek as it may have been on the sitter’s young neck, and even to see it so in the picture if one can. How do we do it? In this case it’s easy. We simply don’t focus so unsparingly on the actual texture but let our gaze move elsewhere so that the skin falls into the periphery. This won’t work if the painted texture is too harsh. We can’t experience Giotto’s female ﬂesh as skin-soft. No manner of focusing or attending will yield that result. But many modes of depiction make it easy to soften the textural appearance. Since our address to pictures consists of many ocular ⁹ Correggio’s Mystical Marriage of St. Catherine in the Louvre serves admirably as an example for both skin and hair (c.1526–7), oil on wood, 105 × 102 cm. A good image is available at: . Click on the image to enlarge.

seeing things in pictures

217

ﬁxes, slides, scans, and repositionings, it sufﬁces that among them there is one that yields the effect best representing the subject as the maker desires. Often there is. The full depictive character of a painting is revealed only by the ensemble of seeings-in it offers and the varying salience or faintness, persistence or transitoriness, of members of the ensemble—including close relatives of Ernst Gombrich’s illusions (Lopes 1996: 8 ff.; 2005: 29 ff.), depth effects a diligent observer can induce as compelling as the endless ascent of the spiral on a barber’s pole. Exploration of a picture involves all of these modes of address appropriately staged, gathered, and reﬂected upon. That they cannot all be packed into a single moment of full and undivided comprehensive salience—a master synthesis—is neither here nor there.¹⁰ A more dramatic case of facture separation than the Ghirlandaio, in a work more impressive than the Stevens, is afforded by Rembrandt’s drawing of Jan Cornelisz Sylvius (see Figure 6.1). The drawing is full of handles for separation seeing-in which a diligent viewer will grasp: the scribbles, the ink strokes heavy and dry, and much more. When we regard the drawing with separation on our mind, one general feature emerges immediately, namely the difﬁculty of imagining substantive counterparts of the graphic features that elicit separation. Many things in the separation subject stand at different depths without visible support. What are we to imagine sustains them? Do they ﬂoat as if suspended in a solution or are they marks on the surface of perfectly non-reﬂecting glass (non-reﬂective from the indicated point of view), or what? What physical constitution are we to ascribe to them? They are clearly marklike, splotch-like, scribble-like, but they stand in pictorial space instead of occupying areas on any detectable surface. We have no physical categories within which to lodge them. This applies to parts as well as to the whole. Seeing-in, whether separation or authorized, is permeated by thought (Wollheim, Hopkins), but here the thought is of a strikingly indeterminate and ﬁnally incoherent sort. It is not one of which we can make an intelligible depiction subject. Not surprisingly this has repelled theorists. Philosophers do not welcome incoherence in an explanation of anything. And since the derivation of the forms in the separation subject to the marks on the surface is evident, it is understandable that they are content ¹⁰ As Walton (1990: 308) says, ‘the experience of seeing-as or seeing-in is not a momentary occurrence but a continuous state’. Just so: it is a continuous state with many phases.

218 john h. brown to consign the problematic forms to the surface merely, where there is no conceptual problem. Yet the forms that matter stand resolutely in pictorial depth, not on the surface.

4. Is Separation Seeing-In Reducible to Inﬂection? If they are so inscrutable, one may wonder, can it be legitimate to posit a separation subject at all? Is it possible that what I have called justiﬁed separation seeing-in is reducible to seeing a real subject looking as if it had the anomalous properties? I think this reduction will not work. It requires an elaboration of the pictorial subject to include a manner of looking. But formidable difﬁculties stand in the way. The fact that we do not normally take these (these particular) subjects in the X-looking-like-Y way is signiﬁcant. To do so would introduce a bizarre twist into the subject which the depictive tradition does not authorize. Moreover, for many examples there is an even tougher impediment. There is no known way for the subject to look as it would have to look on the proposed reduction. In the Stevens example the idea of a normal youth’s headpiece looking like a snaky cordlike tangle overlying a featureless opaque covering is hard—for all one can tell, impossible—to develop into a coherent, visually realizable idea. The same is true of Sylvius looking like the fantastic being in that indecipherably cluttered setting. When the surface marks are as free as in the Stevens or Rembrandt examples, there is no good candidate for an X-looking-like-Y subject. In the case of Giovanna the difﬁculty is not nearly so extreme, but there is also the impropriety of taking Giovannawith-wooden-looking-skin to be the intended subject. Ghirlandaio would never have earned his commission, or deserve his present reputation, if that were the case. Another way of avoiding positing separation subjects comes under the rubric of inﬂection, an idea that has ﬁgured prominently in recent discussions of pictorial seeing, e.g. in Lopes (2005: 123–4). The idea is preﬁgured in a general way in Wollheim (2001: 20) and somewhat more speciﬁcally in Michael Podro (1991, 1998). The term ‘inﬂection’ is not used, but the key ideas are advanced of the surface markings being ‘recruited’ to representational ends and of their interpenetrating or intermingling with properties of the pictorial subject in a way to which there is no analogue in

seeing things in pictures

219

face-to-face experience of a reality corresponding to the subject. Chapters 6 and 7 in this volume, by Hopkins¹¹ and Bence Nanay, moot this idea in some detail, giving to it a yet more speciﬁc meaning, as in this, by Hopkins: inﬂected seeing-in occurs when ‘what is seen in a surface includes properties a full characterization of which needs to make reference to that surface’s design (conceived as such)’ (Chapter 6, Section 2). Hopkins applies this idea to Sylvius’ outstretched hand, following leads provided by Podro (1991: 170 ff.; 1998: 9 ff.), claiming (1) ‘the hand itself seems to be both body part and rising splash of ink’ and (2) ‘what is seen in this picture is a hand composed of rising ink’. From these he concludes: ‘what is seen in needs characterizing in part by reference to properties of the picture’s design’. While I fully endorse Hopkins’s openness to seeing anomalous things in pictorial space, I ﬁnd it hard to believe it can be quite right to see in the inky splotch a hand composed of ink. I suggest that what can be immediately seen in it is a messy hand-like form, at the same time as one sees the splotch as depicting a normal human hand. For what could a hand composed of ink be, rising or still? (Actually the hand is thrusting forwards rather than rising, but let that pass.) Certainly a hand cannot be a splash of ink and the idea of a hand being seen as that seems equally unintelligible. Perhaps a hand might be seen as ink-covered or composed of congealed ink, but no such thing is seen in the picture.¹² Further, I fail to see how the description of what is immediately seen need be taken as necessarily referring to the design. ‘Ink-splotch-like’ or even ‘composed of ink’ certainly need not so refer. Nor is such reference demanded simply because the seeing-in is based on seeing the inky splotch on the surface. If that were so, every mark depictive of a receding form would be inﬂected. Therefore, I cannot see that Hopkins’s criterion of an inﬂected property (one the full characterization of which necessarily makes reference to the surface) is satisﬁed by the cited case. Also one other idea in the inﬂection story, namely that some properties recruited to the subject cannot be found in any conceivable face-to-face ¹¹ Hopkins has suggested to me that inﬂection can be conceived of as a form of separation. So conceived it need not be in conﬂict with my view of separation depending on how the details go. ¹² It is instructive to inquire whether it is possible to clearly depict, within any known depictive tradition, the separation subject of the Rembrandt drawing. Even when one allows for feasible extensions of the known traditions, the answer, I think, is negative because the content is not sufﬁciently coherent and speciﬁc to permit it to be clearly depicted (as the real subject).

220 john h. brown experience, seems not applicable to the real subject, which repels such anomalies. What comes closest to meeting the criteria of inﬂection seems to me to be the following. The signs of particular ways of image-making confer on the drawn or brushed marks the look of dynamic and behavioural characteristics.¹³ Lines or brushstrokes are swift or slow, conﬁdent or hesitant, bold or timid, heavy or light, dry or wet. Calligraphy exploits these descriptively aesthetic properties within the two-dimensionality of the surface without any involvement of pictorial space. So the effect of boldness or timidity is conferred on the lines or brushstrokes and this can be transferred to their immediate analogues in pictorial space. We seem to see this in Rembrandt’s drawing: both mark-like forms and Sylvius himself acquire a look of boldness. For it is doubtful that we could see Sylvius as exhibiting so forceful a personality, or possessed of so assertive a state of mind, without the brio of the brush marks and their three-dimensional counterparts. This is conﬁrmed by the etching for which the drawing was a sketch, since in that he does not come across as forceful (see Figure 6.2). Here we have a case where something like ‘interpenetration’ occurs. But note: what attaches to Sylvius (the true depiction subject) is only the behavioural and temperamental properties. The mark-like conveyors and their distinctive dynamism don’t. These never get beyond the separation subject. So there is no importation into the (true) subject of anything that could not be seen in it face to face—not perhaps in a freeze frame but in the course of a few seconds of live action. The way it is seen differs, but seen it is.

5. Spatial Separation Seeing-In So far we have been considering only separation based on some part of the picture being less than maximally verisimilar of its kind due to its facture. But there is another aspect of separation not mentioned by Hopkins or fully acknowledged by anyone else, but central to my account, which I ¹³ The role of knowledge of the manner of production in drawings has been stressed by Philip Rawson (1987) and Patrick Maynard (2005). The latter usefully develops the detection function of the marks, which in turn serves to prompt the viewer’s imagination of the mark itself as possessed of the qualities of the action: the pressure, the velocity, etc.

seeing things in pictures

221

will call ‘spatial separation seeing-in’. This aspect would operate even if the picture offered no opening for separation within the constraints of what is possible given the medium (including the implements for working it). It operates over all pictures because they are pictures, depending only on the projection system used. Consider Michelangelo’s drawing of one of the nudes from the Sistine ceiling now in the Albertina Collection.¹⁴ What position of the legs do we see in the picture? From the correct viewing station the legs are seen extending out towards the viewer at a roughly forty-ﬁve degree angle. But, as we well know, seen from an oblique angle the effect is different. As we move to our left, the legs swing closer to ninety degrees, and if we move to the right they swing closer to the picture plane. Manet’s dead toreador (The Dead Toreador (1864?), National Gallery, Washington, DC¹⁵) notoriously pivots similarly to right and left as the viewer moves laterally. What legs and bodies do, eyes do also. If they are ever directed towards the viewer, they follow her as she moves laterally. More generally, the space immediately seen in the picture undergoes striking contortions, expansions, and compressions as the picture is viewed from different stations.¹⁶ Yet the depicted subject does none of these things. These anomalies are a direct consequence of the discrepancy of dimensions (two vs three), the projection system (perspective),¹⁷ and the fact that vision is irremediably perspectival. Even if the anomalies are not always noticed consciously, there is no question but that they are available, that they operate at a preconscious level, and that open-minded viewers can be easily taught to notice them. If this were not so, viewers would experience a full illusion of an unchanging subject or else would be incapable of seeing a subject in a picture, i.e. would see the picture only as a surface.¹⁸ ¹⁴ Michelangelo Buonarotti, Nude Image of a Seated Young Man (1510–11?), light and dark red chalks, heightened with white, 16.8 × 18.8 cm. In the Albertina Museum website, search ‘Renaissance’, click ‘Renaissance and Mannerism’. ¹⁵ . ¹⁶ Anyone in doubt as to the perceptual impact of these distortions is invited to walk back and forth in front of the oversized photographs of Jeff Wall. The architectural features of the interior scenes are particularly subject to distortion, as are landscape features such as roadways and ditches depicted as orthogonal to the picture plane. Viewed up close the space seen in the photographs becomes radically compressed in depth, and so forth. Viewers who miss these deformations are not attending closely to what is there to be seen in the works. ¹⁷ Pictures in parallel projection that depict depth are subject to comparable but speciﬁcally different deformations. ¹⁸ I am aware that this ﬂies in the face of the alleged constancy of the pictorial effect (or robustness, as Kubovy 1986: 52 ff. calls it). But the effect of constancy obtains only when pictures are viewed with

222 john h. brown The full menu of ocular and optical peculiarities of picture-viewing compared with those of ordinary perception is well known but has not, to my knowledge, been adequately taken into account by the main theories of depiction. Brieﬂy they are: (D1) Visual experience, binocular or monocular, of pictorial subjects lacks the selective clarity and blur that depends on the accommodation of the eye to distance. (D2) It also lacks the stereoscopic accommodation for distance and therefore any double-imaging of objects viewed out of focus. (D3) No parallax occurs within depicted scenes with change of the viewer’s position: no seeing around objects or sliding of near objects in front of far ones as we move. (D4) From non-optimal points of view the subject seems to deform in response to the changes of the stimulus. This is especially noticeable in architectural pictures and in large pictures with space-marking motifs (orthogonals, diagonals, etc.) Perceptual constancy is then no match for the power of the stimulus. Far from random, the deformations obey strict principles of ‘invariance’ (Gibson 1950: 152–4; 1979: 310–11). (D5) As a result of the foregoing, pictorial space is presented as discontinuous with the viewer’s ecological space and is perceived as such whenever the viewer moves about in the normal area of viewing. (D6) Visual experience when viewing pictures lacks the reduced acuity of perception of things far in the distance, compared with nearer things. (D7) Picture-viewing yields reduced illumination compared with faceto-face viewing of counterparts.¹⁹ D1–D7 cover what the visual array in pictorial seeing offers to our visual systems. How easily noticed these aspects are is another matter. For less than scrupulous attention. Also constancy depends on clear indications of the orientation of the surface, such as the rhomboidal projection of its format to the viewer, almost always reinforced by other environmental markers, or on viewing from so great a distance that the alterations in our visual ﬁeld of the format and environment are slight. Not all features are as robustly inconstant as others. I ﬁnd, for example, that viewing pictures on a slanted surface more readily results in an apparent shortening of the image than the impression of parallel sides of the format yields to an apparent non-rectangularity. ¹⁹ Reduced illumination might be thought a characteristic of the subject, but in fact we do not ascribe it to the real subject. Hence it must be counted an attribute of the mode of presentation. This does not prevent it from being absorbed into the separation subject as one of its attributes.

seeing things in pictures

223

instance, the total scenic depth effect (as opposed to more local ones) in pictures varies greatly with the properties of the picture, the circumstances of viewing, and the address of the viewer. Pictures with strong and consistent depth cues, including chiaroscuro and atmospheric gradients and well-managed colour gradients, are good depth-carriers, especially when the depicted depth is considerable. Surface texture that calls attention to the picture’s surface weakens the depth effect. The size of the picture often matters. Small pictures are in general less potent than large ones, often because optically normal viewers cannot get a well-focused eyeful of the pictorial subject from close enough to the surface. Large ones helpfully ﬁll the viewer’s visual ﬁeld, weakening the saliency of ambient space. Relevant viewing circumstances include how the picture is hung and illuminated. Pictures are often hung too high or too low for viewers to view them from the right height. The way they are illuminated can strengthen or diminish the depth effect. Depth ﬂourishes when the picture surface is brightly but non-reﬂectively lighted and the ambient illumination in the room is kept low. Finally, as to manner of viewing, it is widely known that monocular viewing greatly intensiﬁes the depth effect. But also a manner that can be called looking for depth is important. It consists in part of following out the sequence of (depicted) planes, noticing near and far, keeping tabs on how things look in the periphery as one changes one’s direct gaze, and in general ‘thinking depth’. Also relevant is noticing how the pictorial space squirms as one moves from canonical to non-canonical distances and lateral (or vertical) positions, and how it resumes its proper relations when one reverts to the canonical point of view. Recent vision research shows that pictorial space is perceived as shallower than accords with the subject.²⁰ So viewers register what they may not consciously notice when viewing the picture in the normal way. A related phenomenon that I believe any careful observer can experience is the tendency of farther planes to drift forwards in picture space, a phenomenon found in face-to-face viewing only under exceptional conditions of deceptive alignment or illumination. The result for separation seeing-in is that ²⁰ See Koenderink (1998). As nearly as I can tell, it is also conﬁrmed by the comparatively simple experiments reported by Hagen et al. (1978). Magritte’s famous painting The Human Condition (1933), showing a painting against a landscape which it exactly depicts (), suggests a simple experimental setup for testing the comparative shallowness of pictorial as opposed to real depth.

224 john h. brown we see a shallower depth in the picture than the perspectival indications require us to allot to the real subject, even from the optimal point of view under optimal viewing conditions. This means that advanced pictorial seeing will include recognition of such features as belonging to what can only be a separation subject. A consequence of D1 concerns allegations about the visual ﬁeld or outline or sensational shape that plays a central role in the theories of depiction of Malcolm Budd, Hopkins, and Christopher Peacocke. Each theory describes the shapes (let me lump them together under the rubric of stimulus shapes, with which each has a basic relation) as two-dimensional.²¹ Each theory takes off from the reasonable idea that the stimulus shape commonly relates pictures to the three-dimensional views they represent. What D1 implies is that our experience of the two is importantly different. The stimulus shape abstracted from the ambient visual array in face-to-face perception may be two-dimensional, but its parts are not accessible in sharp focus at a (nearly) single depth, as are those of the shape on the surface of a picture. In ecological perception our focus has to leap out and back following the varying depth of the surfaces that generate the stimulus. This difference in ocular reception of the two is basic to the difference between the pictorial and the perceptual effect. Moreover, it implies that neither stimulus is as a whole two-dimensional but variously extended in depth, each according to the depth of the parts of the distal source. Just how deeply this difference in stimuli of picture viewing as opposed to face-to-face viewing impacts the theories of depiction that use the concepts of visual ﬁeld or outline shape is a question I prefer to dodge. To take up that question for the three theories would lead too far aﬁeld for present purposes. What is sure is that the difference of stimulus helps explain the difference between what the two sorts of visual experience are like, even apart from resemblance of outline or visual ﬁeld shape. For note, no manner of perception of which eyes like ²¹ Hopkins (2003a: 152) deﬁnes ‘outline shape’ of an object in terms ‘of the directions of its various parts from a point in its environment’ corresponding to the point of view (2003b: 367), hence objectively; Budd (1993) deﬁnes ‘visual ﬁeld shape’ as the viewer’s visual world from a point of view with depth removed, hence incorporating whatever properties the viewer ascribes to the scene that do not imply depth; Peacocke (1987) takes visual ﬁeld shape to be in ‘visual sensational space’. Each philosopher endorses the planarity of the shape. Though Hopkins requires relationship to a point of view, he speciﬁcally holds that ‘distance is irrelevant’ (2003a: 152). Thus, the connection with the stimulus, even if not identity, is close. I use ‘stimulus’ rather than Gibson’s ‘ambient visual array’ only because it is shorter and more familiar.

seeing things in pictures

225

ours are capable could eliminate this difference, given the difference of stimulus. The constitutive differences of ordinary and pictorial perception by themselves sufﬁce to ensure one sort of incommensurability between pictorial seeing-in and seeing face to face. At the same time, the peculiarities are such as compromise the resemblance between the spatial relations in the real subject and those in the separation subject. The basic point can also be put in terms of making believe: when we make believe our seeing of the surface is a seeing of the (real) subject, for us to ﬁxate on these properties would defeat the purpose. At the same time, advanced picture viewing is prompted to become conversant with these anomalies to see what values they may yield. That spatial separation seeing-in at the most basic level is afforded by even the most naturalistic picture conﬁrms the ubiquity thesis, since what we see in the design differs anomalously from anything afforded by face-to-face seeing of the actual subject either from given viewing stations or in moving from one station to another. At the same time, the reduction in the depth effect presents signiﬁcant aesthetic advantages, notably a more uniﬁed ﬁeld. Artistic interventions can increase the advantage by further weakening the depth effect, accentuating shallowness, putting more stress on planarity, principally by appropriate use of the resources of facture.²² Doing this is one of the deﬁning tendencies of modernist art. Matisse’s The Egyptian Curtain, discussed below, is a paradigm case of this mode of separation. In traditional modes the interventions more often aim at compensating for the basic deﬁciency of depth, which gives rise to a different range of separation seeing.²³ As the separation subject passes through states of expansion, contraction, and contortion with the viewer’s change of position, in almost all cases one state best accords with the real subject and therefore serves our interest in seeing what the picture depicts. Call this the canonical state of the separation subject. By this I do not mean that other states are to be avoided. Often they cannot be if we want to enjoy the picture at all, for instance when ²² The reverse effect of continuous recessiveness into an indeﬁnite distance, described by Heinrich W¨ollﬂin, is achieved by uninterrupted textural gradients and unstressed contours inter alia (W¨ollﬂin 1950: 73–106). ²³ The well-known practices of enlarging distant features or diminishing the convergence in near scenes are instances. See the discussion of Caravaggio’s Supper at Emmaus in Dubery and Willats (1983: 80).

226 john h. brown it is too small or placed too high. And much sensitive and insightful appreciation is possible from these other viewing stations: ones too close or too far, or laterally or vertically eccentric.²⁴ Still, giving such credit as is due to non-canonical viewing should not blind us to the fact that something is lost when the canonical viewing station is neglected.²⁵ It provides an irreplaceable point of reference. As to artistic signiﬁcance, two facts are worth noting. First, the canonical state of what is (separation) seen in the picture is subject to the artist’s choice of implied point of view.²⁶ Perspective can vary laterally, vertically, and in depth. Thus, in some Dutch architectural painting the canonical point of view may be more to the side than is usual.²⁷ When such a picture is viewed head-on from the usual height and distance, the state of the separation subject is decidedly non-canonical. But that implies no less canonicity, and also no separation beyond the basic. The situation is different for extreme anamorphs: in spite of the fact that the intended subject is better seen from the acute point of view than from any other, the subject itself is not seen nearly so well as the same subject treated conventionally and viewed canonically. The disparity of focus between near and far parts of the anamorph is too great to give us optimal access to the subject. Thus, what a viewer sees in the anamorph exhibits spatial separation seeing-in beyond the basic. ²⁴ A remarkable example of an interpretation that gives equal signiﬁcance to the canonical and one particular non-canonical point of view is proposed by Patrick Maynard of Dürer’s much discussed engraving St. Jerome in his Study (1514) (Maynard 2005: 176–83). This interpretation attributes two different scenes, not just two views, to the picture. ²⁵ Podro (1998: 71, 75) gives testimony conﬁrming this when he observes, regarding the depicted gazes that set up a communicative relation with the viewer: ‘it is only at a certain viewing distance and momentarily that we can imagine ourselves in . . . an exchange of looks. When we go too close we no longer experience their eyes as coordinating their gaze on us and they seem to look behind us.’ Just so. ²⁶ John Hyman (2006: 78) holds that only the line of sight is relevant, but this is clearly wrong, since objects curving into depth reveal less of their margins from close-up than from a distance along the same line of sight. Occlusion patterns among objects at different depths are also altered by distance. Hyman’s idea applies only to cases where parallel projection is used. ²⁷ A good example of this is Saenredam’s The Interior of the Grote Kerk at Haarlem, National Gallery, London, discussed by Dubery and Willats (1983: 87–8). An image is available in that text and on the National Gallery website: . The text on the website erroneously says the artist altered the perspective. What he did was to choose an up-close viewpoint close to the left side. To minimize the considerable distortion resulting from viewing the picture straight away at the normal distance, the artist set its lateral termination in the middle of the ﬁrst and last of the three columns. The ameliorating effect of this is dramatically illustrated when we see an expanded version of the scene, such as is given at under ‘The Illusions in the Picture’.

seeing things in pictures

227

Canonical states may also differ for different parts of a picture. Not infrequently there is more than one implied horizon (i.e. implied viewing level). Then a given part of the subject is presented non-canonically from the point of view that is canonical for another part. Much good can come from these artistic choices, which imply not just eccentricity but inconsistency.²⁸ But whether the picture is consistent or inconsistent, full recognition and enjoyment of its artistry as a depiction depends on giving priority to seeing its subject or its subject parts canonically, even as one maintains an appropriate awareness of non-canonical effects.

6. Implications for the Purposes of Pictures Pictures serve diverse purposes. A central use is that of visually conveying information about the various aspects of a subject, whether that subject is real or ﬁctive. What information is conveyed will depend on the system of depiction, which Lopes and others plausibly take to consist of the commitments and non-commitments regarding what aspects the pictures in that system aim to make recognizable (Lopes 1996: 128). Systems also vary in how detailed and multifarious is the content they convey. ‘Information’ covers both real and make-believe or ‘groundless’ information, the latter applying to ﬁctive subjects. This aspect recognition theory, like its resemblance or seeing-in rivals, is dedicated to basic pictorial content, and in that respect is faithful to what is reasonably taken as the primordial function of pictures. That is, pictures are made in the ﬁrst instance to convey information about actual things, real or prospective, or things believed to be real. Make me a picture of X, someone says. A draftsman complies. It is understood at the outset what sort of thing or scene is to be depicted. Methods of depiction arise in the ﬁrst instance in that context. What counts and what does not depends on what works well enough for the purpose at hand. In this enterprise knowledge of the real world is an essential formative control or reference. Nothing else could possibly have as deep a connection to the pictorial or any other mode of ²⁸ Examples can be found on my website: . For starters, see ‘Getting Deeper into Pictures via Digital Transformation: A Tale of Two Horizons’ and ‘The Role of Monocular Viewing in Pictorial Appreciation’. David Hockney (2001) displays many other examples.

228 john h. brown communication. From such control and the selectivity it enforces comes the normal pictorial subject. But our capacity to see other things in pictures than the subject raises questions not answered by the normal systemic commitments. That the design properties of pictures can and should be seen in ways that disregard the system’s commitments and non-commitments calls for an explanation. Viewers sensitized to depictive systems with a rich and subtle range of content, habituated to an artistic culture that retains the memory of many modes of representation, many styles, many expressions of artistic personality and artistic self-consciousness, and especially a culture that encourages creative diversity, seek to ﬁnd more in pictures than is an authorized part of the subject, and do so appropriately. What seems at ﬁrst sight to be waywardness or idiosyncrasy turns out to be a consequence of advanced pictorial understanding. It arises primarily from the interest in the full and varying impact of graphic (or painterly) marks on our pictorial experience. It asks, what does this ensemble of marks suggest, of all the forms in my repertory of depictables? To what seeing-in experiences does it give access? Thus, design features are taken as an invitation to search out the various depictive suggestions created by the design, to test their strength, their consistency or inconsistency, using criteria based on elements of the very system that generated normal depiction subjects. Once one can see that subject, one can see farther, ‘eccentrically’ and yet with artistic and aesthetic relevance. This use of pictures seems no less valid, indeed just as compelling, for viewers and artists alike, than is blinkered adherence to the real subject. Strictly limiting oneself to the constraints of the depiction subject seems unacceptably restrictive when we ﬁnd ourselves deeply absorbed in the full range of visual experience afforded by the picture itself. The provocation to engage in seeing-in adventures depends greatly on the type of picture. For many pictures where separation is particularly strong, this freewheeling approach ﬁnds itself dealing with things uniquely accessible by way of separation seeing. Not only that. The world of the actual subject becomes less dominant. It tends to recede behind or within the extravagant surface features. Or it declines into bland indeterminacy and loses its allure. The normalizing restraints on pictorial subjects listed by Hopkins and others seem inadequate to account for the interest we take in such pictures. More and more interest gravitates towards suggestions spawned by cryptic texture and problematic spatiality. The main point

seeing things in pictures

229

of the picture seems to be to present something stranger (more lyrical, more surreal, more teasing) than any normal, rule-compliant (even when ﬁctional) world. At the same time these ﬂights presuppose the normal information-transmitting pictorial practice as an underpinning, as poetry presupposes a prose paraphrase. The case for this drift of pictorial seeing beyond the normal is easiest to make for modernist pictorial representations, for all varieties of Cubism as well as less structured sorts of naturalism. Consider an instance of the latter, Matisse’s 1948 painting The Egyptian Curtain. Anomalies leap out from this painting. The curtain is seen as falling in front of the table, which is unusual, since normally curtains are next to the wall. But probably we should count that an oddity of arrangement, hence a property of the authorized subject and not a case of separation. The leaf shapes in the fabric design standing out against the untextured black expanse representing the curtain²⁹ are another matter. We know that they, being mere patterns on the fabric, follow the folds of the curtain. How could they not? But this implication of the authorized subject does not constrain our immediate seeing of the leaves. Their contours twist and turn in space according to a subtle conspiracy of design indications. Thus, the green leaf can be seen to curve along the rounding of the curtain, as it does in the real subject, but also with equal force to extend out from the curtain like a scimitar, which can only be in a separation subject. In so doing, it links up with the pattern of protrusive palm fronds, turning a broad, downward arc into a jauntily upturned, aggressively protrusive blade. Other anomalies are resident in the pattern, comparable to the scimitar, giving the picture a richly sportive suggestiveness. That same green leaf can also be seen (without forcing, without fault) as curling broadside to the viewer’s gaze (another separation twist). Each of these orientations is immediately presented in depth, is seen depth-wise, as the eye plays with the pattern. This is abetted by the perceived ﬂatness of the surface and painterly ﬂattening of the curtain, without which that particular palimpsest of conﬂicting appearances would not be so salient. The conﬂicting readings replace one another as naturally as do the changing but entirely compatible visual impressions that occur as one explores an actual scene, and they too ²⁹ An image of the painting, sufﬁcient for the points I want to make, may be accessed at: . Or Google ‘Henri Matisse Egyptian Curtain’ and select among the many sites.

230 john h. brown leave a residual presence behind. None can elbow the others offstage for long, and those upstaged merely retreat to the margins. Things become even more radical when we ask about the effect of the whole. Does the artist or the sensitive viewer propose that we interpret the painting simply in terms of his hotel room (in Nice, perhaps) and the view outside, which is almost certainly the original setting, the starting point? Anyone who thinks so is invited to imagine sitting as the artist presumably did in that room, looking through the large mullioned window at the crown of a palm tree in the blazing sun set off by a curtain close up on the right and a table with a bowl of pomegranates between window and curtain and about level with the bottom of the window. If one imagines this, working from the painting, and tries to see it as being what the painting means to engage our interest in, one has done all one can to normalize the depiction subject in scenic terms. Does that do justice to the subject of this painting? I think not. When I try the experiment, discrepancies leap out at me. No experience of a normal palm tree, normal table covering, normal pomegranates, normal curtain and wall surfaces, can match up. The dissonance between that commonplace subject and the scene given to us by the painting is irresolvable. Can we take the given scene to disclose the artist’s experience of the real-world scene? But how could Matisse or anyone else experience the tabletop as having, or looking as if it has, that cushy texture, or the palm fronds as being, or looking as if they are, that stubby-leaved? To be sure, these anomalies help introduce an expressive tone into the subject. So let the authorized subject be a mode of experiencing the scene rather than the scene itself. Let that experience be suffused by Matisse’s cherished values of radiance, calm, and delight. I cannot think even that does full justice to the work’s artistic presence and force as long as the scenic part of the subject remains as commonplace as is required by any standard construal of the depiction subject. In this and many other cases the viewer ﬁnds herself confronted by an apparition that is no part of the world of the normal subject, one that resists any set of determinations sufﬁcient to constitute a possible world whether real or ideal. Seen thus, the painting presents us with, or invites us to make-believe and even to elevate into primacy, as its special gift,³⁰ something anomalous ³⁰ In the 2003 Aristotelian Society exchange with Hopkins concerning seeing-in, Wollheim makes two claims bearing on anomalies in pictures (Wollheim 2003: 142 ff.). One conﬂicts

seeing things in pictures

231

and indecipherable. The painting itself seems to imply that the real subject should give way to the separation subject. The same I would say applies to Picasso’s painting of Franc¸oise Gilot in Femme Fleur (1946),³¹ which is discussed by Lopes (1996: 95), and indeed almost every painting in that artist’s mature œuvre. Certainly the conventional remarks we ﬁnd in the literature do not meet the need: for instance, that Picasso’s Vollard (1910)³² portrays the man from multiple viewpoints which are all related in the Cubist manner (Lopes 1996: 126) or even the moderately Gothic description of the same artist’s Girl with a Mandolin (1910)³³ as being ‘the shattered body of a woman’ (Podro 1998: 171). These conventional descriptions hardly touch the fascinating strangeness of the images.³⁴ Of course, from another point of view it is right and proper to disallow or marginalize the anomalies by sensitively responding to the information with my understanding of the work and the other leaves me uncertain of the author’s meaning. The ﬁrst is that we correctly see Parmigianino’s Madonna with the Long Neck (1534) () only when we do not see in it the Madonna as long-necked. That feature, he says, belongs only to what he calls the Presentational how, not the Representational what and not even to the Representational how, since the latter ‘corresponds to a property of the what of representation, possessed either permanently or transiently, whereas the Presentational how does not qualify the what at all’ (2003: 143). That is, it should affect only how the picture is perceived, not how the normally necked Madonna is taken to be represented. This strikes me as an implausible and uncalled-for playing down of the artist’s eccentricity. The second concerns Matisse’s Portrait of Mme Matisse (1905) (), which features a stroke of green on the face. This is explicitly said not to be a property of the Representational what: ‘he was not representing a woman who had a green line down her face’. But it is never explicitly said whether the line might be part of the Representational how, representing a transient state of illumination, say. More generally we are told nothing speciﬁc about the signiﬁcance of these features of the Presentational how in the cases at hand. In consequence I cannot make out how my view of separation seeing relates to the author’s view. Since Wollheim says the Presentational how ‘may reﬂect a range of things from the expressive vision of the artist, through the artistic pressures of the day’ (2003: 143), what I am contending for may conceivably fall under it. Much hangs on whether the picture’s reﬂecting the expressive vision of the artist can be explained without recourse to separation seeing. ³¹ . ³² . ³³ . ³⁴ Much the same must be said of the treatment of pictorial illusion in Penrose (1973), whose account of Cubism, for instance, is a fount of the generalities that are now part of the conventional wisdom without contributing any sharp-edge understanding of any particular work. The only thing we can take from this is the phrase ‘this intricate game of illusion and counter-illusion’ (1973: 255). We have to hold in abeyance the claims concerning Cubists exercising precise control or conducting probing analysis and rigorous reconstruction of objects, until someone cashes them out. Yet along the way he says various apt things: ‘every style adopts forms of illusion appropriate to its attitude toward reality’ (1973: 246); and cites suggestive statements by artists, e.g. this by Delacroix: ‘those things which are most real are the illusions I create in my painting’ (1973: 245). That is more like it.

232 john h. brown such a picture offers as to the normal subject. For, after all, Matisse certainly sat in his studio viewing the palm tree through the window, felt what he felt, and painted accordingly, wielding his brush as he did. Normal trafﬁc in Nice went whirring by, off in the distance; normal intrusions of the household into the painter’s space came and went. The world of the picture can be seen as containing some suitably generalized version of this background, but keeping it in the background in favour of pushing into the foreground a celebration of the joy of the sun on the palm and the unruly energies suggested by the pattern on the curtain. Likewise, Picasso confronted Kahnweiler in his Paris studio and produced a portrait of him, and art historians appropriately scrutinize the cryptic crystalline ambience for any hint of decor buried therein.

7. Implications for Theories of Depiction If pictorial separation phenomena are as pervasive and as signiﬁcant as I have claimed, what consequences, if any, are there for extant conceptions of depiction? I begin with the one offered by Lopes (1996: 152). His deﬁnitions of basic portrayal and depiction pose necessary and sufﬁcient conditions that omit any reference to separation seeing-in or separation subjects. I concede that these deﬁnitions may be adequate for depiction considered in the abstract. That is, in some possible world separation seeingin may not play a signiﬁcant role in the most advanced pictorial seeing attainable in that world. That incapacity would not mean that pictures in that world did not portray or depict. At the same time these viewers’ separation-free experience is not ours. They would lack a visual-imaginative sensibility so basic to our pictorial experience as to be impossible for us to understand them from the inside. This implies that basic depiction and portrayal as we know them are not completely characterized by Lopes’s deﬁnitions. A clause relating to separation seeing-in (recognition of aspects, in Lopes’s terms) is required to do justice to the full range of human pictorial seeing and the pictures that are its products and its delight. For unauthorized seeing-in is a matter of recognitions being triggered, however anomalous and incompletely deﬁned their content may be. A full account of the commitments and non-commitments that deﬁne an artistic

seeing things in pictures

233

depiction system can make appropriate room for well-founded separation subjects. Similarly, Wollheim’s condition concerning the artist’s intended seeingin (successfully provided for) needs comparable expansion to include the separation seeing-in that the artist welcomes or tolerates. Hopkins’s theory of depiction can accommodate the broadband separation side of pictorial experience by enriching the ‘three broad kinds’ of knowledge (1998: 137) to allow separation subjects to pass through the ﬁlter as secondary subjects—not, of course, as real subjects—and for their impact on our total pictorial experience to be given its due.³⁵ Just how fundamental separation seeing-in is to human experience can be appreciated if we take note of its analogue in ordinary (ecological) perception. We are constantly enabled and induced to see something like separation subjects in ordinary perception—to recognize aspects of things in the world that we know are not there. We see false alignments, accelerated convergences, false local colours, and countless other separation aspects which persist even when we are aware of the deception. Separation phenomena of one sort or another are endemic to any visual system of which we can claim ﬁrst-person knowledge, and perhaps to any conceivable system as richly informative as ours. At any rate, we do not know what it would be like for separation to be purged. Further, it is highly doubtful that we would like the result. To the contrary, our experience seems greatly enriched by the prevalence in it of separation ecological seeing. Who would willingly do without the oversized harvest moon near the horizon?

References Budd, Malcolm (1992), ‘On Looking at a Picture’, in Jim Hopkins and Anthony Savile (eds), Psychoanalysis, Mind and Art (Oxford: Oxford University Press). ³⁵ I have not dealt explicitly with Walton’s full theory, which is considerably extended by the admirably detailed and nuanced treatment of temporal factors in pictorial seeing (Walton 2008, Ch. 10). He notes an example of what I have called separation seeing in the case of a picture of children running on a beach. He observes: ‘I suspect that at some level or in some manner, we imagine seeing the kids frozen in a certain, admittedly unnatural position.’ He declines to rule out the possibility that we experience the photograph in at least two ways simultaneously (2008: 184). I take this to open the way towards a fuller recognition of the phenomenon I have explored in this chapter.

234 john h. brown Budd, Malcolm (1993), ‘How Pictures Look’, in Dudley Knowles and John Skorupski (eds), Virtue and Taste: Essays on Politics, Ethics and Aesthetics. In Memory of Flint Schier (Oxford: Oxford University Press). (1995), Values of Art: Pictures, Poetry and Music (London: Penguin). Dubery, Fred and John Willats (1983), Perspective and Other Drawing Systems (London: Herbert Press). Feagin, Susan (1998), ‘Presentation and Representation’, Journal of Aesthetics and Art Criticism, 56/3: 234–40. Gibson, J. J. (1950), Perception of the Visual World (Westport, Conn.: Greenwood Press). (1979), The Ecological Approach to Visual Perception (Boston: Houghton Mifﬂin). Gombrich, Ernst (1960), Art and Illusion: A Study in the Psychology of Pictorial Representation (New York: Pantheon). (1973), ‘Illusion and Art’, in R. L. Gregory and E. H. Gombrich (eds), Illusion in Nature and Art (London: Duckworth). Goodman, Nelson (1968), Languages of Art: An Approach to a Theory of Symbols (Indianapolis: Bobbs-Merrill). Hagen, Margaret (1986), Varieties of Realism: Geometries of Pictorial Representation (Cambridge: Cambridge University Press). Rochelle Glick, and Barbara Morse (1978), ‘The Role of Two-Dimensional Surface Characteristics in Pictorial Depth Perception’, Perceptual and Motor Skills, 46/3: 875–81. Hockney, David (2001), Secret Knowledge: Rediscovering the Lost Techniques of the Old Masters (New York: Viking Studio). Hopkins, Robert (1997), Review of Lopes, Understanding Pictures, British Journal of Aesthetics, 37/3: 283–6. (1998), Picture, Image and Experience (Cambridge: Cambridge University Press). (2003a), ‘What Makes Representational Painting Truly Visual?’, Proceedings of the Aristotelian Society, suppl. vol., 77/1: 149–67. (2003b), ‘Pictures, Phenomenology and Cognitive Science’, Monist, 86/4: 653–75. Hyman, John (2006), The Objective Eye: Colour, Form, and Reality in the Theory of Art (Chicago: University of Chicago Press). Koenderink, Jan J. (1998), ‘Pictorial Relief ’, Philosophical Transactions of the Royal Society, London, ser. a, 356: 1071–86. Kubovy, Michael (1986), The Psychology of Perspective and Renaissance Art (Cambridge: Cambridge University Press). Levinson, Jerrold (1998), ‘Wollheim on Pictorial Representations’, Journal of Aesthetics and Art Criticism, 56/3: 226–33.

seeing things in pictures

235

Lopes, Dominic McIver (1992), ‘Pictures, Styles, and Purposes’, British Journal of Aesthetics, 32/4: 330–41. (1995), ‘Pictorial Realism’, Journal of Aesthetics and Art Criticism, 53/3: 277–85. (1996), Understanding Pictures (Oxford: Oxford University Press). (2005), Sight and Sensibility (Oxford: Clarendon Press). Maynard, Patrick (1994), ‘Seeing Double’, Journal of Aesthetics and Art Criticism, 52/2: 155–67. (1996), ‘Perspective’s Places’, Journal of Aesthetics and Art Criticism, 54/1: 23–40. (2005), Drawing Distinctions: The Varieties of Graphic Expression (Ithaca, NY: Cornell University Press). Peacocke, Christopher (1983), Sense and Content (Oxford: Oxford University Press). (1987), ‘Depiction’, Philosophical Review, 96/3: 383–410. Penrose, Ronald (1973), ‘In Praise of Illusion’, in R. L. Gregory and E. H. Gombrich (eds), Illusion in Nature and Art (London: Duckworth). Podro, Michael (1991), ‘Depiction and the Golden Calf ’, in N. Bryson, M. Ann Holly, and K. Moxey (eds), Visual Theory: Painting and Interpretation (New York: HarperCollins). (1998), Depiction (New Haven: Yale University Press). Rawson, Philip (1987), Drawing, 2nd edn (Philadelphia: University of Pennsylvania Press). Schier, Flint (1986), Deeper into Pictures (Cambridge: Cambridge University Press). Vasari, Giorgio (1965), Lives of the Artists: A Selection Translated by George Bull (Baltimore: Penguin). Walton, Kendall (1990), Mimesis as Make-Believe (Cambridge, Mass.: Harvard University Press). (2002), ‘Depiction, Perception, and Imagination: Responses to Richard Wollheim’, Journal of Aesthetics and Art Criticism, 60/1: 27–40. (2008), Marvellous Images: On Values and the Arts (Oxford: Oxford University Press). White, Christopher (1999), Rembrandt as an Etcher: A Study of the Artist at Work, 2nd edn (New Haven: Yale University Press). W¨ollﬂin, Heinrich (1950), Principles of Art History (New York: Dover). Wollheim, Richard (1980), Art and its Objects (Cambridge: Cambridge University Press). (1987), Painting as an Art (Princeton: Princeton University Press). (1998), ‘On Pictorial Representation’, Journal of Aesthetics and Art Criticism, 56/3: 217–26.

236 john h. brown Wollheim, Richard (2001), ‘On Pictorial Representation’, in R. van Gerwen (ed.), Richard Wollheim on the Art of Painting: Art as Representation and Expression (Cambridge: Cambridge University Press). (2003), ‘What Makes Representational Painting Truly Visual?’, Proceedings of the Aristotelian Society, suppl. vol., 77/1: 131–47. Wright, Lawrence (1883), Perspective in Perspective (London: Routledge & Kegan Paul).

Index Abell, C. 6, 8–10, 83 n.1, 201 aesthetic appreciation 16, 182, 191, 209–10, 214, 216, 225, 228 aesthetic signiﬁcance 11, 16–17, 157, 160 n.3, 164–7, 171, 179 aesthetic value 10, 12 Aglioti, S. 201 ambiguity 20–21, 106–113, 120 perceptual 106–107 representational 107, 109–112 anamorphosis 41–2, 46, 48 Aristotle 14, 128, 134–41 Arnheim, R. 138 attention 14, 19, 39, 60, 185–6, 191–4, 196, 202, 204, 216, 222–3 awareness 11–13, 19, 78, 128–34, 137–41, 143–9, 151, 154–8, 160 n. 3, 163–7, 186–7, 191, 193, 213, 216, 227 aspects of; see aspects of seeing-in; twofoldness Bantinaki, K. 13–15, 173 n. 9 bare bones content; see content, pictorial, bare bones Berm´udez, J. L. 52–4, 78 binding problem 56, 58, 69 Bosch, H. 25, 29 Brown, J. 10–12, 31, 43 n. 5, 104, 124–5, 143 Br¨ucke E. 37 Bruno, N. 201 Budd, M. 5, 67, 131–2, 134, 141, 147, 168, 209, 224 Byzantine icon 42 Campbell, J. 74–6, 193 Carroll, N. 183 C´ezanne 42, 184 Chabris, C. F. 194 Clark, A. 91 n.7, 200 Cohen J. and Meskin A. 85–6, 90 n. 6, 94–5, 102

competition semantic 5, 26, 32, 34–8, 41, 46, 48–9 syntactic 5, 26, 32, 34–8, 41, 44, 46–7, 49 compound substance 14, 128, 134–7, 139, 141–2, 144, 147 conﬁguration pictorial 105–16, 187 retinal 107, 113 conﬁgurational properties; see design properties Constable, J. 36–7 content aspectual 21, 108–111, 118, 120–23, 126 levels of 10, 106–125; see also Double Content Theory of seeing-in 10–18, 152, 154, 160 n. 2, 161, 164, 167–8, 170–72, 183, 192 conﬁgurational 105–6, 111–12; see also aspects of seeing-in, conﬁgurational recognitional 105; see also aspects of seeing-in, recognitional perceptual 20–21, 104–23, 200; see also Double Content Theory pictorial 7, 10–11, 19–21, 25–9, 31–2, 34, 38–40, 43–5, 49, 83–4, 86, 93, 95, 100, 105, 109, 113–6, 120, 123, 152, 186, 227–8; see also Double Content Theory bare bones 3–5, 47–8, 126 ﬂeshed out 3, 47, 126 Corregio 216 cubism 30, 208, 214, 229, 231 Currie, G. 90 n. 6, 92 n. 9 deep decentering 23, 76 Dehaene, S. 200 deixis 22–3, 59, 61, 63–8, 73–5, 77–8 demonstrative reference 21–3

238 index depiction 1–23, 38–41, 45 n. 6, 74, 83–4, 91, 128–31, 138, 145–6, 148, 182, 195, 208, 212–14, 216, 227, 232 system of 28, 29, 43, 227, 233 theories of 2, 26–7, 42 n. 4, 43, 46–50, 67–8, 196–99, 212–3, 215, 222, 224, 232–3 divisive vs. unitary 12–13, 15, 19, 126 phenomenological 2, 5–6, 10, 227 recognitional 2, 5–6, 10, 18, 49, 67, 86, 213 resemblance 2, 6, 10, 83, 212, 227 structural 2–5 depicted object 13–14, 16–17, 19, 21, 29, 31, 36, 37, 40, 42, 45, 52, 62–3, 65, 68–9, 76–7, 128, 130–3, 152, 154–5, 157, 160 n. 3, 168, 170 n. 7, 173 n. 8, 181, 183–6, 188, 190–99, 201–4 external 83–9, 93, 96–7, 99–101 internal 83, 86, 89 depth 209, 214, 217–8, 221, 223–6, 229 design 16–17, 19, 104–9, 112–3, 120, 122, 124, 135 n. 5, 140, 143, 155–78, 181–4, 187–9, 192–9, 202–3 design-content correlations 29, 31 design features 83, 104–5, 155–6, 158–60, 163, 174, 177–8, 182, 192, 195–8, 201–3 214, 215 n. 6, 219, 228 design-scene properties 19, 160 n. 3, 194–8, 202–4 design seeing 183, 187–9, 192, 195–6, 199; see also seeing-in, conﬁgurational aspect of diagrams 26, 32, 35–6, 45 Dilworth, J. 20–1, 104, 106–7, 109, 111, 113, 116, 121–2, 124–5 dorsal visual subsystem 18–20, 19–204; see also motion guiding vision Double-content (DC) Theory 21, 104–23; see also orientational DC theory Dretske, F. 90 n. 6 eccentric properties 11; see also eccentric seeing-in; separation Eimer, M. and Schlaghecken, F. 186 Elkins, J. 27 epistemic properties 82, 95 epistemic roles 81–2, 84, 96 epistemic signiﬁcance 81–2, 93, 95 epistemic value 8, 81–101

Escher M. C. 157 Evans, G. 45 n. 6, 53–5 facture 215–7, 220, 225 Feagin, S. 193 ﬁsheye photos 28–33, 35, 41–4, 48, 126 ﬂeshed out content; see content, pictorial, ﬂeshed out form 14, 128, 134–41, 144, 147–9 Fra Angelico 25, 29 Franz, V. H. 201 Freud, L. 128, 139, 141–4 Gainsborough, T. 37 Gentilucci, M. 201 gestalt 13–4, 137–8, 140, 148 Ghirlandaio 210 n. 2, 215–6, 217 Giacometti 122 Gillam, B. 201 Goldstein, E. B. Gombrich, E. 26, 36–7, 39–41, 49, 112, 152–4, 173, 185 n. 1, 217 Goodale M. 60, 200–1 Goodman, N. 3, 57, 68, 208 n. 1 Greenwald, A. et al. 116 Hagen, M. 223 Haugeland, J. 47, 126 Hobbema 36–7 homonymy principle 136 Hopkins, R. 5, 7–8, 10–13, 16–19, 30, 32 n. 1, 41, 42 n. 4, 43 n. 5, 45–6, 48–50, 67, 83, 92–3, 99, 102, 104–5, 124–6, 134 n. 4, 143, 149, 159 n. 2, 154, 161, 166, 168, 182, 184–5, 189, 195–9, 203, 209–14, 217, 219–20, 224, 228, 233 Hyman, J. 6, 145–6, 226 n. 26 illusion 5, 12, 15, 60–1, 63, 66, 91, 145, 201, 217, 221, 231 inﬂection 16–9, 151–79, 183, 190, 192, 195–6; see also seeing-in, inﬂected existence of 160–3 nature of 154–60 signiﬁcance of 164–79 information 7–9, 22, 54–5, 59, 61, 64–6, 68, 73–5, 82, 85–8, 91, 94–98, 102, 105–113, 118, 188, 227, 229, 231

index intention 7–10, 21, 46, 84, 92–3, 96, 99–100, 111–2, 121, 124–5, 130 n. 3 interpretative practice 26, 28, 36–7, 41 interpretative schemes 26–37, 40–44, 48–50 Jacob, P. and Jeannerod, M. 91 n. 7, 200 Koenderink, J. 223 Kulvicki, J. 3–5, 27, 30, 43, 45–8, 83, 126, 185 landscape 26, 36–7, 44 Leonardo 216 Lessing, G. 45 n. 6 Levinson, J. 186–7, 192–3, 195–6 Lewis D. 86 linear painting 26, 38–41 location 56, 58, 69, 74–5, 77 Lopes, D. 6–7, 13, 15–7, 21–3, 29, 49–50, 63, 67–8, 83, 86, 91, 104, 121, 155–6, 163, 165–70, 173, 181–3, 187–9, 192, 194–7, 200–1, 203, 208 n.1, 209, 213, 217–8, 227, 231–2 Lorrain, C. 37 McGinn, C. 54 Mack, A. 194 Magritte, R. 223 Manet, E. 221 manipulation 100–1 Matisse, H. 42, 225, 229–30, 232 matter 14, 128, 134–9, 147–8 Matthen, M. 22, 55, 57–61, 64–5, 68–9, 72–4, 78, 200 Maynard, P. 220 n. 13, 226 n. 24 medium 11–5, 19, 21, 111–2, 117, 121–2, 124, 126, 128, 130–33, 135 n. 5, 138–49, 187, 211–2, 221 Michelangelo 221 Milner, A. D. and Goodale, M. A. 60, 91, 200 mimesis thesis 166, 168–70; see also puzzle of mimesis misrepresentation 32 n. 1, 93, 100–2 Monet, C. 37 moving pictures 53, 69–77

239

Nanay, B. 60, 91 n. 7, 104, 145–6, 160 n. 3, 165 n. 6, 185, 186 n. 2, 202, 219 optic ataxia 60, 200 orientational DC theory 21, 113–23 concepts 113–23 ﬁeld 114, 117–23 ﬁeld orientation 114, 117–23 intrinsic orientation 115–23 inversion 115–23 OSMI principle 116 uprightness 115–23 outline shape 48–9, 122, 210, 212, 224 painterly painting 26, 38–41 Parmigianino 30–1 Peacocke, C. 5, 168, 224 perceptual ambiguity; see ambiguity, perceptual perceptual constancy 117, 124–5, 186, 188, 191, 221 n. 18, 222 perspective 221, 226 axonometric 28, 48 linear 27–30, 35, 41, 43, 47–8 reverse 28, 41–2, 46, 48–9 phenomenology 5–7, 10, 12, 17, 19, 23, 59, 128–9, 132–3, 142, 145–7, 149, 151, 156, 168, 171, 197; see also twofoldness phi phenomenon 22, 56–8, 69 photographic mechanisms 9, 83–4, 91–3, 98–101; see also photographic process standardization of 9, 90–1, 98–100 photographic process 85, 88–9, 90, 93, 97–102 Picasso 69, 214 n. 5, 231–2 pictorial diversity 25–8, 30, 32, 36, 41–50 pictorial evaluation thesis 166–7 pictorial experience 12–20, 23, 104, 131, 133, 145, 151–70, 172, 174, 176–7, 179, 181–2, 185, 187, 195–6, 208, 212, 214–5, 228, 232–3; see also seeing in pictorial perception 14, 18–23, 129, 130 n.3, 133–4, 139–40, 142–3, 146–9, 185, 208–9, 213–14, 218, 222, 224, 229, 232, 233 n. 35; see also seeing in pictorial practice 11, 144, 212, 214, 224 pictorial seeing; see pictorial perception

240 index picture-production process 9, 28, 85–8, 96–7, 99, 100; see also photographic process reliable 85–8, 90–1, 93, 95, 97–99, 101 picturesque 38–9, 40 Podro, M. 16, 121, 151, 156, 158, 161, 183–4, 197 n. 8, 218–9, 226 n. 25, 231 Polanyi, M. 192 n. 4 presence, feeling of 59, 61, 63–5 Prinz, J. 193 pseudo-twofoldness 188 Purves D. and Lotto, R. B. 110–11 puzzle of mimesis 16–8, 165–170 recognition object 6, 18, 138, 147, 200 of subject 6, 132, 140, 209, 224, 232 of artistry 227 of aspects 232 recognitional abilities 86, 49, 213 reference, demonstrative 52–5, 59, 64, 67, 78 Reimer, M. 54 relational properties 19, 193–6, 202 Rembrandt 105–6, 111–13, 152, 156, 161, 171, 178, 217–8, 219 n. 12, 220 Renaissance 27, 40 Rensink, R. A. 194 representation 1–3, 5, pictorial 3–4, 26–7, 38, 40, 43, 46, 48, 50, 83, 89, 93, 99, 128–31, 133, 148, 185–6, 190, 229 system of 2–4, 26–50 visual 15, 17–9, 21–2, 57–9, 66, 69, 73, 76–7, 154, 172–9, 191–2, 197–9, 203 central 23, 73–6; see also central vision deictic; see deixis representational ambiguity; see ambiguity, representational representational seeing 190 resemblance 6–7, 10, 13, 16, 83, 85, 105, 123, 168–70, 173 n. 10, 210 n. 2, 212, 224–5 experienced 10, 46, 48–9, 212 richness 9, 87–8, 90, 94–5, 97–8 Rock, I. 194 Savedoff, B. 101 Scaltsas, T. 135–6

Schier, F. 6, 49, 86 Scruton, R. 92 n. 9, 84 n. 2 seeing-in 5, 10–20, 22–3, 59, 64, 67–8, 74, 77–8, 104, 106, 121, 123–4, 126, 128–34, 138–9, 143–9, 151–60, 164–79, 181–93, 196–9, 203–4, 209–33 aspects of 12–16, 128, 130–3, 144, 146–7, 152, 169, 188, 232; see also twofoldness conﬁgurational 12, 130–1 recognitional 12, 130, 132, 213 authorized 215, 217 eccentric 11, 212, 214–5, 226–8 inﬂected 17–19, 182–5, 190–9, 202–4 nested 159–60, 175–6 separation 10–11, 143, 210, 213–27, 232–3; see also unauthorized seeing-in aesthetic signiﬁcance of 214, 216–8 spatial 220–7 ubiquity of 214–5 unauthorized 209–10, 232 uninﬂected 17, 159 n. 2, 182–5, 190, 193–9, 202–3 unitary vs. divisive accounts of 126, 167–79, 196–9 semantic competition; see competition, semantic separation 10, 143, 209, 210, 213–20, 220, 226, 228–9, 232; see also separation seeing in subject 210, 212, 214, 217–20, 222, 224–6, 229, 231–3 Seurat, G. 69 Shapiro, M. 41 Siegel, Susanna 54 Simmons, D. J. 194 Snowdon, Paul 77–8 Snyder, J. and Allen, A.D. 99 space 26, 33, 35, 41, 47–8, 62, 68–9, 77, 143, 148, 163, 208, 214–7, 219–23, 229 spatially agnostic informants 94–5 standard of correctness 6–12, 92–3, 102, 130 n. 3 causal 7–8, 10, 92–3 intentional 7–8, 10–11, 92–3 Standard Visual Representation 17–8, 149, 173–9, 198, 203

index Stevens, A. 210, 212, 214–5, 217–8 Strahan, E. et al. 186 Strawson, Peter F. 54–5 subliminal priming 186 surface 3, 8, 13, 15, 17–18, 20–21, 28, 30–1, 33, 43, 47–9, 62–5, 69, 104–9, 111, 113, 129–30, 132–3, 139–41, 143, 145, 151–66, 171–4, 177, 181–2, 184–9, 191–204, 213, 215, 217–25, 229 features 5, 11, 13, 19, 25–6, 29, 31, 33, 36–8, 42–9, 68–9, 129, 139, 140–1, 143, 151, 155, 158–60, 164–5, 171, 210, 213, 216, 217–8, 228 syntactic competition; see competition, syntactic system of pictorial representation; see representation, pictorial, system of

241

van Gogh, V. 25, 34, 166 Vasari, G. 215–6 ventral visual subsystem 18–20, 199–204; see also descriptive vision Vermeer 97, 190 vision: allocentric 69, 72–4 central 73–7 descriptive 22, 60–1, 64–8, 200 direct vs. indirect 77–8 distance 67–8 dorsal and ventral pathways of 60, 66 egocentric 22, 59–61, 64–69, 72–7, 200 impaired 60 monadic vs. relational 75–77 motion-guiding 22, 60–8, 77, 200 visual agnosia 200

transparency 89–92 trompe-l’oeil 15, 59, 62–3, 65, 68, 77–8, 144–6, 148–9, 154–5, 173, 183, 189, 192–3 twofoldness 5, 12–13, 15, 19, 123–134, 137–48, 152–4, 167–79, 185–92, 197, 213

Walton, K. L. 5, 68, 89–92, 101, 152, 161, 169–70, 185, 197 n. 8 Warhol, A. 122–3 Willats, J. 27–8, 35 n. 2 W¨olfﬂin, H. 26, 36, 38–41, 49 Wollheim, R. 5, 7, 12, 13–15, 19–20, 104–5, 121, 123–5, 128–35, 138–9, 141–9, 152, 154, 161, 165, 169–70, 173 n. 10, 181, 185–197 n. 8

unity 128, 134–40, 143–5, 147

Ziff, P. 25