You are on page 1of 265


This page intentionally left blank

Kendall L. Walton

On Values and the Arts


Oxford University Press, Inc., publishes works that further
Oxford Universitys objective of excellence
in research, scholarship, and education.
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 ofces in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam

Copyright 2008 by Oxford University Press, Inc.

Published by Oxford University Press, Inc.
198 Madison Avenue, New York, New York 10016
Oxford is a registered trademark of Oxford University Press
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise,
without the prior permission of Oxford University Press.
Library of Congress Cataloging-in-Publication Data
Walton, Kendall L., 1939
Marvelous images : on values and the arts / Kendall L. Walton.
p. cm.
ISBN 978-0-19-517794-7; 978-0-19-517795-4 (pbk.)
1. Aesthetics. 2. ArtPhilosophy.
3. PhotographyPhilosophy. I. Title.
BH39.W328 2008

Printed in the United States of America

on acid-free paper


This volume and its forthcoming companion, In Other Shoes: Music, Metaphor,
Empathy, Existence, reproduce a number of essays that I have published over the
years and introduce several new ones. All explore topics in aesthetics or philosophy of the arts, broadly conceived, but most of themwell, all of themtake
up issues in other areas of philosophy at the same time: philosophy of mind,
philosophy of language, or value theory. The present collection begins with a
cluster of essays concerning values and the arts.
Many works of art are marvelousmarvelously beautiful, or aesthetically valuable in other ways. I have long been skeptical of the theoretical importance of
notions of aesthetic value, and I still am. Nevertheless, the opening essay of this
volume, How Marvelous!: Toward a Theory of Aesthetic Value, explores a
kind of value that ts common conceptions of aesthetic value surprisingly well,
despite differing greatly from most traditional accounts. One new postscript
cites several precedents, however; another describes different ways in which a
work might be good because it is bad. The Test of Time, an adaptation of
a review of Anthony Saviles excellent book with that title, examines the claims
of durability or longevity as an indication of aesthetic value, and considers what
sort of value this must be if time is a reasonable test for it.
Works of art possess values of other kinds also, and serve them in important and
distinctive ways. How Marvelous! considers how aesthetic value is related to
other varieties, as well as how it differs from them. Morals in Fiction and Fictional Morality and On the (So-Called) Puzzle of Imaginative Resistance focus
especially on moral values, and on works that subvert them as well as ones that
serve them. They address the peculiar conundrums concerning works of ction
with moral dimensions that I noted in Mimesis as Make-Believe, which have been
discussed subsequently under the misleading rubric of imaginative resistance.



Besides marveling at particular works of art, one cannot but nd remarkable

the resources artists have at their disposalthe media they work in and the
techniques they employwhich make possible the production of marvelous
objects and ones that are effective in serving or subverting values of various
sorts. The essays in part II examine the medium of visual representation. Some of
them develop and defend the account of depiction I presented in Mimesis. Others
consider what is special about two important species of depictionphotography
and still pictures (with emphasis on the fact that they are still).
Pictures and Hobby Horses: Make-Believe beyond Childhood, a lecture
designed for general audiences, is an informal introduction to my theory of
depiction and the notion of make-believe it rests on. I include it partly because
it provides relatively easy access to the central features of my views on these topics, for readers not familiar with Mimesis, but also because it approaches them
from a different angle, one that brings out my indebtedness to Ernst Gombrich.
Unlike Mimesis, it reproduces the order in which I originally developed the ideas
it sketches, starting with the problem of how to account for the visualness of
pictures and other depictions, then introducing the theory of make-believe in
order to solve it, and after that exploring applications of the theory to ctions of
various kinds and a wide range of related phenomena.
Transparent Pictures: On the Nature of Photographic Realism initiates and
anchors my attempt to discover what is special about photographs. Photographs, I
say, are aids to vision; on seeing a photograph of a bear, one sees (indirectly) the bear.
This is not the only respect in which photographs differ from pictures of other kinds,
but it is the most important one. I have included several postscripts to this essay clarifying features of my transparency claim that have sometimes been misunderstood,
and indicating directions in which I think further investigation would be fruitful.
I do not do much in these essays by way of replying to objections or assessing rival theories. An exception is On Pictures and Photographs: Objections
Answered, which considers arguments advanced by Nol Carroll and Gregory
Currie against my account of depiction and the transparency thesis.1 Seeing-In
and Seeing Fictionally and Depiction, Perception, and Imagination examine
relations between my account of pictorial representation and that of Richard
Wollheim, in the process clarifying and elaborating my account.
Although much has been written about photography using still photographs
as examples, there are very few discussions of what is special about still photographs as opposed to moving ones, or of the medium of still pictures generally. The depiction of motion and change by means of unchanging marks on a
picture surface, in particular, has been insufciently remarked or investigated.
1. I replied to other objections in an essay not included here, Looking Again through
Photographs: A Response to Edwin Martin (Critical Inquiry 12 [1986], 801808). Curries more recent writings on pictorial representation go in a different direction, utilizing
a notion of perception like imagining.



Experiencing Still Photographs: What Do You See and How Long Do You See
It? rst published in this volume, tackles these topics. Although it concentrates primarily on photographs, having begun as a contribution to a symposium on photography, its conclusions apply to still pictures in general, indeed
to still depictions of all sorts: sculptures, paintings, and drawings, as well as
photographs. As a bonus, reection on still media brings interesting features
of motion pictures into focus.
The two older essays constituting part III treat very general issues concerning
the understanding and appreciation of works of art of many kinds. Categories
of Art and Style and the Products and Processes of Art both address the mossencrusted questions of intentionalism in aesthetic theory. Each nds a role in
appreciation and criticism for facts about the circumstances of a works genesis,
including the artists intentionsbut very different roles in the two cases.
The value of works of art, why they are important and why people esteem
and treasure themnot far from the surface in any of these essaysis especially
evident in Style, which thus connects with the more recent essays of part 1,
especially How Marvelous!
An unmarked theme of several of the essays in this volume, made explicit
only in Experiencing Still Photographs, is the difference between two things
appearing (to be) different, and their appearing differently, a distinction that is
commonly overlookedwhen, for example, people claim that if a forgery and
the original are visually indistinguishable they must be identical aesthetically.
Categories and Style both, in different ways, illustrate this distinction and
underscore its importance.
I have saved for In Other Shoes essays focusing primarily or substantially on empathy and its relatives, especially what I have called other shoe experiences, and
essays on emotional responses to ction, on music, on metaphor, and on existence
claims and the ontological status of ctional entities. This may seem a scattered
assortment, but the essays draw a network of substantial, if sometimes unexpected, connections among these topics.
The choices as to which essays to include in which volume were inevitably somewhat arbitrary. Pictures and Hobby Horses contains informal observations about
empathyabout empathizing with people in picturesand about what later
came to be called mental simulation, topics I address more systematically and thoroughly in In Other Shoes. In Style I argue, in effect, that empathy with artists and
performers is a fundamental ingredient of much appreciation of the arts.
Readers with special interests in music may want to consult Categories
and Style in addition to the essays devoted primarily to music that are collected in In Other Shoes. Musical performance, music making, is a paradigm of
the processes of art discussed in Style. Both Style and Categories bear on
questions concerning musical personae, and musical examples are prominent in
each of them.



While wrestling with notions of aesthetic value in How Marvelous!

I found myself thinking about the nature of sports and the values they may
involve. A new essay to appear in In Other Shoes Its Only a Game!: Sports
as Fictionexamines sports and the experiences of participants and spectators
more directly.
Each of the essays in both volumes is intended to stand on its own, to be understandable apart from any of my other writings. This makes for some unavoidable
overlap among them, especially in introductory sections. Several contain short
sketches of ideas presented more carefully in others or in Mimesis as Make-Believe
before going on to make their distinctive contributions. Minor editorial adjustments aside, the texts of all but two of the reprinted essays are complete and
unchanged (although I have inserted several new footnotes, in square brackets).
The exceptions are Pictures and Hobby Horses, which nds here a more or less
settled form after its varied life as an occasion-driven lecture, and The Test of
Time, which was modied to be independent of Saviles book.
Footnotes in the various essays and references to writings of other scholars constitute an extensive but seriously incomplete record of my intellectual debts.
I should mention especially David Hills and Patrick Maynard, whose stimulation and advice for many decades is inadequately acknowledged in my references
to them. I cant overemphasize the benets of numerous casual conversations and
offhand commentsby many colleagues at the University of Michigan over the
years, by lecture audiences, and by students in courses and seminarswhich not
infrequently pointed toward what I came to see as signicant insights, or alerted
me to problems in my views or infelicities in my formulations of them. Thanks
to Katherine Kuehn, whose detective and diplomatic skills were indispensable in
obtaining permissions for the illustrations and arranging for their reproduction
in this volume. Thanks also for support from the Deans Ofce Discretionary
Fund, University of Michigan College of Literature, Science and the Arts, and
from the University of Michigan Department of Philosophy.


par t i .

aesthetic and moral values

Chapter 1. How Marvelous!: Toward a Theory of Aesthetic Value, 3

Postscripts to How Marvelous! 20
Chapter 2. The Test of Time, 23
Chapter 3. Morals in Fiction and Fictional Morality, 27
Chapter 4. On the (So-Called) Puzzle of Imaginative Resistance, 47

par t i i .

pictures and photographs

Chapter 5.

Pictures and Hobby Horses: Make-Believe

beyond Childhood, 63
Chapter 6. Transparent Pictures: On the Nature of
Photographic Realism, 79
Postscripts to Transparent Pictures: Clarications and To Dos, 110
Chapter 7. On Pictures and Photographs: Objections Answered, 117
Chapter 8. Seeing-In and Seeing Fictionally, 133
Chapter 9. Depiction, Perception, and Imagination: Responses to
Richard Wollheim, 143
Chapter 10. Experiencing Still Photographs: What Do You See and
How Long Do You See It? 157

par t i i i .

categories and styles

Chapter 11.
Chapter 12.
Index, 249

Categories of Art, 195

Style and the Products and Processes of Art, 221

This page intentionally left blank


This page intentionally left blank

Toward a Theory of Aesthetic Value

esthetics is often classied as a branch of value theory. This classication

is curious and in some ways objectionable. Many important issues of aesthetics,
as traditionally practiced, have no direct connection with notions of value or
evaluation. This is as it should be, if aesthetics is the theoretical or philosophical
examination of the cultural institution of art. For there is much to the institution
besides evaluation, and one might argue that it is far more important to understand and appreciate works of art than to decide how good they are.
It is arguable, also, that the interest we do have in evaluating works of art
is a somewhat parochial feature of the cultural surroundings of the ne arts in
Western society during the last several hundred years, that it is much less important in other contexts. We must not assume that all cultures in which people
produce and enjoy or nd satisfaction in what we call works of art even recognize
anything much like our notion of aesthetic value. But there is no denying that
this notion plays an important role in the practices surrounding the ne arts in
recent Western culture, and for that reason alone it deserves attention.
The notion of aesthetic value can look very questionable when we do attend to
it, however. The worries are familiar. There is enormous variety among the works
we take to be of high aesthetic quality, and our reasons for praising them, for
pronouncing them aesthetically valuable, are astonishingly diverse. Some good or
great works stimulate, some soothe, others are disturbingly provocative or upsetting. Some afford intellectual pleasures; others emotional experiencesfullling
emotional experiences in some instances, distressing ones in others. Some works
offer insight or illumination; others catharsis. Some provide escape from everyday
cares; others help us to deal with them. Some require careful study and analysis; others wear their charms on their sleeves. Great works can be exuberant or
gloomy; they can be intense, or serene, or painful, or funny. Aesthetic value


appears to be an incredible grab bag. What justication is there for speaking of a

single kind of value in cases of all of these sorts?
The distinctiveness of aesthetic value, as well as its unity, threatens to evaporate under scrutiny. Formalists such as Clive Bell and Eduard Hanslick who postulate the autonomy of aesthetic value (or musical value) take a heroic course.
Much of what we take to be aesthetically valuable about many works of art seems
thoroughly intertwined with concerns of everyday life, with practical values
of various kinds, with cognitive and moral and religious values. It just does not
seem plausible that what is so wonderful aesthetically about much great poetry,
for instance, has nothing at all to do with the insight we receive from it, or
that the feelings one has in appreciating music aesthetically are entirely unlike
and irrelevant to everyday emotionseven granting that to be informative or to
elicit emotional responses is not thereby to have aesthetic value.
Could it be that aesthetic value supervenes on or is otherwise dependent on
the capacity to provide practical or cognitive or emotional benets of various
kinds? If so, it may be possible to preserve its unity; a single sort of value might
supervene or depend on any of various other kinds of value. And the supervening
or dependent value may itself be distinct from the practical and other everyday
values that it supervenes or depends on.
I will propose an account of aesthetic value along these lines, although the
values on which aesthetic value may depend, on my account, include ones that
are very different from those I have mentioned. I should emphasize from the start
that in offering this account I do not presume to be articulating what people
have always or usually meant by aesthetic value (even during just the last
several centuries in Western culture). It is far from clear that there is any one
thing that people have usually meant by it, even implicitly. But the notion I will
propose ts surprisingly well into the slot that critics and theorists have expected
aesthetic value to ll, and I believe that it is closer to what many have meant than
it may rst appear to be. Most important, however, what I will call aesthetic
value is value of a kind that needs to be recognized no matter what we call it.
Works of art we judge to be aesthetically meritorious characteristically possess
value of this kind, and appreciators cherish them for it.


We must distinguish questions about the aesthetic merit of particular works of
artquestions about whether a particular sonata or novel or movie or fresco is
a great work of art, or merely good, or mediocre, or terrible and whether one
work is better or worse aesthetically than anotherfrom questions about the
value of the cultural institution of art. In what ways does the institution, our
practices of making, displaying, contemplating, appreciating, discussing, criticizing, judging works of art, benet (or harm) people or society? What purposes
does it serve? What are our reasons for engaging in it? One might ask about the


M A RV E L O U S !

evolutionary value of the institution, the survival value of human inclinations to

develop such an institution, or the contributions such an institution might make
to the health and survival of cultures of which they are a part.
However skeptical one might be about the viability or coherence of the notion
of the aesthetic value of works of art, or about its applicability beyond a narrow
range of cultural institutions, questions about the value of institutions in which
paintings, novels, sculptures, cantatas, and so on, are embedded are clearly in
order and clearly within the province of the aesthetician. This does not make
aesthetics a branch of value theory any more than, for example, the philosophy
of science is. For central to the examination of any human institutionscience,
religion, sports, childrens games of make-believeshould be questions about
the point of the institution, what ends it serves, what reasons we have or what
reasons one might have for participating in it.
In asking about the value of an institution, we usually have in mind extrinsic or instrumental value. We want to know what contributions the institution
makes to our lives, what benecial effects it has. But aesthetic value is usually
thought of as intrinsic.1 Of course works of art play a role in the institutions
achievement of its benecial effects. But this is not what their aesthetic value is
understood to consist in; insofar as they are aesthetically valuable they are good
in themselves.
I want to explore another possible difference between the two kinds of value,
or a cluster of possible differences, ones that might be expressed initially by
saying that questions about aesthetic merit are asked from a perspective internal to the institution and that one is participating in the institution when one
judges a painting or a novel to be great, or to be better aesthetically than
another one, that aesthetic value is institution-bound. This thought might be
spelled out in several ways. Perhaps questions about aesthetic merit make sense
only because the institution gives them sense. Perhaps they are legitimate or
appropriate or in order only within the institution. Perhaps it is traditions of
the institution that determine how they are to be answered, what is to count as
aesthetic value. Maybe the answers matter, maybe people appreciate and care
about aesthetic value, only insofar as they accept or buy into the institution.
We can also ask whether the practice of aesthetic evaluation might be essential
to the institution. Could it be that the benets the institution has to offer are
realized only if people participate by judging the value of works of art or asking about their value?
One way to begin to appreciate the institution-bound character of aesthetic
value is to compare other institutions in which the evaluation of particular objects
or events or activities analogous to the aesthetic evaluation of works of art is
relatively unimportant, or which dont even have provision for such evaluation.
1. Or anyway as inherent. See William K. Frankena, Ethics, 2nd edition (Englewood
Cliffs, NJ: Prentice Hall, 1973), pp. 8182.


Participation in childrens games of make-believe doesnt demand or even

encourage judging the value of props, or games (types or tokens), or participants, or acts of participation. One nds or chooses or is given props for a game,
and uses them, but in participating in the make-believe one neednt evaluate the
props or consider how they stack up against others. Children dont endeavor to
perform well, in participating in their games; it seems out of place to judge that
one child did a superb job of playing dolls, but that, unfortunately, anothers
performance was mediocre.
Judgments can be made from an external viewpoint, however. There are benets to be gained from playing games of make-believe. It is fun, entertaining,
pleasurable. One learns about real-world activities, about how to do certain
things, about what it is like to do them and how one feels about them. A particular prop may be especially well suited to achieving these benets, or more
conducive to this than other props are. But this value is instrumental rather than
intrinsic, and so is not analogous to aesthetic value. And one neednt judge the
prop to be good, one neednt recognize its value, in order to benet from it. It is
necessary only that one use it in the game. The same may be true of a particular
game, or a participant, or a childs performance on a particular occasion. Any of
these may be especially conducive to whatever benets one gains, but achieving
the benets usually does not require one to recognize its merit.
Many folk art traditions are much like childrens games of make-believe
in these respects, and unlike the tradition of Western ne art. (Consider the
tradition of singing hymns in religious ceremonies.) People may participate
in singing or dancing or acting, or watch with interest and enjoyment as others do, without it ever occurring to them to ask how good aesthetically the
performance or the work performed is, or whether it is better or worse than
another one. These questions are likely to seem out of place or inappropriate,
at best. And they neednt be raised in order to achieve the benets that folk
art serves.
Contrast sports and other competitive gamesbaseball, chess, pickup sticks,
Ping-Pong, bridge, Monopoly, dominoes, and so on. These are instances in
which values (of a sort) are clearly internal and intrinsic to an institution,
in which evaluation makes sense only within the institution and is essential to
the institutions function. The value in question is that of winning. And we
judge teams and players, moves, plays, and strategies, as being better or worse, as
they are more or less conducive to winning. What constitutes winning is dened
by the rules of the game, by the clause in the rule book that goes, The object of
the game is . . . And it is (primarily, if not exclusively) from a perspective within
the game that one cares about winning and losing. Making these value judgments is essential to the institution. One must recognize winning and losing in
order to play baseball, or even to follow the game. Perhaps participants must care
about winning, at least while they are playing, or pretend to, or anyway try to
win. And no doubt some of the benets of which the institution can boast, the


M A RV E L O U S !

excitement it affords, for instance, depend on its provision for winning and losing and on participants recognizing winners and losers.
The contrast between internal and external judgments, between judgments
of winners and losers, on the one hand, and judgments of players and plays and
games as being conducive to the purposes or benets the institution of sport (or
a particular sport) serves, on the other, is at its starkest here. Besides excitement,
thrills, and entertainment for players and fans, these benets include keeping
kids out of trouble, developing physical and mental skills, providing practice
in handling success and failure, promoting cooperation and self-reliance, providing a safe outlet for aggressive tendencies, enhancing a sense of community,
increasing alumni contributions. (There is a downside also.) Evaluating players
and plays with regard to aptness for winning is not judging them for their contributions to these social or personal goods. And the two often conict. Close
games are more exciting than lopsided ones, and may well contribute more to
the achievement of other social benets. But a superb play, one conducive to winning, by the team or player already in the lead may turn what could have been a
thrilling contest into a dull romp. The general level of play within a league or a
sport has little relation to these social benets. It is unlikely that society is better
off now than it was fty years ago because of the fact that skill levels in sports
are much higher, even if the world is a better place with sports than it would
be without them. (Maybe there is benet in the fact that the trend is upward,
however, even if the absolute level of skill doesnt matter.) The sixth game of the
1975 World Series has been called the greatest baseball game ever playednot
because there was more winning in it than in other games. Like (virtually) all
baseball games, it had just one winner and one loser. And the winner won just
barelythat is part of what made it such a good game. Chess games (either the
play of one of the contestants, or the game as a whole) are sometimes admired for
their beauty or elegance. Among game types, I prefer baseball, Ping-Pong, and
chess to football, boxing, and gladiator contests, and I can give reasons. These
are all external value judgments, and are distinct from judgments of winners and
losers, or of aptness for winning and losing.
Players primary objectives are, standardly, to win, not to further the social
benets the institution is capable of furthering; the institution prescribes playing
to win. But it is by playing to win, frequently, that these benets are achieved.
A game known to be rigged isnt exciting even if the score is close. Athletes do
not usually perform for fans in the way that rock musicians do, even if they depend
on fans just as much for their livelihood and fame. (Performers of classical music
may have objectives somewhat more like those of athletesthey may think of
themselves as serving the music rather than the audience. And this may be what
listeners want and expect.)
The distinction between evaluation of the institution of sport and evaluation of
players and plays as to their aptness for winning, is reminiscent of John Rawlss
distinction between justifying the institutions of promising and punishment, and


justifying actions falling under these institutions.2 The institution denes what
counts as winning and tells participants to try to win (if they want to participate
in the institution), just as the institution of promising determines what constitutes
promising and keeping promises, and tells people to keep their promises. One can
step outside the institution of sport and ask whether having it, with its criteria for
winning and its injunction to try to win, is a good thing, just as one can ask whether
the institution of promising is a good thing, from a perspective outside of it.
There can be values internal to an institution (in one or another sense) without the criteria of value being explicitly stipulated, of course. The institution of
stamp collecting has its own special criteria of excellence, but not because there
is anything like an authoritative rule book spelling out the object (or objects) of
the institution.3 Rareness in stamps is (intrinsically) valuable from a perspective internal to the institution, and to value a stamp because it is rare is to participate in the institution. This fact is recorded in the literature, but it is simply
the practice of stamp collectors that makes it a fact. Other values internal to the
institution are less denite and harder to specify, and many are not even recorded.
(Of course a nonparticipant may appreciate the economic worth of a stamp which
results from the value stamp collectors accord it by virtue of its rarity. This
economic value is of course instrumental.)


The value of winning, in competitive games, will strike us as very unlike aesthetic
value. But there may be more to the comparison than rst appears. The main difference, it seems, is that winning is, in itself, a value only in scare quotes, only within
and relative to the institution. It does not really matter, in general, whether one
wins or loses. Maybe participants in the game only pretend that it matters; there
may be value in engaging in this pretense.4 (Sometimes it does really matter for
extrinsic reasons; winning may boost ones condence or alumni donations. But losing can also have good consequences. It may teach one how to handle failure; it may
force one to pay more attention to things that matter more than sports do.) From a
perspective external to the institution of stamp collecting, the value accorded to
rareness may seem rather arbitrary; there isnt really anything especially good about a
stamps being rare. Aesthetic value, by contrast, is surely real (as real as any values are);
beauty really is a good thing, it seems, and it is good apart from consequences.5

2. John Rawls, Two Concepts of Rules, Philosophical Review 64 (1955), pp. 332.
See especially p. 16.
3. I owe this example to David Hills.
4. [See my Its Only a Game!: Sports as Fiction, in Walton, In Other Shoes (forthcoming).]
5. The sense in which the value of winning and that of rareness in stamps are unreal
needs examination. Sports fans and stamp collectors really do care about winning and


M A RV E L O U S !

Once we have attached scare quotes to the value of winning, however, we nd

analogies to aesthetic value. The (scare-quoted) value of winning is intrinsic, as
aesthetic value is supposed to be. Winning is good in itself (from the perspective within the game), not because of its consequences; whatever constitutes winning is the object of the game. And the value of winning is independent of other
values, of practical values, moral values, economic values, and so on, as aesthetic
value is supposed to be. If what counts as aesthetic value is somehow decreed by
an artistic tradition, as what counts as winning and losing is decreed by the institutions of baseball, tennis, bridge, and so on, we will be able to account for both
the intrinsicness and the distinctiveness of aesthetic value. What about unity?
Is the notion of winning, in competitive games, a unied one? What counts as
winning in different gamesbasketball, chess, horse racingis very different.
Still, we might say that to win is, in every case, to achieve the object of the
game, whatever that happens to be. Maybe aesthetic value has a similar unity in
Lets set aside the notion of aesthetic value, for a moment, and consider
whether there might be anything in the arts at all like winning (and losing)
in sports, anything analogous to the object of a game, any values that might
need scare quotes, that are values only within and relative to an artistic tradition. There are in various artistic contexts (unwritten) understandings that
certain objectives are to be pursued, and there are traditions of valuing their
achievement. A goal of some visual art is the realistic portrayal of a threedimensional world on at surfaces; some artists attempt to come as close as possible to fooling the eye. Composers, in some periods, compete with each other
in the craft of counterpoint, in producing intricate combinations of melodic
lines with their inversions, retrogrades, retrograde inversions, augmentations
and diminutions, and so on, while conforming to the rules of one or another system of counterpoint. Musical performers aim for technical perfection and sometimes put on ashy demonstrations of skill. Some writers endeavor to reproduce

about rareness, it seems. Fans may also desire to desire that a given team win, which on
some accounts constitutes valuing this result. (See David Lewis, Dispositional Theories of Value, Proceedings of the Aristotelian Society, Supplementary Volume 13, 1989, pp.
113137.) After all, they (probably) chose voluntarily to be sports fans and to root for
the teams they root for. And stamp collectors are not forced to pursue their hobby or to
accept its traditional emphasis on rareness. So maybe they want to desire (to possess) rare
stamps. Presumably only what people are disposed to value in relevantly ideal circumstances is really valuable for them. Would the fans and collectors meta-desires survive
cognitive psychotherapy? Would they survive full disclosure and vivid awareness of all
relevant facts? Maybe, if there are extrinsic reasons for buying into the institution, for
being a sports fan or a stamp collector. But will that account for the apparently intrinsic
value, for the fan or the collector in question, of the relevant team winning or of the
collectors possessing rare stamps?



local dialects as faithfully as possible; others take up the challenge of rendering

ordinary speech in established poetic forms. Various other constraints, either
self-imposed constraints or ones inherent in a chosen medium, challenge artists
in other ways. Some artists attempt the trick of telling a story by means of
visual images only, without the help of words. Others dream up devices for
portraying motion in still pictures. John Hollander describes the pantoum verse
form as follows:
There may be any number of quatrains, but, starting with the second one,
they are generated by repeating the even-numbered lines of each as the
odd-numbered ones of the next. The nal line of the poem repeats the opening
one. In addition, a touch of riddle is preserved in that the rst half of each
quatrain is about something wholly different from the second half.6

One cant help but think of constraints imposed by the rules of competitive
games, under which one attempts to achieve the object of the game. Sometimes
the object, in the case of an artistic style, is not very denite, even if the constraints are; satisfying the constraints may itself be difcult enough, and the
object may amount to producing something satisfying them that makes sense
or is interesting.
There is a lot of variety in these examples. In some, the (stated or unstated)
objective and the restrictions under which it is to be pursued may be thought to
serve aesthetic value (whatever that is). No doubt, in increasing the realism of their
portrayals artists sometimes achieve greater aesthetic value. Avoiding parallel
fourths, fths, and octaves in contrapuntal writing in the style of Bach is perhaps
justied by the fact that such parallels sound ugly in that style. (Their ugliness
probably has something to do with their tendency to make the two voices sound
like one, or their establishment of a clunky harmonic rhythm.)
But the connection with aesthetic value is often tenuous. Deceptive realism
and intricate counterpoint certainly do not always make for aesthetic value; they
can have the opposite effect. And aesthetic value can be achieved by deliberately
unrealistic portrayals, or by studied contrapuntal simplicity. Emphasis on goals
like those I mentioned can reduce a work to a sterile academic exercise. Virtuosity in a musical performance, athletic technical facility, may replace inspiration
and insight. Some will speak of art degenerating into craft.
One may have the impression that some of the goals or objectives artists pursue are arbitrary, to a greater or lesser extent, an impression that is reinforced by
the fact that the goals and the restrictions change from generation to generation
and vary from culture to culture and genre to genre. It is as though artists devise

6. John Hollander, Rhymes Reason: A Guide to English Verse (New Haven, Conn.: Yale
University Press, 1981), pp. 4344.


M A RV E L O U S !


puzzles for themselves and try to solve them, or exercise skills just for the heck
of itlike children making up games or contests.7 We may be impressed by
the skill an artist demonstrates in achieving her objectives. But concerning the
objectives themselves, the results, we may be temptedin some instances more
than in othersto ask, So what? What is the point? The achievement may
interest us only to the extent that we buy into the artistic tradition, only to
the extent that we choose to accept the goals the artist was pursuing, the object
of her game, as desirable. This choice may seem arbitrary, in the way a choice
to play a game with certain rules and a certain object is arbitrary. Only if we do
make such a choice will we be impressed by the artists achievement.
I will suggest that what seem to be, what in fact are, arbitrary goals and constraints are often connected more closely than one might think, and in a surprising way, to aesthetic value, a kind of value that transcends the mere achievement
of these goals.


Lets look more closely at the place judgments of value, the making of judgments of value, have within the institution of (ne) art. After listening to a late
Beethoven string quartet or reading War and Peace or watching a performance
of King Lear, I exclaim, That was wonderful! Marvelous! In doing so I am
making a judgment, claiming that the work or performance in question is of
great value, and I am expressing my admiration for it. But my judgment and
my admiration are not just responses to the value I recognize; they are partly
constitutive of it. The value consists in part in the experience of judging the
work or performance highly. It is partly by virtue of eliciting admiration that it
is worthy of admiration. To gain the benet of the works value is to appreciate
it; if I didnt feel admiration for the work (or the artist) or judge it highly, if
I merely felt pleasure or enjoyment as a result of my experience with it, I would
not be appreciating it.
We can reap the benets of many other good things without judging them to
be good or admiring them. A good hoe or a good car will do its job efciently
and well, and benet the user accordingly, regardless of what she thinks about its
value. The car, if it is a good one, will provide dependable and safe and efcient
transportation for years and years even if the beneciary berates it constantly.
The value of the car and the hoe are instrumental. But the same goes for many
intrinsic values. Suppose that it is a value for me that I achieve posthumous

7. Historical circumstances make certain goals salient in a given cultural context,

no doubt, and help to explain their adoption, even if they are arbitrary in the sense that
neither having nor achieving them is in itself more valuable than having or achieving
alternative goals would be.



fame, or that my friends not sneer at me behind my back. (Perhaps this is so by

virtue of the fact that, after cognitive psychotherapy or whatever, I would desire
it, or desire to desire it.) I will never be in a position to praise my posthumous
fame, and I may think, paranoically, that my friends do sneer at me behind my
back. Nevertheless, I may in fact possess what is in fact an intrinsic good for
meforthcoming posthumous fame and friends who dont sneer. Could I fail
to realize that these are values for me, as well as failing to realize that I possess
them? Probably; I may not have sorted out the desires or counterfactual desires
which (on some accounts) determine what is good for me. So I could possess the
value without either judging it to be valuable, or judging that I am in possession
of itwithout saying anything like How marvelous! How wonderful!
Much the same goes also for some desirable experiences, and for things whose
(inherent, if not intrinsic) value lies in the fairly direct production of pleasurable
experiencesthings like a hot shower, or a walk around the block. One neednt
think the shower or the walk or ones experience of it is anything special, or even
entertain questions about how good it is, in order to enjoy it. One can enjoy it
without appreciating it. (Actively denigrating the shower or the walk might
ruin ones enjoyment, however.) Compare enjoying a folk song or a game of
make-believe without thinking itor the performance of the song, or the props
in the game, or ones own or another participants activitiesis special. One
might enjoy the shower or the walk more if one does think it is special, however.
Perhaps in that case ones enjoyment is in part aesthetic.
Aesthetic value arguably consists in a capacity to elicit in appreciators pleasure of a certain kind, pleasurable experiences (or experiences of enjoyment, satisfaction, gratication?). But unlike some pleasures produced by hot showers
and walks around the block, aesthetic pleasures include the pleasure of nding something valuable, of admiring it. One appreciates the work. One does not
merely enjoy it; one takes pleasure or delight in judging it to be good. One
marvels at it. Again, I dont think I would enjoy Beethovens C# Minor Quartet
as I do if I didnt have my admiration of it to delight in, if I were not inclined
to exclaim, How marvelous! How wonderful! The quartet doesnt deliver up
what it has to offer, not all of it anyway, unless we give the work the credit due
to it. Its value consists in part in its propensity to induce observers to judge it
valuable, and to enjoy doing so.
The owner of a hoe or a car might, in addition to hoeing her crops or driving her car, appreciate and admire how marvelously suited the hoe or car is to
its task. This gives her a certain enjoyment in using this tool (and perhaps just
in owning it), on top of the assistance it provides in the cultivation of her crops
and in getting herself from one place to another. It seems not unreasonable to
describe this enjoyment as aesthetic appreciation.
There is more than winning and losing in baseball. A team may win a game,
but win ugly. This may amount to winning in a manner that does not elicit
admiration. A fan rooting for the ugly winner may like the result, but her


M A RV E L O U S !


enjoyment will lack what we might call the aesthetic dimension of pleasurable
admiration. To play a beautiful game of chess, whether one wins or loses, is to
play in a way that does elicit admiration, and the delight that goes with it.
In addition to laughing as a result of watching a comedians act, one may
notice and admire the elegant, masterful means by which the laughs are elicited,
the comedians superb balance of sense and absurdity, the exquisite timing of
his delivery, his sheer cleverness, and so on. One appreciates the artistry of the
comedians routine in a way that goes beyond merely nding the jokes funny. (It
is likely that the appreciation of the comedians technique makes his jokes even
funnier than they would otherwise be, however.)
Reading a profoundly perceptive poem may benet one cognitively. One may
come to new understandings about life or even acquire bits of wisdom from ones
experience with the work, and one may enjoy achieving this illumination. But the
reader may also appreciate, he may admire with pleasure, the poets perceptiveness and insightfulness and her skill in presenting profound truths in a vivid and
convincing manner. Then the readers enjoyment is (in part at least) aesthetic.
Another reader might enjoy the cognitive benets but without admiring the
poem (or the poet) and hence without the pleasure of admiration. He may not
give the work (or the poet) credit for the wonderful new insights he obtains while
reading. He may think he achieved them himself, that he came to his conclusions
on his own while being stimulated only accidentally by the poem. (The poem
might in fact have been designed to have exactly this effect, of course.) The reader
acquires knowledge and enjoys doing so, but he does not experience the pleasure
of admiration. He, arguably, does not benet aesthetically from the experience.
( John Cages objective may be, in part, to deaestheticize our experiences, to get
us to enjoy sounds themselves without admiring them or their creators.)
People have a natural tendency to enjoy the experience of admiring things.
An evolutionary explanation of why this is so would seem not hard to come by.
But admiration is not necessarily pleasurable. Sometimes admiration is without
delight, respect grudging. A disturbing but perceptive novel may benet us
cognitively, we may learn from it and realize that we do, but without taking
pleasure in admiring it, without saying How marvelous! A person who reacts
to a cartoon with hostility may know that she is getting medicine she needs; she
may realize that the cartoonist is skillfully and perceptively forcing her to see a
painful truth. But she may hate the cartoon for it. She may admire or at least
approve of it, but without enjoyment.
As a rst stab, lets dene aesthetic pleasure as pleasure which has, as a component, pleasure taken in ones admiration or positive evaluation of something;8 to
be pleased aesthetically is to note somethings value with pleasure. This makes

8. My suggestion is that we regard pleasure taken in the object as part of ones aesthetic
pleasure if it is combined with pleasure taken in ones admiration for the object.



aesthetic pleasure an intentional state, not just a buzz or a rush caused by experiencing a work of art. The pleasure of a hot shower or a walk around the block is
presumably an intentional state also. One takes pleasure in something; the pleasure attaches in part to ones awareness of something. But one is not pleased by
the shower or the walk in the way I am pleased by Beethovens C# Minor Quartet,
unless one takes pleasure not only in the shower or the walk or ones experience of
it, but also in ones experience of admiring it, in ones judging it to be good.
Certain modications of this account of aesthetic pleasure are in order. A person might take pleasure of a self-congratulatory sort in admiring something;
one might pat oneself on the back, with delight, for ones sophisticated and
subtle taste in recognizing the things merit. This pleasure would seem not to
be aesthetic. The needed restriction is something like this: Aesthetic pleasure is
not just pleasure taken in my admiration of something, but in its getting me to
admire it. We may also want to broaden the denition considerably. As I will
suggest shortly, we may want to count pleasure taken in certain attitudes other
than admiration, as aesthetic. But lets leave the proposal in its simple form for
now: aesthetic pleasure is pleasure which has as a component pleasure taken in
ones admiration of something.
Aesthetic value will no doubt have something to do with the capacity to produce
aesthetic pleasure. But rather than dening it simply as a tendency to elicit aesthetic pleasure, or to do so in appropriately constituted and positioned observers,
I would propose requiring that, for something to have aesthetic value, there must
be a certain propriety in taking aesthetic pleasure in it; it must be reasonable or
apt or make sense to do so. I leave aside the question of how to spell out this
propriety.9 As a model, think of humor. If something makes all of us laugh, we
might nevertheless deny that it is funny on the ground that we shouldnt laugh at
itbecause it is in bad taste, or is racist, or whatever.
We may sometimes be bamboozled or tricked or deceived into admiring something which does not merit our admiration, and take delight in admiring it.
These may be cases in which our aesthetic pleasure is inappropriate, and hence
the thing does not possess the aesthetic value it seems to possess. A close examination or analysis of the work might convince us that it is shallow and unworthy
of our appreciation, and we may then prefer not to appreciate it, not to admire
it, even if such admiration would be enjoyable. One might argue that the moral
reprehensibility of Leni Riefenstahls Triumph of the Will makes it improper to
admire it with pleasure, and so undercuts the aesthetic value it would otherwise
have had.10 This gives us a way of understanding how moral and aesthetic value
can interact, while still taking them to be distinct.

9. Allan Gibbard, Wise Choices, Apt Feelings (Cambridge, Mass.: Harvard University
Press, 1990).
10. [Cf. Morals in Fiction and Fictional Morality, chap. 3, this volume.]


M A RV E L O U S !


Understanding aesthetic value in this way enables us to accommodate its

diversity while locating a common thread. The thread is the pleasure taken
in admiring things. The diversity lies in what we admire things for. We may
admire a work for the way it soothes us, or excites us, or provokes us, for the
intellectual pleasures it affords, or the emotional ones, for the insight it provides
or the manner in which it does so, for the way it enables us to escape the everyday cares of life, or the way it helps us to face life, and so on and on. But none of
these grounds for admiration itself constitutes the works aesthetic value. If we
take pleasure in admiring the work for whatever we admire it for, this pleasure is
aesthetic. And if such pleasure is properly taken in the work, this constitutes the
works aesthetic value. In this way aesthetic value is distinct from, yet dependent
on, whatever value we admire the object for. If the value we admire it for is a
practical one, this practical value underlies but does not constitute the aesthetic


Lets explore the diversity a little, the values for which one might enjoyably
admire something.
An appreciators enjoyable admiration, usually if not always, involves not only
recognizing a things valuerecognizing the marvelous job it does of opening
our eyes to important truths, for instance, or how wonderfully suited it is for
providing safe and efcient transportation; ones admiration also involves recognizing the creators accomplishment, the talent and skill a person demonstrated
by producing something with this value. Admiration is paradigmatically, if not
essentially, an attitude we have in part toward people.
One can admire a persons talent or skill in accomplishing an objective whether
or not one thinks much of the objective. I may think that hitting, with a stick,
a spherical missile traveling 90 mph, or winning a baseball game, is neither here
nor there, that the world is not a better place for it (neither because winning or
hitting a fastball is good in itself, nor because it has good consequences); yet

11. Kant asks whether in the judgment of taste the feeling of pleasure precedes or
follows the judging of the object (Critique of Judgment, 9). On my account of aesthetic
value, the answer is (with qualications) both. If aesthetic pleasure is in part pleasurable
admiration, pleasure taken in judging something highly, it comes (logically, if not temporally) after this judgment. The prior judgment is not a judgment of aesthetic value (not a
judgment of taste) in the cases I have described so far, but in other cases it is, as we shall
see shortly; sometimes we take pleasure in judging something to be excellent aesthetically. Such bootstrapping is typical of central instances of aesthetic value. Aesthetic judgments are, however, judgments of an objects capacity to produce pleasurepleasurable
admirationand it is normally by experiencing this pleasure that one detects the objects
capacity to provide it; thus does aesthetic judgment follow feelings of pleasure.



I may admire the athletes accomplishmentthe skill and concentration and

strategy that make for the win. And I may take pleasure in my admiration.
(Maybe it helps if I somehow pretend that the result is desirable. I need not base
my admiration on the thought that the abilities which enable the players to win
are ones that would be genuinely useful in other contexts.) Of course, if I think
baseball is a waste of time and energy, that it diverts attention and resources
from the important problems of life, if I think that winning, or even playing, is
undesirable, this may prevent me from admiring the players accomplishment, or
from admiring it with pleasure.
So among the values that can underlie aesthetic value are ones we might consider to be arbitrary, scare-quoted. We may take pleasure in admiring someone
for the accomplishment of an arbitrarily chosen objective. The pleasure constitutes a real value, even if the result accomplished, in itself, does not. We might
admire an artists skill in painting bubbles convincingly, and take pleasure in
admiring it, whether or not we think convincingly painted bubbles are themselves a good thing (aesthetically good or good in some other way). We may
admire with pleasure a composers accomplishment in producing intricate counterpoint, or a poets skill in satisfying the conditions of the pantoum, whether or
not we think these feats have any merit themselves.
Compare a photograph of a bubble with Jean-Simon Chardins painting of a
boy blowing soap bubbles. The convincing photographic portrayal might have
some interest for us in itself. But we will not have the same pleasurable admiration for the photographers achievement that we have for Chardinsknowing,
as we do, that it takes relatively little skill on the photographers part to capture
convincingly the bubbles transparency as well as its partial reection of light,
its perfectly rounded form, and so on. (One might be awestruck by the medium
of photographypeople were, when photography was inventedbut that is
different.) Consider also computer-generated counterpoint.
Even if the values underlying aesthetic value are arbitrary and scare-quoted,
it doesnt follow that the work itself isnt really valuable. It has the desirable
capacity to induce in appreciators pleasurable admiration, although this capacity belongs not to the physical work itself, but to the work understood in a
certain wayas an artists attempt to accomplish certain possibly arbitrary
Dont such demonstrations of skill sometimes conict with aesthetic value?
Dont they sometimes make for academic and uninspired art, and doesnt admiration of skill sometimes interfere with genuinely aesthetic appreciation? Yes,
but this doesnt prevent aesthetic value from consisting in a capacity to induce
pleasurable admiration. Academicness in art sometimes amounts to the demonstration of a relatively mundane skill (perhaps one that has become mundane by
the constant attention of artists in a certain tradition), a skill that any moderately
talented and persevering artist might now accomplish, and one that does not
induce much pleasurable respect or admiration. Sometimes technical virtuosity


M A RV E L O U S !


replaces inspiration, or the demonstration of insight or ingenuity, of a kind that

we would admire more and with more pleasure than the most impressive technical abilities.
Insofar as the values for which appreciators admire a work consist in the
achievement of arbitrary goals, there is a clear sense in which aesthetic value
is likely to be institutional, tradition bound. The arbitrary goals are set by
the institution. There is nothing like a rule book specifying the object of the
game, but there are understandings within an artistic tradition about what
goals artists are expected to pursue. Only if the appreciator is familiar enough
with the tradition to recognize the goals and goes along with them, only if she
in this way buys into the institution, will she admire their achievement and
take pleasure in doing so. One must also be familiar enough with the task to
appreciate its difculty, and this may come from familiarity with the institution. We may have here a partial explanation of the peculiar susceptibility of
art to charges of fraudulence, and what can seem the fuzzy boundary between
fraudulence and profundity. If we buy into the institution with its arbitrary
objectives, we say How marvelous! If we focus on their arbitrariness, we call
The value underlying a works aesthetic value can itself be aesthetic. We may
admire something for its capacity to elicit pleasurable admiration, and take pleasure in admiring it for this. Then we have aesthetic value which is not only
distinct from practical or arbitrary values, but also independent of them. Bootstrapping of this sort amounts to art for arts sake. Often bootstrapping is only
partial. One may admire something, with pleasure, for its capacity to produce
insight, for instance, and also admire it, with additional pleasure, for its capacity
to elicit the former admiration. I have some temptation to dene aesthetic value
as necessarily involving an element of bootstrapping. This would help to exclude
cases in which the pleasure one takes in ones admiration is of a self-congratulatory
sort. Bootstrapping requires that one admire the work for its capacity to elicit
admiration; admiring oneself for ones admiration doesnt sufce.

I have gradually been stretching the word admiration out of shape. I spoke
of admiring things for their practical (e.g., cognitive) values and for their aesthetic value, and also of admiring people for the achievement of difcult even
if arbitrary objectives. And there is more variety to come. I suggest replacing
admiration with a family of related terms. What will remain is at least this:
that aesthetic pleasure consists in pleasure taken not just in an object or person
itself, but in an attitude one has toward an object or person, the attitude being
either admiration or something else.
Sometimes our attitude toward what we take to be aesthetically valuable is
better described as one of awe or wonder than one of admiration. This is especially



true in the case of aesthetically regarded natural objects, things that are not
the product of human activity and do not call for recognition of a persons
achievement. (One may of course respond to works of art with awe or wonder as
well.) The aesthetic value of sunsets, alpine meadows, waterfalls, and owers may
consist (in part) in our taking pleasure in the awe or wonder we feel toward them.
This enables us to explain the fact that aesthetic appreciation of natural objects
seems, pretheoretically, to be similar to, but also signicantly different from,
much of our aesthetic appreciation of works of art. The difference lies in the
fact that pleasure is taken, in one case, in admiration for a persons accomplishment and, in the other case, in the rather different experience of awe or wonder,
which need not involve recognition of a human accomplishment. The similarity
lies in the fact that in both cases pleasure is taken in an attitude one has toward
something (and in whatever similarity there is between the attitudes of awe or
wonder, and admiration). In the case of both natural objects and works of art, one
may be surprised by the object or by particular features of it, or one may experience feelings of familiarity and recognition, and one may take pleasure in these
experiences, and in other as well. These responses may be components of ones
awe or wonder or admiration.
Terrible events or activities may demand of us a certain respect or awe, if not
admiration, and we sometimes delight in feeling this respect or awe, even while
genuinely regretting the terrible occurrences. (This kind of case is to be distinguished from one of grudging admiration, when one does not take delight in
ones admiration.) Something like this may occur when we experience what has
been called the sublime. One might also delight in judging something negatively,
or in being revolted or annoyed by it or in nding it offensive or repugnant.12
Revulsion can coexist with delight in responding with revulsion. (Things we
love to hate.) Some anti-art may be designed to elicit reactions of this sort. And
they help to account for the attraction painful works of more traditional kinds
sometimes have for us, works that provoke in us negative rst-order responses.
Revulsion is very different from admiration, although it might involve a certain
awe or wonder. In any case, the pleasure is, again, pleasure in experiencing an
attitude toward something.

12. Daniel Jacobson suggested this possibility. In an interesting paper on The Pleasures of Tragedy (American Philosophical Quarterly 20 [1983], pp. 95104), Susan Feagin
argues that the pleasure we derive from tragedies is a meta-response, arising from our
awareness of, and in response to, the fact that we do have unpleasant direct responses to
unpleasant events as they occur in the performing and literary arts (p. 98). Some of the
metaresponses she has in mind are self-congratulatory pleasures (We nd ourselves to be
the kind of people who respond negatively to villainy, treachery, and injustice [p. 98]),
and so might not be indications of aesthetic value. But they are still pleasures derived
from tragedy. See also my Mimesis as Make-Believe: On the Foundations of the Representational
Arts (Cambridge, Mass.: Harvard University Press, 1990), 7.3.


M A RV E L O U S !


Sometimes, of course, we are pained or displeased by the experience of judging

something negatively. And we may judge it even more negatively as a result of
this displeasureand so be all the more displeased.13 Displeasure and disapproval may thus feed on and reinforce each other, as pleasure and admiration do
in positive cases. It is as though the stakes for the artist are continually being
But whether or not one enjoys being revolted or judging a work negatively,
one might at the same time admire it for its capacity to produce revulsion or
to elicit a negative judgment, and one might enjoy admiring it for this. If we
understand the artists objective to be to disgust the appreciator or to provoke
negative judgments, we may admire with pleasure his achievement in accomplishing this end. The kind of aesthetic value that consists in a capacity to elicit
pleasurable admirationwhat I am inclined to regard as the central or paradigmatic variety of aesthetic valuecan thus coexist with, and indeed depend on, a
capacity to disgust or irritate or evoke negative judgments. And we may, accordingly, judge the work positively. We may or may not feel a conict between the
positive judgment and a negative one on which it depends. If we do, we will feel
conictedas we sometimes do in any case, especially when we confront some
of the more disturbing contemporary works.


It is an open questionand by no means an easy onehow the conception of
aesthetic value I have outlined sorts with traditional ones, how close it comes to
capturing what people have meant in speaking of aesthetic value (or beauty,
or elegance, or sublimity) in contexts of one kind or another, whether critics
have something like it in mind when they pronounce works of art to be good
or bad, masterpieces or failures. I do not doubt that some entrenched notions of
aesthetic value have little to do with the account I have offered. And the possession of what I am calling aesthetic value is certainly not our only reason for
valuing paintings and plays and symphonies. But I, for one, am sure that a signicant portion of the enjoyment I receive from many works of art derives from
my admiring them or judging them to be valuable in one way or another. My
enjoyment depends on my assessing the work positively, on my being moved to
declare, How marvelous! The word appreciation ts this enjoyment nicely,
suggesting as it does not just pleasure felt in response to the work but a recognition of its worth. If appreciation is understood to be central to the aesthetic, we
might expect the capacity to elicit pleasurable admiration to qualify with little
strain as aesthetic value.
Appreciation may not be quite the right word for pleasure taken in ones
experiences of awe or wonder or, especially, shock or irritation or revulsion, and
13. I am indebted here to Arthur Danto.



capacities to induce these pleasures may strike us as constituting less than paradigmatic instances of aesthetic value. But such capacities are values, ones that
are prevalent in works of art. And they are akin to paradigmatically aesthetic
value in being capacities to induce pleasure in ones attitudes toward things.
Moreover, as we noted, they are values on which paradigmatically aesthetic value
is sometimes based, values for which one pleasurably admires things. They may
be reasons for declaring, How marvelous!14

Since composing How Marvelous! I have come across a number of observations by other writers which seem to me to point in the direction I took in that
essay, or at least can be construed as doing so.
Thomas Reid wrote, in 1785, that
Beauty is found in things, so various, and so very different in nature, that it is
difcult to say wherein it consists, or what there can be common to all the objects in
which it is found. . . . What can it be that is common to the thought of a mind, and
the form of a piece of matter, to an abstract theorem, and a stroke of wit? . . . There
seems to be no identity, nor even similarity, between the beauty of a theorem and the
beauty of a piece of music, though both may be beautiful. . . .
If there be nothing common in the things themselves, they must have some
common relation to us, or to something else, which leads us to give them the
same name.
All the objects we call beautiful agree in two things, which seem to concur
in our sense of beauty. 1st, When they are perceived, or even imagined, they
produce a certain agreeable emotion or feeling in the mind; and 2dly, this
agreeable emotion is accompanied with an opinion or belief of their having some
perfection or excellence belonging to them.1

14. This paper benefited from comments by Arthur Danto, David Hills, Peter
Railton, Alicyn Warren, and Stephen Yablo, and from discussion at the 1991
National Endowment for the Humanities Summer Institute for Aesthetics and the
Histories of Art.
1. Thomas Reid, Essays on the Intellectual Powers of Man (Cambridge, Mass.: MIT
Press, 1969), pp. 779780. (Originally published 1785, Edinburgh.)



M A RV E L O U S !


Reid then asks whether the pleasure we feel in contemplating beautiful

objects may have any necessary connection with the belief of their excellence.
He leaves this question hanging, but seems inclined toward a positive answer.
Much more recently, Clement Greenberg argued that nothing can be experienced esthetically without a value judgment, nothing can be experienced esthetically except through a value judgment. Esthetic experience is constituted by


The notion of things being enjoyable because they are so bad, which my account
helps to explain, is commonplace. Mozarts Ein Musikalischer Spass is a paradigmatic instance in which pleasure is taken in the masterful accomplishment of
otherwise negative ends:
In the hands of a bad composer, lines in parallel thirds, sixths, and tenths can
produce intolerable tonal ambiguities and cacophonous dissonances. This is
especially true if, as sometimes happens, two linear progressions occur at once,
each counterpointed in parallel motion. Mozart made use of this possibility in a
wonderfully dreadful passage in the Musical Joke.3

The listeners pleasure is in part the pleasure of admiration. We understand

Mozart to have been aiming for dreadfulness, and admire his success in achieving
it so splendidly.
By contrast, the reviewer who declared the movie Bubble Boy to be such a
unique mess that its awfulness becomes weirdly enjoyable, and described its
theme as spectacularly unsubtle, probably was not admiring the skill and talent of the moviemakers, with pleasure or without; presumably the moviemakers
were not aiming for awfulness. Rather, the reviewer enjoys something like awe
or amazement at how awful the thing turned out to be despite the efforts of its

2. Clement Greenberg, The Experience of Value, in Homemade Esthetics: Observations

on Art and Taste (New York: Oxford University Press, 1999), p. 59. But it is pretty clear
that, for Greenberg, only judgments of aesthetic value make for aesthetic experience. And
he appears to identify value judgments with liking and disliking.
3. Carl Schachter, A Commentary on Schenkers Free Composition, in Schachter,
Unfoldings: Essays in Schenkerian Theory and Analysis (Oxford: Oxford University Press,
1981), p. 204. My italics.
4. Bob Campbell (Newhouse News Service), Bubble Boy So Awful, Its Enjoyable,
Ann Arbor News, 25 August 2001.

This page intentionally left blank


he test of time is a deeply ingrained tenet of folk wisdom. The works of art
which last through the centuries, it is assumed, are those which are great. But
why assume this? What does longevity have to do with greatness? What warrant
is there for taking longevity as proof of greatness? These questions constitute the
framework of Anthony Saviles book.
Once we reect on it, plausible ways of justifying the test of time are easy to
come by. Indeed, what is striking, and even embarrassing, is their abundance and
diversity. This should make us suspect ourselves of confusion. The test of time
may be several importantly different tests masquerading as one, not all of them
equally defensible. Only if we have a clear grasp of the differences can we expect
to use any of them intelligently. Is it simply that considered opinions are more
to be trusted than snap judgments, and hence that we are better able to evaluate works which have been around for a while? Thus, if our assessment of a work
remains favorable after many years or centuries it is an assessment we can be condent of. Or is it that in order to last, a work must remain alive through changes
of local conditions? It must speak to different cultures, to people in different
historical contexts, to people with different sensibilities and different concerns;
it must have something to say to everyone? This multifacetedness, perhaps, is
what constitutes greatness. Savile rejects these two suggestions, but argues that
the test of time, properly construed, is legitimate and is an essential feature of art
theory. He begins by noting Rostovtzeffs observation concerning the Emperor
Diocletianthat his reforms in administration, judicial proceedings, and nancial
In 1982 Anthony Savile published an important and richly fascinating study,
The Test of Time: An Essay in Philosophical Aesthetics (Oxford University Press). The present
essay, taken from a review of this book, originally appeared under the title Degrees of
Durability (Times Literary Supplement, 18 February 1983).




and military organization stood the test of timeand later introduces more
mundane examples such as that of a coat which is assessed highly because it wears
well. Savile takes these tests of time to be relatively unproblematic, and directs his
efforts towards exploring the analogies between them and tests of time in the arts.
But the nonartistic cases need to be examined with some care. The coat analogy
especially is rather seriously misleading when used as a model for understanding
the test of time in the arts.
Things which endure have the capacity to endure. Sometimes this capacity
is itself a virtue, because longevity is desirable. A long-lived coat or bicycle or
lawn mower doesnt have to be replaced or repaired. That is an advantage, and
the capacity to lastdurabilityis a virtue in such artefacts. The coat which
survives years of use has proved itself to be durable; hence we judge it highly.
Durability is a virtue of administrative and judicial procedures as well. Changes
bring confusion, disruption, instability. The very fact that Diocletians reforms
had the capacity to survive is one reason to think well of them. But it is not the
only reason, and not the only one which their survival supports. Their endurance
presumably demonstrates that they were reasonably fair and reasonably effective
in achieving societys goals; otherwise we might expect them to have been challenged and overthrown. So, the test of time can be construed very differently in
this case from how it was in the case of the coat. The merit of the coat to which
its survival attests would go unrealized if, because of re or shipwreck or some
other outside interference, it came to a premature demise. In that event a less
durable coat would have served just as well. But fairness and effectiveness in
achieving societys goals are values quite apart from their contribution to longevity. Even if an administrative system is doomed to early destruction by invasions
of Mongol hordes, it is desirable that it be fair and effective while it lasts. The
survival of Diocletians reforms is evidence for qualities which are virtues independently of their survival value.
The coat analogy suggests that durability in works of art is itself a virtue, and
the one which the test of time tests for. But it becomes clear as the argument progresses that this is not Saviles considered opinion. We will surely want to side
with him here as against the analogy. The value of a great work is not unrealized
if its life span is prematurely cut short. If, but for outside interference, a work
would have lasted through the centuries it is presumed to have qualities which
are to be valued even for the short time it actually endured. If durability is itself
a virtue in works of art it is not the only one which the test of time is supposed to
indicate. Indeed, it is not clear that durability is, in general, an aesthetic virtue at
all, still less that it is constitutive of greatness. Glassware which is fragile is not
thereby less good aesthetically than more hardy pieces; if anything, the reverse
is true. To be sure, this is physical fragility, and Savile rightly insists that the
durability relevant to the test of time in the arts is the ability to survive in our
attention. But broken glasses are not likely to last long in our attention. It is a
pity to break a glass; its survival is valued. But that doesnt mean that the glass



itself is better aesthetically for having the capacity to survive, either physically
or in our attention.
The idea that great works are ones which accommodate very different sensibilities and that that is why they endure deserves especially close consideration. The
fact that a work of one age accidentally ts well in later contexts, that it contains
elements which, as it happens, can be reinterpreted in a way later appreciators
will nd interesting and valuable, is no credit to the work itself and no ground
for judgments of greatness. Neither is ambiguity or multifacetedness designed
into a work the essence of greatness. Adding a magnifying glass, saw, toothpick,
tweezers, and corkscrew to a simple pocketknife does not elevate it toward greatness. Greatness is not a conglomeration of lesser merits. What is important is not
the number of functions a work can perform but something more like how well
it performs them. But then why dont great works wither when, in a later age,
the functions that they perform so well are no longer in demand? If endurance
bespeaks mere multifacetedness, it loses its connection with greatness. To pass
the test of time, Savile contends, is not to survive by opportunistically changing colors as changing cultural conditions require but to survive under a constant
appropriate interpretation.
But Savile neednt deny that greatness involves a certain multifacetedness.
What is important is that there be deep connections among the various faces
of a great work. Perhaps a great work is one which at some very basic level has
a single meaning which manifests itself in different ways in different cultural
contexts. A work which on the surface speaks of Napoleon may on a deeper level
speak of political power and, on a deeper level still, of power relationships in
general. So, it may with equal force engage those whose interests are contemporary politics, or power struggles involved in interpersonal relationships, or even
perhaps conicts within a given persons own psyche. Its underlying insight may
be so abstract as to be scarcely formulable. Yet it is this which ties together the
more obvious particular interpretations critics give to it in different ages and
which is ultimately responsible for its ability to accommodate itself to changing
climates. The fact that the works treatment of particular local issues is based
on some such fundamental insight gives its treatment of them a kind of depth
which is arguably the essence of greatness. This depth is to be valued even when
it happens not to result in longevity.

This page intentionally left blank


Works of art from previous ages or from other cultures may contain or embody
ideas that we nd strange or disagree with. We take some differences in stride,
but sometimes we objectthe content we disagree with ruins our pleasure
and we take it to be grounds for judging the work negatively. In the nal ve
paragraphs of Of the Standard of Taste,1 David Hume attempts to locate
this difference. We are not or shouldnt be bothered by representations of
out-of-date fashions, he says. Where any innocent peculiarities of manners are
representedlike princesses carrying water from the spring, or ruffs and fardingales in pictures of our ancestorsthey ought certainly to be admitted; and
a man who is shocked with them, gives an evident proof of false delicacy and
renement. We are happy to overlook what we take to be factual mistakes.
Speculative errors . . . found in the polite writings of any age or country . . .
detract but little from the value of those compositions. But moral differences
are quite another matter, according to Hume. We do not, and should not,
tolerate in a work ideas of morality and decency that we nd repugnant.
[Although] I may excuse the poet, on account of the manners of his age,
says Hume, I never can relish the composition. Morally reprehensible ideas
constitute deformities in the work.
Hume has a point hereactually more than one. Thats the trouble. Our rst
task will be to disentangle them. I will begin with the simpler and more obvious

1. David Hume, Of the Standard of Taste, in Essays Moral, Political and Literary
(Indianapolis: Liberty Classics, 1987), pp. 245249.




strands and work toward the messier and more interesting ones. Some of the
strands have clear afnities with the objections to painting and poetry that Plato
expressed in the Republic, and have been much discussed since then; others are
quite different from these. Questions will arise, as we sort things out, about
what exactly Hume had in mind. Often there will be no clear answer. But there
is a varied landscape richly deserving of exploration, in the general direction in
which he gestured.

If someone advocates a moral position we nd reprehensible or tries to get us to
feel or to act in a way that violates our moral convictions, naturally we object.
We refuse to think or feel or act in the way we are asked to, and we are likely
to respond to the assertion or request or demand with disgust. The assertion or
request or demand may come in an ordinary statement or a lecture or sermon or
newspaper editorial. But people also make reprehensible claims or demands by
writing poems, by telling stories, by creating ctions.2 Hume says that where
vicious manners are described, without being marked with the proper characters
of blame and disapprobation; this must be allowed to disgure the poem, and to
be a real deformity. His thought is probably that such a work in effect condones
the vicious manners, that it condones behaving viciously in real life. If a story has
as its moral or message the idea that the practice of genocide or slavery is morally
acceptable, or that it is evil to associate with people of other races, of course we
object, just as we would to a newspaper editorial that advocates genocide or slavery or condemns interracial friendships. Works of either kind will arouse disgust,
and we will judge them negatively.
What kind of defect in the work is this? A moral one, obviously. But not, some
would say, an aesthetic one. Hume doesnt speak specically of aesthetic value.
But he appears to have in mind values that are not themselves narrowly speaking
moral, which the presence of morally repugnant ideas in a work may undermine.
Morally repugnant ideas may so distract or upset us that we are unable to appreciate whatever aesthetic value the work possesses. Disgust with the celebration
of the Nazi Party and its values in Leni Riefenstahls Triumph of the Will may
prevent us from appreciating or even noticing the lms cinematic beauty. But
maybe the beauty is there nonetheless; maybe the works moral failings merely
interfere with the enjoyment of its beauty. (They might outweigh its aesthetic
value, if the two kinds of value are commensurable.) If so, we should consider it
unfortunate that we are psychologically unable to bracket our moral concerns in
order to appreciate the work aesthetically. Given that the work exists and has the

2. Hume mentions poetry specically in these paragraphs, but his essay concerns
works of other sorts as well, especially other works of literary ction.



moral deformities and aesthetic merits that it has, it is too bad that awareness of
the former interferes with enjoyment of the latter.
In many instances we do not take this attitude, however. Rather than regretting our inability to appreciate the work aesthetically, we may feel that we
dont want to; we may be unwilling even to try to look beyond our moral concerns in order to enjoy the works beauty, as though the beauty itself is tainted.
Perhaps our thought, sometimes, is that we dont want to prot (aesthetically)
from moral depravity. (The realization that the pyramids were built by slave
labor might ruin ones enjoyment of them.) This thought will make more or
less sense depending on the extent to which we think the depravity contributes to our potential aesthetic enjoyment. If a works beauty lies in the elegant manner in which it expresses certain thoughts, the thoughts provide the
opportunity for the elegance, and to enjoy the beauty will be to prot from the
expression of the thoughts.3 But the cinematic or formal beauty of the shots
of Hitlers airplane ying through the clouds, in Triumph of the Will, may be
entirely independent of the lms moral depravity. They would be no less beautiful if they were embedded in an unobjectionable context, and a viewer who is
somehow unaware of the lms message would have no difculty appreciating
them aesthetically.
In either case, the way still seems open to regard the work as possessing aesthetic
value. But that is something we seem sometimes to deny, precisely because of
moral failings. Compare a racist joke or a political cartoon that makes a point we
nd offensive. We may declare pointedly that it is not funnyprecisely because
its message is offensive. To laugh at it, we may feel, would amount to endorsing its message, so we refuse to laugh. Even judging it to be funny may feel like
expressing agreement. Perhaps it isnt just that our disgust with the message of
Triumph of the Will interferes with our ability to appreciate it aesthetically. To
allow ourselves to enjoy even its cinematic or formal beauty may be to endorse
or concur with its praise of Hitler and the Nazis, in this sense to enter into the
sentiments Riefenstahl is expressing. We might express our unwillingness to do
this by declaring that the lm is not beautiful.
We must not simply assume that this declaration is to be taken literally
(although I doubt that much is to be gained by deciding this question). One
might reasonably hold that the lm is beautiful and the cartoon funny, but that
admitting this, as well as allowing ourselves to enjoy the beauty or the humor,
amounts to subscribing to the works evil messageso we dont admit it. Even so,
there is a closer connection between moral and aesthetic value than some would
allow. No amount of squinting or compartmentalizing could make appreciation

3. See my How Marvelous!: Toward a Theory of Aesthetic Value, Journal of Aesthetics and Art Criticism, special issue on Philosophy and the Histories of the Arts, 51(3),
1993. [Reprinted in this volume.]



of the aesthetic value morally acceptable. If the works obnoxious message does
not destroy or lessen its aesthetic value, it nevertheless renders this value morally
inaccessible. That may be counted as an aesthetic as well as a moral defect; it is a
circumstance that is unfortunate from an aesthetic point of view.
What about the contrast that Hume insisted on between ideas concerning
morality and ideas of other kinds, in works of art? Maybe works serve less frequently as vehicles for assertions about factual matters than moral ones. To
describe vicious manners in a story without marking them with the proper
characters of blame and disapprobation is not always to condone them, of course,
but in stories of some kinds it is likely to be. Stories about fairy godmothers or
time travel, however, rarely have as their messages the claim that there actually
are fairy godmothers or that time travel is a real possibility, even if the story does
not mark such ideas as not to be believed. Perhaps readers are more in the habit
of looking for moral messages than for nonmoral ones in literature.
But ctions do sometimes serve to assert or convey information about nonmoral matters. An historical novel may be expected to get the historical events
right, at least in broad outline, and it may have as one of its objectives informing
readers about them. If it gets things wrong we may complain. And we will not
necessarily object less strenuously than we would to a work we take to be advocating a moral attitude we disagree with. The assertion of factual falsehoods is
sometimes a serious matter (sometimes for moral reasons, sometimes for reasons
that are not clearly moral). And we wont mind winking at what we take to be a
relatively trivial moral claim with which we disagree.
The assertion of factual falsehoods in a story, when it matters, may distract
us from appreciating the work aesthetically. I am less condent that appreciating
the work aesthetically or judging it to be aesthetically good will often be felt as
endorsing whatever factual claims we take it to be making.

Not all works have messages or morals (even on rather generous construals of
these notions). Many contain or embody or express, in one way or another, ideas
we may nd morally repugnant, but without going so far as asserting or advocating them. The response some works call for is more one of imagining than one of
acceptance or belief. A story might encourage or induce appreciators to imagine
taking up a certain moral perspective or subscribing to certain moral principles
without recommending that they actually do so. One obvious way to induce such
imaginings is by portraying sympathetically and with understanding a character
who accepts the perspective or principles in question. The story might at the
same time encourage readers to disagree with the character; the author may make
it clear in her story that she rejects the moral views her character subscribes to.
If we nd the perspective presented in a story offensive enough, we may object
even to imagining taking it up. We might refuse to empathize with a character



who accepts it, to put ourselves imaginatively in her shoes. We usually dont
inch at imagining accepting as true nonmoral propositions that we rmly
believe to be false: the proposition that there is a ring that makes its wearer
invisible, or that a village in Scotland appears and disappears every hundred
years. But the difference is not as large as it appears to be.
Why should we resist merely imagining subscribing to a moral perspective
we consider offensive? One familiar explanation is that such imaginings may,
subtly or otherwise, tend to encourage one actually to subscribe to it. I am sure
there is some truth to this. Suppose I am taken to a cricket match. Finding the
event disappointing as ballet, I think I would enjoy it more if I rooted for one
team or the other. But I have no reason to prefer either team. Still I want to have
a desire about the outcome. So I pick one of the teams arbitrarily, by ipping a
coin, and then set out to imagine wanting it to winpretending to myself that
it matters. At rst this isnt very satisfying and it doesnt help much to make
the match exciting. My imaginings are too deliberate and articial, and I am
too vividly aware that I have no real reason for my imagined preference and that
only a coin toss sent me in one direction rather than the other. But I follow the
same team throughout the season, and my imaginings become less deliberate and
seem more natural. Eventually, I nd myself actually wanting my chosen team to
win, and rather unaware of the fact that I have no good reason for wanting it to
(although I may admit this if asked).4
If in an ordinary case like this, imagined experiences of believing, desiring,
and feeling can, over time, lead to the real thing, one should expect that, whatever combination of beliefs, desires, and feelings, or dispositions thereto, constitute accepting certain moral principles or a certain moral perspective, imagining
accepting them can have some tendency to induce one actually to do so. So if a
story presents, even just for imaginative understanding, a moral perspective we
consider repugnant, we may rightly be wary about entering into the imagining.
We still do not have a very substantial difference between moral ideas in works
of art that we disagree with and nonmoral ones, however. Advertisers and political propagandists know that getting people to imagine believing a factual proposition can nudge them toward believing it. We wont resist much if the matter
is of little importance to us. It wont hurt me much to believe falsely that Brand
A paper towels are softer and more absorbent than Brand X (if they are in fact
comparable in quality and price). But when it does matter I do resist. I may want
not to imagine that people of one race are genetically less capable in a certain
respect than people of another. And I may object to a novel in which it is ctional
that this is so, one that asks readers to imagine this. My objection in this case
is based on moral considerations, although the proposition I avoid imagining
is not itself a moral one. In other cases my concern is prudential. I might avoid

4. David Lewis suggested to me that he had an experience something like this.



reading a historical novel I know to be inaccurate, while preparing for a history

examination, for fear it might confuse my knowledge of the historical events.

Concern about being inuenced to believe what we want not to believe does not
explain very much of the resistance we feel to imagining contrary to our beliefs.
Even when our convictions are so secure that there can be no real danger to them,
we may strenuously resist imagining them to be mistaken. Hume seems to suggest that it is when we are sure of our moral convictions that we reject works
containing contrary ideas.5 Imaginings can have undesirable and even dangerous
effects which, although cognitive in character, are not happily characterized, in
ordinary folk psychological terms, as inducing false beliefs. Here is a distinctly
nonmoral example.
I am lost in the woods and mistaken about which direction is which. A look at
my compass sets me straight. But I am still turned around; it still seems to me that
that direction is north, even though I know it is not. Lets say that I remain disoriented. In order to correct my orientation, to bring it into line with my knowledge
and belief, I actively imagine north being the direction I know it to be, I picture to
myself my house, New York, the Pacic Ocean where I know they are. Eventually
my orientation, my picture of my surroundings, turns around to match reality.
Although ones orientation is distinct from ones beliefs and can vary independently of them, it has a lot to do with the organization, salience, and accessibility
of what one believes. It is much easier for me to gure out which road leads home
when I am correctly oriented than when I am not, even while I am looking at my
compass. And if I walk without thinking when I am disoriented, my feet may
take me in the wrong direction. So it is important that my orientation, as well
as my beliefs, be correct.
Perhaps orientation is a matter of imagination, of possessing a certain imaginative picture or map of ones surroundings. In any case, explicit imaginings can
affect ones orientation; it was by imagining things as they are that I corrected my
orientation. Imagining what I know to be false can have the opposite effect. I may
avoid imagining north to be where I think east is for fear doing so might disorient
me, even if there is no danger to my knowledge of which direction is which.
We may have similar reasons to resist imagining accepting moral principles
or perspectives which we consider mistaken or wrong. Even if we are entirely
condent in our judgment and see no real possibility that any imagining will

5. Where a man is condent of the rectitude of that moral standard, by which he

judges, he is justly jealous of it, and will not pervert the sentiments of his heart for a
moment, in complaisance to any writer whatsoever (Hume, Of the Standard of Taste,
p. 247).



change our minds, we want our instincts to be in line with our convictions. That
makes it easier to decide what actions accord with our convictions, and more
likely that, when we act without thinking, we will do what we believe to be
right. Adopting even in imagination a moral view that I reject in reality, allowing myself to think and feel in imagination as though my convictions were different from what they actually are, might change my moral orientation; it might
in this sense pervert the sentiments of my heart, even if it doesnt change my
convictions. The more condent I am of my convictions, the more strenuously
I will resist anything that might pry my moral orientation away from them.
Works of art may evoke imaginings which can affect ones orientation. If they
threaten to induce an orientation that conicts with what we believe concerning some matter we take to be important, we object. (We sometimes object to
metaphors for similar reasons.)6
It is possible that this concern is especially important in the moral realm. I can
certainly engage in a lot of imagining about fairies and goblins and time travel
and magic rings without having to worry about my orientation with regard to
these matters being distorted. (I suppose the child who nds himself afraid to
walk home at night after watching a horror movie, though he knows full well
that the monsters he saw are conned to the world of the movie, suffers such a
distortion.) But the example of ones sense of direction shows that it is not only
in moral instances that concerns about orientation apply.

It has not been hard to nd explanations for appreciators objections to works
of art that contain ideas about morality they consider repugnant; the reasons
I have mentioned are neither surprising nor unfamiliar. But we have not made
much progress in validating the asymmetry that Hume insisted on between the
moral and the nonmoral content of works of ction. In Mimesis as Make-Believe,7
I suggested that such an asymmetry obtains at the level of mere representation,
that is, when it comes to ascertaining what is true-in-the-ctional-world, quite
apart from what we might take to be the works message or moral or any ambition or tendency it might have to change or reorganize our beliefs or attitudes
or behavior or instincts. My suggestion was, very briey, that when we interpret
literary and other representational works of art we are less willing to allow that
the works ctional worlds deviate from the real world in moral respects than in

6. For an account of what a perspective induced by a metaphor might consist in, see
my Metaphor and Prop Oriented Make-Believe, European Journal of Philosophy 1(1),
April 1993. See also Richard Moran, Seeing and Believing: Metaphor, Image and Force,
Critical Inquiry 16, Autumn 1989.
7. Kendall L. Walton, Mimesis as Make-Believe: On the Foundations of the Representational
Arts (Cambridge, Mass.: Harvard University Press, 1990), pp. 154155.



nonmoral ones. I associated this point with Humes remarks in the paragraphs
before us. But I have since come to think that, although some of what Hume
says can be construed as aiming in this direction, my point in Mimesis is distinct
from and independent of much of what Hume seems to be getting at. I suspect,
however, that Hume had something like this point vaguely in mind when he
constrasted objectionable moral ideas in literary works with nonmoral ones.
We go about deciding what is ctional, or true-in-a-ctional-world, in many
instances, in much the way we go about deciding what is the case in the real world.
We make similar inferences, utilizing much the same background information and
exercising similar sensitivities and intellectual abilities. We often judge characters
feelings, motivations, and personalities on the basis of what they do and say, for
instance, as though they were real people. We make use of whatever knowledge of
human nature we may think we possess, and any relevant life experiences we have
had. We sometimes put ourselves into characters shoes to understand from the
inside what they may be feeling or thinking, as we do in the case of real people.
This is what one would expect insofar as the construction of ctional worlds is
governed by what I called the Reality Principle. Crudely glossed, the Reality Principle says that we are to construe ctional worlds as being as much like the real
world as possible, consistent with what the work directly indicates about them.
We are entitled to assume that ctional characters, like real people, have blood
in their veins, that they are mortal, and so onunless the story contains explicit
indications to the contrary. On reading a story we note what it says explicitly
about characters and events, andinsofar as the Reality Principle appliesask
what would be the case in the real world if all this were true.
The Reality Principle applies much less frequently than one might have
supposed, and it is easy to underestimate the extent to which considerations special to the interpretation of works of ction or certain genres of ction, considerations without analogues in investigations of the real world, come into play when
we decide what is ctional. Some exceptions to the Reality Principle occur when
the author held beliefs about reality which we know to be mistaken. A medieval
storyteller describes a character as recovering from disease after being treated
by bloodletting, and expects listeners or readers to assume that (ctionally) the
treatment cured him. Shall we disagree, since we know bloodletting to be ineffectual? I think we may well prefer to go along, to understand the story as we
know the teller meant it to be understood. Otherwise it may lose its point. We
may allow that, in the ctional world, bloodletting cures disease (even though
the story does not directly or explicitly establish that this is so), despite our
certainty that this is not so in the real world.8
8. One might in this case prefer what I called the Mutual Belief Principle (which follows suggestions of David Lewis and Nicholas Wolterstorff). There is an enormous range
of cases in which nothing even approximating either of these principles seems to apply.
See Walton, Mimesis as Make-Believe, pp. 161169.



When it comes to moral matters (moral principles anyway), however, I am

more inclined to stick to my guns, and it seems to me that most interpreters are
also. I judge characters by the moral standards I myself use in real life. I condemn
characters who abandon their children or engage in genocide, and I dont change
my mind if I learn that the author (and the society he was writing for) considered
genocide or abandoning ones children morally acceptable, and expected readers
to think this is so in the world of the story. If the author is wrong about life, he
is wrong about the world of his story. I dont easily give up the Reality Principle,
as far as moral judgments (moral principles) are concerned.
Can an author simply stipulate in the text of a story what moral principles
apply in the ctional world, just as she species what actions characters perform?
If the text includes the sentence In killing her baby, Giselda did the right thing;
after all, it was a girl or The village elders did their duty before God by forcing
the widow onto her husbands funeral pyre, are readers obliged to accept it as
ctional that, in doing what they did, Giselda or the elders behaved in morally
proper ways? Why shouldnt storytellers be allowed to experiment explicitly
with worlds of morally different kinds, including ones even they regard as morally obnoxious? There is science ction; why not morality ction?
I am skepticalskeptical about whether ctional worlds can ever differ morally from the real world. Of course people in ctional worlds can subscribe to
moral principles we recognize as repugnant. Evil characterscharacters who by
our lights have twisted notions of moralityabound in the pages of ction. An
entire society in the world of a novel, the entire population of a planet, might
accept the practice of genocide as legitimate or condemn interracial marriage as
contrary to nature. But can it be ctional that they are right? Can we reasonably judge it to be ctional that genocide is legitimate or interracial marriage
a sin, while insisting that the real world is different? Can we accept that what
would be virtue in the real world is, in a ctional world, vice, or vice versa?9, 10
I have learned never to say never about such things. Writers of ction are a clever

9. Some may take the position that one has no right to pass judgment on the moral
principles accepted in another society, that anthropologists, for instance, should not condemn practices that accord with the moral code of the agents culture even if they conict
with the anthropologists own moral code. Extending this tolerance to ctional as well as
actual societies does not make the ctional world different morally from the real one.
10. I am using the language of moral realism here, but I do not mean to beg any questions in its favor. Antirealists may insist on reformulating the problem, but that wont
make it disappear. If there are no such things as moral propositions, it wont be ctional
either that slavery is just, or that it is unjust. But antirealists will have to explain what
look like judgments readers make about the moral qualities of the actions of ctional
characters. And they will have to make sense of the embedding of sentences expressing moral judgments in larger contexts, including In the story . . . contexts, as well as
conditionals, etc. I do have hope that some variety of antirealism will make the problem
more tractable.



and cantankerous lot who usually manage to do whatever anyone suggests cant
be done, and philosophers are quick with counterexamples. But in this instance
counterexamples are surprisingly difcult to come by.
A readers likely response on encountering in a story the words, In killing
her baby, Giselda did the right thing; after all, it was a girl, is to be appalled
by the moral depravity of the narrator.11 The sentence probably serves to express
the narrators moral sentiments, not the moral reality of the ctional world. If
it were ctional that infanticide for the purpose of sexual selection is morally
acceptable, readers would be called on to imagine that the sentiment expressed
is proper, that Giselda did indeed do the right thing. They would be barred
from imaginatively condemning either her or the narrator, although they might
be aware of the repulsion they would feel concerning such practices in the real
world. (A reader of science ction may remind herself that demonic geniuses
from outer space are not actually invading the earth and that travel in time is not
possible, while imagining otherwise.) This strikes me as a seriously inadequate
characterization of the experience a reader would be likely to have. The reader
will imaginatively condemn the narrators endorsement of infanticide, not allowing that he is right even in the ctional world in which he exists.
Some narrators are said to be omniscient. This usually means that whatever,
ctionally, they say is, ctionally, true. (It is usually not ctional that they are
omniscient.)12 Why shouldnt narrators sometimes be omniscient, in this sense,
about morality? Then from the fact that ctionally the narrator declares infanticide or ethnic cleansing to be permissible we could conclude that, ctionally, it is
permissible. In real life some people do sometimes accept another persons judgments about moralitychildren believe their parents, occasionally, the faithful
trust religious leaders, disciples follow gurus. Why shouldnt there be conventions allowing a narrator this authority in certain instances? I am happy to go
along with an omniscient narrator who informs me that there are grifns or
fairies or that someone travels in time. But I jealously guard my right to decide
questions of virtue and vice for myself, even in a ctional world. It is as though
I would be compromising my actual moral principles, should I allow that different moral principles hold in a ctional world. The moral sentiments expressed
by narrators are just that, it seems: their own personal moral sentiments. We are
free to disagree, even though it is the moral nature of the ctional world, not the
real one, that is in question.
Is there always a narrator to take the rap? If a literary ction containing a
statement in praise of ethnic cleansing has no narrator whose sentiments it
can be understood to express, will there be any alternative to understanding
11. By narrator I mean a character in the work world who, ctionally, utters the
words of the text. I have in mind what in Mimesis as Make-Believe I called reporting
narrators, as distinguished from storytelling narrators.
12. See Walton, Mimesis as Make-Believe, 9.3.



it to characterize the ctional world itself? I do not rule out the possibility
of narrator-less literary ctions, but it is not easy to nd clear instances, even
hypothetical ones. And the very fact that a text expresses a denite moral
attitude may give us reason to recognize a narrator. Words expressive of praise
or blame cry out to be attached to a (possibly ctional) personanything, it
seems, to avoid allowing them to characterize the moral nature of a ctional
A better place to look for narrator-less ctions is in pictorial representations.
Pictures do not generally present someones (ctional) report about events or
states of affairs; they portray the events or states of affairs themselves. The
spectator, typically, imagines perceiving the events or states of affairs for herself, not being told about them (or even shown them) by someone. (There are
exceptions, of course.) But how can a picture portray moral facts, the obtaining of certain moral principles, explicitly or directly? These arent the sorts
of states of affairs one perceives. A picture may depict a mixed race couple
walking arm in arm, or a slave master beating a slave. But then it is up to us,
the spectators, to decide on the moral attributes of these actions. I go by my
own moral sense, the one I use in real life. I take it to be ctional that there
is nothing wrong with the interracial friendship, and that the beating of the
slave is abhorrent.
Suppose the picture of the interracial couple is titled Shame! or Sin! Here,
nally, we have words in a work which probably are not to be attributed to a
(reporting) narrator. The words of the title are not themselves part of the ctional
world; it probably isnt ctional that anyone is using them to characterize the
behavior of the couple. But there is a tradition of allowing titles to contribute to
what is ctional in the world of a picture. Paul Klees Singer of Comic Opera (1923)
depicts a woman, but the image itself doesnt establish that she is a singer, let
alone a singer of comic opera. Only the title makes this ctional. Does the title of
the picture of the interracial couple establish that it is ctional that the couples
behavior is shameful or sinful? I doubt it. Maybe the artist, in giving the picture
its title, intended or expected this to be ctional.13 Even so, I will insist that it
is not, that ctionally there is nothing shameful or sinful in what the couple is
doing. The title amounts to an interpretation of the picture which we are free to
disagree with, not an authoritative pronouncement establishing a feature of the
ctional world. The disgusting sentiment expressed in the title can be attributed
to the artist who chose it, or possibly to an implied or apparent or ctional artist
(a storytelling narrator), rather than taking it to establish the moral reality of the
ctional world.
13. This may be clear even if there is no title. Activities may be depicted in a gloried
manner indicating the artists approval, her belief that it is ctional that they are admirable, and her approval of similar behavior in the real world. (Compare social realistic
styles of depiction.)



If ctional worlds ever differ morally from the real world, I suspect that this will
be so when the moral character of the ctional world is presented implicitly or
indirectly rather than by explicit stipulation, and when it is part of the background rather than the focus of the work.
I appreciate and value many works that in some way presuppose or are based on
moral perspectives I dont entirely share. I think all of us do; otherwise there would
be little for us to appreciate. Unlike Triumph of the Will, whose obvious main purpose is to further an obnoxious moral and political agenda and cannot but inspire
disgust, some works merely presuppose or take for granted certain moral perspectives without in any way advocating them, or even addressing or intending to raise
the question of their propriety. These moral perspectives then serve as a resource,
as part of the setting in which the author pursues other, more specically aesthetic
objectives. If we disagree with the perspective, we might consider reliance on it to
be a defect in the work, even an aesthetic defect, but this doesnt always prevent us
from recognizing and appreciating the aesthetic qualities that result.14
I may understand a ctional event to be tragic, or ironic, or absurd, or poignant.
I may think of a character as noble, or as ridiculous. The ending of a story may
strike me as a happy one,15 or as one of unmitigated tragedy, or as uncomfortably
ambiguous, or as constituting a tting denouement to the events that preceded
it. I may think that a character does, or does not, in the end, get her comeuppance. Such aesthetically important perceptions are inevitably linked to certain
values, often certain moral principles or perspectives; it is in light of a particular
moral attitude that an event strikes me as tragic, or a character ridiculous, or an
ending tting.
The nature of the link is hard to pin down. Does it have to be ctional that
the relevant moral principles are true in order for it to be ctional that certain
events are tragic or ironic? Does appreciating the tragedy or irony commit us to
recognizing the ctionality of those principles? If so, when we disagree with the
principles we may have to judge that the ctional world differs morally from the
real one. But there are other possibilities. The tragic or ironic nature of ctional
events might derive from the fact that ctionally some or all of the characters
(perhaps including the narrator) accept moral principles with which we disagree,
without its being ctional that they are true. Appreciation might require respect
or sympathy for the characters moral attitudes. It might even require that we
imagine agreeing with them, that we imagine sharing these attitudes ourselves
without requiring us to judge it to be ctional that they are true. Perhaps we
neednt even take it to be ctional that the events are tragic or ironic; it may be

14. I am indebted here to David Hills.

15. This doesnt mean simply that the characters end up happy. An unhappy villain
doesnt prevent the story from ending happily.



enough to realize that the author (or storytelling narrator) meant them to be so
taken, and to respect or sympathize with him.
These are subtle and difcult questions which call for careful critical attention to examples of many different kinds. But we have a mystery on our hands
in any case. Whether or not ctional worlds can ever differ morally from the real
world, it seems clear that they dont as easily or as often as one might expect. We
recognize the ctionality of ordinary empirical propositions and even propositions stating scientic laws, which we consider false, far more readily than we
do that of moral principles which we reject. Authors just do not have the same
freedom to manipulate moral characteristics of their ctional worlds that they
have to manipulate other aspects of them. Why is this? The reader will not nd
a denitive answer in this essay. But progress can be made by ruling out some
kinds of explanations which might initially seem plausible, and we will come to
understand the puzzle better in the process.

Propositions that are true-in-the-world-of-a-story, ones I call ctional, are (in a
nutshell) propositions readers of the story are to imagine.16 We may nd it distasteful, morally objectionable, to imagine that interracial friendships are sinful
or that slavery is morally acceptable. I noted our resistance to imagining accepting moral principles we disagree with or disapprove of. Surely we would resist
imagining those moral principles themselves, imagining them to be true. So we
are unwilling to imagine what we are called upon to imagine, if it is ctional
that interracial friendships are sinful or slavery acceptable.
This doesnt help. It does not explain why anyone should resist allowing that
these propositions are ctional. To recognize it to be ctional in a story that
slavery is morally acceptable would be merely to recognize that the story calls
for imagining this. We dont have to go ahead and actually do the imagining.
We might decide not to go along with the story, or not even to read it, precisely
because it does ask us to imagine that slavery is acceptable, because it makes this
ctional. A person who objects to imagining that the Holocaust was a hoax, or
that Abraham Lincoln was secretly a slave trader, may be unable or unwilling
to appreciate a story in which this is so. But this wont prevent her from recognizing that it is ctional in the story that the Holocaust didnt occur or that
Lincoln traded in slaves. We might as well suppose that one cannot allow that a
newspaper editorial advocates ethnic cleansing if one nds the practice of ethnic
cleansing disgusting. It is not clear that moral objections to imagining moral
principles we nd repugnant have anything to do with the resistance I think
most of us feel to recognizing such principles to be ctional.

16. Walton, Mimesis as Make-Believe, 1.5.



Is this resistance essentially moral in character at all? Do we object morally to
recognizing it to be ctional that slavery is morally acceptable? The resistance is
of a piece, it seems to me, with an unwillingness to recognize the ctionality of
certain propositions about matters we dont feel strongly about, including ones
that do not involve morality.
Consider a really dumb joke, like this one: Knock, Knock. Whos there? Robin.
Robin who? Robbin you! Stick em up!17 It is not easy to see how it could be ctional that this joke is hilariously funny (in circumstances just like ones in which, in
the real world, it would be dumb), how one could reasonably allow it to be hilarious
in a ctional world, while thinking that it is actually dumb. The same goes for a
nonjoke such as A maple leaf fell from a tree (said in no special context). This isnt
funny in the real world, and it is not clear how one could create a ctional world in
which it is funny (without supplying a special context which would make it funny
in the real world as well). If in a story a comedian tells one or the other of these
jokes and the author simply writes explicitly in the text that it is hilariously funny,
I expect that I would attribute a juvenile or an incomprehensible sense of humor
to the narrator, and stick with my own judgment that the joke is not funny. I insist
on applying my own sense of humor, the one I use in the real world, to the ctional
world, as I do my own standards of morality. It may be ctional that the comedians
audience and other characters in the ction are amused, of course; they may be rolling in the aisles. I can admit that it is funny for them while judging that their reaction is inappropriate. I dont rule out the possibility of fancy counterexamples, cases
in which there are special reasons for allowing ctional worlds to differ from the real
one with respect to what makes for humor, but the fact that the counterexamples
would have to be fancy needs explaining.
Whether either the dumb joke or the nonjoke is funny is hardly a question that
arouses the passions or that we much care about, and it neednt have anything
much to do with morality (although some jokes do). It is not passion, moral
passion or any other kind, that drives my reluctance to let it be ctional that it
is funny. I have no moral objection to recognizing this to be ctional. What is
crucial, I believe, is that being funny or not funny supervenes or depends in a
certain way on the natural characteristics determine what is funny and what
is not. I suspect that it is particular relations of dependence, which properties
determine in the relevant manner which others, that cannot easily be different in
ctional worlds and in the real one. Why this is so, and what kind of determination or dependence is involved, is still a mystery.
I invite readers to experiment with their intuitions about various other examples.
Can different aesthetic principles obtain in ctional worlds as compared to the
real one? Can what counts in the real world as a jagged or angular or awkward
17. Thanks to Jenefer Robinson.



line be owing or graceful in a ctional world (when relevant aspects of background and context are the same)? Can what in the real world makes for elegance
or profundity or unity or bombast or delicacy be different in a ctional world?
Those who take the mental to supervene on the physical may consider whether
one might judge it to be ctional that a given mental state supervenes on certain
physical ones, if one does not think it actually does.
Moral properties depend or supervene on natural ones and, I believe, in the
relevant manner (whatever that is); being evil rests on, for instance, the actions
constituting the practices of slavery and genocide. This, I suggest, is what
accounts (somehow) for the resistance to allowing it to be ctional that slavery
and genocide are not evil.
If I am right about this, the present point is very different from those I discussed earlier. We may judge a work to be morally defective if it advocates moral
principles we nd repugnant, or if it invites or has a tendency to induce us to
imagine accepting them. (This moral failing might constitute or contribute to
an aesthetic one.) If a novel endorses slavery or encourages even imaginative
acceptance of it we will loathe it with something of the loathing we have for the
institution of slavery. The more we abhor moral principles which a work promotes, the more objectionable we nd it.
Refusing to understand it to be ctional that slavery is morally acceptable is
not in itself to nd the work defective. But if the author meant this to be ctional,
her failure to make it so may be responsible for failings in the work. The very fact
that an author tries to do something she cant bring off, if the attempt is evident
in the work, can be disturbing or disconcerting to the appreciator. And insofar as
other objectives the author meant to accomplish in the work depend on its being
ctional that slavery is legitimate, she will have been unsuccessful in accomplishing them. We may be unable to regard the hero of the story as heroic or his
downfall tragic if, contrary to the authors intentions, we judge him to be morally
despicable.18 This may not only destroy the storys excitement and dull our interest in it, it may also ruin the storys formal properties, the shape of the plot.
These are not moral defects in the work, however, but aesthetic ones, and we
dont loathe it for failing to make it ctional that slavery is legitimate, with the
loathing we direct toward slavery. Indeed, this failure is if anything a point in
the works favor, from a moral perspective. (But we may condemn the author
for attempting to make this ctional in the work.) Our negative feelings about
slavery do play an indirect role in the recognition of these aesthetic failings; it is
because we nd slavery repugnant that we judge it to be evil, that we recognize
being evil to supervene on the practice of slavery. And that, I am suggesting, is
why we disallow its being ctional that slavery is not evil.
18. We are not interested in the fortunes and sentiments of such rough heroes: . . .
And . . . we cannot prevail on ourselves to . . . bear an affection to characters, which we
plainly discover to be blameable. Hume, Of the Standard of Taste, p. 246.



Where do we stand in the attempt to nd something special about our reaction

to moral ideas that we disagree with in works of art? Our reluctance to allow
moral principles we disagree with to be ctional is just an instance of a more
general point concerning dependence relations of a certain kind. But it does
distinguish moral principles from propositions about ordinary empirical matters
of fact and also from scientic laws, which (usually) do not state dependence relations of the relevant kind.

We still need an explanation of why we should resist allowing ctional worlds
to differ from the real world with respect to the relevant kind of dependence
relations. My best suspicion, at the moment, is that it has something to do with
an inability to imagine these relations being different from how we think they
are, perhaps an inability to understand fully what it would be like for them to
be different.
This seems, initially, a most unpromising proposal. Some say that contradictions, logical or conceptual impossibilities, are unimaginable. Imaginability is
supposed to be a test for possibility. But the propositions that slavery is just, and
that the two jokes mentioned earlier are hilariously funny, are surely not contradictions. Moreover, even contradictions can apparently be ctional, although it
takes some doing to make them so. The time travel portrayed in some science
ction stories is contradictory; there are pictorial contradictions in William
Hogarths False Perspective, in etchings of M. C. Escher, and in an assortment of
familiar puzzle pictures.
How can contradictions be ctional? Sometimes a work makes it ctional that
p (prescribes the imagining of p), and also makes it ctional that not-p. Then the
conjunction, p and not-p, may be ctional by virtue of the ctionality of its conjuncts.19 It is not clear that a similar strategy will work for the proposition that
the institution of slavery is just and proper, that this can be separated into distinct
components, each of which can unproblematically be made ctional. It might be
ctional that a persons behavior on a given occasion was morally acceptable,
and also that her behavior on that occasion consisted in beating a slave (just as it
might be ctional that a person was simultaneously living in twentieth-century
Chicago and in sixteenth-century Italy). But this doesnt make it ctional that
she was behaving morally by virtue of the fact that her behavior consisted in beating a slave. It still may be difcult or impossible for that to be ctional, because
it is difcult or impossible to imagine its being true.
19. There may then be a prescription to imagine the conjuction, even if that cant be
done. Some might prefer not to regard the conjunction as ctional at all, but the ctional
world will still be contradictory in the sense that the conjunction of what is ctional is
a contradiction.



Do contradictions or obvious conceptual impossibilities get to be ctional in

other ways? If a work portrays Philip II of Spain and the Guises as a three-headed
monster, or fascism as an octopus, it would not seem that the ctionality of
these impossibilities derives from the ctionality of their components. But are
these conceptual impossibilities ctional at all; are we to imagine that Philip and
the Guises are (literally) a three-headed monster, or that fascism is an octopus?
Perhaps what is ctional is merely that there is a three-headed monster, or an
octopus, and in making this ctional the work expresses a thought about Philip
and the Guises, or fascisma thought one would express in uttering the obvious
Is it difcult or impossible, for those of us who abhor slavery and genocide,
to imagine engaging in these activities to be morally proper? We are capable
of imagining accepting or subscribing to moral principles that in fact we reject, it
seems. And we can imagine experiencing the feelingsfeelings of disgust, or
approvalthat go with judging in ways we think mistaken. Most of us remember holding moral views we have since come to renounce. We know what it is
like to subscribe to them, and we can still imagine doing so. A person who has
undergone a conversion from one moral perspective to another may not want to
put herself in her previous shoes; she may nd it painful even to imagine thinking and feeling in the ways she previously did. She may be unable to bring herself
to imagine this; it may require a great effort in this sense, just as sticking pins
into a photograph of a loved one does. But certainly she could imagine this if she
wanted to; otherwise why would she dread doing so? Sometimes we are able to
understand and empathize with people who hold moral views we have never held
or even been seriously tempted by, and this empathy is likely to involve imagining subscribing to these moral views ourselves. An important function of literary
works is to facilitate such empathy by presenting characters with various moral
perspectives in a sympathetic light.
But there are limits to our imaginative abilities. It is not clear that I can, in a
full-blooded manner, imagine accepting just any moral principle I am capable of
articulating. I cant very well imagine subscribing to the principle that nutmeg
is the summum bonum and that ones highest obligation is to maximize the
quantity of nutmeg in the universe. (Some will put this by saying that I dont
know what it would be like to hold this moral view.) I can entertain the supposition
that I accept this principle, as one would in thinking about conditional propositions or in using reductio ad adsurdum arguments. But I have argued that
ctionality involves a more substantial sense of imagining than this.20 I have no
difculty imagining nding the Knock Knock joke related earlier funny. It is
the sort of joke I once appreciated, and I know and empathize with people now
who would appreciate it. But I have trouble with the nonjoke about the maple

20. Walton, Mimesis as Make-Believe, pp. 1921.



leaf. Perhaps with effort and ingenuity I could dream up a way of thinking about
it in which it would strike one as funny. But there is a sense in which I cant now
imagine nding it funny. People who do laugh at it would mystify me in a way
that people who laugh at the Knock Knock joke do not.
I know what it is to be amused. Cant I just put that notion together in imagination with the idea of the story about the maple leaf, and imagine being amused
by the story? I am suggesting that full-blooded imagining of this may require
not just conjoining these two thoughts but imagining a way in which the story
amuses me. (Compare: a person may be incapable of imagining an instance of
justied true belief which is not an instance of knowledgeuntil having read
the Gettier literature he learns how this can be so, how to imagine it. And he
might know, on authority, that this is possible and still not be able to imagine
it. A contemporary of Columbus may be unable to imagine traveling west and
arriving in the east, until she thinks of the possibility that the earth is round.)
We are still very far from the explanation we are after. For it is not only those
propositions concerning morality or humor I have difculty imagining accepting,
that I am reluctant to recognize as ctional. I resist allowing it to be ctional
that the Knock Knock joke is funny, or that moral principles I can, apparently,
imagine accepting are true.
But can I imagine not only accepting or believing a moral principle which
I actually disagree with and feeling appropriatelycan I imagine being justied
in accepting or believing it? Can I imagine its being true?21 A work in which it
is ctional that genocide is morally permissible would be one that calls for imagining that genocide is morally permissible, not just imagining accepting this
to be so. I nd myself strangely tempted by the thought that although I might
imagine the latter, I cannot imagine the former.22
Alternatively, we might reconsider the idea that I can imagine believing, accepting as true, moral propositions I now reject. Maybe the attitude I imagine having,
when I remember my earlier moral self or empathize with others, falls short of
belief or acceptance. A sensitive portrayal of the Maa or of colonial plantation
owners might enable me to imagine desiring and feeling in many respects as they
do. And I can imagine being amused by the Knock Knock joke. (This already
distinguishes it from the maple leaf story.) But (rst-order) desires and feelings
dont constitute moral commitments, and being amused does not itself amount
to understanding the joke to be funny. On some accounts one needs to take a certain attitude toward ones desires or feelings or amusement, to endorse or desire

21. Again, I am not committed to the propriety of this realist formulation.

22. Richard Moran raised this possibility in Art, Imagination, and Resistance,
a talk he presented at the meetings of the American Society for Aesthetics in 1992.
Maybe it isnt quite as strange as it seems. It is arguable that I can imagine believing
that Ortcutt is not identical with Ortcutt, or that water is not H2O, but that, knowing
what I know, I cant imagine either of these propositions being true.



them or regard them as proper or appropriate.23 Perhaps one must also take an
attitude of endorsement toward the second-order attitudes, or at least not take a
negative attitude toward them. At some point in the series one may nd oneself
able to imagine refusing to endorse an attitude but unable to imagine endorsing
it; maybe this happens when I in fact reject the moral principles in question or
consider the joke not to be funny. This inability may be akin to my inability to
imagine being amused by the tale of the maple leaf. And perhaps it amounts to
an inability to imagine accepting a moral position that I actually reject.
There are loose ends in this sketchy story, and insecure links. I dont know
whether it can be made to work. And even if it were to succeed in establishing
that people are, always or sometimes, unable to imagine, in a signicant sense,
accepting moral positions they reject, it may not be obvious how this explains
ouror anyway myreluctance to allow moral principles I disagree with to be
ctional. The line of thought I have just outlined is worth pursuing, I believe,
but I wont be too surprised if we nd ourselves back at square one.
Hume had no idea how many worms lived in the can he opened. I have left
most of them dangling, but at least I have begun to count them. That, I hope,
is progress.24
23. See for instance Harry Frankfurt, Freedom of the Will and the Concept of a
Person, Journal of Philosophy 68(1), January 14 1971; Allan Gibbard, Wise Choices, Apt
Feelings (Cambridge: Harvard University Press, 1990); and David Lewis, Dispositional
Theories of Value, Proceedings of the Aristotelian Society, supplementary volume 63, 1989.
24. I am grateful for conversations with Allan Gibbard, Daniel Jacobson, Eileen John,
Richard Moran, Peter Railton, Gideon Rosen, Alicyn Warren, and especially David
Hills. Richard Morans Art, Imagination, and Resistance, on which I commented, was
also very helpful, in addition to renewing my interest in this topic.
[In response to helpful conversations with Daniel Jacobson, I have made a couple of
clarifying corrections in the text of this paper. Jacobsons In Praise of Immoral Art
[Philosophical Topics 25(1), Spring 1997, David Hills, editor] explores perceptively, and
more thoroughly than I do, the cluster of issues concerning relations between art and
morality that occupy sections 14 of the present essay. See also the discussions by Nol
Carroll, Berys Gaut, and Matthew Kieran that Jacobson cites. An important recent examination of the question which is the main focus of the present essay, whether ctional
worlds can differ morally, that is, with respect to what moral principles obtain, from the
real world, is Tamar Szab Gendlers The Puzzle of Imaginative Resistance, Journal of
Philosophy 2 1997.]

This page intentionally left blank

( SO-C A L L ED )

n Mimesis as Make-Believe, I happened upon a surprising peculiarity in

interpretive practice, a curious reluctance to allow ctional worlds to differ in
fundamental moral respects from the real world as we understand it.1 It seemed
to me that this might have been what David Hume was getting at in the nal
ve paragraphs of On the Standard of Taste, although this attribution now
strikes me as highly questionable. Revisiting the topic in a later essay, Morals
in Fiction and Fictional Morality,2 I emphasized that there is actually a tangled
nest of importantly distinct but easily confused puzzles in the vicinity, several
of which can be traced uncertainly to Humes observations.3 My strategy in
addressing them was disentangle-and-conquer. I do not claim to have successfully
completed the conquest in my previous forays, nor will I do so now. But I did do

This essay began as a postscript to Morals in Fiction and Fictional Morality (see note
2 below), but got out of hand. Thanks to Tamar Gendler and Shaun Nichols for very
helpful comments.
1. Kendall L. Walton, Mimesis as Make-Believe: On the Foundations of the Representational
Arts (Cambridge, Mass.: Harvard University Press, 1990), pp. 154155. In the terms of
my discussion in Mimesis, the peculiarity consisted in a strange insistence on the Reality Principle of implication, for deciding whether moral propositions of certain sorts are
2. Kendall L. Walton, Morals in Fiction and Fictional Morality, Proceedings of the
Aristotelian Society, Supplementary Volume 68 (1994): 2750. [Reprinted as chapter 3 of
this volume; page references are to this volume.]
3. Tamar Gendler examines Humes comments in some detail. Gendler, Imaginative Resistance Revisited, in The Architecture of the Imagination: New Essays on
Pretense, Possibility, and Fiction, ed. Shaun Nichols (Oxford: Oxford University Press),
pp. 149173.




some untangling, and without that there is no hope of conquest. I now see that
there are even more strands to separate than I recognized previously. So more
untangling is in order.
Several of the puzzles, and amalgamations of them, travel together in the recent
literature under the name the puzzle of imaginative resistance.4 This unfortunate
appellation ignores the puzzles multiplicity, and is misleading in other respects
as well. There are perceptive and illuminating discussions in the literature nonetheless, with some promising suggestions about solutions. One contributor who
is careful to distinguish the main strands of the tangle is Brian Weatherson.
Weatherson recognizes four related puzzles, and examines three of them.5 The
differences between the three are small, he saysI disagree about thatbut he
separates them clearly and gives them useful labels. I will look at two of these
threethe imaginative puzzle and what he calls the alethic one, which I will
rename the ctionality puzzle. But I will begin with the one he doesnt examine,
the aesthetic puzzle.


The rst several sections of Morals in Fiction survey, briey, the neighborhood
of the aesthetic puzzle.6 If a work of art is objectionable on moral grounds, does
this diminish or destroy its aesthetic value?
4. Of those who use this name, Gregory Currie, Ian Ravenscroft, and Shaun Nichols
address only the imaginative puzzle; Stephen Yablo focuses on the ctionality one; Tamar
Gendler, Derek Matravers, and Richard Moran have some of both in mind. See Currie,
Desire in Imagination, in Conceivability and Possibility, ed. Tamar Gendler and John
Hawthorne (Oxford: Oxford University Press, 2002), pp. 201222; Currie and Ravenscroft, Recreative Minds: Imagination in Philosophy and Psychology (Oxford: Oxford University Press, 2002); Nichols, Just the Imagination: Why Imagining Doesnt Behave Like
Believing, Mind and Language 21, no. 4 (2006): 459474; Yablo, Coulda, Woulda,
Shoulda, in Conceivability and Possibility, pp. 441492; Gendler, The Puzzle of Imaginative Resistance, Journal of Philosophy 2 (2000): 5581; Matravers, Fictional Assent and
the (so-called) Puzzle of Imaginative Resistance, in Imagination, Philosophy, and the Arts,
ed. Matthew Kieran and Dominic Lopes (London: Routledge, 2003), pp. 91106; and
Moran, The Expression of Feeling in Imagination, Philosophical Review 103, no. 1 (1994):
75106. In his response to my Morals in Fiction, Michael Tanner touches only on the
aesthetic puzzle; see Morals in Fiction and Fictional Morality, Proceedings of the Aristotelian Society, Supplementary Volume 68 (1994): 5166.
5. Brian Weatherson, Morality, Fiction and Possibility, Philosophers Imprint 4, no. 3
(2004): 127.
6. This issue has enjoyed a urry of discussion recently, although it is certainly not
new. Cf. Nol Carroll, Moderate Moralism, British Journal of Aeshtetics 36 (1996): 223
237; Mary Devereaux, Beauty and Evil: The Case of Leni Riefenstahls Triumph of the
Will, in Aesthetics and Ethics: Essays at the Intersection, ed. Jerrold Levinson (Cambridge:
Cambridge University Press, 1998), pp. 227256; Berys Gaut, The Ethical Criticism
of Art, in Aesthetics and Ethics, 182203; Daniel Jacobson, In Praise of Immoral Art,





I did not say nearly enough about the various kinds of moral objections one might
have to a work (or a joke, or a cartoon, or a metaphor), and much of the literature
on this topic is lacking in this respect as well. Two broad categories are obvious: (a)
Some works are vehicles by virtue of which the artist expresses sentiments or advocates a moral point of view that we may nd objectionable. People sometimes make
reprehensible claims or demands by writing poems, by telling stories, by creating
ctions, I observed.7 We may condemn the sentiments or point of view and their
expression whether or not we think the artist has any chance of making converts or
persuading anyone. (b) Alternatively, we may worry, as Plato did, that a work will
have morally deleterious effects, whether or not they were intended or envisaged by
the artist. We may criticize a work for encouraging immoral attitudes or behavior
or unwanted feelings in audiences; one might even fear succumbing oneself. And we
may complain about a works likely indirect consequences; its sales might line the
pockets of a distributor who will bankroll evil causes, for instance.
Other moral objections are of neither of the above kinds. One who objects to the
Egyptian pyramids because of their construction by slave labor, or to European high
art on the grounds that it was made possible by an obscene concentration of wealth
among the royalty or the clergy, neednt presume either that their creators were
advocating morally obnoxious views or that the works might have unfortunate consequences. According to H. L. Mencken, the last movement of Beethovens Eroica
is not only voluptuous to the last degree; it is also Bolshevistic. Try to play it with
your eyes on a portrait of Dr. Coolidge. You will nd the thing as impossible as eating ice-cream on roast beef.8 Mencken need not have supposed that Beethoven was
endorsing Bolshivism, or that the Eroica nale stands any chance of promoting it.
My own intuitions about these sketchily characterized examples, as to whether
their (presumed) moral faults affect their aesthetic merit, are fuzzy, but insofar
as I do have inclinations, they go in different directions. I am less inclined to
think that morally undesirable consequences, especially relatively indirect ones,
detract from a works aesthetic value than that serving as a vehicle whereby the
artist advances morally obnoxious claims (at least if this is evident in the work)
does. Andsupposing that the Bolshevism that Mencken hears in the Eroica is
really there, and is morally objectionableI am more inclined to accept that it
lessens the works aesthetic value than that the construction of the pyramids by
slave labor does. A single answer to the question of what if any bearing moral
failings have on aesthetic value is not in the ofng.

Philosophical Topics 25, no. 1 (1997): 155199; Eileen John, Artistic Value and Opportunistic Moralism, in Contemporary Debates in the Philosophy of Art, ed. Matthew Kieran
(Oxford: Blackwell, 2005), pp. 331341; Matthew Kieran, Art, Imagination, and the
Cultivation of Morals, Journal of Aesthetics and Art Criticism 54, no. 4 (1996): 337351.
7. Walton, Morals in Fiction, p. 28.
8. H. L. Mencken, Music and Sin, in Prejudices: Fifth Series (New York: Alfred
A. Knopf, 1926), p. 295.



I did not mean to propose any such answer in Morals in Fiction, or even to
decide at all denitely about any particular case. My main purpose in discussing
the aesthetic puzzle, beyond providing an overview of the issues involved, was
to clarify its relation to, and distinctness from, the others. What I called an aesthetic defect, in connection with Triumph of the Will, is simply a circumstance
that is unfortunate from an aesthetic point of view, namely the fact that the
lms moral reprehensibility is likely to prevent people from appreciating it aesthetically.9 It remains an open question whether or not the lm possesses aesthetic
merit, whether its moral faults destroy or lessen its aesthetic value, or merely
render its aesthetic value unavailable or inaccessible.10 This question is not itself,
I think, a very interesting or important one. Likewise for the question whether
a racist joke which, lets suppose, ought not to be laughed at, is in fact funny, or
whether the complaint That isnt funny! should be taken as the literal truth.
What is important and interesting is the fact that moral failings in works of
art do sometimes (not always) impede aesthetic appreciation, and the various
ways in which they do. All of the varieties of moral defects I mentioned may do
this, although the interference takes different forms in different cases. We may
be unwilling to appreciate a work, or even experience it, because we think that
doing so would itself be morally objectionable: in appreciating it we would be
proting from slave labor, or letting ourselves in for temptation, or (in effect)
declaring allegiance to or openness to an obnoxious moral perspective, or contributing to the works undesirable indirect consequences. Sometimes a works
negative moral qualities may be so overwhelming or painful or guilt inducing or
distracting that we are simply unable to appreciate it even if we are willing. It is
also important to recognize thatas Plato famously observedimmoral works
of many varieties often are exceedingly powerful despitemaybe even because
oftheir immorality; they may mesmerize or move us, even against our will.
Of course, if authorities censor them or if we refuse even to experience them,
their moral aws will have prevented their being appreciated.
How and why and when moral failings have inhibiting effects, as well as when,
why, and how they are effective nevertheless, are rich areas for empirical investigation. I barely touched on them in Morals in Fiction. And I didnt even
approach the important normative questions about whether, in various instances,
one ought to avoid enjoying a work, whether, for instance, it is wrong to laugh at
a racist joke or to marvel at the pyramids. None of these questions requires that
we decide whether moral defects lessen a works actual aesthetic merit.
When moral considerations prevent us from appreciating a work of ction,
this is often, though not always, because we are unable or unwilling to imagine
9. Walton, Morals in Fiction, p. 30.
10. Thanks to Daniel Jacobson for insisting that I clarify this point. Cf. his In Praise
of Immoral Art. He also makes a good case for the idea that moral aws are sometimes
aesthetic merits.





in the way the work calls for. This is the link between the aesthetic puzzle and
the imaginative one. But insufcient attention to the distinction between them
may make it seem that the imaginative puzzle concerns only matters having to
do with morality. It doesnt. And neither does the ctionality puzzle.11
It is worth pointing out, also, that the aesthetic puzzle itself has nonmoral analogues. I borrow an example from Frank Sibley: An apparently abstract photograph
which strikes us as beautiful may be impossible to appreciate once we learn that it
is a photograph of lesions on a human body, or gangrene, or ulcers. Or we may not
want to enjoy it then, even if we can. It is less likely that we will think we ought
not let ourselves appreciate it, that doing so would be morally objectionable. Sibley
suggests that it may true to say that such a photograph is beautiful, nevertheless,
that it possesses one kind of beauty anyway (predicative beauty, as opposed to
beauty as a photograph of gangrene, i.e., attributive gangrene-beauty).12


The most easily confused of the several tangled strands are the ctionality and the
imaginative puzzles. The ctionality one, the focus of my observations in Mimesis,
is the most perplexing of the bunch. We easily accept that princes become frogs
or that people travel in time, in the world of a story, even, sometimes, that blatant
contradictions are ctional. But we balkI do anyway, in some instances, and it is
evident from the literature that I am not aloneat interpretations of stories or other
ctions on which it is ctional that (absent extraordinary circumstances) female
infanticide is right and proper, or that nutmeg is the summum bonum, or that a
dumb Knock Knock joke is actually hilarious. Why the difference? This is the
ctionality puzzle. The imaginative puzzle concerns not what is or isnt ctional,
but what we do or do not imagine. These are different; I may recognize that something is ctional, true in the world of a story, without actually imagining it, or
imagine something that I take not to be ctional.13 People are sometimes unwilling
or unable to engage in certain imaginings. Why? This is the imaginative puzzle.
11. Section 8 of Walton, Morals in Fiction aims to establish that the ctionality
puzzle extends to nonmoral propositions. My concentration on moral matters earlier in
the essay may have, misleadingly, suggested the contrary, however. Weatherson (Morality, Fiction and Possibility) and Yablo (Coulda, Woulda, Shoulda) have much more to
say about the scope of the ctionality puzzle.
12. Frank Sibley, Aesthetic Judgments: Pebbles, Faces, and Fields of Litter, in
Approach to Aesthetics, ed. John Benson, Betty Redfern, and Jeremy Roxbee Cox (Oxford:
Oxford University Press, 2001), pp. 176190.
13. Walton, Morals in Fiction, p. 39. Imagining a proposition is more than merely
recognizing or entertaining or understanding or formulating it or supposing it to be so.
Cf. Walton, Mimesis as Make-Believe, pp. 1921; also Gendler, Imaginative Resistance
Revisited, in Nichols, Architecture of the Imagination.



Tamar Gendler has contributed a rich and interesting essay treating the
ctionality and imaginative puzzles. She slides back and forth between them,
however, characterizing what needs explanation sometimes as a failure to imagine, sometimes as resistance to taking something to be ctional. In several places
she writes of resistance to making-believe, which might be construed either
way.14 She acknowledges the ambiguity but contends that conating the two
readings is legitimate (at least in some contexts) on the grounds that what is
true in a story is what the author gets the (appropriate) reader to imagine, if
(appropriate) readers are unable or unwilling to [imagine that p], they will be
unwilling or unable to [accept that p is ctional].15 This will be so if an appropriate reader is simply one who imagines what is ctional. But then a reader
may resist reading appropriately, while acknowledging that that is what he is
resisting; he may refuse to or fail to imagine what he recognizes to be ctional.
Alternatively, a reader may deny that the proposition in question is ctional,
resist the notion that to read appropriately is to imagine it. These resistances or
failures are different, and demand different explanations.
Shaun Nichols has pointed out to me that there may be an epistemological
link between imaginings and judgments of ctionality; a person who is unable
or unwilling to imagine something may be a poor judge of whether it is to be
imagined, whether it is ctional. It is largely because a work induces us to imagine something, in many instances, that we judge it to be ctional. (What I nd
myself imagining seeing when I look at a picture, for example, what I see in it,
is likely to be what it depictsassuming that I am a normal or properly qualied observer.) If I am incapable of imagining a particular proposition or dont
allow myself to imagine it, this test for ctionality will not be available to me.


To avoid begging questions, lets formulate the imaginative puzzle as neutrally as
possible: Sometimes people do not engage in imaginings that one might expect
them to. These may be imaginings of propositions that are ctional in a given
work, or were clearly meant to be ctional, or imaginings a work seems apt for
inducing, or ones that appear to be appropriate or called for or likely in situations
not involving a work of ction.
When people fail to imagine as expected, is this because they are unable to do
so, or because they are unwilling to or refuse? Both answers have been proposed

14. Gendler, Puzzle of Imaginative Resistance, pp. 6263, 66, 75, 79. But see her
more recent reections in Imaginative Resistance Revisited. I avoid using makebelieve as a verb in my own writings because of its ambiguity. Other expressions in the
literature which are susceptible to a similar ambiguity include what we accept and imagine as ctionally true, ctionally assenting to, and imagining a ctional world.
15. Gendler, Puzzle of Imaginative Resistance, p. 58n6.





in the literature on imaginative resistance, and there are instances of both

kinds. Sometimes we refuse, for any of a variety of reasons. And some imaginings
are difcult or impossible to bring off, the difculty or impossibility having different explanations in different cases. The imaginative puzzleitself only part of
the tangled nest we are addressingdivides into many.
I mentioned some reasons people may have for declining to engage in an imagining; there are others as well, not all of them moral reasons. Imagining can sometimes
lead to belief or acceptance, so one may avoid imagining subscribing to a moral perspective that one considers pernicious or reprehensible (or to a false factual claim),
for fear of succumbing to it.16 The danger, I should add, may be merely that the
person will take it more seriously than she thinks she should, or regard it as less than
utterly absurd, as an alternative to be considered; even that, she may think, would
be bad enough. My orientation example suggests that imaginings can encourage
behavior in accordance with a point of view one rejects, even if one continues to
reject it.17 It is likely that the more condent a person is of her convictions, the more
she will want not to be oriented differently, not to be induced by imagining to act
contrary to them. But the less condent she is the more susceptible she might feel,
the more danger she might think there is of her being corrupted.
Some of the other reasons one may have for eschewing imaginings are related to
these, but some are rather different. Imaginings not connected at all with a point
of view or an attitude one is intent on rejectingreliving a terrifying experience,
for instance, or imaginatively anticipating one that one greatly fearsmay be
just too painful to endure. Many of us nd imaginative experiences elicited by
some violent movies or tragic literature unpleasant or intolerable. What makes
the experience painful may be not so much what is imagined as the manner in
which it is, the vividness of ones imaginative experience, induced by a vividly
realistic portrayal. But there may be certain horrendous scenarios that one simply
cannot imagine in a tolerably detached manner.
Bernard Williams presents the possibility of a person who regards certain
courses of action as unthinkable, or who thinks it insane to consider what one
ought to do in some hypothetical bizarre and monstrous situation:
16. Walton, Morals in Fiction, p. 31. A number of empirical studies bear this out.
See Robert Cialdini, Systematic Opportunism: An Approach to the Study of Tactical
Social Inuence, in Social Inuence: Direct and Indirect Processes, ed. Joseph P. Forgas and
Kipling D. Williams (Philadelphia, Pa.: Psychology Press, 2001), pp. 2539, and the
studies he refers to; Amos Tversky and Daniel Kahneman, Availability: A Heuristic
for Judging Frequency and Probability, Cognitive Psychology 5:207232; and Frederick Bacon, Credibility of Repeated Statements: Memory for Trivia, Journal of Experimental Psychology: Human Learning and Memory 5 (1979): 241252. Gregory Currie and
Tamar Gendler both cite relevant empirical work. See Currie, The Moral Psychology of
Fiction, Australian Journal of Philosophy 73, no. 2 (1995): 250159; and Gendler, The
Puzzle of Imaginative Resistance.
17. Walton, Morals in Fiction, pp. 3233.



It could be a feature of a mans moral outlook that he regarded certain courses

of action as unthinkable. . . . Entertaining certain alternatives, regarding them
indeed as alternatives, is itself something that he regards as dishonourable or
morally absurd. But, further, he might equally nd it unacceptable to consider
what to do in certain conceivable situations. Logically, or indeed empirically
conceivable they may be, but they are not to him morally conceivable. . . . For
him, there are certain situations so monstrous that the idea that the processes
of moral rationality could yield an answer in them is insane: they are situations
which so transcend in enormity the human business of moral deliberation that
from a moral point of view it cannot matter any more what happens. Equally,
for him, to spend time thinking what one would decide if one were in such a
situation is also insane, if not merely frivolous.18

If a work of ction encourages readers to imagine performing actions of certain

kinds (by encouraging them to empathize with a character who performs them, for
instance), the person Williams has in mind may put it down. And one may refuse
to read a novel if it demands simply having to choose, in imagination, between
monstrous alternatives in a bizarre situation. It need not be part of Williamss
suggestions that, in either instance, the person fears being lured to adopt, in real
life, a point of view or to take on an attitude she wants to avoid, or to behave as
though she did. Faced, in imagination, with a dreadful choice, one can and probably will imagine deciding, as best one can, in accordance with ones actual moral
principles. If, as might happen, all of the alternatives would be wrong on ones
actual moral principles, there may be pressure to revise ones principles, though
not necessarily to adopt principles contrary to ones better judgment.
Gendler wants to trace the source of [our unwillingness to imagine morally
deviant situations] to a general desire not to be manipulated into taking on
points of view that we would not reectively endorse as authentically our own.19
The impression of being manipulated may increase ones aversion, no doubt. But
a susceptible person might worry that the imagining itself, regardless of how
it is induced, will tempt him to adopt a point of view he wishes to avoid or to
give it more credence than he thinks it deserves. My aversion to sticking pins
into a portrait of a loved one and so, inevitably, imagining harming her does
not depend on an impression that someone is directing my imagining. And my
aversion need not (though it might) involve a worry that imagining thus will
encourage or nourish a desire actually to harm the person.
These armchair observations about instances in which one might refuse to imagine
in certain ways are, of course, subject to empirical conrmation or disconrmation,
18. Bernard Williams, A Critique of Utilitarianism, in Utilitarianism, For and
Against, by J. J. C. Smart and Bernard Williams (Cambridge: Cambridge University
Press, 1973), p. 92. Emphasis in the original.
19. Gendler, Puzzle of Imaginative Resistance, p. 56. See also p. 79.





although I dont think they are likely to be very controversial. They are claims only
about what sometimes happens, and they are not very specic. In any case, it seems to
me obvious and unsurprising that people do, in various kinds of situations and for
a variety of understandable reasons, resist engaging in certain imaginings, however
much there is to learn about the details of when and how and why this is so. This
branch of the imaginative puzzle is not very puzzling.
There is considerably more mystery, not to mention confusion, about what
might be difcult or impossible to imagine and why. Much has been written
about whether logical or metaphysical impossibilities can be imagined, whether
imaginability is a good test for possibility, and also about the prospects for imaginings that go beyond ones prior experiencesimagining something blue or
the taste of vegemite, if one hasnt seen anything blue or tasted vegemite, or
imagining what it is like to be a bat.20 I am not prepared to deny that impossibilities can be imagined.21 And I take no stand on whether, or how far, a persons
imaginings might transcend her prior experience. Gregory Currie offered the
intriguing suggestion that what he calls desire-like imaginings, in contrast to
belief-like ones, are especially difcult to bring off when they do not accord with
ones actual desires.22 I am not yet convinced that these are fundamentally different kinds of imaginings. We need to understand how this difference sorts with
the difference, which I and others recognize (though I dont understand it as well
as I would like to), between imagining that . . . (including imagining that I desire
such-and-such) and imagining X-ing (e.g., imagining desiring . . . , imagining
feeling . . . , and also imagining believing . . . ). Also, we need an explanation of
why desire-like imaginings should be difcult.
If asked to imagine experiencing a series of tones descending in pitch while
remaining at the same pitch, we are likely to be stumped. We cannot imagine
such an experience, one might suppose, because the experience itself is impossible. But it is not impossible. Shepard tones are series of tones that seem to
descend in pitch while remaining at the same pitch.23 And once one has heard
Shepard tones, one is likely to be able to imagine the experience, to call up an
auditory image of the Shepard tones from memory. (What seems impossible
initially is an instance of imagining X-ing, in a sense that doesnt reduce to
propositional imagining. There is no difculty at all in imagining that one
20. Cf., e.g., the essays collected in Gendler and Hawthorne, Conceivability and
21. Cf. Walton, Mimesis as Make-Believe, pp. 3234, 64.
22. Currie, Desire in Imagination, in Gendler and Hawthorne, Conceivability
and Possibility. Peter Carruthers raises objections to Currie, in Review of Currie and
Ravenscroft, Recreative Minds, Notre Dame Philosophical Reviews 11, no. 12 (2003). Currie considers a different suggestion in The Capacities That Enable Us to Produce and
Consume Art, in Kieran and Lopes, Imagination, Philosophy, and the Arts, pp. 293304.
23. Roger Shepard, Circularity in Judgments of Relative Pitch, Journal of the
Acoustical Society of America 36, no. 12 (1964): 23462353.



enjoys an experience as of a tone descending in pitch while remaining at the

same pitch.)


It would be surprising if the ctionality and imaginative puzzles were not linked
in some important way, since ctionality is dened in terms of imagining. That
they are is claimed or assumed in much of the literature on imaginative resistance, even (or especially) when the two puzzles are not clearly differentiated.
But it is not easy to say what the link is, and there is disagreement about which
branch of the imaginative puzzle connects with the ctionality one.
A proposition is ctional if it is to be imagined, if a story or other work of
ction prescribes imagining it. We might think of what is ctional as what
appreciators, qua appreciators of the work in question, ought to imagine. But
this encourages a misconception about how the puzzles are related: Since ought
implies can, one might suppose, we ought to imagine something only if it is
possible to do so; so we will rightly refuse to judge a proposition to be ctional
if we are unable to imagine it. But this ought is a conditional one, which does
not imply can. One ought to imagine p if one is to fully appreciate the work
in question, that is, full appreciation requires imagining p. But imagining
p and full appreciation might not be possible. Those who think we cannot
imagine metaphysical impossibilities or blatant contradictions neednt deny
that they can be ctional, that some works enjoin imagining them.24 It is
generally agreed that (at least some) blatant contradictions and metaphysical
impossibilities can be ctional. In arguing against what she calls the impossibility hypothesisthe hypothesis that imaginative resistance is explained
by the fact that the relevant scenarios are conceptually impossible and hence
unimaginableGendler claims that impossibilities can be imagined. She may
well be right. But if by imaginative resistance here she means resistance to
accepting the scenarios in question as ctional, as she seems to, this claim is
Imaginative resistance, Gendler thinks, is due primarily to our unwillingness
to imagine certain propositions, not an inability to do so.25 Assuming, again, that
it is the ctionality puzzle she has in mind here, this is unsatisfactory; unwillingness to imagine something does not account for resistance to judging it to be
ctional. A person who refuses to imagine that Mother Teresa is a drug dealer,
or that the holocaust is a hoax, may accept with no hesitation whatever that this
is true in the world of a story; she might condemn the story or refuse to read it
24. Kathleen Stock concurs that what is ctional need not be imaginable. See Stock,
The Tower of Goldbach and Other Impossible Tales, in Kieran and Lopes, Imagination,
Philosophy, and the Arts, p. 108.
25. Gendler, Puzzle of Imaginative Resistance, pp. 56, 74, 79. But see also 73n25.





just because it does make this ctional. To use one of Gendlers examples: I may
be unwilling to imagine that my beloved Aunt Ruth looks like a walrus;
I may simply not want to notice the way in which her forehead juts forward . . . or
the way that her eyes bug out, or the fact that . . . lines beneath her nose . . . look
a bit like tusks.26 Yet I may recognize all too well that an unattering portrait
depicts her as looking thus, or that someone might or did write a story in which,
ctionally, she was locked up in a zoo because she was virtually indistinguishable from an escaped walrus. I may refuse to look at the picture or read the story,
given that it does make this ctional, so as to escape the invitation or inducement to engage in the uncomfortable imaginings.
My best suspicion as to why we resist allowing ctional worlds to differ
from the real world when we do, I said, is that it has something to do with
an inability to imagine [certain kinds of dependence relations] being different
from how we think they are, perhaps an inability to understand fully what it
would be like for them to be different.27 This is not an endorsement of the
impossibility hypothesis Gendler objects to, since I hold neither that conceptual impossibilities in general are unimaginable, nor that what is unimaginable
cannot in general be ctional. What seems to me to be important is a very particular kind of imaginative inability, one that attaches to propositions expressing
certain sorts of supervenience relations, which the imaginer rejects. This is barely
a beginning. But both Brian Weatherson and Steven Yablo suggest plausible
ways of developing this line of thought.28


The untangling I have undertaken above and in Morals in Fiction should make
evident how misleading it is to apply this label to the nest of issues treated in
that essay. Pluralizationthe puzzles of imaginative resistancewould bring
out the multiplicity of the strands, but its mischaracterization of most of them
would be glaring. The aesthetic puzzle, which concerns the relation between
aesthetic and moral values, involves resistance to imagining only very indirectly.
The resistance constituting the ctionality puzzle is not imaginative resistance,
resistance to imagining, but resistance to accepting that something is ctional.
Among the imaginative puzzles, those concerning what we cannot imagine are
not puzzles of imaginative resistance; inability is not resistance. This leaves only
our unwillingness to engage in certain imaginings in certain circumstances, which
26. Gendler, Puzzle of Imaginative Resistance, p. 380.
27. Walton, Morals in Fiction, p. 42.
28. Weatherson, Morality, Fiction and Possibility, pp. 1618, 2124. Yablo
proposes that the ctionality puzzle applies to propositions involving response enabled, or
grokking predicates (e.g., oval). See Yablo, Coulda, Woulda, Shoulda.



is aptly described as imaginative resistance. But although this unwillingness is

important and is surrounded by important unanswered questions, it is not itself
particularly puzzling.
The ctionality puzzle, especially, certainly is puzzling; indeed it is much
more than a puzzle. Calling it that (as I have done) suggests that it is relatively
supercial, subject perhaps to a quick and denitive solution, if not an easy one.
This is far from the case. As anyone immersed in it can testify, to wrestle with
the ctionality puzzle is to enter, by a side door, absolutely fundamental mysteries about the nature of concepts, supervenience relations, response dependence,
normative judgments and the imagination.
I have no complaints about the word of in the puzzle of imaginative
Bacon, F. T. 1979. Credibility of Repeated Statements: Memory for Trivia. Journal of
Experimental Psychology: Human Learning and Memory 5:241252.
Carroll, N. 1996. Moderate Moralism. British Journal of Aeshtetics 36:223237.
Carruthers, P. 2003. Review of Currie and Ravenscroft, Recreative Minds. Notre Dame
Philosophical Reviews 11(12).
Cialdini, R. B. 2001. Systematic Opportunism: An Approach to the Study of Tactical
Social Inuence. In Social Inuence: Direct and Indirect Processes, edited by J. P. Forgas
and K. D. Williams, pp. 2539. Philadelphia, Pa.: Psychology Press.
Currie, G. 2003. The Capacities That Enable Us to Produce and Consume Art. In
Imagination, Philosophy, and the Arts, edited by M. Kieran and D. Lopes, pp. 293304.
London: Routledge.
. 2002. Desire in Imagination. In Conceivability and Possibility, edited by
T. S. Gendler and J. Hawthorne, pp. 201222. Oxford: Oxford University Press.
. 1995. The Moral Psychology of Fiction. Australian Journal of Philosophy 73(2):
Currie, G., and I. Ravenscroft. 2002. Recreative Minds: Imagination in Philosophy and Psychology. Oxford: Oxford University Press.
Devereaux, M. (1998). Beauty and Evil: The Case of Leni Riefenstahls Triumph
of the Will. In Aesthetics and Ethics: Essays at the Intersection, edited by J. Levinson,
pp. 227256. Cambridge: Cambridge University Press.
Gaut, B. (1998). The Ethical Criticism of Art. In Aesthetics and Ethics: Essays at the
Intersection, edited by J. Levinson, 182203. Cambridge: Cambridge University Press.
Gendler, T. S. 2000. The Puzzle of Imaginative Resistance. Journal of Philosophy 2:5581.
. 2006. Imaginative Resistance Revisited. In The Architecture of the Imagination:
New Essays on Pretense, Possibility, and Fiction, edited by S. Nichols, pp. 149173.
Oxford: Oxford University Press.
Gendler, T. S., and J. Hawthorne, eds. 2002. Conceivability and Possibility. Oxford: Oxford
University Press.
Hume, D. 1965. Of the Standard of Taste. In Of the Standard of Taste and Other Essays,
pp. 324. Indianapolis: Bobbs-Merrill.
Jacobson, D. 1997. In Praise of Immoral Art. Philosophical Topics 25(1): 155199.
John, E. 2005. Artistic Value and Opportunistic Moralism. In Contemporary Debates in
the Philosophy of Art, edited by M. Kieran, pp. 331341. Oxford: Blackwell.





Kieran, M. 1996. Art, Imagination, and the Cultivation of Morals. Journal of Aesthetics
and Art Criticism 54(4): 337351.
Matravers, D. 2003. Fictional Assent and the (so-called) Puzzle of Imaginative Resistance. In Imagination, Philosophy, and the Arts, edited by M. Kieran and D. Lopes,
pp. 91106. London: Routledge.
Mencken, H. L. 1926. Music and Sin. In Prejudices: Fifth Series, pp. 293296. New York:
Alfred A. Knopf.
Moran, R. 1994. The Expression of Feeling in Imagination. Philosophical Review 103(1):
Nichols, S. 2006. Just the Imagination: Why Imagining Doesnt Behave Like Believing. Mind and Language 21(4): 459474.
Shepard, R. N. 1964. Circularity in Judgments of Relative Pitch. Journal of the Acoustical Society of America 36(12): 23462353.
Sibley, F. 2001. Aesthetic Judgments: Pebbles, Faces, and Fields of Litter. In Approach
to Aesthetics, edited by J. Benson, B. Redfern, and J. R. Cox, pp. 176190. Oxford:
Oxford University Press.
Stock, K. 2003. The Tower of Goldbach and Other Impossible Tales. In Imagination,
Philosophy, and the Arts, edited by M. Kieran and D. Lopes, pp. 107124. London:
Tanner, M. 1994. Morals in Fiction and Fictional Morality. Proceedings of the Aristotelian
Society, Supplementary Volume 68:5166.
Tversky, A., and D. Kahneman. 1973. Availability: A Heuristic for Judging Frequency
and Probability. Cognitive Psychology 5:207232.
Walton, K. L. 1990. Mimesis as Make-Believe: On the Foundations of the Representational Arts.
Cambridge, Mass.: Harvard University Press.
. 1994. Morals in Fiction and Fictional Morality. Proceedings of the Aristotelian Society, Supplementary Volume 68:2750. [Reprinted as chapter 3 of this volume.]
Weatherson, B. 2004. Morality, Fiction and Possibility. Philosophers Imprint 4(3): 127.
Williams, B. 1973. A Critique of Utilitarianism. In Utilitarianism, For and Against,
by J. J. C. Smart and B. Williams, pp. 75150. Cambridge: Cambridge University
Yablo, S. 2001. Coulda, Woulda, Shoulda. In Conceivability and Possibility, edited by
T. S. Gendler and J. Hawthorne, pp. 441492. Oxford: Oxford University Press.

This page intentionally left blank


This page intentionally left blank

Make-Believe beyond Childhood

ake-believe is not just for children. Many adult activities are best understood
as continuations of childrens make-believe, and can be illuminated by comparing
them with games of dolls, cops and robbers, and hobby horses. One adult activity
that involves make-believe is that of making and looking at pictures. Verbal texts
also involve make-believe, in some instances. But to be a picture is essentially,
I claim, to have a role of a certain kind in certain sorts of games of make-believe.
What are pictures? How does a picture of a man differ from the word man? In
a nutshell, pictures are props in visual games of make-believe.1
In Meditations on a Hobby Horse, Ernst Gombrich compared pictures to
a simple hobby horse, a stickperhaps with a wooden head attached, but
This essay is a version of a lecture which I presented in various forms on various occasions, initially as part of the Stieren Distinguished Lecture in the Arts at Trinity University in 1991. A drastic abbreviation of it appeared in Art Issues 21 ( January/February
1992), as Make-Believe, and its Role in Pictorial Representation, pp. 2227. A somewhat longer variant was published in Philosophic Exchange (1994) under the title MakeBelieve, and its Role in Pictorial Representation and the Acquisition of Knowledge.
A different short version, Make-Believe and the Arts, appeared in Aesthetics, by Susan
Feagin and Patrick Maynard (New York: Oxford University Press, 1997), pp. 288296,
but with only one of the illustrations. The present essay differs substantially from these
previously published ones, but it combines most of what I consider worthwhile in each
of them. I have retained the informal lecture style in all the printed versions, including
this one.
1. The theory of make-believe that I sketch here is developed much more thoroughly
in Mimesis as Make-Believe: On the Foundations of the Representational Arts (Cambridge,
Mass.: Harvard University Press, 1990). For a more complete statement of my count of
depiction, see esp. chap. 8. [See also the other essays on depiction in this volume.]




perhaps just a plain stickon which a child rides around the house. Gombrich
considered and rejected describing this stick as an image of a horse, an imitation of [a horses] external form. He also considered and rejected thinking of it
as a sign that signies or stands for or refers to a horse, or to the concept horse.
Pictures also, he suggested, are not to be thought of in either of these ways. He
proposed thinking of pictures and hobby horses, rather, as substitutes. A hobby
horse substitutes for a horse; a picture of a man substitutes for a man.2
Meditations on a Hobby Horse, famous though it is, has been largely
ignored. It is fair to say that most discussions of pictorial representation during
the last forty years have proceeded in one or the other of the two directions Gombrich advised against. There are resemblance theories of representation (some
more sophisticated than others). And there are semiotic theories, such as that
of Nelson Goodman, who declares atly that denotation is the core of representation.3 Even Gombrichs own later work, including Art and Illusion, has
been understood by some to advance the idea that pictures are imitations of the
external forms of objects. Others nd in it the conception of pictures as symbols
or signs that signify or stand for what they are pictures of.4 Neither interpretation is entirely without justice. But Gombrichs original characterization of
pictures as substitutes, and his comparison of pictures with hobby horses, was
on the right track.
Two central thoughts stand out in Gombrichs reections on pictures and the
hobby horse. First, he emphasizes that art is creation rather than imitation.
The child makes a train either of a few blocks or with pencil on paper, he
observesshe doesnt imitate or refer to a train; she makes one.5 All art is imagemaking and all image-making is rooted in the creation of substitutes.6 But is it
mere substitutes that the image maker creates? Gombrich described the child as
making a train out of blocks or on paper, not a substitute for a train. To cement
the uncertainty he states: By its capacity to serve as a substitute the stick
becomes a horse in its own right, it belongs in the class of gee-gees and may
even merit a proper name of its own.7 What is it that the artist creates when she
draws a mana man or a substitute for a man?
The second central idea that Gombrich derives from the association of pictures
with hobby horses is an emphasis on function rather than form. The rst hobby
horse was . . . just a stick which qualied as a horse because one could ride on it.
2. Ernst Gombrich, Meditations on a Hobby Horse, in Meditations on a Hobby Horse
and Other Essays (London: Phaidon Press, 1963), pp. 13.
3. Nelson Goodman, Languages of Art, 2d ed. (Indianapolis: Hackett, 1976), p. 5.
4. See David Summers, Real Metaphor: Towards a Redenition of the Conceptual
Image, in Norman Bryson, Michael Ann Holly, and Keith Moxey, eds., Visual Theory:
Painting and Interpretation (New York: HarperCollins, 1991), pp. 234235.
5. Gombrich, Hobby Horse, p. 3.
6. Gombrich, Hobby Horse, p. 9.
7. Gombrich, Hobby Horse, p. 2.



Any ridable object could serve as a horse. A ball represents a mouse to a cat,
he says. And to a baby, who sucks its thumb as if it were a breast, the thumb
represents a breast. The ball has nothing in common with the mouse except that
it is chasable. The thumb nothing with the breast except that it is suckable.8
Function rather than form.
But the distinction between function and form may seem to be just where
hobby horses and pictures diverge. Yes, a mere stick with hardly any of the form
of a horse, just enough to be ridable, serves as a horse. But pictures capture the
appearance of the things they picture. One doesnt ride a picture of a horse; one
looks at it. A single object can have more than one function, however. One function of a horse is to be ridden, but another function, which some horses have for
some people, is to be looked at. Maybe pictures of horses substitute for horses as
objects of vision. But I am getting ahead of myself.
Much of what Gombrich said in spelling out the analogy between hobby horses
and pictures is blatantly and straightforwardly false. (This might be one reason
why his early essay was ignored.) The notion that the stick is (literally) a horse,
or that a picture of a man is (literally) a man, is as blatant a falsehood as one can
nd. The stick is a stick; the picture is a picture. Nevertheless, as Gombrich
observes, it is perfectly ordinary for perfectly sane people to point to a picture of
a man and say, in all seriousness, That is a man. It is also perfectly natural for a
perfectly normal child to point to the stick and say, This is a horse.
Are these just short ways of saying, That is a substitute man or This is a
substitute horse, it being understood that substitutes are not the real thing? But
the hobby horse is not much of a substitute for a horse. Had Paul Reveres horse
been sick the night of the British attack, he could hardly have made do with a
hobby horse borrowed from a neighborhood child. Not even a wonderfully realistic hobby horse with a carved head and carpet tacks for eyes would have enabled
him to beat the British to Concord. Hobby horses are not ridable, not really; so
they cant really substitute for actual horses. And if someone wants to look at a
horse, a picture of a horse is not a very satisfactory replacement. To see a picture
of a horse is not to see a horse, not really. And the viewer of the picture does not
even enjoy an illusion of seeing a horse. In all but the rarest of cases it is perfectly
obvious that what one is seeing is a at surface with marks on it, not a horse.
The children in Jonathan Eastman Johnsons The Old Stagecoach (gure 5.1)
have something better than sticks to use for horses; some of them play the
parts themselves. But children are not really horses any more than sticks are.
They are not much better than sticks for ridingPaul Revere couldnt have
replaced his sick horse with a neighbors child any more successfully than with
the childs hobby horse. And even four children cant pull a stagecoach very far.
Not really.

8. Gombrich, Hobby Horse, p. 4.



Figure 5.1 Jonathan Eastman Johnson (American, 18241906), The Old Stagecoach,
1871. Oil on canvas, 36 1/4" 60 1/9". Milwaukee Art Museum, Layton Art Collection,
Gift of Frederick Layton, L1888.22.

But the children in this picture have created a ctional worldthe world of
their game of make-believe. Within this world there are horsesreal ones, not
substitutes; and they really do pull the stagecoach. Lets say that it is ctional,
ctional in the world of the game of make-believe, that real horses are really
pulling the stagecoach. Speaking in the real world, I must say that the horses are
merely real-in-the-world-of-the-game, that it is only ctional that they are real.
But if I could get inside the ctional world myself and speak there, I could say
that the horses are real, period.
The children you see are in the ctional world. It is ctional, true-in-theworld-of-the-game, that some of them are riding in a coach pulled by real horses.
And they can say, within the game, Those are real horses (if they feel it necessary to belabor the obvious). It is only when we stand outside the game, when
parents are talking about the fun their children are having with the old broken
down stage coach, for instance, that saying That is a horse is a blatant falsehood. Yes, Paul Revere cannot replace an ailing real horse with either a hobby
horse or a child. But that is because the British attack comes in the real world. If
the British attacked in the world of make-believe, a child might ride off on his
hobby horse or on another childon whatever in the world of the game counts
as a real horseto spread the alarm.
Pictures have worlds also. There is a ship in the world of Stanelds On the Dogger
Bank (gure 5.2)a real ship, not a substitute. From my position in the real world
I have to tell you that this isnt really a real ship. Here in the real world we have
nothing but a picture consisting of colored marks on a at surface, a picture of a



Figure 5.2 Clarkson Staneld (English, 17931867), On the Dogger Bank, 1846. Oil
on canvas. Reproduced by permission of V&A Images / Victoria and Albert Museum.

ship; it is only ctional that there is a real ship here. But if I could somehow get
inside the picture, inside the picture world, I could then say That is a real ship.
Gombrichs analogy between pictures and hobby horses now seems in jeopardy. A child playing with a hobby horse belongs to the world of her game of
make-believe. But the spectator of a picture does not belong to the world of the
picture. Real people can and do get inside make-believe worlds. But all we can
do with picture worlds is observe them from outside. Maybe cartoon characters
can get into pictures. The character in gure 5.3 does. Cartoon characters are not
always bound by the laws of logic and metaphysics. But logic or metaphysics
seems to bar real people like you and me from entering picture worlds.



Figure 5.3 Mischa Richter. The New Yorker Collection 1965. From cartoonbank.
com. All rights reserved.

But wait! How did the ship get into Stanelds picture? Maybe I can get in by
the same way (gure 5.4). I brought my son along to help me paddle. It really
is me in the picture world. It is ctional, true in the world of the picture, that
I, Kendall Walton, am paddling a canoe in heavy seas close to a small sailing
ship. I got into the picture world in almost the same way the ship did. It was
painted in; I was pasted in, and that is just as good.9 While I am in the picture
world I can with perfect appropriateness declare the ship to be realas I do.
But this is disappointing, and not just because I ruined a nice picture. I am not
present in the picture world in the way a child playing hobby horses is present in
the Wild West world of her game of make-believe. The trouble is that I am still
here in the real world, giving a lecture on the nature of pictorial representation.
And I am looking not at a ship, but just at a picture, a picture of myself looking
at a ship.
The difference in the two ways of being in ctional worlds is partly this: What
is in the picture world depends on the picture, on a pattern of shapes and colors
on a at surface. But what exists in a game of make-believe depends on the children who are playing the game, as well as on properties of the stick and other
props. It is because of the pattern of colored shapes on the page, because of the
extra shapes caused by doctoring the picture, that my son and I are paddling a
canoe in the picture world; where I really am and what I am actually doing now
are irrelevant. But it is because of what the child is actually doing, because she
is straddling the stick and jumping around the house, that she belongs to the
world of her game and, in that world, rides a horse.
Another difference is this: I could be mistaken, when I look at the picture,
about whether it is really me in the picture world. I might have to look closely to
recognize myself. And even if I think I do, the picture might portray, not me but
someone who looks exactly like me. I could trace the history of the photograph
that was pasted onto the reproduction of Stanelds painting: Who was the camera aimed at when the photograph was taken? Obviously I could make a mistake
about that. If the picture has a title that includes my name, it is still possible that
the name in the title refers to someone else with the same name. By contrast, it
9. Actually, I am not in the world of Stanelds picture; we now have a different picture.
But I am in a ctional world, the world of this new picture.



Figure 5.4 Clarkson Staneld (English, 17931867), On the Dogger Bank, 1846. Oil
on canvas. Reproduced by permission of V&A Images / Victoria and Albert Museum.
Permission to digitally alter this image has been granted by the Victoria and Albert

does not seem possible for the child playing hobby horses to be mistaken about
the fact that it is she herself, not someone else, who, in the world of the game, is
riding a big black stallion.
Maybe instead of trying to squeeze myself into a picture, I can make the picture
world bigger, big enough to include me where I am. It will have to expand in the
third dimension, like this (gure 5.5):
This gentleman is not in the picture world proper, inside the frame, but there
is a larger world extending in front of the picture that includes both him and the
saguaro cactuses in the picture. He has the right kind of presence in this world:



Figure 5.5 Drawing by F. B. Modell. 1951, 1979, The New Yorker Magazine, Inc.

it is by virtue of his actually standing in front of the painting that it is ctional

in the expansion of the picture world that the desert sun casts a shadow behind
him. And he can hardly be wrong in identifying himself as the person who, in
the larger ctional world, is looking at the cactus.
This may, however, seem as fantastic as the earlier cartoon, beyond the capacity
of real world mortals. Most painted suns arent brilliant enough to cast actual
shadows into the real world. But the idea was not to make ctional things real;
our thought was to get the actual spectator into a ctional world, to expand the
picture world around the spectator. Caravaggios Bacchus (gure 5.6) is a real-life
picture whose world really does expand to include you and me. Bacchus offers
you a drink. You may not be able to take the glass of wine from his hand, but
even before you do he has you in a ctional worldnot the world of the painting
proper, but a larger world that includes both you and what is in the picture. It is
ctional in this larger world that Bacchus offers you a glass of wine. What makes
this ctional is the fact that you are actually looking at the image on the page.
And you will have no doubt that it is you, not someone else, to whom Bacchus
offers the wine. By placing yourself in front of the picture you put yourself in
position to be the recipient of Bacchuss offer.
Think of this larger world as the world of a game of make-believe in which
the picture is a prop. There is a parallel with the childs hobby horse. When
the hobby horse leans unused in the corner of a room, we can think of it as, by
itself, establishing a ctional world something like the world of a picture (or



Figure 5.6 Michelangelo Merisi da Caravaggio, Bacchus. Scala/Art Resource, New York.

a sculpture). There is a real horse in that world, but a child playing checkers
on the other side of the room does not belong to it. When the child takes the
stick and uses it as a prop in a game, the world of the hobby horse expands into
a world of a game of make-believe, and in this world the child rides the horse.
The larger world is established by the prop, the stick, together with what the
child does with it.
Normally, spectators dont do anything with pictures as physical as riding them;
museums have rules about not touching paintings. But we do look at pictures, and
looking at Caravaggios Bacchus in the normal manner lets one in for an offer of a
drinkin the world of the game of make-believe. (We are tempted sometimes to



play more physical games with pictures. A portrait of a despised politician makes
a wonderful prop in a game in which we, ctionally, throw darts at him.)
Bacchus is a special case. Looking at most pictures does not make it ctional
that one is offered a drink. But it is ctional not only that Bacchus offers you a
drink but also that you see him. And depending on the manner in which you
examine the picture, it may be ctional that you look into his eyes, or that you avert
your gaze; that you identify and count the fruit in front of him, or that you fail
to notice the fruitall this in the world of the game with the picture.10
In looking at Stanelds seascape we expand the picture world, which itself contains a ship oundering in the sea, into a larger world of make-believe in which
we see the ship. We use the picture as a prop in a game in which it is ctional, by
virtue of our actually looking at it in the way we do, that we see a ship. It may be
ctional also that we examine the rigging, or watch the sailor in the stern trying
to retrieve the broken spar from the sea, or stare apprehensively at the wave in
the background that is about to lift the ships bow high in the air.
So I can, after all, while examining the picture as I speak in a lecture hall, say
See that ship? Its a real one!provided that in saying this I am participating
in the game of make-believe, speaking within the world of my game. Just as
straddling a stick and jumping around establishes a ctional world in which
one rides a horse, looking at a picture establishes a ctional world in which one
observes things of the kind the picture depicts.
We now have a better way of understanding what it means to call the stick or
a picture a substitute. The stick is neither a real horse nor can it really be used as
a horse; one cant ride it. But it can be used in a game of make-believe within
which it is real and is really ridable. The picture is used in games in which it is
ctional that one really does see a real ship.
Games of make-believe are imaginative activities. As they climb on and in and
around the old stage coach, the children do not just observe that it is ctional
that the stage is moving at high speed, drawn by four horses, that one of them
lets call him Rodneyis handling the reins, and so forth. They also imagine all
this to be true.
A mere spectator of the game may imagine this as well, of course. So what is
the advantage of participating in the game? In part, it is the fact that participants
imagine about themselves. Rodney imagines that he, Rodney, is driving a stage.
But this is not all. He also imagines driving a stage. Imagining doing something
or experiencing something is not the same as imagining that one is doing or experiencing it. Remember the canoe expedition my son and I took into the Staneld
painting. As I looked at the doctored picture noting within its frame the photographic image taken on a canoe trip on the Mississagi River, I imagined that
I, Kendall Walton, was paddling a canoe with my son in dangerously heavy seas

10. [But see Still Photographs, chap. 10, this volume.]



near a battered sailing ship. But I did not imagine paddling a canoe in dangerously heavy seas. What I imagined doing was watching myself paddle a canoe in
heavy seas. This is the main reason why my excursion into the picture world was
disappointing, why my presence there was less satisfying than the presence of
children in their games of make-believe. The child playing with his hobby horse
does not merely imagine that he is riding a horse; he imagines riding one.11
Besides the overt physical participation I have considered so far, children
participate verbally and psychologically in games of make-believe. Rodney shouts
directions to the horses: He really does shouthe does make loud vocal noises
and in doing so he makes it ctional that he shouts to the horses. He imagines
shouting to the horses, and he imagines of the noises he actually emits that they
are his shouts to the horses.
Psychological participation is especially important. It is ctional that Rodney
is thrilled and a little nervous as he strains to control the team, and maybe it is
ctional that he swells with pride at the momentous responsibility entrusted
to him of taking the stage safely to its destination. He really is tense and
excited. And it is in virtue of this that ctionally he is tense and excited. He
is not really proud of his responsibility for the stage; he realizes perfectly well
that he doesnt actually have that responsibility, that he is only playing a game.
But he does really experience a swelling sensation as he imagines bearing this
responsibility. It is partly this sensation that makes it ctional that he swells
with pride in the importance of his position. Aware of his swelling sensations,
he (spontaneously) imagines them to be swellings of pride in his responsibility
for the safety of the journey.
Where do the swelling sensations come from? What causes Rodneys feelings
of tension and excitement? These actual feelings result from his imaginings,
from his imagining, vividly, driving the stage, looking out for bandits, bearing
the responsibility for the safety of the stage and its passengers. There is a complex interplay between Rodneys actual feelings or sensations and his imaginings;
they interact with and feed each other. His vivid imagining of his momentous
responsibility stimulates actual swelling sensations, which he imagines to be
feelings of pride in his responsibilities.
Spectators of paintings participate psychologically, as well as visually and verbally, in games of make-believe in which the pictures are props. I feel tension
as I notice the enormous waves in Stanelds seascape and the ships disarray,
and I interpret this tension as a combination of fear for the safety of the ship
and awe at the power of the sea. I really do feel a certain tension, as I look at the
picture. I dont really fear for the ship, since I know that what is before me is not
a ship but a painting. But it is ctional in my game that I see a real ship and
see the difculty it is having in high seas. I imagine seeing this, and I imagine

11. I examine these two varieties of self imagining in Mimesis, section 1.4.



fearing for the ships safety. My actual feelings of tension are incorporated into
my imaginative experience: I imagine these actual feelings to be feelings of a
combination of fear for the ship and awe at the power of nature.
Compare a dream in which you are on your way to school, and the school bell
rings while you are still two blocks away. This means a tardy slip and half an
hour of detention at the end of the day. On waking from the dream, you realize that the school bell was really the sound of your alarm clock, and that you
still have an hour before classes begin. The sound of the alarm was actual and
you really did hear it while you were dreaming, but you interpreted it in
your dream as the school bell. You imagined hearing the school bell, and you
imagined what actually was the hearing of the alarm to be your hearing of the
school bell.
Pictures, as I said, are props in visual games of make-believe. A picture of a
turtle is a prop in games in which viewers imagine seeing a turtle, and imagine
their actual visual experience of the picture to be their seeing of a turtle. It is
ctional, in the world of the game, that a turtle is an object of their vision. Most
accounts of pictorial representation recognize only the world of the picture, and
have the viewer standing outside that world and observing it. Theories differ
as to the manner in which a picture picks out the propositions constituting its
world. Some say it does so by virtue of resemblance or similarity; the picture
resembles states of affairs of the kind the propositions it picks out expressa
picture of a turtle resembles or looks like a turtle. Others say conventions of
some sort are involved. (These correspond roughly to Gombrichs two rejected
alternatives.) In either case, the viewers job is to ascertain what propositions
the picture picks out, what is true in the world of the picture, by noting the
relevant resemblances or by adducing the relevant conventions.
Here is an example to demonstrate the inadequacy of understanding picture
perception as simply a matter of ascertaining what is true in the picture, however that is done. Consider two lms of a roller coaster ride. Both were made by a
camera attached to the last car of the roller coaster. In one case, the camera is hung
from a support in such a way that it remains aligned with the horizon even when
the car rolls from side to side. In the other case the camera is attached rigidly to
the roller coaster so as to tip back and forth as the car does. In the rst lm, the
horizon remains horizontal on the screen, and one sees the roller coaster sway to
the right and the left. In the second lm, the image of the roller coaster remains
upright on the screen, while the horizon tilts. Lets add that both lms have
circular rather than rectangular images on the screen. The two lms contain
exactly the same information; the world of the picture is the same in both cases.
We could make a showing of one indistinguishable from a showing of the other
just by rotating the projected image at the appropriate times.
But the viewers experiences of the two lms will surely be very different.
The viewer of the one made by the rigidly attached camera has the impression
of riding in the roller coaster, of swaying dangerously right and left as the car



goes around turns. The viewer of the other lm has the impression of watching
the swaying roller coaster from a stable position outside of it. The viewer of the
former is more likely than the viewer of the latter to feel sick. The difference lies
in the spectators games of make-believe and their experiences of imagining seeing. The spectator of one lm imagines seeing the roller coaster from a perspective xed relative to the careening roller coaster. The spectator of the other lm
imagines seeing the same roller coaster careening in the same manner, but from
a perspective xed relative to the earth and detached from the roller coaster.
Words are not pictures. And the difference is much more fundamental than is
suggested by saying that words and pictures are simply symbols or signs of different kinds. Words do not necessarily have anything to do with make-believe
at all. If you tell me that San Antonio is the site of the battle of the Alamo,
you are just conveying to me a piece of information. Your words do not call for
imaginings on my part at all like the imaginings a child engages in when she
rides a hobby horse or the imaginings of spectators when they look at pictures.
Some words do elicit visual imaginings, and they may be designed to do so. But
this doesnt amount to their serving as props in visual games of the relevant
kind; it doesnt make them pictures. A vivid description in a travel brochure of
mountains in New Zealand might induce me to form images of enormous snow
covered peaks, owing glaciers, mountain meadows and streams, spring wildowers. But I probably do not imagine of my actual perception of the text that
it is a perception of mountains, glaciers, and meadows.
Language used ctionallyin novels and stories and theater, for instanceis
used as a prop in games of make-believe, though (in the case of novels and stories) not generally visual games. Spectators at a performance of Romeo and Juliet
engage in make-believe in which, ctionally, they not only watch Juliet and
Romeo but also listen to their words. Novels and stories are props in games in
which readers learn about adventures of various kinds. They dont (in the makebelieve world) observe these adventures, but, typically, learn of them from the
testimony of a narrator. The words of many novels and stories are substitutes
not for people and events of the kinds they describe but for serious reports about
them. The text of Gullivers Travels is, ctionally, the text of the journal of a ships
physician, a certain Lemuel Gulliver. We imagine, of our actual reading of the
novel, that it is a reading of such a journal, and we imagine learning from it
about Gullivers adventures in exotic lands.
The make-believe games involving literary ctions, like those in which pictures serve as props, have psychological dimensions. The reader of Anna Karenina
does not merely note that it is ctional that Anna is unfaithful to her husband,
suffers the disapproval of society, and is nally driven to throw herself under
the wheels of a train. It is ctional in the readers game that he learns about all
this, that he sympathizes with Anna, and suffers with her. He imagines learning
about an actual Anna, and imagines sympathizing with and grieving for her.



Spectators of Romeo and Juliet who shed actual tears as they watch the play, interpret them, in their game, as tears of grief for the characters; they grieve for Romeo
and Juliet, in imagination, and imagine their actual tears to be tears of grief. Where
did the tears come from in the rst place? They result from the spectators vivid
imaginings of the tragedy and of the sufferings endured by Romeo and Juliet. The
vivacity of the imaginings depends to a considerable extent on the skill with which
the actors portray the tragedy, of course. A bad performance will fail to elicit vivid
imaginings and actual tears that can be imagined to be tears of grief.
What is the point of all this make-believe? It consists largely in the imaginings
that props elicit in participants, in their imagining seeing, or reading about,
or learning about, or knowing about, events of this or that sort, and imagining
feeling one way or another about them. By engaging in these imaginings we
enrich our understanding of the kinds of experiences we are undergoing, in
imagination, and of the situations we imagine experiencing.12
It is usually characters, people inside pictures and novels, who have the interesting experiences. Appreciators just watch. It is a character who must choose
between love and duty, or who is shipwrecked alone on a desert island, or who
suffers bereavement, or is condemned to die. Appreciators, in the worlds of the
games they play with the work, observe or read about or learn about the characters dilemma or his experiences on the desert island. In reading Yukio Mishimas
Death in Midsummer, I imagine learning about the tragic drowning of three
children and about how their parents respond to it. But the experience of reading the story does not help me to understand only what it is or might be like to
learn about such tragedies befalling other people; it is likely to give me insight
into what it is or might be like to suffer such a tragedy oneself, to lose ones own
children. How does this happen? A quick answer is that I empathize with the
parents in the story. This involves imagining myself in their shoes, imagining
suffering bereavement myself, and responding as they do. But I imagine this,
I empathize with them, as a result of imagining learning about their tragedy and
noting how they deal with it.
Van Goghs lithograph Sorrow (gure 5.7) is, in obvious respects, much less
explicit and detailed than Mishimas story. We have no way of knowing why the
woman is sorrowful. And the picture is more suggestive than explicit concerning her expressive behavior. We dont even see her face. All we have to go on is

12. Imaginings serve cognitive ends in a wide variety of more mundane instances as
well. Here is one kind of case: If you have two right-hand gloves whose mates are lost, can
you make a right into a left by turning it inside out? Try it in imagination. Imagine peeling the glove off your right hand so that it turns inside out, and then tting it onto your
left hand. Yes, it ts! It is crucial to the success of this experiment that one imagine seeing
or feeling the glove turned inside out and then tting onto your left hand. Just imagining
that it has been turned inside out, or that one sees or feels it, doesnt do the trick.



Figure 5.7 Vincent van Gogh, Sorrow. Lithograph. Reproduced by permission of Van
Gogh Museum Enterprises B.V.

her hunched posture. Perhaps we empathize with her, imagining ourselves to

be sorrowful in the way we take her (ctionally) to be. But perhaps not. I am
not sure that I actually imagine being sorrowful myself, when I contemplate
the picture. I do imaginatively respond to the woman, however, in ways that
are not easy to articulate. By imagining feeling as I do toward the woman,



I imaginatively understand her. And this imaginative experience gains for me an

understanding of what a particular kind of sorrow is like.
All this began with the expansion of the picture world into a world of makebelieve big enough to include the perceiver as well as the contents of the picture
world. Rather than merely standing outside Van Goghs lithograph and imagining what it depicts, imagining a sorrowful woman sitting hunched with her head
and arms resting on her knees, I imagine seeing her and observing her sorrow.
This leads to imagining feeling about her and for her, and perhaps with her, in
ways that enable me imaginatively to understand her sorrow. Thus I come to
understand what it is like to feel this way. None of this would be possible if pictures were simply imitations of visual forms, or if they were just signs signifying
or standing for things of the kind they represent. None of this would be possible
if pictures were not, like hobby horses, props in games of make-believe in which
people participate visually, and also psychologically.

On the Nature of Photographic Realism

Photography and the cinema . . . satisfy, once and for all and in
its very essence, our obsession with realism. . . . The photographic
image is the object itself.
Andr Bazin, The Ontology of the Photographic Image
Every photograph is a fake from start to nish.
Edward Steichen, Ye Fakers

Photographs and pictures of other kinds have various strengths and weaknesses.
But photography is commonly thought to excel in one dimension especially, that
of realism. Andr Bazin and many others consider photographs to be extraordinarily realistic, realistic in a way or to an extent which is beyond the reach of
paintings, drawings, and other handmade pictures.
This attitude is encouraged by a rich assortment of familiar observations. Photographs of a crime are more likely to be admitted as evidence in court than
paintings or drawings are. Some courts allow reporters to sketch their proceedings but not to photograph them. Photographs are more useful for extortion; a

Work on this paper was aided by fellowships from the Rockefeller Foundation and the
Stanford Humanities Center. I wish to thank audiences at a number of universities for
helpful criticisms of earlier versions. Those whose observations had particular inuence
on the shape of the result include John G. Bennett, Robert Howell, David Lewis, Patrick
Maynard, Christopher Peacocke, and Stephen White.



Figure 6.1 Francisco Goya y Lucientes (Spanish, 17461828), Tanto y ms (All this
and more); Fatales consequencias de la sangrienta guerra en Espaa con Buonaparte. Y otros
caprichos enfaticos (Disasters of War), plate 22. Etched 1810; whole series rst published posthumously 1863. Etching, lavis, and burin; working proof. Catalogue Raisonn: Harris 142, I, 3; Delteil 141. Platemark: 16.2 25.3 cm (6 3/8" 9 15/16"). Sheet:
21.8 31.7 cm (8 9/16" 12 1/2"). Museum of Fine Arts, Boston. 1951 Purchase Fund,
51.1648. Photograph 2004 Museum of Fine Arts, Boston.

sketch of Mr. X in bed with Mrs. Yeven a full color oil paintingwould cause
little consternation. Photographic pornography is more potent than the painted
variety. Published photographs of disaster victims or the private lives of public
gures understandably provoke charges of invasion of privacy; similar complaints
against the publication of drawings or paintings have less credibility. I expect
that most of us will acknowledge that, in general, photographs and paintings
(and comparable nonphotographic pictures) affect us very differently. Compare
Francisco Goyas etchings The Disasters of War with the Civil War photographs
by Mathew Brady and his associates (see, for example, gures 6.1 and 6.2). It is
hard to resist describing the difference by saying that the photographs have a
kind of immediacy or realism which the etchings lack. (This is not to deny that
the etchings might equal or surpass the photographs in realism of some other
sort, and it is certainly not to claim that the photographs are better.)
That photography is a supremely realistic medium may be the commonsense
view, butas Edward Steichen reminds usit is by no means universal.
Dissenters note how unlike reality a photograph is and how unlikely we are



Figure 6.2 Timothy H. OSullivan, Death on a Misty Morning, 1883. Photograph.

Library of Congress.

to confuse the one with the other. They point to distortions engendered by
the photographic process and to the control which the photographer exercises
over the nished product, the opportunities he enjoys for interpretation and
falsication. Many emphasize the expressive nature of the medium, observing
that photographs are inevitably colored by the photographers personal interests,
attitudes, and prejudices.1 Whether any of these various considerations really do
collide with photographys claim of extraordinary realism depends, of course, on
how that claim is to be understood.
Those who nd photographs especially realistic sometimes think of photography as a further advance in a direction which many picture makers have taken
during the last several centuries, as a continuation or culmination of the postRenaissance quest for realism.2 There is some truth in this. Such earlier advances
1. Perhaps the best recent defense of this dissenting view is that of Joel Snyder
and Neil Walsh Allen, Photography, Vision, and Representation, Critical Inquiry 2
(Autumn 1975): 143169; all further references to this work, abbreviated PVR, will
be included in the text.
2. See Andr Bazin, The Ontology of the Photographic Image, What Is Cinema?
trans. Hugh Gray, vol. 1 (Berkely and Los Angeles: University of California Press, 1967),
p. 12; all further references to this work, abbreviated OPI, will be included in the text.
See also Rudolf Arnheim, Melancholy Unshaped, Toward a Psychology of Art: Collected
Essays (Berkeley and Los Angeles: University of California Press, 1967), p. 186.



toward realism include the development of perspective and modeling techniques,

the portrayal of ordinary and incidental details, attention to the effects of light,
and so on. From its very beginning, photography mastered perspective (a system
of perspective that works, anyway, if not the only one). Subtleties of shading,
gradations of brightness nearly impossible to achieve with the brush, became
commonplace. Photographs include as a matter of course the most mundane
details of the scenes they portraystray chickens, facial warts, clutters of dirty
dishes. Photographic images easily can seem to be what painters striving for
realism have always been after.
But photographic realism is not very special if this is all there is to it:
photographs merely enjoy more of something which other pictures possess in
smaller quantities. These differences of degree, moreover, are not differences
between photographs as such and paintings and drawings as such. Paintings can
be as realistic as the most realistic photographs, if realism resides in subtleties
of shading, skillful perspective, and so forth; some indeed are virtually indistinguishable from photographs. When a painter fails to achieve such realism up
to photographic standards, the difculty is merely technological, one which,
in principle, can be overcomeby more attention to details, more skill with
the brush, a better grasp of the rules of perspective. Likewise, photographs
arent necessarily very realistic in these sort of ways. Some are blurred and badly
exposed. Perspective distortions can be introduced and subtleties of shading
eliminated by choice of lens or manipulation of contrast. Photographic realism
is not essentially unavailable to the painter, it seems, nor are photographs automatically endowed with it. It is just easier to achieve with the camera than with
the brush.
Bazin and others see a much deeper gap between photographs and pictures
of other kinds. This is evident from the marvelously exotic pronouncements
they have sometimes resorted to in attempting to characterize the difference.
Bazins claim that the photographic image is identical with the object photographed is no isolated anomaly. He elaborates it at considerable length; it
is echoed by Christian Metz; and it has resonances in the writings of many
Such wild allegations might well be dismissed out of hand. It is simply and
obviously false that a photographic image of Half Dome, for example, is Half
3. Here is more from Bazin:
Only a photographic lens can give us the kind of image of the object that is capable
of satisfying the deep need man has to substitute for it something more than a mere
approximation, a kind of decal or transfer. The photographic image is the object itself,
the object freed from the conditions of time and space that govern it. (OPI, p. 14)
The photograph as such and the object in itself share a common being, after the fashion
of a ngerprint. Wherefore, photography actually contributes something to the order of
natural creation instead of providing a substitute for it. (OPI, p. 15)



Dome. Perhaps we shouldnt interpret Bazins words literally.4 But there is no

readily apparent nonliteral reading of them on which they are even plausible. Is
Bazin describing what seems to the viewer to be the case rather than what actually is the case? Is he saying that, in looking at photographs, one has the impression, is under an illusion, of actually seeing the world, that a photographic image
of Half Dome appears to be Half Dome?
There is no such illusion. Only in the most exotic circumstances would one
mistake a photograph for the objects photographed. The atness of photographs,
their frames, the walls on which they are hung are virtually always obvious and
unmistakable. Still photographs of moving objects are motionless. Many photographs are black-and-white. Even photographic motion pictures in living color
are manifestly mere projections on a at surface and easily distinguished from
reality. Photographs look like what they are: photographs.
Does our experience of a photograph approach that of having an illusion more
closely than our experiences of paintings do, even though not closely enough
to qualify as an illusion? Possibly. But this is not what Bazin means. If it were,
theater would qualify as even more realistic than photography. Theater comes
as close or closer to providing genuine illusions than lm does, it would seem.
There are real esh-and-blood persons on stage, and they look more like the

And see Christian Metz, Film Language: A Semiotics of the Cinema, trans. Michael Taylor
(New York: Oxford University Press, 1974): The cinema is the phenomenological art
par excellence, the signier is coextensive with the whole of the signicate, the spectacle its
own signication, thus shortcircuiting the sign itself (p. 43).
The claim that the photographic image is identical with the object photographed
has resonances in Helmut Gernsheims observation that the camera intercepts images,
the paintbrush reconstructs them (quoted by Charles Barr, Cinemascope: Before and
After, in Film Theory and Criticism: Introductory Readings, ed. Gerald Mast and Marshall
Cohen, 2nd ed. [New York: Oxford University Press, 1979], p. 144); in Erwin Panofskys
dictum, The medium of the movies is physical reality as such (Style and Medium in
the Motion Pictures, in Film Theory and Criticism, p. 263); and in the frequent characterization of photographs as duplicates or doubles or reproductions or substitutes
or surrogates (see, e.g., Roger Scruton, Photography and Representation, Critical
Inquiry 7 [Spring 1981]: 577603).
4. Stanley Cavell prefers not to take Bazin and Panofsky literally. The truth in what
they say, he suggests, is that a photograph is of the world (of reality or nature),
whereas [a] painting is a world. In explanation, he observes that one can always ask, of
an area photographed, what lies adjacent to that area, beyond the frame. This generally
makes no sense asked of a painting (The World Viewed: Reections on the Ontology of Film,
enlarged ed. [Cambridge, Mass.: Harvard University Press, 1979], pp. 24, 16, 24, 23).
But photographs typically have their own (ctional) worlds, as do paintings. And since
paintings frequently portray actual scenes, they, like photographs, are often of the real
world. We can ask, concerning a painting of an actual scene as well as a photograph, what
there is in reality outside the portion depicted. Indeed we can also ask, in both cases,
what the ctional world is like beyond the frame. Smoke within a frame may indicate
(ctional) re outside it.



people portrayed than do plays of light and dark on a at screen. But Bazin
regards the fact that photographs are produced mechanically as crucial to their
special realismand theatrical portrayals are not produced mechanically (see
OPI, pp. 12 and 14). (Erwin Panofsky explicitly contrasts lm with theater, as
well as with painting.)5
Bazin seems to hold that photographs enjoy their special status just by virtue
of being photographs, by virtue of their mechanical origins, regardless of what
they look like. No matter how fuzzy, distorted, or discolored, no matter how
lacking in documentary value the [photographic] image may be, it shares, by
virtue of the very process of its becoming, the being of the model of which it is
the reproduction; it is the model (OPI, p. 15).
To add to the confusion, let us note that claims strikingly similar to Bazins
observations about photography, and equally paradoxical, have been made concerning painting and other handmade representations, the very things Bazin
and others mean to be distinguishing photography from!
When we point to [a painted] image and say this is a man [s]trictly speaking
that statement may be interpreted to mean that the image itself is a member
of the class man. . . . [A stick which a child calls a horse] becomes a horse in
its own right, it belongs in the class of gee-gees and may even merit a proper
name of its own.6
[A wooden robin poised on a bird-feeding station] does not say: Such is a
robin! It is a robin, although a somewhat incomplete one. It adds a robin to
the inventory of nature, just as in Madame Tussauds Exhibition the uniformed
guards, made of wax, are . . . intended . . . to weirdly increase the staff of the

What, then, is special about photography?

There is one clear difference between photography and painting. A photograph
is always a photograph of something which actually exists. Even when photographs portray such nonentities as werewolves and Martians, they are nonetheless photographs of actual things: actors, stage sets, costumes. Paintings neednt
picture actual things. A painting of Aphrodite, executed without the use of a
model, depicts nothing real.8 But this is by no means the whole story. Those

5. See Panofsky, Style and Medium in the Motion Pictures, pp. 248 and 260.
6. E. H. Gombrich, Meditations on a Hobby Horse or the Roots of Artistic Form,
Meditations on a Hobby Horse, and Other Essays on the Theory of Art (London: Phaidon
Press, 1963), p. 2.
7. Arnheim, The Robin and the Saint, Toward a Psychology of Art, p. 325. Italics in
8. See Scruton, Photography and Representation, p. 579.



who see a sharp contrast between photographs and paintings clearly think that it
obtains no less when paintings depict actual things than when they do not, and
even when viewers fully realize that they do. Lets limit our examples to pictures
of this kind. The claim before us is that photographs of Abraham Lincoln, for
instance, are in some fundamental manner more realistic than painted portraits
of him.
I shall argue that there is indeed a fundamental difference between photographs and painted portraits of Lincoln, that photography is indeed special, and
that it deserves to be called a supremely realistic medium. But the kind of realism most distinctive of photography is not an ordinary one. It has little to do
either with the post-Renaissance quest for realism in painting or with standard
theoretical accounts of realism. It is enormously important, however. Without a
clear understanding of it, we cannot hope to explain the power and effectiveness
of photography.

Painting and drawing are techniques for producing pictures. So is photography.
But the special nature of photography will remain obscure unless we think of it
in another way as wellas a contribution to the enterprise of seeing. The invention of the camera gave us not just a new method of making pictures and not just
pictures of a new kind: it gave us a new way of seeing.
Amid Bazins assorted declarations about photography is a comparison of the
cinema to mirrors. This points in the right direction.9 Mirrors are aids to vision,
allowing us to see things in circumstances in which we would not otherwise be
able to; with their help we can see around corners. Telescopes and microscopes
9. But Bazin was fuzzy about what direction this is. The screen, he says, puts us
in the presence of the actor. It does so in the same way as a mirrorone must agree
that the mirror relays the presence of the person reected in itbut it is a mirror
with a delayed reection, the tin foil of which retains the image. . . . In the lm about
Manolete . . . we are present at the actual death of the famous matador. (Theater and
CinemaPart Two, What Is Cinema? pp. 9798)

Obviously, spectators of a lm of a matador are not in the presence of the matador, nor
does it seem to them that they are. Indeed, Bazin himself apparently agrees, as he continues:
While our emotion may not be as deep as if we were actually present in the arena at that
historic moment, its nature is the same. What we lose by way of direct witness do we not
recapture thanks to the articial proximity provided by photographic enlargement?
(Theater and Cinema, p. 98; my emphasis)

Cavell also suggests comparing photographs with mirrors (see The World Viewed, p. 213).
E. M. Zemach discusses aids to vision more generally (see Seeing, Seeing, and Feeling,
Review of Metaphysics 23 [Sept. 1969]: 324).



extend our visual powers in other ways, enabling us to see things that are too
far away or too small to be seen with the naked eye. Photography is an aid to
vision also, and an especially versatile one. With the assistance of the camera,
we can see not only around corners and what is distant or small; we can also see
into the past. We see long deceased ancestors when we look at dusty snapshots
of them. To view a screening of Frederic Wisemans Titicut Follies (1967) in San
Francisco in 1984 is to watch events which occurred in 1967 at the Bridgewater
State Hospital for the Criminally Insane. Photographs are transparent. We see the
world through them.
I must warn against watering down this suggestion, against taking it to be a
colorful, or exaggerated, or not quite literal way of making a relatively mundane
point. I am not saying that the person looking at the dusty photographs has the
impression of seeing his ancestorsin fact, he doesnt have the impression of seeing
them in the esh, with the unaided eye. I am not saying that photography supplements vision by helping us to discover things that we cant discover by seeing.10
Painted portraits and linguistic reports also supplement vision in this way. Nor is
my point that what we seephotographsare duplicates or doubles or reproductions
of objects, or substitutes or surrogates for them. My claim is that we see, quite literally, our dead relatives themselves when we look at photographs of them.
Does this constitute an extension of the ordinary English sense of the word
see? I dont know; the evidence is mixed.11 But if it is an extension, it is a
very natural one. Our theory needs, in any case, a term which applies both to
my seeing my great-grandfather when I look at his snapshot and to my seeing
my father when he is in front of me. What is important is that we recognize a
fundamental commonality between the two cases, a single natural kind to which
both belong. We could say that I perceive my great-grandfather but do not see him,
recognizing a mode of perception (seeing-through-photographs) distinct from
visionif the idea that I do perceive my great-grandfather is taken seriously. Or
one might make the point in some other way. I prefer the bold formulation: the
viewer of a photograph sees, literally, the scene that was photographed.
Slippery slope considerations give this claim an initial plausibility. No one
will deny that we see through eyeglasses, mirrors, and telescopes. How, then,
would one justify denying that a security guard sees via a closed circuit television
10. Siegfried Kracauers talk of photographys revealing reality could be taken as making this point (see Theory of Film: The Redemption of Physical Reality [London: Oxford University Press, 1960], p. 28). And so could Arnheims claim that by its very nature . . . the
motion picture tends to satisfy the desire for faithful reports about curious, characteristic,
exciting things going on in this world of ours (Film as Art [Berkeley and Los Angeles:
University of California Press, 1957], p. 34).
11. We speak naturally enough of seeing Johnny Carson on television, of seeing Charlie Chaplin in the movies, and of hearing people over the telephone and in recordings.
We may also, naturally enough, deny that a person has seen Johnny Carson if he has seen
him only on television, for example.



monitor a burglar breaking a window or that fans watch athletic events when
they watch live television broadcasts of them? And after going this far, why not
speak of watching athletic events via delayed broadcasts or of seeing the Bridgewater inmates via Wisemans lm? These last examples do introduce a new element: they have us seeing past events. But its importance isnt obvious. We also
nd ourselves speaking of observing through a telescope the explosion of a star
which occurred millions of years ago.12 We encounter various other differences
also, of course, as we slide down the slope. The question is whether any of them
are signicant enough to justify digging in our heels and recognizing a basic
theoretical distinction, one which we might describe as the difference between
seeing (or perceiving) things and not doing so.13

12. Some nd the notion of seeing the past too much to swallow and dismiss talk of
seeing long concluded events through telescopes as deviant or somehow to be explained
away (see Alvin I. Goldman, Perceptual Objects, Synthese 35 [ July 1977]: 269, and
David Lewis, Veridical Hallucination and Prosthetic Vision, Australasian Journal of
Philosophy 58 [Sept. 1980]: 241242). If seeing the past is allowed, one might worry that
having a memory image of something will qualify as seeing it. Zemach accepts this consequence (see Seeing, Seeing, and Feeling, pp. 1516). But it probably can be avoided,
at least for most memory images. Many, if not all, memory images are based on ones
own earlier beliefs about the object, in a manner relevantly similar to the way in which
the visual experiences of the viewers of a painting are based on the painters beliefs about
the object. So one does not see through the memory image for the same reason that one
does not see through paintings. But, if we are to speak of seeing-through-photographs,
we may have to allow that when an image of something one saw previously, but did not
notice, pops into ones head, one sees it again. I do not nd this result distressing. For any
who do, however, or for any who reject the possibility of seeing the past, there is another
way out. Suppose we agree that what I call seeing-through-photographs is not a mode
of perception. We can always nd a different term. The sharp break between photography
and other pictures remains. We still can say that one sees present occurrences via a television monitor but not through, for instance, a system of simultaneous sketching. This is a
signicant difference. And ones access to past events via photographs of them differs in
the same way from ones access to them via paintings.
13. The slippery slope may make it hard to avoid sliding farther in another direction
than some would like. When we look at fossils or footprints, do we see or perceive ancient
marine organisms or ancient animals feet? I repeat that my point neednt be made in
terms of vision or perception. One might prefer to introduce a new notion, to speak of
being in contact with things, for instance, when one either sees them with the naked
eye or sees mirror images or photographs or fossils or footprints of thembut not when
one sees drawings of them (see Patrick Maynard, The Secular Icon: Photography and the
Functions of Images, Journal of Aesthetics and Art Criticism 42 [Winter 1983]: 155169).
It may not be desirable for our theory to recognize, in addition, a more restricted notion
of perceiving or seeing, one which better ts the cases in which we use these everyday
expressions; there simply may be no such natural kind. We should be prepared for the
possibility that there is no very important distinction which even approximates the difference between perceiving things, in any everyday sense, and not perceiving themthat
what we need is a radical reorganization of our concepts in this area.



Mechanical aids to vision dont necessarily involve pictures at all. Eyeglasses,

mirrors, and telescopes dont give us pictures. To think of the camera as another
tool of vision is to de-emphasize its role in producing pictures. Photographs are
pictures, to be sure, but not ordinary ones. They are pictures through which we
see the world.
To be transparent is not necessarily to be invisible. We see photographs themselves when we see through them; indeed it is by looking at Titicut Follies that
we see the Bridgewater inmates. There is nothing strange about this: one hears
both a bell and the sounds that it makes, and one hears the one by hearing the
other. (Bazins remarkable identity claim might derive from failure to recognize
that we can be seeing both the photograph and the object: what we see are photographs, but we do see the photographed objects; so the photographs and the
objects must be somehow identical.)
I dont mind allowing that we see photographed objects only indirectly,
though one could maintain that perception is equally indirect in many other
cases as well: we see objects by seeing mirror images of them, or images produced by lenses, or light reected or emitted from them; we hear things and
events by hearing the sounds that they make. One is reminded of the familiar
claim that we see directly only our own sense-data or images on our retinas.
What I would object to is the suggestion that indirect seeing, in any of these
cases, is not really seeing, that all we actually see are sense-data or images or
One can see through sense-data or mirror images without specically noticing them (even if, in the latter case, one notices the mirror); in this sense they
can be invisible. One may pay no attention to photographic images themselves,
concentrating instead on the things photographed. But even if one does attend
especially to the photographic image, one may at the same time be seeing, and
attending to, the objects photographed.
Seeing is often a way of nding out about the world. This is as true of seeing
through photographs as it is of seeing in other ways. But sometimes we learn
little if anything about what we see, and sometimes we value the seeing quite
apart from what we might learn. This is so, frequently, when we see departed
loved ones through photographs. We cant expect to acquire any particularly
important information by looking at photographs which we have studied many
times before. But we can see our loved ones again, and that is important to us.

What about paintings? They are not transparent. We do not see Henry VIII
when we look at his portrait; we see only a representation of him. There is a sharp
break, a difference of kind, between painting and photography.
Granted, it is perfectly natural to say of a person contemplating the portrait
that he sees Henry VIII. But this is not to be taken literally. It is ctional, not



true, that the viewer sees Henry VIII.14 It is equally natural to say that spectators
of the Unicorn Tapestries see unicorns. But there are no unicorns; so they arent
really seeing any. Our use of the word see, by itself, proves nothing.
A photograph purporting to be of the Loch Ness monster was widely published some years ago. If we think the monster really exists and was captured
by the photograph, we will speak comfortably of seeing it when we look at the
photograph. But the photograph turned out not to be of the monster but (as
I recall) of a model, dredged up from the bottom of the lake, which was once used
in making a movie about it. With this information we change our tune: what we
see when we look at the photograph is not the monster but the model. This sort
of seeing is like the ordinary variety in that only what exists can be seen.
What about viewers of the movie (which, let us assume, was a straightforward
work of ction)? They may speak of seeing the monster, even if they dont believe
for a moment that there is such a beast. It is ctional that they see it; they actually see, with photographic assistance, the model used in the making of the lm.
It is ctional also that they see Loch Ness, the lake. And since the movie was
made on location at Loch Ness, they really do see it as well.
Even when one looks at photographs which are not straightforward works of
ction, it can be ctional that one sees. On seeing a photograph of a long forgotten family reunion, I might remark that Aunt Mabel is grimacing. She is not
grimacing now of course; perhaps she is long deceased. My use of the present
tense suggests that it is ctional that she is grimacing (now). And it is ctional
that I see her grimacing. In addition, I actually see, through the photograph, the
grimace that she effected on the long past occasion of the reunion.
We should add that it is ctional that I see Aunt Mabel directly, without photographic assistance. Apart from very special cases, when in looking at a picture
it is ctional that one sees something, it is ctional that one sees it not through
a photograph or a mirror or a telescope but with the naked eye. Fictionally, one
is in the presence of what one sees.
One such special case is Richard Shirleys beautiful lm Resonant (1969),
which was made by lming still photographs (of an elderly woman, her house,
her belongings). Sometimes this is obvious: sometimes, for example, we see the
edges of the lmed photographs. When we do, it is ctional that we see the
house or whatever through the photographs. But much of Resonant is fascinatingly ambiguous. The photographs are not always apparent. Sometimes when
they are not, it is probably best to say that ctionally we see things directly.
Sometimes we have the impression of ctionally seeing things directly, only
to realize later that ctionally we saw them via still photographs. Sometimes,
probably, there is no fact of the matter. Throughout, the viewer actually sees

14. The reader can get a better idea of what I mean by ctionality from my Fearing
Fictions, Journal of Philosophy 75 ( Jan. 1978): 527.



still photographs, via the lm, whether or not he realizes that he does. And he
actually sees the woman and the house through the photographs which he sees
through the lm.
We now have uncovered a major source of the confusion which infects writings about photography and lm: failure to recognize and distinguish clearly
between the special kind of seeing which actually occurs and the ordinary kind
of seeing which only ctionally takes place, between a viewers really seeing
something through a photograph and his ctionally seeing something directly. A
vague awareness of both, stirred together in a witches cauldron, could conceivably tempt one toward the absurdity that the viewer is really in the presence of
the object.

Lets look now at some familiar challenges to the idea that photography differs essentially from painting and that there is something especially realistic
about photographs. Some have merit when directed against some versions
of the thesis. They are irrelevant when the thesis is cashed out in terms of
The objection that a photograph doesnt look much like the actual scene,
and that the experience of looking at a photograph is not much like the experience of observing the scene in ordinary circumstances, is easily dismissed.
Seeing directly and seeing with photographic assistance are different modes
of perception. There is no reason to expect the experiences of seeing in the
two ways to be similar. Seeing something through a microscope, or through
a distorting mirror, or under water, or in peculiar lighting conditions, is not
much like seeing it directly or in normal circumstancesbut that is no reason to deny that seeing in these other ways is seeing. The point is not that a
photograph shows us . . . what we would have seen if we had been there ourselves. Joel Snyder and Neil Allens objections to this view are well taken
but beside the point (PVR, p. 149, and see pp. 151152). It may be ctional
not that viewers of the photographs are shown what they would have seen but
that they are actually there and see for themselves. Here, again, the confusion
is caused by not distinguishing this from the fact that they actually do see via
the photograph.
If the point concerned how photographs look, there would be no essential
difference between photographs and paintings. For paintings can be virtually
indistinguishable from photographs. Suppose we see Chuck Closes superrealist
Self-Portrait (gure 6.3) thinking it is a photograph and learn later that it is a
painting. The discovery jolts us. Our experience of the picture and our attitude
toward it undergo a profound transformation, one which is much deeper and
more signicant than the change which occurs when we discover that what we



Figure 6.3 Chuck Close, Self-Portrait, 1968. Acrylic on canvas. Collection, Walker
Art Center, Minneapolis, Minn. Art Center Acquisition Fund, 1969. Photo courtesy
of Art Center.

rst took to be an etching, for example, is actually a pen-and-ink drawing. It is

more like discovering a guard in a wax museum to be just another wax gure.
We feel somehow less in contact with Close when we learn that the portrayal
of him is not photographic. If the painting is of a nude and if we nd nudity
embarrassing, our embarrassment may be relieved somewhat by realizing that



the nudity was captured in paint rather than on lm. My theory accounts for the
jolt. At rst we think we are (really) seeing the person portrayed; then we realize
that we are not, that it is only ctional that we see him. However, even after this
realization it may well continue to seem to us as though we are really seeing the
person (with photographic assistance), if the picture continues to look to us to be
a photograph. (In the case of the nude, this may account for the continuation of
some of our original feelings of embarrassment.)15
We have here a case of genuine illusion. It really does look to us as though we
are seeing someone via the medium of photography, and at rst we are fooled.
This is not the sort of illusion which so often is attributed to viewers despite
overwhelming evidence that it almost never occurs. It does not appear to us that
we see a person directly, one standing right in front of us.
We have genuine illusions also when we do see through a photograph but what
we see through it is not what it seems to be. Figure 6.4 is a photograph through
which we see not people but a life-sized sculpture. Illusions of this kind are commonplace in lm, and they contribute importantly to viewers experiences. A
detective in a movie surprises two thugs, pulls a gun, res, and they drop. The
viewer seems to be seeing these events via the lm. He does see one man, an actor,
approach two others, draw a gun, and pull the trigger. But he doesnt see the one
kill the others, since what was photographed was not an actual killingthe bullets were blanks, and the blood, ketchup. Still, the scene looks as though it were an
actual killing which was lmed. The obvious considerations against the idea that
a killing occurs in the viewers presence are irrelevant to the illusion I have described.
The sharp edges of the illuminated rectangle, the obvious atness of the screen, the
fuzziness of some images, the lack of color do nothing to keep it from seeming to
the viewer that he is seeing an actual killing via a photographic lm of it.
There are some superrealist paintingsDouglas Bonds Ace I (Figure 6.5),
for instancewhich have distinctly photographic stylistic traits but are rather
obviously not photographs. Their photographic character is more pretense than
illusion. It doesnt seem to the viewer that he sees through the photographs, but
it may be ctional that he does. It may be ctional that Ace I is a photograph
through which one sees a group of men walking in front of Pasadena City Hall.
The debate about whether photography is special sometimes revolves around
the question of whether photographs are especially accurate. Some contend that
photographs regularly falsify colors and distort spatial relationships, that a photograph of a running horse will portray it either as a blur, which it is not, or as
frozen, which it also is notand of course there is the possibility of retouching
in the darkroom. It remains to be seen in what sense photographs can be inaccurate.
15. Here is an analogous example: suppose a proud parent hears what he takes to be
a recording of Johnny playing the piano and then learns that it is actually someone else
mimicking Johnnys piano playing. He thought he was hearing Johnny play, via the
recording, but he wasnt. Initially he swells with pride in little Johnny, then is deated.

Figure 6.4 John deAndrea, Man with Arms around Woman, vinyl polychrome, 1976.
Photo Courtesy O. K. Harris Works of Art, New York.

Figure 6.5 Douglas Bond, Ace I, acrylic on canvas, 1967. Gallerina La Medusa,
Rome: Photo courtesy of Gallerina.




Yet misleading they certainly can be, especially to viewers unfamiliar with them
or with photographs of a given kind.
But why should this matter? We can be deceived when we see things directly.
If cameras can lie, so can our eyes. To see something through a distorting mirror
is still to see it, even if we are misled about it. We also see through fog, through
tinted windshields, and through out-of-focus microscopes. The distortions or
inaccuracies of photographs are no reason to deny that we see through them
(see, for example, Figure 6.6).
To underscore the independence of accuracy and transparency, consider a theatrical portrayal of actual events, an acting out in a courtroom of events that led
to a crime, for example. The portrayal might be perfectly accurate. Jurors might
gain from it much correct information and no misinformation. Yet they certainly
do not see the incident via the portrayal.
Is the difference between photographs and other pictures simply that photographs are generally more accurate (or less misleading), despite occasional lapses,
that the photographic process is a more reliable mechanism than that of drawing or painting, and that therefore there is better prima facie reason to trust
photographs? I doubt it. Consider a world in which mirrors are so exible that
their shapes change constantly and drastically and unpredictably.16 There seems
no reason to deny that people see through these mirrors, notwithstanding the
unreliability of the mechanism. Perhaps the mechanism is not a knowledge-producing one.17 If a person looks into a mirror and forms beliefs, on the basis of
what he sees, about the things reected in it and if those beliefs happen to be
true, perhaps his beliefs do not constitute knowledge. But this does not mean that
he does not see the reected things.
Some objections focus on the idea that photographs owe their special status
to their mechanical, automatic origins, whereas paintings are handmade.
What is crucial is supposed to be the involvement of a person in the process. Several writers have managed to imply that people dont make photographs.18 In any

16. This example is a relative of Lewiss case of the loose wire (see Lewis, Veridical
Hallucination and Prosthetic Vision, p. 244).
17. See Goldman, Discrimination and Perceptual Knowledge, Journal of Philosophy
73 (18 Nov. 1976): 771791.
18. William Henry Fox Talbot, inventor of the calotype, claimed for the Lacock Abbey
in Wiltshire the distinction of being the rst building that was ever yet known to have
drawn its own picture (The Pencil of Nature [London: Longmans, Brown, Green & Longmans, 1844 46], n. to pl. 15). Bazin credits photography with completely satisfying
our appetite for illusion by a mechanical reproduction in the making of which man plays
no part. . . . For the rst time an image of the world is formed automatically, without the
creative intervention of man (OPI, pp. 12, 13). The fundamental peculiarity of the
photographic medium, says Arnheim, is the fact that the physical objects themselves
print their image by means of the optical and chemical action of light (On the Nature
of Photography, Critical Inquiry 1 [Sept. 1974]: 155).

Figure 6.6 Andr Kertsz, Distortion # 157, 1933. Andr Kertsz. Photo courtesy
of the artist.




case the remarkable realism of photographs is considered to derive not from what
they look like but from how they come about.
On this point I agree. Why is it that we see Lincoln when we look at
photographs of him but not when we look at his painted portrait? The answer
requires an account of seeing (or better, an account of perception in general).
I would subscribe to some variety of causal theory: to see something is to have
visual experiences which are caused, in a certain manner, by what is seen. Lincoln
(together with other circumstances) caused his photograph and, thus, the visual
experiences of those who view it. This does not yet answer our question. For
Lincoln caused his portrait as well as his photograph. The difference lies in the
manner of the causation.
Putting things together, we get this: part of what it is to see something is to
have visual experiences which are caused by it in a purely mechanical manner.
Objects cause their photographs and the visual experiences of viewers mechanically; so we see the objects through the photographs. By contrast, objects cause
paintings not mechanically but in a more human way, a way involving the
artist; so we dont see through paintings.
Objections leap to the fore. Photographs are made by people: The [photographic] image is a crafted, not a natural, thing (PVR, p. 151). Photographers
and painters just use different tools in making their pictures, it seemsone uses
a camera and the other a brush. In what sense, then, are our visual experiences
caused mechanically when we look at photographs and not when we look at
Objectors frequently add that photographs do not present us with things as
they really are but rather with the photographers conception or interpretation of
them, that what we get from a photograph is not our own view of the world but
his. A photograph, no less than a painting, has a subjective point of view.19
All this is beside the point. The manner in which things cause my visual
experiences when I see them is not one which rules out a causal role for human
beings. People often show me things and in other ways induce me to look this
way or that. They affect what I can see or how I see itby turning the lights
on or off, by blowing smoke in my eyes, by constructing and making available
eyeglasses, mirrors, and telescopes. Why not say that photographers, by making
photographs, show me things and also enable me to see them? Surely that does
not mean that I dont really see them.

19. See H. Gene Blocker, Pictures and Photographs, Journal of Aesthetics and Art
Criticism 36 (Winter 1977):
Photographs most certainly do not escape subjectivity. . . .
Through the selection of subject, angle, amount, and direction of light, background,
sharpness of focus, and light-dark contrastin all these ways the photographer represents
the object from a subjective point of view, expressive of feeling and mood. (p. 158)



When I see, I may well get a sense of someone elses conception or interpretation of what I see. If you point out something to me, I know that you consider it
worth pointing out. I learn by seeing, when others affect my vision, what things
are objects of their fears and fetishes, what they value, and what they deplore. It
may not be inappropriate to speak of seeing things through their eyes. Yet I do
see those things myself. Photography can be an enormously expressive medium
Andr Kertszs Distortion # 157 (Figure 6.6) is certainly expressivebut this
expressiveness does not render photographs opaque. If expressiveness is the mark
of art, photographys credentials are beyond question. In Triumph of the Will,
Leni Riefenstahl, by careful selection and editing, interprets for us the Nazi
Party Congress of 1934; she presents it as she construes it. It does not follow
that we ourselves do not see Hitlers airplane descending through the clouds,
the thousands of marching troops and cheering spectators, and Hitler delivering
tirades, even if the lm fosters misconceptions about the things we see, inducing
us to believe, for example, that the people we see were more enthusiastic about
Hitler than they actually were. We can be aware, even vividly aware, of both the
medium and the maker without either blocking our view of the object.
A nal worry is that photography makes use of conventions, conventions
which are built into the construction of the camera and our photographic processing techniques.20 There is nothing sacrosanct about the system of perspective
used in photography, it is argued; we just happen to have incorporated the one
we did into the photographic process. Doesnt this mean that the conventions
of photography get between the viewer and the objects photographed, that the
viewer must know the language of photography and read its symbols, and
that therefore he cannot be said to see the objects through the photographs? Not
at all. We could have a convention to the effect that mirrors used in certain contexts are to be warped in a certain manner (for example, convex mirrors which
enable drivers to see around dangerous corners). The convention must be understood or internalized for one to read properly the mirror images. Nevertheless,
one sees things through the mirrors.

With these objections laid to rest, it is time to tackle directly the question of
what it is about photographs that makes them transparent. The reason why we
see through photographs but not paintings is related to a difference in how we
acquire information from pictures of the two kinds. Suppose an explorer emerges
from a central African jungle with a batch of photographic dinosaur-pictures,
purportedly shot in the bush and processed straightforwardly. The pictures
(together with background information) may convince us that there is a dinosaur
20. See Blocker, Pictures and Photographs, p. 161, and PVR, pp. 156 and



lurking in the jungle. Alternatively, suppose that he emerges with a sheaf of

dinosaur-sketches, purportedly drawn from life in the eld. Again, we may be
convinced of the existence of a dinosaur. Perhaps the photographs are more convincing than the drawings, but they neednt be. That is not the crucial difference
between them; we might have better reason to trust the drawings than the photographs. The important difference is that, in the case of the sketches, we rely on
the picture makers belief that there is a dinosaur in a way in which we dont in
the case of the photographs.
The drawings indicate to us what was in the jungle by indicating what the
artist thought was there. We have reason to believe that the artist set out to draw
what he saw and that he is a competent draftsman. Since the sketches show a
dinosaur, we judge that he thought he saw one. Taking him to be a reliable
observer, we judge that the dinosaur he thought he saw was actually there. We
trust his judgmentour information about the dinosaur is secondhand.
We dont need to rely on the photographers judgment in the same way. We
may infer that he believes in the dinosaur, knowing that he was looking through
the viewnder when the pictures were taken. We might even assume that it is
because he believed there was a dinosaur that the photographs exist or are as they
arewe may assume that he aimed the camera where he did and snapped the
shutter when he did because he thought he spotted a dinosaur. But no such inferences or assumptions are required for our judgment of the dinosaurs existence.
Even if we know or suspect that he didnt see the dinosaur, that he left the camera
on a tripod with an automatic triggering device, for instance, we may still infer
the existence of the dinosaur from the photographs. In fact, if the photographs do
convince us that he believed in the dinosaur, they do so because they convince us
that there was a dinosaur, not the other way around.
We do need to make certain assumptions if we are going to trust the photographs: that the camera was of a certain sort, that no monkey business was
involved in the processing, and so on. These may require our accepting the say-so
of the photographer; we may have to trust him. And it could be that we are being
taken for a ride. It is easy to see that this sort of reliance on the photographer
does not mean that we do not see through his photographs. In order to trust
the evidence of my senses, I must always make certain assumptions about them
and the circumstances in which they operate: that they are not inuenced by
hallucination-inducing drugs, that they are not being fed misinformation by an
evil neurosurgeon, and so forth. I might rely on someone elses word in making
these assumptions; I might consult a benecent doctor. If he assures me that the
system is operating normally, and it is, then I am seeing (or perceiving), notwithstanding my reliance on him.
The manner in which we trust the photographer when his photographs convince us of the existence of the dinosaur differs signicantly from the manner
in which we rely on the artist when we are persuaded by his sketches. Both
sets of pictures have a counterfactual dependence on the scene in the jungle.



In both cases, if the scene had been differentif there had been no dinosaur,
for examplethe pictures would have been different (and so would our visual
experiences when we look at them). This is why, in both cases, given that the pictures are as they are, we can judge that the scene was as it was. But why are these
counterfactuals true? A difference in the scene would have made a difference in
the sketches because it would have made a difference in the artists beliefs (and
hence in the way he sketched or whether he sketched at all). But that is not why
a difference in the scene would have made a difference in the photographs. They
would have been different had the scene been different even if the photographer
believed, and so aimed and snapped his camera, as he actually did. Suppose that
the picture makerartist or photographeris hallucinating the dinosaur which
he attempts to portray. The artists sketches will show a dinosaur nonetheless,
but the photographs will not. What the sketches show depends on what the artist thinks he sees, whether or not he is right; the actual scene in the jungle is, in
this way, irrelevant to how his pictures turn out. But if the photographer thinks
he sees a dinosaur and acts accordingly, what his photographs show is determined
by what is really there before him, regardless of what he thinks. The artist draws
his hallucination; the camera bypasses the photographers hallucination and captures what is in the jungle.
A persons belief can be relevantly based on someone elses even if he doesnt
realize that it is. If what convinces me of the dinosaurs existence is a painting
which I take to be a photograph, I may suppose mistakenly that my belief is
independent of the picture makers and that I see the dinosaur. My grounds for
my belief do not include his belief. But still, the absence of the dinosaur would
have made a difference in the picture only because it would have made a difference in the artists belief. Unbeknown to me, my belief is (relevantly) dependent
on his, and I am wrong in thinking I see the dinosaur.
Not all theories of perception postulate a strong link between perceiving and
believing.21 We neednt assume such a link. The essential difference between
paintings and photographs is the difference in the manner in which they, not the
beliefs of those who see them, are based on beliefs of their makers. Photographs
are counterfactually dependent on the photographed scene even if the beliefs
(and other intentional attitudes) of the photographer are held xed.22 Paintings
which have a counterfactual dependence on the scene portrayed lose it when
the beliefs (and other intentional attitudes) of the painter are held xed. Both the
beliefs and the visual experiences which the viewer derives from a picture are
dependent on the picture makers beliefs in whichever manner the picture itself
21. See Fred I. Dretske, Seeing and Knowing (Chicago: University of Chicago Press,
1969), chap. 2.
22. In some cases the important conditional counterfactual dependence which distinguishes opaque pictures from transparent ones may be not so much on the picture
makers beliefs as on his visual experience, or his thoughts, or possibly his intentions.



is. In order to see through the picture to the scene depicted, the viewer must have
visual experiences which do not depend on the picture makers beliefs in the way
that paintings do. We can leave open the question of whether, to be seeing the
scene, the viewer must have beliefs about it and what connection there may be
between his visual experiences and his beliefs.23
A familiar pair of science ction examples may help to convince some that I am
on the right track.24 Suppose that a neurosurgeon disconnects Helens eyes from
her optic nerves and rigs up a device whereby he can stimulate the optic nerves
at will. The doctor then stimulate Helens nerves in ways corresponding to what
he sees, with the result that she has visual experiences like ones she would have
normally if she were using her own eyes. Let us add the assumption that the doctor
is conscientious about feeding Helen correct information and that she has every
reason to trust him. Helen seems to be seeing things, and her visual experiences are
caused by the things which she seems to see. But she doesnt really see them; the
doctor is seeing for her. This is because her visual experiences are based on his in
the way I described. It is only because differences in scenes make for differences in
the doctors beliefs that they make for differences in her visual experiences.
Contrast a patient who receives a double eye transplant or a patient who is
tted with articial prosthetic eyes. This patient does see. He is not relying in
the relevant manner on anyones beliefs about the things he sees, although his
visual experiences do depend on the work of the surgeon and on the donor of the
transplanted eyes or the manufacturer of the prosthetic ones. In real life, cataract patients owe their visual experiences to others. All of our visual experiences
depend on acts of omission by those who have refrained from altering or destroying our visual organs. Obviously these facts do not blind us.

The intuitions I have been appealing to are of a piece with those underlying
H. P. Grices distinction between natural and nonnatural meaning.25 Spots meanN
(mean naturally) measles, he says, and the ringing of the bell of a bus meansNN
(means nonnaturally) that the bus is full. Grice would say, no doubt, that if the
explorer did indeed capture an actual dinosaur on lm, his photographs meanN
23. In special cases photographs may be causally but not counterfactually dependent
on the scene. Then there may be no hope of learning about the scene from the photograph: the photograph would have been as it is even if the scene had been different. But
one still sees the scene through the photograph. Perception is to be understood in terms
of causation rather than counterfactuals, if the former doesnt reduce to the latter (see
William K. Goosens, Causal Chains and Counterfactuals, Journal of Philosophy 76 [Sept.
1979]: 489495).
24. These examples are adapted from Lewis, Veridical Hallucination and Prosthetic
Vision, pp. 243244. But Lewis does not see a sharp difference between the two cases.
25. See H. P. Grice, Meaning, Philosophical Review 66 ( July 1957): 377388.



that there is a dinosaur. One characteristic of natural meaning is this: the fact that
something meansN that p entails p.26 Black clouds mean (meanN) rain only if they
are in fact followed by rain. If the rain doesnt come, that isnt what the clouds
meant. This gives us a sense in which photographs are necessarily perfectly accurate. If there was no dinosaur, then the photograph does not meanN that there
was one, no matter what it looks like. One who knew enough about the camera
used in making a photograph, how the lm was processed, and other relevant
circumstances could infer with perfect accuracy about the objects photographed.
This alone does not distinguish photographs from other pictures. Presumably, if
I know enough about an artistabout his beliefs, desires, attitudes, capacities,
and such, or his physiological makeupI could infer accurately, from his drawings, about what was in front of him when he drew (see PVR, pp. 159162).
But Grices distinction brings out a difference between the two cases. A sketch of
a dinosaur does not meanN that there was a dinosaur, even if there was one. The
sketch is not necessarily accurate in this way.
The essential accuracy of photographs obviously does not prevent them from
being misleading. It affects instead how we describe our mistakes and how we
think of them. Consider a photographic portrait of Twiggy, made with the help of
a bowed mirror, which appears to show her with a huge paunch. If viewers are misled, it is not because of a divergence between what the pictures meansN and reality.
Their mistake is about what the picture meansN. It meansN not that Twiggy is fat
but that she is skinny, as one who knew about the mirror could ascertain.
To think of photographs as necessarily accurate is to think of them as especially
close to the facts. It is not to think of them as intermediaries between us and the
facts, as things that have their own meanings which may or may not correspond
to the facts and which we have to decide whether or not to trust. To interpret a
photograph properly is to get the facts.
Synder and Allen claim that the way in which a photograph is made has little
to do with the way we normally interpret it (PVR, p. 159). Presumably, they
would say that we interpret the photograph of Twiggy as meaning that she is
fat, regardless of the fact that it was made with a distorting mirror. There is some
truth in this. We may take the photograph to meanN that Twiggy is fat; it may
look to us as though it meansN that; Twiggy may appear to us to be fat when we
see her through the photograph. Perhaps, also, the photograph makes it ctional
that she is fat, and it might even meanNN that she is. None of these facts force
us to deny that the picture meansN not that Twiggy is fat but that she is skinny.
Photographs, as bearers of natural meaning, are necessarily accurate. And our
realization that they areeven when we are unsure of or mistaken about what
they meanNprofoundly affects our experience of them.
The fact that something meansNN p does not entail p. It is connected instead
with the notion of someones meaning p by it. Nonnatural symbols are thought of
26. See Grice, Meaning, p. 377.



as intermediaries which stand between us and the facts. We ascertain what the
symbols mean, from which we learn what was meant by them (which neednt be
the same as what the symbols mean), and we must judge whether what is meant
by them is true. Our access to things via nonnatural symbols is thought of as less
direct than our access via natural ones.
Drawings and paintings are sometimes nonnatural symbolsbut not always.
Pieter Brueghel probably did not intend viewers of Childrens Games to learn
what games were played in the sixteenth century by recognizing his intention
that they do so. Still, the meaning of the picture is enough like nonnatural meaning for us to see its difference from photographs. The beliefs about childrens
games in the sixteenth century which the painting induces are based on the beliefs
of the painter, if not on his communicative intentions.

The distinction between transparent and opaque pictures will provoke a variety
of intriguing examples. Some of them show that this distinction does not coincide neatly with our usual differentiation between photographs and nonphotographic pictures. Some suggest that there are degrees of transparency, while others
suggest that a picture can be transparent in certain respects and opaque in other
respects. In some instances the question of whether a picture is transparent probably has no determinate answer.
There are pictures which are drawn or painted by people but in a mechanical
manner of one sort or another. One may attach a piece of transparent paper to
a window and trace the outlines of the objects seen through it. One may copy
a photograph, conceivably without even recognizing what it is a photograph
of, or paint over a photograph, matching the brightness of each spot of the
original.27 One might use a directional light meter and ll in the squares of
a grid with shades of gray corresponding to the readings it gives of the various parts of a scene, or one might dispense with the light meter and estimate
the brightnesses by eye. There are also doodles done automatically, while the
doodlers mind is on other things.28 Some such mechanically executed drawings
are probably transparent.29
27. It was not uncommon in the mid-nineteenth century to paint portraits over photographs (see Aaron Scharf, Art and Photography [Harmondsworth: Penguin Books Ltd.,
1968], p. 44).
28. I owe the last two examples, respectively, to Robert Howell and to George Wilson.
29. It is time to confess that the Chuck Close example (Figure 6.3) is not as clear-cut
as I implied. Close made many of his works by projecting a photograph on the canvas
and painting over it. If this is how his Self-Portrait was executed, its opacity may be questionable. My point, of course, is unaffected. If Self-Portrait had been painted in the usual
manner, it would denitely be opaque, and the viewer who comes to believe that it was so
painted after having assumed it to be a photograph experiences the jolt I described.



Figure 6.7 Jerry N. Uelsmann, Symbolic Mutation, 1961. Jerry N. Uelsmann.

Photo courtesy of the artist.

Are any photographs opaque? What about ones which are devised largely in
the darkroomby combining negatives, retouching, burning out unwanted
images, manipulating exposure and contrast, using lters, and so on (see, for
example, Figure 6.7)? Some have maintained that such photographic constructions are essentially similar to paintings.30 The darkroom artist exercises as
much control over the nished product as painters do; his work seems no more
mechanical or less human, although his tools and materials are different. The
paradigms of transparent pictures would seem to be not the work of professional
photographers but casual snapshots and home movies made by doting parents
and wide-eyed tourists with assists from Kodak.
Photographic constructions do differ importantly from snapshots, but to lump
them with paintings would be a big mistake. There is the extreme case of a
photograph made by exposing photographic paper, dot by dot, with a ashlight,

30. Scruton remarks that if a photographer proceeds to paint things out or in, to
touch up, alter, or pasticher as he pleases, . . . he has now become a painter (Photography
and Representation, pp. 593594).



to make a pointillist-style rendition of Lincoln, for example. This is drawing with

a ashlight; one doesnt see Lincoln through the picture. But consider more common darkroom techniques such as combining negatives and manipulating contrast.
We see a person through a photograph of him no matter how lightly or darkly it is
printedeven if it falsies the brightness of the person or the brightness relations
of his partsalthough we may not see the state of affairs of his being illuminated in
a certain way. If a photograph apparently showing Deng Xiaoping conversing with
Yasir Arafat was made by combining negatives of each, the viewer does not see the
event of their conversing, even if they were conversing when the two photographs
were taken. But he does see Deng, and he does see Arafat. Most photographic constructions are transparent in some of their parts or in certain respects. If a viewer doesnt
know how a photograph was made, he wont know what he is seeing through it and
what he isnt. But he will probably realize that he is seeing some of the things or events
or states of affairs which the picture portrays, even if he does not know which ones,
and this realization signicantly colors his experience. His experience is not unlike
that of seeing a white shape and wondering whether one is seeing a ghost.
It may seem to the viewer, moreover, that he is seeing everything that the
photograph portrays even if he is not and even if he knows that he is not. Many
photographic constructions appear to be transparent even in respects in which
they are not, and this gives them a sort of realism which obviously nonphotographic pictures lack.
The viewer of Jerry Uelsmanns Symbolic Mutation (Figure 6.7) hardly has
the impression of seeing a hand fused with a face, however; it is too obvious that
the picture was made from two negatives. In other cases sophisticated viewers
may judge simply from the slickness of a photograph that it is likely to have
been manipulated in one way or another in the darkroom, even if they dont
spot the seams. As a result, their impression of seeing through the picture may
be weakened. This is one reason why some lmmakers have deliberately tried
to mimic the crudity of home movies, using hand-held cameras, purposefully
bad focus, and so on (for example, John Cassavetess Shadows [1960]). These
techniques sacrice any possibility of producing the illusion that the viewers are
face-to-face with the characterswhich is hardly a live possibility anywayin
favor of a more convincing illusion of seeing the characters through the photographs.
This reconciles the immediacy which is claimed for such techniquesthe feeling
they provide of intimacy with the objects portrayedwith the obvious sense of
contrivance that they engendertheir calling attention to the medium. Emphasizing the medium is usually regarded as a way of distancing appreciators from
the world portrayed. In this case it has just the opposite effect.

A certain conception of the nature of perception is beginning to emerge: to
perceive things is to be in contact with them in a certain way. A mechanical



connection with something, like that of photography, counts as contact, whereas

a humanly mediated one, like that of painting, does not. Perceptual contact
with things has rather less to do with acquiring knowledge about them than has
sometimes been supposed.
We may be approaching a necessary condition for seeing through pictures and
for perception in general, but we are far from having a sufcient condition.
Imagine a machine that is sensitive to the light which emanates from a scene
and that produces not pictures but accurate verbal descriptions of the scene. The
machines printouts are surely not transparent; in looking at them, one does not
see the scene which the machine translated into words. Yet the printouts are made
just as mechanically as any photographs are.
It is easy to say that the reason why we dont see through such mechanically
generated descriptions is that we dont see them as the scene they describe; perhaps we are incapable of seeing them this way. If one fails to see a photograph as
Dwight Eisenhower, or as a person, or as anything but a collection of blotches on
a at surface, we might deny that one sees Eisenhower through the photograph.
One doesnt see Eisenhower, perhaps, unless one notices him, in some appropriate
sense (although it isnt necessary to recognize him as Eisenhower or even as a
person). But this doesnt help without an account of seeing-as and an explanation
of why our not seeing the descriptions as the scene should make a difference. Nor
will it help to declare that only pictures, not representations of other kinds, can
be transparent. We need to know why the machines printouts dont qualify as
pictures and why nonpictures cant be transparent.
Investigating things by examining pictures of them (either photographs or
drawings) is strikingly analogous to investigating them by looking at them
directly and disanalogous to investigating them by examining descriptions of
them. One such analogy concerns what is easy and what is difcult to ascertain
and what mistakes the investigator is susceptible to. The numerals 3 and 8
are sometimes easily mistaken for each other. So when reading about a tree which
is actually 85 feet high, one might easily take it to be 35 feet high. This mistake
is much more likely than that of thinking it is 85.00001 rather than 85 feet
high. The reverse is true when we look at the tree directly or examine a picture
of it. A house is easily confused with a horse or a hearse, when our information
comes from a verbal description, as is a cat with a cot, a madam with a madman,
intellectuality with ineffectuality, and so on. When we confront things directly or
via pictures, houses are more apt to be confused with barns or woodsheds, cats
with puppies, and so forth.
It would be much too hasty to conclude that it is simply differences of this sort
which disqualify investigating a scene through mechanically generated descriptions as seeing it. Different mistakes are likely when we see under conditions of
dim illumination from those that are likely with bright illumination. (Colors are
especially hard to ascertain in dim light; outlines may be easier to distinguish then
than in bright light.) If there were such a thing as seeing-through-descriptions,



we should expect that the mistakes one is susceptible to when seeing in that
manner would differ from those one is susceptible to when seeing in other
ways. There is a deeper point to be madeone about perception in general, not
just vision.
There are important correspondences between the way we perceive (whether
directly or with photographic assistance) and the way the world really is (or the
way we think of it as being, but I will postpone this caveat temporarily). I do not
mean that the results of perception conform to facts about the world, that things
have the properties we perceive them to have. Nor do I mean that our percepts or
sense-data resemble what they are percepts or sense-data of. Rather, the structure
of the enterprise of perceiving bears important analogies to the structure of reality. In this sense we perceive the world as it is.
The mistakes a perceiver is susceptible to correspond to similarities among
things themselves. Things which are easily confusable perceptually, difcult to
discriminate, are things which really are similar to each other in some respect,
more similar than things which are less easily confusable. An 85-foot tree resembles one which is 85.0001 feet high more closely than it does a 35-foot tree.
Houses are more like barns and woodsheds than horses or hearses. Things with
different shades of red are more like each other (in color) than they are like green
things. In fact, the degree of similarity explains the likelihood of confusion. It
is because of the similarity between 85- and 85.00001-foot trees that they are
difcult to distinguish. The correspondence between similarity and perceptual
confusability is intrinsic, I suggest, to the notion of perception. A process of
discrimination counts as perceptual only if its structure is thus analogous to the
structure of the world. When we perceive, we are, in this way, intimate with
what is perceived. This goes a long way toward explaining our feeling of closeness to things which we see through photographs.
We are not similarly intimate with the world when we investigate it through
descriptions, even mechanically generated ones. Descriptions scramble the real
similarity relations. Houses are not much like horses or hearses. The difculty of
distinguishing a house from a hearse when we are reading about it is due not to
the nature of the house and hearses but to facts about the words used to describe
them. So we think of the words as getting between us and what we are reading
about, as blocking our view of it, in a way that photographs and sense-data do
not block our view of what they are photographs or sense-data of. The structure
of discrimination by means of mechanically generated descriptions does not correspond to the structure of the world and, so, does not qualify as perception.
Are things easily confusable in perception really similar in some respect? Scientic investigation may suggest otherwise. Perceived colors dont correlate precisely
with wavelengths of reected light. Environments which feel similarly or even
indistinguishably cold may differ considerably in temperature, with compensating
differences in humidity and wind. One might take this to mean that the correlation between how things affect us perceptually and how things are in themselves is



less than perfect. Or one might recognize propertiesvisible colors and perceived
coldwhich are distinct from wavelengths of reected light and temperature and
with respect to which the correlations do hold. In any case, we think of easily confusable objects as being similar, despite our awareness of the scientic facts. And
perhaps it is this that is intrinsic to perception. If scientic research should uncover
massive breakdowns in the presumed correlations and if, after reecting on these
results, we no longer even thought of easily confusable things as being similar,
I doubt that we would or should continue to speak of perceiving them.
Some question the very notion of real similarity. Resemblance is only a matter
of how we think of things, it is argued; similarity is relative to ones conceptual
scheme. In that case it will have to be what we think of as similaritieswhat
similarities there are relative to ones conceptual schemewhich corresponds
to difculty of perceptual discrimination. But this will sufce. We dont think
of houses as being especially similar to horses or hearses; so discrimination by
means of mechanically generated descriptions is not perceptual.
Why do we regard the things we do as being similar? Sometimes, I suggest,
precisely because they are easily confused (when examined in ways which otherwise count as perceptual). It is because visually discriminating among paint
chips of various shades of pink is relatively difcult that we think of them as
resembling each other. So facts about our discriminative capacities might be said
to create similaritiessimilarities relative to our conceptual scheme, which on
the present suggestion is the only kind that there isthereby establishing the
relevant correlations.31
It now looks as though mechanically generated descriptions could, in the
right circumstances, be transparent. Suppose that we used description-generating
devices regularly to investigate the world. Perhaps this would affect what we
think of as similarities, thereby changing our conceptual scheme. We might
recognize such properties as apparent-via-description-generating-devices houseness and
apparent-via-description-generating-devices hearseness and regard these properties as
analogous to visible colors, as characteristics of things themselves in virtue of
which they can be alike, not just as capacities to affect us through the devices. In
that case difculty of discrimination by means of description-generating devices
would be correlated with what we think of as similarities. So we might well
think of ourselves as seeing through the descriptions, andespecially if there is
nothing to real similarity among things except being thought of as similarwe

31. This seems to turn on its head our earlier suggestion that it is similarities among
things that make them difcult to discriminate perceptually. But we can have it both
ways. What count as similarities for us, what respects of resemblance there are relative
to our conceptual scheme, is determined (partly, anyway) by which discriminations are
easy to make and which are difcult, given our usual modes of (what otherwise count
as) perception. The fact that certain things are similar in these respects explains the difculty of discriminating them.



might really be seeing through them. Perhaps the mechanically generated

descriptions would then be transparent.
We are quickly becoming entangled in some of the deepest problems philosophy has to offer. Nevertheless, it should be clear from our recent speculations
that there are fundamental differences between pictures and descriptions of a
kind which plausibly allow mechanically generated picturesphotographsto
be transparent even though, apart from unusual circumstances like those just
imagined, mechanically generated descriptions would not be. This challenge to
the transparency of photographs is defused.
We have learned that perceptual contact with the world is to be distinguished
from two different sorts of nonperceptual access to it: access mediated by intervening descriptions as well as access via another person. The common contrast
between seeing something and being told about it conates the two. When
someone describes a scene to us, we are doubly removed from it; contact is broken both by the intervention of the person, the teller, and by the verbal form of
the telling. Perceptual contact can itself be mediatedby mirrors or television
circuits or photographs. But this mediation is a means of maintaining contact.
Viewers of photographs are in perceptual contact with the world.

What is photographic realism? Transparency is not the whole story. Realism is
a concept with many faces, and photography wears more than one of them. We
must not forget how adept photography is at portraying subtleties of texture,
shadow, and reection; how effortlessly it captures the jumbled trivia of ordinary life; how skillfully it uses perspective. The capacity of photography as it is
now practiced to reveal reality is especially important. Photographic evidence
is often very reliablehence its usefulness in court proceedings and extortion
plots. This is no automatic consequence of the mechanicalness of the photographic process, however. It derives rather from the fact that our photographic
equipment and procedures happen to be standardized in certain respects. (They
are not standardized in all respects, of course, so we have to be selective about
what conclusions we draw from photographs. We can usually say little beyond
gross approximations about the absolute illumination of a scene, for example, on
the basis of a photograph, since shutter speeds, lm speeds, and lens apertures
are so variable.)
But photographys various other talents must not be confused with or allowed
to obscure its remarkable ability to put us in perceptual contact with the world,
an ability which can be claimed even by a fuzzy and badly exposed snapshot
depicting few details and offering little information. It is thisphotographys
transparencywhich is most distinctively photographic and which constitutes
the most important justication for speaking of photographic realism.



Clarications and To Dos
Some commentators nd my transparency claim seriously counterintuitive,
and take this to be a substantial or even decisive count against it. I do not nd
it counterintuitive at all.1 So shall we argue about whether it really is counterintuitive? It would be better simply to recognize that intuitions are largely reections of ones currently internalized theoretical commitments (there being no
such thing as entirely pretheoretical intuitions), and that whatever authority one
accords them amounts to resistance to theory change simply because it is theory
change.2 Then we can look at the theory in question, the transparency thesis, and
assess it on its merits. Confusion or uncertainty about the content of the theory is
partly responsible for the divergence of intuitions. Clarication is in order.
Reliance on intuition is especially tempting if one takes the transparency thesis to be about the current meaning of the English word see.3 But it isnt.

1. Not after thinking through the reasons I offered in favor of it, and I dont think
I found it especially counterintuitive initially. Dominic Lopes and Eddy Zemach hold
that we see things by looking at paintings as well as photographs of them. See Lopes,
Understanding Pictures (New York: Oxford University Press, 1996), pp. 179193; Zemach,
Real Beauty (University Park: Pennsylvania State University Press, 1997), pp. 195196.
2. A certain minimal dose of this resistance is reasonable. Theories that we have internalized can be expected to have something going for them.
3. The thesis would be, presumably, that the current meaning of see is such that I see a
frog (for example) is literally true when I observe a photograph of one, true in the same
sense in which it is true that people see frogs directly or through mirrors or telescopes.
(Call this the linguistic thesis.) Speakers intuitions about what it is appropriate to say in
what circumstances, as well as what they do say when, are among the data that the theory
of which this thesis is a part must accommodate. (This doesnt require the theory to accept
their intuitions as true.) But ordinary speakers can hardly be expected to have intuitions,
let alone reliable ones, about the truth or falsity of the linguistic thesis, involving as it does
the esoteric notions of literalness and sameness of meaning. Fluency in the language doesnt
require an explicit grasp of these notions, however useful they might be to linguists and
philosophers of language in understanding, constructing theories about, the practices of
speakers and hearers. Nor is an assessment of the thesis to be read off in any simple way
from what speakers say when and what they judge to be appropriate utterances in what
circumstances. People do say things like I see a frog when looking at a photograph (or a
painting, for that matter) of a frog, and they sometimes characterize what is said (or meant,
or expressed) as true. Whether I see a frog is to be understood literally, in such cases,
and in the same literal sense in which people see frogs directly, is quite another matter.



My project is theory construction, not conceptual or linguistic analysis.4 At the

heart of the transparency theory is recognition of a fundamental respect in which
observing a photograph of something is like seeing it directly or in a mirror or
through a telescope, and unlike observing a painting of it. I can think of no more
natural way of expressing these relations than to say that we see things through
photographs, as well as through mirrors and telescopes and directly, and to
deny (in the same breath and the same spirit) that we see things through paintings of them. This is how I expect the person in the street would express the
point I am making, leaving it to linguistic theory to decide, if it can and wants
to, whether she is thereby adjusting the meaning of see. But as I insisted in
Transparent Pictures, I dont mind at all if one prefers to express the transparency claim differently, to say that we perceive, or schmee, or perschmeive, or whatever,
things through photographs and in the other ways, but not through paintings
and drawings.
I may have sown confusion by insisting that we see, quite literally, our dead
relatives themselves when we look at photographs of them (p. 86). This might
encourage the impression that my claim is about the current (literal) meaning
of the English word, although in the very next paragraph I explicitly deny that
it is. The point of this insistence, as I explained, was to rule out construing the
thesis as a claim that we see things through photographs only in a derivative
sense, one different from the sense in which we see things directly, that, for
example, we merely seem to see objects when we look at photographs of them.
To avoid thus watering down the thesis, I say that we see things through photographs, and see things directly, as well as through mirrors and telescopes, in a
(single) literal sense of see, without worrying about whether this is a new sense
of the word.

I made two different, independent claims in Transparent Pictures, both of
which can be stated without deciding whether to describe viewers of photographs as seeing the photographed objects. One of them is what I just characterized as the heart of the transparency theory:
(I) There is a natural kind which includes seeing photographs of things as well
as seeing them directly and through mirrors and telescopes, and so forth, but
not seeing handmade pictures of them.

4. I emphasized this especially in Looking Again through Photographs: A Response

to Edwin Martin, Critical Inquiry 12, pp. 805806. For more on these methodological
issues, see my AestheticsWhat? Why? and Wherefore? Journal of Aesthetics and Art
Criticism 65, no. 2 (April 2007), 147161.



There are other natural kinds in the neighborhood as well, of course, some of them
overlapping this one. (I) does not rule out the possibility that there is a smaller natural
kind, like the above except that it excludes seeing through photographs. I expressed
scepticism about this possibility, suggesting that the slope descending from direct,
face to face seeing is slippery all the way to (what I call) seeing through photographs; hence my challenge to readers to nd a signicant stopping place before the
photography cases. My second claim, less important than the rst, is:
(II) There is not a natural kind, not a very signicant one anyway, comprising
seeing things directly and through mirrors and telescopes, and so on, but
excluding seeing photographs as well as handmade pictures of them.

Most critics of Transparent Pictures target (II), leaving (I) untouched.5 They
attempt to meet my challenge, to nd grounds for classifying direct and mirrorand telescope-assisted seeing together which do not apply to seeing photographs.
Most of them also restrict the word see to the former cases, preferring to say
that we do not see things when we see photographs of them. But this terminological preference is irrelevant to the issue at hand, since the claim they are
objecting to is not about the word see.
I am not convinced that any of these attempts succeed.6 But if one or another
does, we will want our theory to mark the natural kind in question. One way to
mark it would be to say, as critics do, that we see things directly and through
mirrors and telescopes, but not through photographs. If this is our choice, we
will need another way of characterizing the larger natural kind specied in (I);
that would be just ne with me. But if (II) is right, it isnt clear what place see
might have in our theory (if we use it at all) except to mark this larger kind. So,
although neither (I) nor (II) needs to be stated in terms of seeing, both together
make it exceedingly natural to use see in the way I prefer. Hence my decision
to express (I), in conjunction with (II), by saying that we see things when we
see photographs of them. I did not, however, emphasize sufciently the distinctness and independence of (I) and (II).
Whatever the fate of (II), (I) stands. (I) by itself sufces to defeat the pervasive
and sometimes passionately defended idea that there is no fundamental difference
between photographs and handmade pictures, that photographers and painters

5. This includes the arguments of Carroll and Currie that I discuss in On Pictures
and Photographs: Objections Answered (this volume). Currie recognizes the distinction
between (I) and (II), and in effect endorses (I).
6. Some probably or certainly succumb to counterexamples (not always very obvious
ones)uncontroversial instances of direct or assisted seeing which turn out, on a given
proposal, to be classied with seeing photographs of things. Some have a distinct ad hoc
avor, xing on relatively unimportant grounds for separating seeing photographs of
things from uncontroversial ways of seeing them.



simply use different tools and techniques in creating their pictures. As I demonstrated in section 4 of Transparent Pictures, (I) is entirely consistent with the
obvious fact that both photographs and handmade pictures (most of them anyway) are created deliberately by persons and portray things in ways that express
the interests and attitudes of their makers. Transparent Pictures does an end
run around these and other considerations typically adduced in efforts to debunk
the natural impression that photographs differ fundamentally from paintings
and drawings. It accounts for and defends this impression by showing that seeing
photographs of things, in contrast to seeing handmade pictures of them, is very
much like ordinary instances of seeing thingshowever we choose to express
this fact, and whether or not (I) is true.


The transparency of photographs is not essentially connected to any thesis about
their epistemological value. In attributing this special kind of realism to them
I was not aiming to explain the supposed value of photographic evidence. A contrary impression may have been derived, on hasty reading, from my observation
early in Transparent Pictures that photographs are commonly thought to be
superior to other pictures as sources of information. And I did argue that the way
photographs inform viewers about photographed objects, when they do, is different from the way other pictures inform viewers about depicted objects when
they do. But nothing follows about which kind of picture is more informative or
more reliably so. Depending on the circumstances, either may be better evidence
for propositions of one kind or another than the other is. What is particularly
important, however, is that the special interest we have in photographs, their
putting us in what I called perceptual contact with objects, does not depend
on their providing us much if any information about them. Remembering that
transparency concerns direct object seeing, not seeing that such and such is the
case, may help keep it properly separated from epistemological matters.


We must not forget that photographs are pictures, depictions in my sense, as well
as being transparent.7 A rich area for further investigation that I addressed only
briey concerns the relations and interactions between the depictive nature of
photographs and their transparency, between their service as props in visual games
7. Most photographs, anyway. Given Patrick Maynards useful denition of photography as a technology for marking surfaces, some photographs are not pictures. Cf.
Maynard The Engine of Visualization: Thinking Through Photography (Ithaca, N.Y.: Cornell
University Press, 1997), pp. 3, 2021. Transparent Pictures is concerned only with
photographs that are pictures.



of make-believe and their role as aids to vision. This requires keeping clearly in
mind the fact that what viewers of a photograph imagine seeing is sometimes,
but not always, what they actually see through it. As Patrick Maynard puts it, a
photograph may be a photograph of one thing, but a depiction of something else.8
There is another twist that I did not mention in Transparent Pictures.
Observing a family photograph of Judy Garland, one imagines seeing her
directly, face to face, while actually seeing her indirectly, via the photograph. The
viewer of The Wizard of Oz imagines seeing Dorothy in the magical kingdom of
Oz directly, while actually seeing Judy Garland and the movie set via the photographic lm. But the moviegoer is likely also to imagine seeing Garland and the
movie set directly, face to face, as the viewer of the photograph does. (Imagining
this requires, I presume, that the thought of seeing them occurs to him, that he
not be absorbed exclusively in the world of Dorothy and Oz. His actual indirect
seeing of Garland and the movie set does not depend on his entertaining any
such thought; a person might see something even if she has no idea that that is
what she is seeing.) So the spectator may imagine seeing Garland and the movie
set face to face, and also imagine seeing Dorothy and the Land of Oz face to face.
These two imaginings are not, of course, acts of participation in the same game
of make-believe; the spectator doesnt recognize a single ctional world in which
Dorothy searches for the Wizard of Oz and Garland acts in a movie.
Salman Rushdie indicates a special case with further complications:
Staring at a photograph [a still from The Wizard of Oz], I realized I was not
looking at the stars at all, but at their stunt doubles, their stand-ins.9

Rushdie was actually seeing the stand-ins via the photograph, even before
he realized that he was. Once he realized this, he probably imagined seeing the
stand-ins directly. I suspect that he also imagined seeing the stars directly, after
his realization as well as before. (The stand-ins can be understood to be representing the stars representing the characters.) All along, I presume, Rushdie
imagined seeing Dorothy, the Tin Man, and the Cowardly Lion (directly).

Another topic eagerly waiting to be explored in light of the transparency thesis
is that of digital photography. As digital technologies increasingly replace

8. Maynard, The Engine of Visualization, p. 114.

9. Salman Rushdie, The Wizard of Oz (London: British Film Institute, 1992), p. 45.
Incidentally, it does not seem at all obvious to me that Rushdie doesnt mean that he is
literally looking at, and seeing, the stand-ins, in perfectly ordinary senses of these terms.
And it would strike me as much less natural to speak this way if Rushdie were looking at
a painting or drawing of the scene on the movie set, rather than the photograph.



lm, readers should be aware that Transparent Pictures focuses entirely on

the latter.
Whether and in what respects a digital image is transparent will depend
on what if any manipulation occurred in its processing, as it does in the case
of lm photographs. The important difference between the two technologies,
for our purposes, is the fact that electronic manipulation is vastly easier (and
easier to disguise in the nal product) than darkroom doctoring, and so is far
more common and far more likely in particular cases. Digital images, then,
are ordinarily more likely to be opaque, or opaque in a given respect, than
lm photographs. What is perhaps even more signicant is that digital technology makes it relatively easy to create illusions of transparency like that of
Chuck Closes Self Portrait, to fool viewersat least viewers whose experience
is informed primarily by familiarity with lm mediainto thinking they see
things via images when they dont, or to make it seem to them that they do
even when they know better.
But viewers catch on. Familiarity with digital techniquesespecially, I think,
practice digitally manipulating images ourselvesalters our experience. Not
only do we, on a cognitive level, become properly skeptical about the transparency of the images we see, our perceptual experiences of them change as well.
The illusion tends to vanish. It no longer seems to us, at least not as robustly as
it did, that we see objects through the images, even when we do. Our experience
of digital images becomes more like that of paintings.
The pervasiveness of digital techniques and our familiarity with them transforms our understanding and experience of traditional lm photographs as
well as digital ones, especially since we often cant tell and may not know
which method was used in a particular instance. Mathew Bradys Civil War
photographs are obviously old, having begun with exposures of reected light
on light sensitive surfaces more than a century before the development of digital photography. But a print of a Brady photograph on display today may
have been made by scanning an original to produce a digitized version, then
reproducing it on lm and printing it in a darkroom. Observing a given photographic print, we may have no idea whether there was a digital stage in its
processing, a stage in which it was highly susceptible to transparency-limiting
manipulation. Informed by implicit or explicit awareness of this possibility,
a viewers impression of seeing battleeld scenes via the photograph is likely
to be substantially diminished, compared to her impression of the same print
prior to the digital age.
There are also, of course, obvious but important effectscontingent ones,
largely independent of transparencythat the easy alterability of digital images
has on their reliability as evidence and their powers of persuasion. Digitization may either undermine a pictures credibility or enhance it, depending on
the circumstances. Manipulation can correct for misleading features of freshly
downloaded images, and we may sometimes have good reason to think that



the manipulator tried to do just that, and succeeded, even if the manipulation
renders the image opaque in certain respects. Again, the presence and prevalence
of digital images affect our experience and understanding of traditional photographs as well. In deciding what we might or might not learn from the latter,
we must now take into account the possibility that their processing included a
digital stage.

Objections Answered

early all lms are representational; more specically, they are visual or depictive
representations, pictures. And the vast majority of lms are photographic depictions.
The depictive and the photographic are two of the most fundamental categories that
need to be explained if we are to understand the medium of lm. I have elsewhere
offered an account of depiction: picturesboth still pictures and moving onesare
props in visual games of make-believe. By this I mean, in part, that in looking at
a picture the spectator imagines seeing what it portrays.1 I have also argued that
photographs are special among pictures in that they are transparent: to look at a photograph is actually to see, indirectly but genuinely, whatever it is a photograph of.2
Both of these claims have elicited a wide variety of reactions, some sympathetic and some sceptical. Nol Carroll and Gregory Currie are among the more
thoughtful sceptics, and have discussed these matters with special attention to
the medium of lm. In what follows I consider their objections and examine
alternatives that they propose. I should add that Currie endorses the main features of my theory of representation and incorporates them in his own. My focus
now is on points of disagreement.

The question of what pictures are, and in particular how they differ from descriptions, is a lot more difcult than it seems. Given how obvious it is that a picture
1. Kendall L. Walton, Mimesis as Make-Believe: On the Foundations of the Representational
Arts (Cambridge, Mass.: Harvard University Press, 1990).
2. Kendall L. Walton, Transparent Pictures, Critical Inquiry 11: 2 (1984), 246277
[reprinted as chapter 6 of this volume]. See also Looking Again through Photographs: A
Response to Edwin Martin, Critical Inquiry 12 (Summer 1986), 801880.



of a mountain and the word mountain are animals of very different kinds, the
nature of the difference is disconcertingly elusive. Many attempts to account for
it manage to do little more than point to the fact that there is a difference.3 This
much seems clear, however: there is something especially visual about pictorial representation and depiction of other kinds. Currie observes that painting,
theater, and lm are visual media.4 But how so? Pictures are to be looked at. But
so are written words, and we use our eyes on graphs and diagrams as well.
The answer lies in the particular nature of the visual experiences that pictures
provide. What is distinctive about these experiences shows in our ways of talking
about them. We speak of seeing an ox when we look at a picture of an ox. We
may point to the area of the canvas that our eyes are xed on and say, There is
an ox there, or, That is an ox. Remarks like these are not appropriate when we
look at the word ox or at a written description of an oxthe Blue Ox in the
Paul Bunyan stories, for instance.
Such remarks, when one is observing a picture, are not to be taken literally. To
look at a picture of an ox is not actually to see an ox, and it is (virtually always)
obvious to the viewer that he is not seeing one.5 The person who points to the
canvas saying that is an ox knows that there is only a piece of canvas there.
Nevertheless, seeing a picture of an ox involves thinking of oneself as looking
at an ox. We can put this by saying that one imagines seeing an ox, as one looks
at the picture.6 I do not mean that one deliberately undertakes to imagine this.
Rather, one nds oneself imagining it, more or less automatically, as a result
of perceiving the picture. In watching a lm, the images on the screen induce
spectators to imagine seeing the characters and events that are portrayed. And we
imagine seeing them from a certain perspective or point of view, one determined

3. This is true of many traditional attempts to dene depiction in terms of visual

resemblance. The point is not that visual resemblance is not part of what makes pictures
pictures, but that the main work of devising an informative account consists in specifying what kind of resemblance is involved and how it is involved. I hold that the relevant
kind of resemblance is to be explained in terms of imagining or make-believe. And
nothing is lost by simply explaining depiction in terms of imagining or make-believe,
without mentioning resemblance.
4. Gregory Currie, Image and Mind: Film, Philosophy and Cognitive Science (Cambridge:
Cambridge University Press, 1995), 169, 181.
5. Unless it is a photograph of an ox (as well as a photographic picture of one). But
I see an ox said while observing such a photograph, understood in one natural way, is
not literally true.
6. Imagining seeing an ox is not reducible to imagining that one sees an ox. Cf.
Walton, Mimesis as Make-Believe, 1.4 Imagining about Oneself. Carroll neglects this
distinction, and Currie explicitly collapses it: I take it to be a distinctive thesis of classical lm theory that cinema encourages a certain kind of imagining which I have called
imagining seeing: imagining that you are seeing the ctional events of the lm, and
seeing them from the point of view of the camera. (Image and Mind, 168)



by the position of the camera, or rather by features of the screen images that
result from the position the camera was in when the lm was photographed.
This is not the whole story. Words sometimes stimulate readers to imagine
seeing. A reader of the Paul Bunyan stories might call up a visual image, and
might describe his experience by saying, I see an ox. But this comment (taken
in a non-literal way, of course) does not characterize an aspect of the readers
visual experience of the text, as it does ones experience of a picture of an ox;
the readers eyes might be closed when he calls up the image. And it will not be
appropriate for the reader to point to the words on the page and say (in the same
spirit that one does while pointing to the picture), An ox is there, or, That is
an ox. Ones perception of the text is merely a cause of an experience involving the thought of seeing an ox. In the case of picture perception, not only does
looking at the picture induce us to imagine seeing an ox, we also imagine our
actual visual experience, our perceiving the relevant part of the canvas, to be an
experience of seeing an ox.
Carroll and Currie are concerned with the rst part of my account, the idea
that the perception of picturesmoving pictures in particularinvolves imagining seeing.
Currie claims that if viewers were to imagine themselves to be seeing the things
and events portrayed by a lm, they would have to imagine being in various
bizarre situations, undergoing strange and unlikely transformations, and enjoying magical modes of access to the ctional happenings.7 We do not ordinarily
engage in these latter imaginings when we watch movies, he thinks, so we cannot ordinarily be imagining seeing what they portray. In watching a sequence
of shots from different points of view we will imagine seeing from different
points of view. According to Currie, this means that the spectator will have to
imagine jumping around from one place to another. In the scene in Hitchcocks
The Birds in which Melanie Daniels crosses Bodega Bay in a hired boat, the
transitions between the rst three shots would require [the spectator] to imagine
her position shifted instantly through ninety degrees twice, around the edge of
the bay. . . . The transitions between 5 and 14 would then have her imagine herself shifting back and forth nine times between Melanies own position . . . and
different points on the shore, all within the space of a minute or two.8
This is simply not so. Nothing is easier than, rst, to imagine watching a boat
from the shore, and then to imagine observing it from on board, without ever
imagining moving or being transported somehow from the shore to the boat.
One just imagines the one visual experience, and then imagines the other. In
deciding where to go to observe a launching of a space shuttle you might imagine watching the moment of blast-off from one vantage point, and then imagine
7. Currie, Image and Mind, 172.
8. Currie, Image and Mind, 177.



watching it from another. There is no requirement that you imagine being in

both places at once. The principle here is simple: one need not imagine the conjunction of other things that one imagines. Imagining p and imagining q does
not have to involve imagining that both are true. If viewers of The Birds imagine
watching the boat in Bodega Bay from the shore, and then, in the next shot,
imagine being on the boat, it does not follow that they must ever have imagined
even having been in both places, let alone moving from one place to the other.
Even if the spectator does imagine being on shore at one moment and on board
the boat a moment later, this does not require imagining moving or being transported from the shore to the boat, imagining changing locations. One need not, in
ones imaginative experience, follow out the implications of what one imagines.9
This is especially obvious if, as Currie thinks, imaginings are necessarily occurrent
mental events, if there is no such thing as dispositional imagining.10 Obviously,
when thoughts occur to us, many of their consequences do not.
Currie concedes that one neednt imagine the logical consequences of what
one imagines,11 effectively undercutting many of his objections to the idea that
in watching a lm one imagines seeing what is portrayed. He contends that if in
watching The Birds the viewer imagines seeing from the perspective of the camera, one of the shots would require her to imagine herself suddenly in the water
by the boat. This wrongly assumes that she is required to work out the consequences of what she imagines. The viewer is likely not even to consider the
question of where one would have to be in order to see the boat from the point
of view she imagines seeing it from.12 (Also, the fact that a sudden change in the
images on the screen induce spectators suddenly to imagine a certain event does
not mean that they must imagine the event occurring suddenly.)
Do I imagine myself on the battleeld, mysteriously immune to the violence around me? Currie asks rhetorically. The answer is: probably not. One just
imagines seeing the violence of the battle from a particular perspective, if that is
what the lm portrays.
In the case of subjective shots, shots portraying the view of a character in
the ction, Currie thinks that on the imagining seeing hypothesis the spectator would have to identify himself with the character, to imagine that he is the
character.13 This is not so. We imagine seeing things from a certain point of

9. Cf. Walton, Mimesis as Make-Believe, especially ch. 4.

10. Currie, Imagination and Simulation, in Martin Davies and Tony Stone (eds.),
Mental Simulation: Evaluations and Applications (Oxford: Blackwell, 1995), 160.
11. Currie, Image and Mind, 177.
12. Given the propositions that a work initially induces us to imagine, which of their
consequences are we also to imagine and which of them can we ignore? The answer varies
from medium to medium and genre to genre. We should not expect a simple systematic
formula. Cf. Mimesis as Make-Believe, 165166.
13. Currie, Image and Mind, 174176.



view, noticing certain aspects of them, and so forth. And we understand that
what we imagine seeing is what ctionally the character sees; we imagine that
that is what the character sees. The lm thus shows us what the characters visual
experience is like. This need not involve identifying with the character, imagining ourselves to be him.
Nevertheless, we do, sometimes, identify in one sense or another with characters, empathize with them, and subjective shots often encourage such identication.14 I leave open the question of whether this involves imagining oneself
to be the character. It certainly involves simulating the experience of characters,
including their visual experience. And even when we do not empathize with the
prominent attitudes and objectives of a character (when in watching subjective
shots from the point of view of a homicidal maniac our sympathies are with the
victim, for instance15), there is no reason why we shouldnt simulate their visual
experiences. Subjective shots induce exactly this kind of simulation.16
If we are to imagine ourselves seeing ctional things and events when we
watch a lm, we shall have to imagine that our visual powers are strangely
restricted, and that what I see depends in no way on our own decisions. The
camera is often placed to restrict our view of the action for dramatic purposes,
and in these cases one would often like to see more or see differently. But if we
are imagining ourselves to be seeing the ction itself, what are we to imagine
concerning the source of this restriction?17 There is a confusion here between
restrictions on what one imagines seeing, and imagining there to be restrictions
on what one sees.18 If what I imagine seeing depends not on my decisions but on
the images projected on the screen, this does not force me to imagine that my
decisions have nothing to do with what I see. Nothing prevents me from imagining that I could see something different by looking in a different direction or
stepping around an obstruction, even if the screen images guiding my imaginings do not allow me to imagine actually doing so. And if I should imagine that
my vision is restricted, I neednt imagine anything at all about why or how this
is so, about the source of the restriction. Nor must I imagine that the restriction
is a strange one.
Currie argues for one exception to the idea that one need not imagine the logical consequences of what one imagines: You cannot imagine, of a certain scene
represented to you onscreen, that you are seeing it, but not that you are seeing
it from any point of view. This is so, he says, because the concepts of seeing

14. Cf. Walton, Mimesis as Make-Believe, 3234, 255.

15. Currie, Image and Mind, 176.
16. Currie himself says that in watching Stagecoach we see Dallas the prostitute from
Lucys point of view (Image and Mind, 175, my italics).
17. Currie, Image and Mind, 172173.
18. Cf. Walton, Mimesis and Make-Believe, 359, and 6.4 Restrictions on Participation. The restrictions I discuss obviously do not make it ctional that one is restricted.



and of point of view are linked more intimately than by entailment alone. His
point is obscure. To see is to see from a point of view, he says; there is no such
thing as nonperspectival seeing.19 Lets agree. And let us grant, for the sake of
argument, that the connection is so intimate that one cannot imagine seeing
without imagining that there is a point of view from which one is seeing, that
ones seeing is perspectival. This is innocuous.20 But Currie seems to mean something much stronger: that to imagine seeing is necessarily to imagine seeing
from some particular perspective, that is, that there must be a particular perspective such that one imagines seeing from that perspective. This is certainly not
sonot at least if, as Currie seems to think, to imagine seeing is just to imagine
that one sees. I can easily imagine that I see something without there being a
particular point of view which I imagine that I see it from. I need not imagine
that I see from above, or that I see it from the side, or that I see it from nearby,
or from afar.
I argued that imagining seeing does not reduce to imagining that I see.21 And
perhaps imagining seeing, or the kind of imagining seeing depictions provoke,
is imagining seeing from one or another particular perspective.22 Even so, there
is no paradox. What Currie nds strange is not the idea that spectators of lms
imagine seeing from a given perspective, but the idea that they imagine certain
consequences of their seeing from the perspective in question: their being in the
water, being immune to bullets, changing position, and so on. Imagining seeing
from the relevant perspective does not require imagining these consequences.
The perspective might be dened simply as being a certain approximate distance
and direction from the object seen. And it is likely that the imaginer, or a person
who actually sees something from a certain perspective, cannot specify in words
even the distance and direction from which he sees or imagines seeing.23
19. Currie, Image and Mind, 178. Italics in original.
20. It is also questionable, especially if Currie is right in assuming that all imagining
is occurrent. Must one who imagines seeing be imagining occurrently that his seeing is
perspectival? (Not imagining that ones seeing is perspectival is not the same as imagining that ones seeing is not perspectival, of course.)
21. See note 6.
22. But I am skeptical. See Walton, Mimesis as Make-Believe, 8.7.
23. Curries claim of a more-intimate-than-entailment connection between seeing and
points of view is part of an argument against an Imagined Observer Hypothesis weaker
than the one he attributes to the classical lm theorists, Panofsky and Balzs in particular. The weaker alternative postulates a kind of purely visual imagining, unconnected
with any imaginings about where we are seeing from or how it is that we are able to
see, one that does not involve imagining ourselves placed anywhere in the scene, or as
undergoing any changes of position (Image and Mind, 177). The mistake here is in thinking that an alternative that is weaker in the ways mentioned would have to postulate an
incoherent imagined nonperspectival seeing (169). It is not clear that, on a charitable
reading, Belasz and Panofsky have to be understood as holding the strong view he
outlines, rather than a more reasonable one of the kind I have been defending.



Currie presents another objection: Suppose I am watching a movie in which

the murderer enters unseen. On the imagining seeing hypothesis, he thinks, it
would have to be true that I imagine that there is an unseen murderer which
I see. But it is implausible to suppose that in the case described the audience
is called on to imagine something contradictory. I refer the reader to Mimesis as
Make-Believe, where I considered exactly this objection.24 What I say here will be
brief, and is not meant to replace my discussion there.
In the rst place, it is simply a mistake to suppose that if the spectator imagines seeing something and also imagines that that thing is not seen by anyone,
he must be imagining a contradiction. Imagining p and imagining not-p do not
entail imagining the conjunction, p and not-p. Curries hypothetical movie might
be understood to induce viewers to imagine seeing a murderer creeping silently
into a building, and to induce this imagining as a way of indicating what the
viewers are to imagine occurring without being seen. If, understanding this, they
do imagine that the murderer is unseen, they neednt ever have imagined that
the murderer is both seen and unseen.
The main point to be noticed, however, is that appreciators regularly are
required to imagine incompatible and otherwise conicting propositions, in any
case, even apart from any imagining seeing, and that they do imagine these
propositions, ordinarily, without feeling any particular tension or sense of paradox.25 The idea that the spectator imagines seeing things which he also imagines
to be unseen introduces no special difculties, and constitutes no reason to reject
the imagining seeing hypothesis.
It is a commonplace that dreams often contain paradoxeswhat on reection,
on awakening, we recognize as paradoxeswhich are not felt as such while we
are dreaming. Here is one example:
Last night I had a dream . . . Mrs. Terry . . . told us that Marion and Florence were
at the theatre, the Walter House, where they had a good engagement. In that
case, I said, Ill go on there at once, and see the performanceand may I take
Polly with me? Certainly, said Mrs. Terry. And there was Polly, the child,
seated in the room, and looking about nine or ten years old: and I was distinctly
conscious of the fact, yet without any feeling of surprise at its incongruity, that
I was going to take the child Polly with me to the theatre, to see the grown-up
Polly act! Both picturesPolly as a child, and Polly as a woman, are, I suppose,

24. See especially 6.6 Seeing the Unseen, which relies on principles developed in
4.5 Silly Questions, and elaborated elsewhere in Mimesis. My discussion in Mimesis
was in response to an early statement of the objection by Nicholas Wolterstorff, directed
to an earlier presentation of my account of depiction.
25. It is not hard to nurture a sense of paradox, however, even cases like that of the
unseen murderer. One can make them feel like M. C. Eschers prints and some absurdist
stories which emphasize the conicts in what they ask us to imagine.



equally clear in my ordinary waking memory: and it seems that in sleep I had
contrived to give the two pictures separate individualities.26

In Mimesis I described numerous incongruities in what we imagine in appreciating the most ordinary representational works of art and discussed various ways of
treating them, incongruities that involve only imaginings that Currie will surely
allow. What is important is that no additional theoretical resources are needed to
accommodate (apparent) incongruities arising from viewers imagining seeing,
and we should be neither surprised nor dismayed that there are such.
If imagining seeing is not the key to the notion of depiction, what is? Nol
Carroll thinks there is an easy way of understanding pictorial representation,
without invoking make-believe or imagining seeing. Pictorial or depictive representations are those whose subjects we recognize by looking (rather than by
reading or decoding).27 This is not a solution. It is by looking that we recognize
what (written) names and descriptions refer to. Yes, we read them, so this must
not be the kind of recognition by looking Carroll has in mind. But what is
the difference? This is itself the heart of the problem. What is it to perceive pictures and how does picture perception differ from reading? (It is not a matter of
ascertaining what is portrayed immediately or noninferentially or automatically.
Reading is often immediate, noninferential, and automatic. Nor is it a matter of
how much or what kind of training is required; we want to know how picture
perception and reading themselves differ, regardless of how they or the capacities
for them may have come about.)
Gregory Curries answer is more elaborate, but no more successful. What
makes the experience of cinema, painting and the other pictorial media an essentially visual one is that it gives rise to perceptual imaginings. Poetry and the
novel, on the other hand, give rise to symbolic imaginings28 What are perceptual imaginings? We imagine that things have a certain appearance, when
we see a lm or picture of them, he says. But of course we will imagine this on
reading a novelists description of their appearance as well. Three further characteristics are supposed to make the imaginings elicited by paintings and lms
perceptual: (a) We imagine things possessing certain clusters or bunches of
features, corresponding to those we might perceive something as possessing
both color and shape, for instance, rather than just one or the other. By contrast,
if we had been reading a novel we might, at a certain point, have read something that prompted us to imagine that the characters eyes were blue. [But] we

26. Lewis Carroll, quoted in S. D. Collingwood, The Life and Letters of Lewis Carroll.
(1899). From Stephen Brook, The Oxford Book of Dreams, 202203. Italics in original.
27. Nol Carroll, Critical Study: Kendall L. Walton, Mimesis as Make-Believe,
Philosophical Quarterly 45: 178 ( Jan. 1995), 97.
28. Currie, Image and Mind, 184.



would be in no position to imagine anything about their shape, because we are

told nothing about shape. (b) We imagine very specic features. When we see
the screen we imagine that the characters eyes are exactly that shape, that colour,
and that size in relation to the characters other features, whereas if we were
reading a novel we would be in no position . . . to imagine that the [characters]
eyes are some specic shade of blue. And (c) our imaginings are sensitive to
ne variations in the perceptual qualities of the stimulus (e.g., the lm image).
If what we saw on the screen were shaped or coloured in a slightly different
way, what we would then have imagined about the characters features would
have been correspondingly different. Written descriptions are not similarly sensitive, he says. And some features are irrelevant: It would be a matter of indifference to our imaginings, moreover, whether the text was composed in this type
face (or size, or colour) or that one.29
The imaginings elicited by novels and stories do not differ as sharply as Currie suggests from those elicited by pictures. And differences of the kinds he
describes do not begin to account for the distinctively perceptual nature of our
experiences of pictures. A monochrome drawing or print will prompt us to imagine the shape of a persons eyes but tells us nothing about their color. Novels
often do describe both the shape and color of a characters eyes, and we could
introduce words specifying both shape and color (e.g., blare = blue square)
without making descriptions containing them the least bit pictorial. There are
words and phrases specifying particular shades of color (burnt sienna, the
color of x), and a description of a character as being exactly ve feet eleven and
one-quarter inches tall requires the reader to imagine his height more precisely
than a picture is likely to. If typefaces, sizes, or colors of linguistic symbols, or
ne differences in their shapes, were semantically signicant and so affected what
appearance the reader of a novel is supposed to imagine a character having, it is
hard to believe that the experience of reading would thereby be even slightly
more like the experience we describe as seeing a man, which perceivers of a
picture of a man enjoy.30
The inadequacy of Curries account of depiction is especially apparent when
he tries to explain the fact that pictures depict things from certain points of
view or perspectives, without endorsing what is surely the most natural way
of doing this: understanding a picture to induce an experience one thinks of as
seeing them from that point of view. Currie thinks that a picture, a cinematic
image of a man, for instance, induces perspectival imaginings, but he means
by this no more than that it induces spectators to imagine that the man has a

29. Currie, Image and Mind, 184. See also pp. 182183.
30. Curries account fails for much the same reasons that Nelson Goodmans does,
with which it has considerable afnity. But Goodman doesnt claim to be dening a
distinctively visual or perceptual notion of representation.



certain appearance from a certain perspective.31 Beliefs can be perspectival in the

same sense, as Currie insists. One may believe that Uncle Albert appears thus
and so from a certain perspective. Holding this belief is nothing like the kind
of perspectival perceptual experiences we have when we look at pictures and
other depictions, and neither does imagining that Uncle Albert has a particular
appearance from a certain perspective. Perspectival imagining, in Curries weak
sense, doesnt begin to account for these experiences.

I turn now to photography, and to my claim that photographic pictures are
The idea that photographs have a mechanical connection with what they are
photographs of, that they differ fundamentally in this respect from drawings,
sketches, and paintings, which are humanly mediated, and that because of this
photographs somehow put us in closer contact with the world than handmade
pictures do, has been a constantly recurring theme in discussions of photography. It persists in the face of determined objections, and despite the difculty
of spelling it out coherently. We hardly need to be reminded that most photographs, like pictures of other kinds, are made by people, and that they reect the
photographers interests, desires, vision, and so on. Dont photographs, like other
pictures, put us in contact, in the rst instance, with a human beings conception of reality, rather than reality itself? Isnt photography just another method
people have of making pictures, one that merely uses different tools and materialscameras, photosensitive paper, darkroom equipment, rather than canvas,
paint, and brushes? And dont the results differ only contingently and in degree,
not fundamentally, from pictures of other kinds? I answered that the difference is
indeed fundamental, that (with some qualications) photographs are transparent
and handmade pictures are not, and that this difference is entirely compatible
with the fact that photographs, like paintings, result from human activity and
reect the picture makers interests, intentions, beliefs, and so on.
Nol Carroll and Gregory Currie misconstrue the transparency thesis in one
important respect. Both take transparency to be incompatible with representation. According to Carroll: [Walton denies] that [documentary] photographs
are representations, preferring to think of them as prosthetic devices, like binoculars, that enhance our ability to see whatever they are photographs of. So for
Walton there is no imagining seeing when it comes to this sort of photograph.32
My position is that photographs, documentary photographs included, induce
imagining seeing and are representations (depictions, pictures), in addition to

31. Currie, Image and Mind, 188.

32. Nol Carroll, Critical Study, 97. See Currie, Image and Mind, 5051, 71, 72.



being transparent.33 In viewing a photograph of a class reunion, for instance, one

actually sees the members of the class, albeit indirectly via the photograph, but
at the same time one imagines seeing them (directly without photographic assistance). In the case of non-documentary lms, what we actually see (the actors
and the movie set) may be different from what we imagine seeing (the characters, a murder, a chariot race). As I emphasized in Transparent Pictures, the
combination of actual and imagined seeing, and interaction between the role of
photographs as aids to vision and their role as representations, is one of photographys most important and intriguing characteristics. To construe transparency
as excluding imagining seeing is to miss out on it completely.
The question, then, is not whether photographs are representations, pictures,
but whether they are pictures of a special kind. Currie agrees with most of what
I said on this score. He agrees that photographs differ from other pictures in
the respect which, I hold, prevents the latter from being transparent. Photographs are counterfactually dependent on the scenes they portray: if the scene
had been different the photograph would have been different. The same is often
true of paintings, in particular when the artist painted from life aiming to portray accurately what he saw. Butthis was my main pointa painting from life
depends counterfactually on the scene because the beliefs of the painter depend
counterfactually on it. The counterfactual dependence of a photograph on the
photographed scene, by contrast, is independent of the photographers beliefs. It
is because a difference in the scene would have affected the painters beliefs about
what is there, that it would have made the painting different. But a difference
in what is in front of the camera would have made the photograph different even
if it didnt affect the photographers beliefs. The painter paints what he thinks
he sees. The photographer captures with his camera whatever is in front of it,
regardless of what he thinks is there.34
Currie concurs with all of this. As he puts it, photographs have natural
counterfactual dependence on the photographed scenes, whereas handmade
paintings possess only intentional counterfactual dependence on what they
portray. He also agrees that this makes for a signicant similarity between
seeing a photograph of something and seeing the thing itself in the ordinary

33. See Walton, Transparent Pictures, 85, 88 (this volume), and Mimesis as MakeBelieve, 88, 330, 331 n. Patrick Maynard makes a good case for regarding certain things
as photographs but not pictures. Cf. The Engine of Visualization: Thinking Through Photography (Ithaca, NY: Cornell University Press, forthcoming [1997]). I was concerned only
with what he calls photographic pictures.
34. This paragraph is a gloss on my more precise formulation in Transparent Pictures, 262265. Carroll missed the point about counterfactual dependence. Paintings
are not counterfactually dependent on the objects they depict, he says. (Nol Carroll,
Towards an Ontology of the Moving Image, in Cynthia A. Freeland and Thomas
E. Wartenberg [eds.], Philosophy and Film [New York: Routledge, 1995], 70).



manner, and a respect in which both differ from seeing a painting of the thing.
Ones visual experience is naturally counterfactually dependent on the thing
when one either sees it directly or sees a photograph of it, but not when one sees
a painting of it.35
What Currie disagrees with is my decision to bring out this similarity by
regarding viewers as actually seeing things when they see photographs of them.
In saying this I was not especially concerned to be faithful to the ordinary sense
of the word see (if there is such a thing). So I could almost declare victory at
this point, as far as Currie is concerned, and dismiss the remaining disagreement
as terminological rather than substantive. But my transparency claim reects
more than just the natural counterfactual dependence of photographs on photographed scenes. I did not settle on an account of what it is to see something.36
But even without such an account, comparisons with other aids to vision show
how natural it is to think of seeing a photograph as a way of seeing the thing, as
akin to seeing it directly.
In Transparent Pictures I presented a challenge: we see things in (through)
mirrors. We also see them with the aid of telescopes and microscopes. Why
not regard live television as an aid to vision as well? And if we do, it is hard
to see why we should not regard photography similarly. There are differences
among these various means of access to things, of course. The challenge is to
specify a difference that justies denying that we see via some of these devices
but not others, one that allows mirrors, telescopes, and microscopes, at least, to
be transparent, as surely we must, while excluding photography. I gave reasons
in Transparent Pictures for taking the difference just outlined between photographs and handmade pictures as a reason to draw the line between them, to
deny that the latter are transparent while allowing that the former are. I have
not seen a compelling rationale for drawing the line earlier, somewhere between
mirrors and photography.
Carroll and Currie take up the challenge. With ordinary seeing we get information about the spatial and temporal relations between the object seen and
ourselves. . . . Call this kind of information egocentric information. . . . Photographs, on the other hand, do not convey egocentric information.37 I can orient
my body spatially to what I see, either with the naked eye or through a telescope

35. Currie also reiterates my reasons for dismissing as irrelevant several considerations
that are often adduced against the transparency thesis and the idea that photographs are
special (Image and Mind, 5658).
36. Neither the natural counterfactual dependence of ones visual experiences on
objects, nor that plus a correspondence between similiarity relations and the likelihood of
discriminatory errors, is sufcient for seeing an object. So I do not subscribe to the account
of transparency Currie outlines in Image and Mind, 63. I did make some tentative suggestions about what else is required. Cf. Looking Again through Photographs, 804.
37. Currie, Image and Mind, 66.



or microscope. But when I see a photograph I cannot orient my body to the photographed objects. The space of the objects is disconnected phenomenologically
from the space I live in.38
There can be no doubt that an important function of vision in human beings
is to provide information about how things we see are related temporally and
spatially to us. The ability to see evolved in humans and other animals primarily, no doubt, because such information is so important for survival.39 But why
suppose that seeing occurs only when this function is actually served? If the
capacity to feel pain evolved in humans and other animals mainly as an indicator
of damage to the body and as a device to prevent behavior that would exacerbate
the damage, it certainly does not follow that pain felt when there is no bodily
damage (e.g., when neural pain receptors are stimulated articially) is not really
pain. And the egocentric information that is important to survival is primarily
information about an organisms immediate surroundings, yet the capacity that
has evolved allows us to see stars and other remote objects. An account of what
it is to see should explain how seeing enables organisms to acquire information
about their environment. There is no reason to assume that it must limit seeing
to cases in which that is done.
Carroll and Currie agree that mirrors are aids to vision, that we literally see an
object when we see it in or through a mirror. Consider an array of mirrors relaying the reection of a carnation to a perceiver. Suppose that it is not evident to
the perceiver how many mirrors are involved or how they are positioned, so he
has no idea what direction the carnation is from him or how far away it is. Does
he see the carnation through the mirrors? Surely he does.
Currie bravely bites the bullet. Although we normally see through mirrors,
Currie claims, when there is a confusing iteration of mirrors, such that egocentric spatial information is lost, we do not.40 Carroll will have to agree. I do not
speak of literally seeing the objects in question, he says, unless I can perspicuously relate myself spatially to themunless I know where they are in the space
I inhabit.41
If this result is not bad enough, consider a variation of the example. Suppose
I see a carnation in the ordinary way, right in front of my eyes. But suppose that
there are lots of mirrors around, or I suspect that there are. None of them actually affects my perception of the carnation, but I cannot tell that they do not;
I think I may be seeing the image of a carnation reected in one or many mirrors.

38. Nol Carroll, Ontology, 71. See also Nol Carroll, Critical Study, 9798. Nigel
Warburton made a similar point in Seeing through Seeing through Photographs,
Ratio, ns 1 ( June 1988), 6474.
39. See Nol Carroll, Theorizing the Moving Image (New York: Cambridge University
Press, 1996), 6263.
40. Currie, Image and Mind, 70.
41. Nol Carroll, Ontology, 71.



So I have no idea where the carnation is in relation to me. Currie and Carroll are
forced to deny that I see the carnation at all!
This does not exhaust the peculiar blindness that the Carroll-Currie conception
of seeing would induce in us. Curries blanket contention that photographs supply no egocentric information at all goes too far. He does acknowledge, in a footnote, that photographs can serve, along with information from other sources, in
an inference to egocentric information.42 But he insists that this doesnt count,
that it doesnt render the photographs in question transparent. Now, however, it
will be hard for him to make room for the facta fact he endorsesthat single,
simple mirrors are transparent, in normal instances. The person who sees something in a mirror is likely to know where the reected object is in relation to
him, but only by relying on information from other sources. He must take into
account facts about the reective properties of mirrors. The spatial orientation of
the mirror through which he sees is crucial, and this may not be apparent from
his current visual experience. He may not be able to see even that there is a mirror; he may not see its edges or any other sign of it. And if the mirrors edges are
visible, their signicance is clear only in light of other background information,
information he acquired from previous experience. Moreover, when no mirrors
are involved or even suspected, when, in the most ordinary of cases, I see a carnation in front of my eyes, my egocentric knowledge that it is there depends
on the realization that I am not seeing it through a mirror. It is a commonplace
that what we learn from perception, in general, egocentric information included,
depends on a wealth of background information not available from the perception itself, information by means of which we interpret perceptual cues.
Yes, the background information may be internalized, rather than consciously
appealed to; the perceiver may not need to pay attention to the cues and explicitly draw the inference.43 But this is true in the case of photographs, as well as
those of mirrors and ordinary unassisted seeing. And it doesnt matter whether or
not conscious inferences are made, anyway. If I must consciously gure out that
an object I see is in front of me, or behind me, using the information that there
is or is not a mediating mirror, I am not seeing that it is where it is. But I am still
seeing it. The transparency thesis is a thesis about direct object seeing, not about
seeing that something is the case. When I see a photograph of a carnation I see
the carnation, whether or not I see that it bears such and such spatial relations to
me. (Seeing an object may require seeing that something is true of it; this may follow from a requirement that to see something is in a relevant sense to recognize
it.44 But it does not require seeing that egocentric facts obtainfacts about its
spatial relation to oneself.)

42. Currie, Image and Mind, 66.

43. See Walton, The Dispensability of Perceptual Inferences, Mind (1963), 35767.
44. Walton, Looking Again through Photographs, 804.



The Carroll-Currie proposal amounts to ad hoc linguistic legislation, although

it was not intended that way. So understood it is not pointless. See, in Carrolls
and Curries unusual sense, does at least mark out an important class of cases
(not a sharply delineated one, and not one that separates photographs neatly
from other pictures). But the same is true of see as I understand it. Carroll and
Currie have provided no reason for preferring their construal of the word to one
on which (most) photographs are transparent.
My proposal may or may not be a departure from ordinary language, but if
it is it is an especially natural one. We do speak of seeing Uncle Fred when we
see a photograph of him. Sometimes we say things like this in the same spirit in
which we speak (nonliterally) of seeing Fred while looking at his painted portrait. (Both are pictorial representations.) But sometimes the spirit is very different. In explaning why the Star tabloid planted a photographer on a neighboring
rooftop to catch political operative Richard Morris with a call girl, news editor
Dick Belsky remarked, We wanted to see it with our own eyes.45 A sketch of
the liaison, even by the most credible artist-reporter, would surely not have satised the desire Belsky expressed.
Who is to have proprietary rights to the word see is not the issue, however.
My proposal was meant to bring out the important similarities and differences
that I sketched above (and explained more fully in Transparent Pictures)
especially the kinship which seeing a photograph of something bears to other
ways of seeing it, and seeing a painting of it does not. Other terminology might
serve this purpose. But the restricted notion of seeing that Carroll and Currie
recommend risks losing sight of these similarities and differences. Carroll
himself, following a suggestion of Francis Sparshott, mentions an alternative to
his special sense of see which does not run this risk. We might describe our
experience of lm, when we have no clear sense of the spatial relations the photographed objects bear to us, as alienated vision.46 In calling this vision Sparshott
allows that what Carroll and Currie count as seeingunalienated visionis
only one variety. Seeing through photographs and through confusing batteries of
mirrors, notwithstanding any alienation, is another.
In their discussion of transparency, Carroll and Currie focus almost entirely
on the role of perception in acquiring information. One of the larger objectives
of Transparent Pictures was to show that information gathering is not the
only important function of perception. We sometimes have an interest in seeing
things, in being in perceptual contact with them, apart from any expectations of
learning about them. This interest helps to explain why we sometimes display
and cherish a photograph of a loved one (or a movie star or athelete or personal
hero), even a fuzzy and badly exposed photograph, long after we have extracted

45. Newsweek, 9 Sept. 1996, p. 36.

46. Nol Carroll, Ontology, 71; Theorizing, 62.



any interesting or important information it might contain, and why we may

sometimes prefer such a photograph to a realistic painting or drawing that is
loaded with information. We value the experience of seeing the loved one (even
indirectly), the experience of being in perceptual contact with him or her, for its
own sake, not just as a means of adding to our knowledge.47

47. Some readers may be interested in how I would treat several examples that Currie
thinks make trouble for the transparency thesis:
(a) Suppose scenes cause visual experiences not directly, but only with the mediation
of a Malebranchian God, who in his benevolence acts to maintain [the] counterfactual
dependence we observe (Image and Mind, 62). Currie thinks we would still see things, in
this case, and hence that natural counterfactual dependence is not necessary for perception. I think it is at least as plausible that, in the situation imagined, see would pick
out a different natural kind from the one it actually picks out, and that there would
then be instances of the former but not the latter. Also, it is not easy to be sure that any
inclination to say that seeing occurs, in this exotic example, does not depend on thinking
of the Malebranchian God as something more like a force of nature than a human intentional agent, even if we describe it in intentional terms (as we do computers).
(b) Suppose two clocks are linked mechanically, so their hands always move in tandem.
Do I see one of the clocks by looking at the other? (Image and Mind, 65) No, and for a reason that Currie himself endorses in connection with another example. Suppose a persons
eyes lack lenses, and unfocused light stimulates his retinas. He sees mere homogeneous
elds of white or black or grey, depending on the intensity of the incident light. Does he
see the objects that reect light to his eyes? No. The reason, I argued, is that his visual
experiences are not richly enough counterfactually dependent on the reecting objects,
and Currie concurs (Looking Again through Photographs, 803804; Image and Mind,
57). Only the intensity of light reected by the scene affects his visual experience. Likewise in the clock example: Only the position and/or movement of the hands of the second
clock affects the visual experiences of the person looking at the rst one. This is not
enough for him to qualify as seeing the second clock (or even its hands). If the example
is changed so that the rst clock is dependent in many respects on the second one (and
other conditions are met), I would recommend speaking of seeing the second by looking
at the rst. Even so, the perceiver will not know he is seeing the second clock, and may
not even have the impression of seeing it, unless he realizes that the rich counterfactual
dependence obtains.
(c) Do I see heat, by looking at the column of mercury in a thermometer? (Image and
Mind, 6364.) Not, I think, if I explicitly infer how hot it is from the length of the column. (In that case there would seem to be nothing that I see to be true of the heat. I could
not be said to recognize or notice it, in a sense that, arguably, is required for perceiving
it. Cf. Looking Again through Photographs, 804.) And not if, as is plausible, heat is
by denition something that can only be felt, not perceived in other ways. Currie thinks
that seeing heat is not ruled out by denition. He thinks we might see heat if things
looked darker the hotter they were. I am skeptical. In any case, we would not see heat if
we have to infer, explicitly, how hot things are from how bright or dark they appear. Suppose we dont have to, suppose the lightness or darkness functions simply as a perceptual
cue. If this counts as seeing heat, then surely we would be seeing heat if the height of the
mercury in a thermometer serves as the cue instead.


ichard Wollheims writings on pictorial representation combine philosophical

enquiry into the nature of the mediumenquiry involving issues of philosophy
of mind and language, as well as aestheticswith examination of the place that
painting and other visual arts have in our lives and critical observations about
individual works and particular artistic styles. He brings to this multifaceted
enterprise a rare combination of philosophical sophistication and aesthetic perceptiveness. The philosophical side of Wollheims work is dominant in Art and
Its Objects; aesthetic and critical considerations come to the fore in Painting as an
Art.1 But both are evident in both books, and all of his work on the visual arts is
sensitive in both directions.
It is not surprising that an aesthetically sensitive theory of pictorial representation, of what it is for something to be a picture, should emphasize the phenomenology of the experience of looking at and appreciating pictures. At the heart of
Wollheims theory is a special kind of visual experience that he calls seeing-in.2
One sees a woman in a drawing of a woman, and Henry VIII in Holbeins portrait.

1. Richard Wollheim, Art and Its Objects, 2nd ed. (Cambridge: Cambridge University
Press, 1980); hereafter cited in text as AO. Richard Wollheim, Painting as an Art: The
Andrew W. Mellon Lectures in the Fine Arts (Princeton, N.J.: Princeton University Press,
1987); hereafter cited in text as PA.
2. Wollheim also makes use of a notion of a standard of correctness in his account of
depiction, though I will not discuss it here. In place of this, my own account employs the
idea of a works possessing a certain function in a given social context. The artists intentions have a less essential role in my theory than they do in Wollheims. See my Mimesis
as Make-Believe: On the Foundations of the Representational Arts (Cambridge, Mass.: Harvard
University Press, 1990), p. 52.




By according a central place to this visual experience, Wollheim accounts for the
special visual nature of the medium, and accommodates the intuitively evident
contrast between pictorial representations and verbal symbols, between depiction and description. In emphasizing this contrast, Wollheim follows in the
spirit of Peirces distinction between icons and symbols, signs that are linked
to their objects by virtue of, respectively, resemblances and conventions. Wollheim explains depiction in terms of seeing-in, rather than resemblance (although
some resemblance theorists might utilize the notion of seeing-in in specifying a
special kind of resemblance relevant to depiction); but his account of depiction
shares the intuitive plausibility of resemblance theories, while avoiding their
most glaring difculties.
I heartily endorse the basic motivation of Wollheims project. A primary objective of my own account of depiction, which treats pictures as props in visual
games of make-believe, is to clarify and give proper weight to the idea that
pictorial representation is a genuinely visual medium. Although my theory and
Wollheims are very different, their central tenets are more complementary than
conicting, and they are better regarded as allies than as rivals. The make-believe
theory can be understood to provide a way of explaining Wollheims fundamental notion of seeing-in, which, to my mind, he leaves seriously underexplained.
I shall not argue for the make-believe theory here or spell it out in any detail.3
But I shall examine several features of Wollheims discussion and suggest how
the two theories can be made to mesh.
What is seeing-in? Wollheim describes it as an experience characterized by the
distinctive phenomenology of twofoldness: an experience with two aspects, a
recognitional aspect and a congurational one. The viewer attends simultaneously to what is seen and to features of the medium. When I look at the
representation of a woman, . . . on the one hand, I recognize or identify a woman,
and, on the other hand, I am aware of the marked surface.4
Wollheim insists that these are not two distinct experiences occurring simultaneously, but rather, two aspects of a single experience.5 It is not entirely clear how
in general experiences are to be individuated, or what the difference is between
two experiences and two aspects of a single experience. Wollheim describes the
two aspects as distinguishable but inseparable (PA, p. 46). But the point cannot be that neither can occur without the other. The congurational aspect, at
least, can occur without the recognitional one; a viewer might be aware of the
lines of a drawing of a woman without recognizing the woman. Whether the

3. I develop my account of depiction most fully in Mimesis as Make-Believe, esp. ch. 8.

4. Wollheim, Imagination and Pictorial Understanding, Proceedings of the Aristotelian Society, Supplementary Volume 60 (1986), p. 46. See also Painting as an Art, p. 73.
5. Wollheim, Imagination and Pictorial Understanding, pp. 4647; Painting as an
Art, p. 46.

S E E I N G - I N A N D S E E I N G F I C T I O N A L LY


kind of recognizing involved in seeing-in can occur without the congurational

aspect is harder to decide, pending clarication of what kind of recognizing it is.
(One certainly need not pay attention to the conguration of lines and shapes in
a picture in order to recognize a woman, any more than one must pay attention
to a friends facial features in order to recognize him in the esh.) The important
point may be that, when one looks at the picture in the expected manner, in
addition to recognizing the woman and also observing the painted surface, one
experiences relations between the features of the painting and what is seen in it.
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 identied representationally (AO, p. 216).6
I would urge that the viewer does not merely come to realize, as a result of perceiving both the marks on the surface and the image of a woman, how the marks
work to produce the image (indeed, one may not be explicitly aware of this);
rather, the viewer sees how they do. And seeing this involves both seeing the
marks and recognizing the woman.
Twofoldness is important. I am sure that it has a lot to do with the interest
that visual representations have for us. But the experience of seeing-in is hardly
explained by pointing out that it involves the phenomenology of twofoldness.
An explanation in terms of the two aspects of twofoldness is only as good as our
understanding of the aspects, and the recognitional one is mysterious. In what
sense does a spectator, on viewing a painting of a woman, recognize or identify
a woman? One does not literally do so, of course, there being no actual woman
there to recognize or identify. Neither does there appear to be a woman there; it
does not seem to the viewer that he is recognizing an actual woman. Wollheim
rightly denies that in viewing pictures we experience illusions, and he emphasizes
the discontinuity between recognizing a boy in a picture and recognizing one in
the esh (PA, pp. 4647). Until we understand better what the recognitional
aspect of seeing-in amounts to, we will not have explained what pictures are.
In Painting as an Art, Wollheim associates the recognitional aspect of seeing-in
with the experience of seeing depth in a at surface. When seeing-in occurs, two
things happen: I am visually aware of the surface I look at, and I discern something standing out in front of, or (in certain cases) receding behind, something
else (PA, pp. 4647).7 The experience of seeing depth in a at surface is familiar
enough, as is that of seeing a woman in a design, but being familiar with a phenomenon is not the same as understanding it. Again, it is not that one actually
observes one thing to be in front of another, nor does it seem to one that one does.
Except in unusual cases, the surface is and appears to be at.

6. Wollheim speaks of a particular kind of reciprocity between the two aspects, to

account for naturalistic representation (PA, p. 73).
7. See also Wollheim, Painting as an Art, pp. 60, 62.



Wollheims purpose in connecting seeing-in with seeing depth in at surfaces

is not so much to clarify seeing-in as to indicate the range of cases in which it
occurs. In viewing many paintings often regarded as nonrepresentational, one
sees one plane or shape or line in front of another. Many of the works of such
artists as Hans Hoffmann, Piet Mondrian, and Mark Rothko demand seeingin, Wollheim claims, and qualify as representational along with portraits and
landscapes, although the former are not gurative as the latter are. I think
Wollheim is right about this.8 My worries concern the idea that seeing depth
is necessary for seeing-in. If seeing-in is limited to cases in which one sees depth
in a at surface, our perception of sculptures, of theatrical performances, and of
the ag and target paintings of Jasper Johns (which portray at things on at
surfaces) would appear not to qualify.9 Wollheim seems willing to exclude sculpture, theater, and Jasper Johns paintings from the class of representations in his
sense (AO, pp. 225226). It is his right to use representation in a narrower
sense than others might, of course. But surely what he calls representations are
instances also of a larger genus which includes many sculptures, most theatre,
and Jasper Johns works, and in addition such nonvisual but perceptual representations as musical portrayals of the sounds of galloping horses and birdsong.
Whatever ones terminological preferences, we need a theory that will clarify
what representational pictures have in common with depictions (as I prefer to
call them) of these other sorts, as well as the ways in which they differ.
In all these cases appreciators participate in what I call perceptual games of
make-believe: visual games in the case of paintings, sculptures, and Jasper Johns
canvases, auditory ones in the case of representational music,10 and games that
are both visual and auditory in the case of theatre. It is ctional, in ones game,
that one sees a woman or one plane in front of another or a target or ag, or
that one hears galloping horses or the singing of birds, or that one watches Lear
pacing the oor and listens to his ragings. Depictions are (to put it very briey)
things whose function in a given social setting is to serve as props in sufciently
rich and vivid perceptual games of make-believe.
Participation in these games involves (actually) perceiving the work in a special way, a way imbued with certain imaginings. It is crucial that we understand
clearly the nature of the imaginings. The idea that one simply imagines a horse
upon seeing a picture of a horse is a non-starter. As Anthony Savile notes, There
is a world of difference between being brought to imagine something by seeing

8. But I propose a more substantial way of distinguishing, among representational

paintings, those that are gurative from those that are not than he does. See Mimesis as
Make-Believe, pp. 5457.
9. At least it would appear that one does not see ags and targets in Jasper Johnss
10. More narrowly, in the case of what I call depictive music. See Mimesis as MakeBelieve, pp. 3337.

S E E I N G - I N A N D S E E I N G F I C T I O N A L LY


this mark or that [on the canvas] and being brought to see something in a picture by seeing this mark or that.11 A vivid description of a horse may induce a
reader to imagine a horse, without her seeing a horse in the letters on the page.
Seeing-in is a perceptual experience, one that goes beyond perceiving the marks
on the canvas. Imagining a horse is not itself a perceptual experience, even if it is
a result of perceiving the marks, and perceiving the marks is just that, even if it
causes one to imagine a horse.
Is it that, rather than merely imagining a horse, one imagines the brown mass
of color that one perceives on the canvas to be a horse? Savile says that the viewer
does not do this, and surely he is right again.12 But the viewer does, in addition
to imagining a horse, imagine seeing a horse. And she imagines her actual perceiving of the canvas to be an act of perceiving a horse. (She does not imagine her
perceiving to be both a perceiving of the canvas and also a perceiving of a horse,
of course; she imagines of the perceiving which is in fact a perceiving of the canvas that it is a perceiving not of the canvas but of a horse.) Imagining seeing a
horse is imagining in a rst-person manner (not just imagining that one sees a
horse). In addition, the perceiver imagines this from the inside.13
Engaging even in this special kind of imagining is not sufcient for seeing
a horse in the picture, however. One could look at the picture and then, in a
separate (nonperceptual) act, imagine in the manner I have just described. This
would not be seeing a horse in the picture. We should note, in the rst place,
that there is no good reason to insist that imaginative acts must be deliberate
or under the subjects control.14 Dreams are obvious counterexamples, and so are
many of the imaginings that make up daydreams. The imagining of a viewer
who sees a horse in a picture is not deliberate, but a spontaneous response to the
marks on the canvas; she just nds herself imagining in a certain manner as she
looks at the picture. And she is best regarded not as seeing the picture and also
engaging in this spontaneous imagining, but as enjoying a single experience that

11. Anthony Savile, Imagination and Pictorial Understanding, Proceedings of the

Aristotelian Society, Supplementary Volume 60 (1986), p. 21. In attempting to show that
imagining is not an essential ingredient of normal picture perception, Savile argues that
we fully understand trompe lil works and see in them what they are pictures of, even in
the rare cases in which we are deceived by them. But since imagining something to be
thus and so is incompatible with my experiencing it to be thus and so and also with my
taking myself so to experience it, imagining cant be involved in these cases (ibid., pp.
2122). I nd the assumption that the perceiver who is fooled by a trompe lil painting
sees in it what it portrays and the assumption that imagining is incompatible with being
fooled or experiencing a hallucination both highly questionable.
12. Savile, Imagination and Pictorial Understanding, p. 21.
13. I discuss what it is to imagine in a rst-person manner and from the inside in
Mimesis as Make-Believe, 1.4.
14. See Savile, Imagination and Pictorial Understanding, p. 23.



is both perceptual and imaginative, her perception of the picture is colored by

the imagining. (Probably she enjoys a succession of experiences, each of which
is both perceptual and imaginative.) The experience of recognizing an (actual)
tree as a tree is not a combination of a pure perception and a judgment that what
one perceives is a tree. It is rather a perceptual experience that is also a cognitive
one, one colored by the belief that what one is experiencing is a tree. Likewise,
to see a horse in a design is to have a perceptual experience colored by imagining ones perception to be of a horse, a perceptual experience that is also an
imaginative one.
If we call this experience of imaginative perception one of seeing a horse in the
picture, it is seeing-in of a kind that occurs also when one sees busts of emperors, theatrical productions, Jasper Johnss paintings, and Hoffmanns Pompeii. In
an analogous sense one hears Lears ravings in the voice of the actor portraying
him and birdsong in the notes of Beethovens Pastoral Symphony. This gives us
a broad notion of perceiving-in, on which one might base the inclusive notion of
depiction or (perceptual) representation that we want. One can then proceed to
investigate differences among the many varieties of depictions. For instance, in
some casessculptures and works like Hoffmanns Pompeiiit is plausible that
the representational work itself or part of it is an object of ones imagining. Perhaps we imagine one portion of Hoffmanns canvas to be in front of another or a
block of marble to be an emperors head, whereas we do not imagine a stretch of
painted canvas to be a woman or a horse. Our perceptual games of make-believe
and the imaginative perceptual experiences that participation in them involves
vary in many other ways as well.
Thinking of seeing-in in the way I have sketched is the key to understanding the twofoldness that Wollheim rightly stresses. The recognitional aspect of
seeing a woman in a picture consists, roughly, in the viewers perception of the
picture being bound up with his imagining, in the manner I described, seeing a
woman. The congurational aspect comes into play not just because perceiving
the marks on the surface induces this imagining, but because the imagining is
about that perceiving; one imagines of ones perceiving of the marks that it is a
perceiving of a woman. Thus the two aspects of the experience are intertwined:
the imagining partially constitutive of the recognitional aspect has as an object
the perception that constitutes the congurational aspect. Of course, the viewer
imagines also of his perceptions of particular features of the design (particular
lines, patches of color) that they are perceptions of particular features of a woman
(tousled hair, penetrating eyes). Thus one observes connections between the
marks and the image of a woman.
In Imagination and Pictorial Understanding Wollheim claims that (leaving aside a certain vapid sense of imagination) imagination has no necessary
part to play in the perception of what is represented. His principle reason
for holding this . . . is that we have a perfectly good explanation of how we perceive representations without invoking imagination, an explanation in terms of

S E E I N G - I N A N D S E E I N G F I C T I O N A L LY


a very specic visual capacity, namely, seeing-in.15 It is true that one can give
an explanation, of a sort, of depiction without saying anything about imagining.
One can point to the familiar phenomenon of seeing-in, and then characterize
depiction in terms of it. But that ignores the question of what seeing-in is. We
have no right to assume that seeing-in does not involve imagining; that in order to
explain it, rather than merely point it out, and hence to give a full account of the
perception of pictures, one would not have to bring in imagining. That, I argue,
is indeed the case. Even some of Wollheims own observations seem to point in
this direction. He associates an experience in which the recognitional aspect of
seeing a boy in a stained wall is emphasized to the extent that the congurational
aspectand hence twofoldnessis lost with the experience of visualizing the
boy in the minds eye (PA, p. 47). Isnt visualizing a boy engaging in a certain
kind of imagining? Wollheim considers it plausible that the most primitive
instances of the perceptual capacity with which seeing-in is connected . . . are to
be found in dreams, daydreams, and hallucinations (AO, p. 217). Surely dreams
and daydreams, in any case, are exercises of the imagination.
Wollheim points out that although dreams and daydreams may anticipate
seeing-in or be continuous with it, they lack a crucial element. Actual seeing-in occurs when the relevant visual experiences cease to arise simply in
the minds eye: visions of things not present now come about through looking
at things present (AO, pp. 217218), as when one follows Leonardos advice
to look at damp-stained walls . . . and discern there scenes of battle or violent
action and mysterious landscapes (AO, p. 218). To do this is to engage in
the kind of visual game of make-believe that I described, one in which actual
things serve as props; and this involves imagining ones perceptions of various
features of the stained wall to be perceivings of, for instance, various parts of
a battleeld.
Wollheims idea seems to be that seeing-in is a sui generis kind of experience
which does not admit of explanation in other terms (beyond pointing out the
phenomenology of twofoldness). That this is so cannot be assumed without argument, and it is in any case an unsatisfying conclusion. We would like to be able
to understand seeing-in by relating it to other phenomena if we can. My proposal
not only links the kind of seeing-in that Wollheim recognizes, that which occurs
in the perception of pictures, to our experiences of sculpture and theatre and
to the hearing-in that listeners of music occasionally engage in; it also links
seeing-in, in explicitly specied ways, to other imaginative experiences: to visualizing, dreaming, day-dreaming, and childrens games of make-believe. This
does not involve denying that seeing-in is an experience of a very special kind.
Seeing-in differs signicantly from other exercises of the imagination, as well as
from other perceptual experiences. But it need not remain mysterious. We can

15. Wollheim, Imagination and Pictorial Understanding, p. 46.



say what is special about it; we can specify how it differs from, as well as how it
is similar to, other imaginative experiences and other perceptual ones.
One consequence of holding to [a psychological account of pictorial representation], Wollheim observes, is that it sets me against all those schools of
contemporary thinking which propose to explain pictorial meaning in terms like
rule, convention, symbol system, or which in effect assimilate pictorial meaning
to something very different, which is linguistic meaning (PA, p. 44). Pictorial
meaning and linguistic meaning are indeed very different. Depiction is not just
another language like English or Hungarian or Tagalog. Nor is it merely a language (or symbol system) with conventions satisfying certain special conditions
like Goodmans density and repleteness requirements. No such conditions will
themselves account for the perceptual nature of depiction. On this Wollheim and
I are in full agreement.
But Wollheim apparently holds that the perceptualness of depiction is incompatible with pictorial meaning being conventional or involving conventions or
rules (PA, p. 361, n. 21). This is highly questionable. A lot depends on what is
meant by conventions or rules, of course. If the conventionalists idea is supposed to be that one rst observes the picture and then, in a separate act, gures
out what it depicts by applying a rule or convention, then there is no special
visual experience involved beyond merely ascertaining the relevant features of
the canvas; one does not see something in the picture,16 nor does one participate
appropriately in a visual game of make-believe. But this is not even how linguistic conventions normally work. We automatically recognize a word in a familiar
language as meaning what it does; we do not rst ascertain the shapes of the letters and then apply the relevant convention to gure out what it means. Ones
visual experience of a swastika may be conditioned by the conventional associations determining its meaning. It looks terrifying, ominous, horrible. Yet there is
clearly an important sense in which its meaning is conventional.
This, of course, does not constitute the special kind of perceptualness peculiar
to depictions. Swastikas are not pictures. My point is merely that if it is true that
ones responses to pictures are conditioned by conventions, this fact is not incompatible with ones responses being thoroughly perceptual experiences. In Imagination and Pictorial Understanding Wollheim holds that the visual experience
of seeing-in is conditioned by the cognitive stock that the spectator holds,
and speaks of perception being permeated by cognition.17 Ones internalized
awareness of rules or conventions may be among the cognitions that permeate
ones perception of a symbol. This does not make the experience, colored by this
cognition, any less a perceptual one.

16. Wollheim made essentially this point in On Drawing an Object, in On Art and
the Mind (1973), p. 25.
17. Wollheim, Imagination and Pictorial Understanding, p. 48.

S E E I N G - I N A N D S E E I N G F I C T I O N A L LY


Do people have to learn to perceive pictures, or are we born with the capacity
to do so? If this nature/nurture question is what the issue of the conventionality
or naturalness of pictures comes down to, its answer is irrelevant to the issue
at hand. Wollheim emphasizes nature; Goodman nurture. Surely the truth lies
somewhere in the middle. I have no doubt that our ability to read pictures
depends in part on natural, inborn propensities, and in part on abilities acquired
as a result of experience, but I have little idea how much is nurture and how
much nature. Picture perception is a visual experience in any caseone involving
participation, of the kind I have described, in visual games of make-believe
regardless of how much learning went into our capacity to enjoy that experience.
The imaginings that infect our perception of the picture are no less intimately a
part of it if we had to acquire through experience the ability to perceive in a way
colored by those imaginings.
Much of my argument has amounted to applications of the familiar idea that
there is no such thing as an innocent eye, pure perception unsullied by other
cognitions. If we were to understand visual experience in so narrow a sense that
participating in games of make-believe in the ways I claim viewers of pictures do
will not count as such, we might as well declare ourselves blind, for it is likely
that on such a narrow conception nothing would count as visual experience. On
a more reasonable construal of visual experience, the fact that our experience of
pictures is bound up with imaginings and possibly conditioned by internalized
conventions will not count against their qualifying as fully visual. Painting is a
visual art, and depiction is a visual medium. The make-believe theory explains
how this is so.
There are analogies between the issues I have been discussing regarding the
perception of pictures and questions about appreciators emotional responses to
ction. I have argued elsewhere that when Charles, a typical lmgoer, watches a
horror lm in which a ferocious green slime attacks the camera (the spectator),
he is not genuinely afraid of the slime, but rather is participating in a game of
make-believe in which it is ctional that he is afraid.18 Some commentators (not
Wollheim) have attributed to me the astonishing thesis that Charless reaction
to the lm is not a genuinely emotional one, or even that, in general, appreciators are not genuinely moved by ction.19 That Charles does not experience
genuine emotion follows only if fear of the slime is the only emotion he could

18. See Mimesis as Make-Believe, 5.2 and 7.1, which supersede my earlier discussion
in Fearing Fictions, Journal of Philosophy 75 (1978), pp. 526.
19. See Bijoy Boruah, Fiction and Emotion: Rationality, Belief and Emotional Response to
Fiction (New York: Oxford University Press, 1988), p. 66; Nol Carroll, The Philosophy
of Horror, or Paradoxes of the Heart (New York: Routledge, 1990), pp. 6979; and David
Novitz, Fiction, Imagination, and Emotion, Journal of Aesthetics and Art Criticism 38,
no. 3 (Spring 1980), esp. p. 288, n. 2.



be experiencing, and it obviously is not. Perhaps some think that the fact that
Charles is engaging in an imaginative activity or that he is doing so because of
certain conventions would be somehow inconsistent with his feeling genuine
emotion.20 But there is no need to saddle ourselves with such exotic assumptions. (It is helpful to remember that appreciators imaginings are likely not to
be deliberate, even if they imagine in accordance with internalized conventions,
just as the experiencing of emotions is not, straightforwardly, something that
one deliberately does.) I see no reason why we should not count the experience
I describe of ctionally fearing the slimeor the experience of ctionally grieving for Anna Karenina, for instanceas an emotional as well as an imaginative
one; it may be intensely emotional. The heart is no more innocent than the eye,
and there is no more justication for thinking that imagination or the functioning of internalized conventions is incompatible with or dilutes emotional experiences than that it is incompatible with or dilutes perceptual ones.
The appreciators response to ction may involve other genuine emotions as
well. Charles may be genuinely disgusted by the lm, or even fear it, while and
possibly as a result of undergoing the experience of ctionally fearing the slime.
And the work may induce or revive genuine emotions in the appreciator directed
toward other things: fear of dangers the appreciator thinks might exist in the
real world, pity for real people in situations perceived as analogous to Anna
The make-believe theory is designed to explain the experience of being caught
up emotionally in a story and the special visual nature of pictorial representation; it certainly does not deny that there is such an experience or that depiction
is especially visual. In claiming it to be ctional but not true that Charles fears
the slime, I open the way to understanding his experience to be a genuinely
emotional one, notwithstanding his full realization that there is no slime and
no danger, and I explain why it is so natural to describe him as being afraid of
the slime. In arguing that, on viewing a picture of a woman, it is ctional that
one sees a woman, I make it possible for us to understand the viewers experience
as being a genuinely perceptual one which is richer than merely perceiving the
marks on the canvas. And I explain why it is so natural to speak of the spectators
seeing or recognizing or identifying a woman, despite the fact that there
neither is a woman there nor does there appear to be one.
20. I do not claim without qualication that there are conventions involved. See
Mimesis as Make-Believe, pp. 38, 401, 3012.

D E P I C T I O N,
Responses to Richard Wollheim

ichard Wollheim holds, famously, that pictorial representation is to be

understood in terms of a visual experience of a special kind, which he calls seeing-in, an experience that suitable spectators enjoy when they look at pictures.
On viewing a picture of a re engine, one sees a re engine in the marks on
the surface of the picture. I have argued that pictures are essentially props in
visual games of make-believe of a certain kind, and that the crucial perceptual
experiencewhich I am happy to call seeing-inis an imaginative as well as a
perceptual one. The viewer imagines seeing a re engine as she looks at a picture
of one, imagining her actual visual experience to be of a re engine. Wollheim
and I have carried on an intermittent dialogue in the course of developing and
explaining our respective theories of depiction, commenting on one anothers
views and on the relations between them. I continue the dialogue in this essay,
concentrating now on Wollheims On Pictorial Representation, the most recent
statement and defense of his theory, as well as his Painting as an Art.1

Wollheim endorses (with minor reservations) Albertis observation that the
painter is concerned solely with representing what can be seen, which he takes
1. Richard Wollheim, On Pictorial Representation, Journal of Aesthetics and Art Criticism 56 (1998): 217233; Richard Wollheim, Painting as an Art: The Andrew W. Mellon
Lectures in the Fine Arts (Princeton, N. J.: Princeton University Press, 1987). My main contributions to the discussion are Kendall L. Walton, Mimesis as Make-Believe: On the Foundations of the Representational Arts (Cambridge, Mass.: Harvard University Press, 1990), chap. 8;
Kendall L. Walton, Seeing-In and Seeing Fictionally [chap. 8, this volume]; and Kendall
L. Walton, On Pictures and Photographs: Objections Answered [chap. 7, this volume].




to express a constraint on the scope of representation: only what is visible can

be represented.2 The seeing that Alberti has in mind is, surely, ordinary visual
perception of particular existing objects (or events), what we might call seeing
things face-to-face.
The things that can be seen and so can be represented, Wollheim observes,
include both objects and events. Some of them are particular objects or events,
while others are objects or events merely of a particular kind. So we can have
a representation of Madame Moitessier [Ingress 1851 portrait], or a representation of a young woman behind a bar, perhaps a young woman of some specicitybut no particular young woman (Eduard Manets La Prune, ca. 1877).3
Is what is represented, in the latter case, something that can be seen? I would
expect Alberti to point out that it is possible to see particular existing things of
this kind, actual young women behind bars, even though nothing would count
as seeing (face-to-face) the woman in the picture, the woman the picture represents. And I expect that he would take this to satisfy the principle that painters are concerned solely with representing what can be seen. What cannot be
represented (pictured) are presumably things such as the average price of oil in
the 1970s, magnetic elds, and Cartesian egos.4
Wollheim chooses not to understand the constraint on what can be represented
in this manner. There are two different ways of seeing things, he explains: one
can see things face-to-face, and one can see things in a marked surface. Some
things can be seen only in the second manner, but that is all that is required for
them to be representable. Representation does not have to limit itself to what can
be seen face-to-face: what it has to limit itself to is what can be seen in a marked
surface.5 Objects or events that are merely of a particular kind are among
the things that can be seen in a marked surface but not face-to-face, Wollheim
claims. So the nonparticular woman represented by La Prune is representable
not because particular things of that sort can (could) be seen face-to-face, but
because this nonparticular one can be seen in a marked surface.
Wollheim appears committed to the view that there really are nonparticular
women (and nonparticular battles, etc.)special kinds of things that can be seen
in a special way. It is not clear how serious he means this commitment to be.6
2. Wollheim, On Pictorial Representation, p. 223.
3. Wollheim, On Pictorial Representation, p. 223; cf. Wollheim, Painting as an
Art, p. 69.
4. There is room for skepticism about this. Perhaps such things can be depicted, misrepresented, as being visible. Robert Hopkins has interesting things to say about this in
Explaining Depiction, Philosophical Review 104 (1995): 429431, and Picture, Image
and Experience: A Philosophical Inquiry (Cambridge: Cambridge University Press, 1998),
p. 168. Cf. Walton, Mimesis as Make-Believe, pp. 229330.
5. Wollheim, On Pictorial Representation, p. 223.
6. Perhaps Wollheim means to be avoiding this commitment when he says, Representations that are of things merely of some particular kind cannot sustain answers to

D E P I C T I O N , P E R C E P T I O N , A N D I M A G I N AT I O N


But it is not clear, either, what alternative he would endorse, or what alternatives
are open to him. People do sometimes speak of seeing pink elephants, arguably without implying that there are any such. (One might question whether
this is a literal use of see; I prefer to think of it as short for seem to see.)
But Wollheim obviously does not regard seeing-in as hallucinating, as a kind of
visual illusion, and he is obviously right not to do so.7
Are his nonparticular women ctional entities? If so, the puzzles about their
ontological status are familiar, at least (even more familiar than those concerning objects of hallucinations). And some, I among them, would argue that we
need not suppose that there really are such things while acknowledging, indeed
insisting on, the convenience of speaking as though there are. I would expect
Wollheim to be unsympathetic to this suggestion, as it encourages regarding
imagination as more central to depiction than he seems willing to allow. But
in Painting as an Art he is amenable to construing There are peasants there
uttered in front of a picture of haymakers as an exercise in make-believe.8
He sharply contrasts There are peasants there with I see peasants (also
uttered in front of the picture of haymakers), however, insisting that the latter
expresses a genuine perceptual judgment not involving make-believe.9 This
sharp contrast is intuitively unattractive, to say the least, especially since the
peasants that one sees are surely (as it were) none other than the ones that are
there. It is not clear how Wollheim will account for the naturalness of comments such as There are peasants there, whom I see and There are peasants
there; I can see them. He wants to insist on the fact that the viewer enjoys
a genuine visual experience, which grounds the visual nature of depiction, not
just an imaginary or make-believe one. But this fact is in no danger, not on my
account in any case. For although I deny that the viewers experience is, literally,

the question, Which object? Which event? or, Which woman? Which battle? (On Pictorial Representation, p. 223). But the sense in which they cannot is very unclear. These
questions invite any number of reasonable answers: That one, The one in the picture,
The one so-and-so is now looking at, The one wearing the fancy hat, The one looking over her shoulder. Which if any of these answers is informative will of course depend
on the context. Whether they are true when taken literally depends on ones theory of
ction. In Painting as an Art, Wollheim notes that of course, This is a picture of Venus
does not admit of existential generalization (p. 361, n.16).
7. The representational content of a painting is often, and totally misleadingly,
referred to as its illusionistic content, Wollheim remarks. And he speaks of the unjustied assimilation of the representational to the illusionistic or the imitative (Richard
Wollheim, On Formalism and Pictorial Organization, Journal of Aesthetics and Art
Criticism 59 [2001]: 131132).
8. Wollheim, Painting as an Art, p. 361, n.21.
9. Wollheim, Painting as an Art, p. 361, n.21. On my theory, it is ctional in the world
of the picture that there are peasants but not that I see them. Both statements express ctional truths in the spectators game world. Cf. my Mimesis as Make-Believe, pp. 293304.



one of seeing peasants, I do not for a minute deny that it is an actual visual
experience. What is merely imagined is that this visual experience has peasants
as its object.
So why not allow that There are peasants there and I see peasants both
involve make-believe? To return to Manets La Prune, we imagine seeing a
woman whom we imagine to be there. This makes life a lot easier. For the seeing
that I merely imagine being engaged in is perfectly ordinary, face-to-face seeing,
and it is seeing of a perfectly ordinary, particular woman, indeed an existing
onethat being the only kind of woman there can be. There is no need to recognize seeing of a special kind, directed on a peculiar and otherwise unseeable
object.10 I do (genuinely) enjoy a special kind of visual experience, but it is one
that is understood in familiar termsin terms of really seeing the picture surface
(face-to-face) and imagining this seeing to be of a woman (an ordinary one).
Viewers of the portrait of Madame Moitessier enjoy an experience of just this
kind also, the only difference being that in that case there is a woman whom one
imagines seeing.
The reader will have noticed that, although Wollheim takes La Prune to represent a woman merely of a particular kind, there is an obvious sense in which
it represents a perfectly ordinary, particular woman, or (this may or may not
amount to the same thing) it represents a woman as being perfectly ordinary and
particular. Put differently, the woman in the world of the picture is, in that
world, an ordinary particular womanindeed an existent one. This observation
is awkward for the proponent of seeing-in as Wollheim characterizes it. Conceivably, he might stick to his guns, reiterating that what is represented, what a
(suitable) perceiver sees in the marked surface, is actually a woman merely of a
particular kind, while allowing that the perceiver sees this nonparticular woman
(in the marked surface) as a particular one. The picture would thus be understood
to misrepresent a woman merely of a particular kind as being a particular woman.
We need not dwell on the unattractiveness of this suggestion.
I claim a further advantage for my way of dealing with what Wollheim calls
depictions of things merely of a certain kind: it generalizes readily to works
other than pictures. Stories and novels often portray nonparticular persons and
nonparticular objects and events of other sorts, in whatever sense La Prune
does. There is no special visual experience, seeing-in, which takes such things
as objects. Are they objects of a special experience of reading, one we might call
reading-in, or reading aboutin, even though they cannot be read about in
the ordinary mannerpresumably the manner in which we read about actual
people in newspapers?

10. Those who think that there really is something (the woman in the picture) that
I imagine seeing still have the problem of saying what sort of thing this is, although they
neednt say that it is something that can actually be seen.

D E P I C T I O N , P E R C E P T I O N , A N D I M A G I N AT I O N


It is far better to say that readers of novels and stories imagine ordinary particular people. Sometimes they imagine reading, in an ordinary newspaper manner, about such people. Often (depending on the nature of the story or novel)
they imagine hearing verbal reports of them, or learning about them, or anyway
knowing about them. What is imagined is in any case (exotic examples aside)
ordinary things cognized in ordinary ways.11

In his marvelously rich and perceptive explorations of pictorial representation,
Wollheim has surprisingly little to say about the perspectives or points of view
from which things are depicted. He distinguishes between the foreground and the
background of various paintings, this being, of course, a matter of the pictures
perspectives.12 He describes the fundamental experience of seeing-in as seeing
one thing in front of another.13 He notes that in Nicolas Poussins Rinaldo and
Armida, Rinaldos face is some-what turned towards us.14 But he does not, so
far as I know, spell out what it is for a depiction to be from one point of view
rather than another, or as we sometimes put it, what it is to depict something as
seen from a certain perspective. How might he do this?
It is safe to assume that he would want to account for this in terms of the
visual experiences of suitable observers; so do I. His comment about Rinaldos
face being turned toward us comes in a paragraph describing what we see in
the picture. So perhaps a pictures depictive point of view consists in what the
suitable spectator sees in it.15 How can this be? What we see in the picture is
Rinaldo and various of his propertiesthe position of his head and arm relative to his body, his being asleep, and so on. The perspective is not among his
properties. He does have relational properties that we seehis head being
turned away from Armida, for instance, whom we also see in the picture, a
property that is not constitutive of the pictures perspective. Wollheim says
that Rinaldos head is turned slightly toward us. Do we see this relational property in the picture surface? We do not see ourselves in it, obviously. Do we see
Rinaldo turned toward observers (ourselves?) who, although not themselves

11. In some cases, there is no specic mode of cognitive access such that readers imagine knowing about a person in that manner. Nevertheless, they probably imagine that
their cognitive access to the person is in some ordinary manner or other.
12. Wollheim, Painting as an Art, pp. 210, 215, 218, 223, and 234235.
13. Wollheim, On Pictorial Representation, p. 221; Paintings as an Art, p. 46.
14. Wollheim, Painting as an Art, p. 195.
15. In other places, Wollheim might be taken to suggest that perspective is a matter
of how things are depicted, although he does not explain what this amounts to. He refers
to the point of origin from which something is painted (Painting as an Art, p. 130).
And he speaks of what the picture represents as it represents it (ibid.).



seen in the picture, are understood to be there? Wollheim does allow that
certain paintings have a representational content in excess of what they represent, of what can be seen in them; there may be a gure in the represented
space [but] not the part of it which is represented.16 In some cases there are
unrepresented spectators, what he calls the spectator in the picture, whom
the viewer, the spectator of the picture, identies with, imagines from the
inside.17 But even in these cases the viewer is not part of the representational
content of the picture, not in the represented space;18 she merely identies,
imaginatively, with someone who is. And most pictures do not contain a spectator in the picture anyway.19 Yet most or all pictures depict things from a
certain point of view.20 I do not see how this can be explained in terms of what
is seen in the picture.21
The perspective from which one sees something, in cases of ordinary visual
perception, is a matter of the point in space, relative to the object seen, from
which one sees it. (This usually has consequences for what is seen, of course.
But it would be a mistake to identify the experience of seeing from a particular
perspective with the properties of the thing that one sees.) The viewer of Rinaldo
and Armida is actually at a certain place relative to the pictureseven feet from
it and slightly to the left of center, for instance. It is from this position that one
sees Rinaldo in the picture surface. But this location in space does not correlate
with ones perspective on Rinaldo in the sense in which his face is turned toward
us, the sense in which it is the perspective from which he is depicted. To change
ones position relative to the canvas, to move closer to it, for instance, or farther
to the left, does not affect ones point of view in the latter sense. (This is why we
can say that the picture depicts Rinaldo from a certain point of view; we cannot
normally say this about freestanding sculpture.)
Shall we say that the markings on the canvas are such that, from wherever the
viewer is actually positioned, what she sees in the picture surface is Rinaldofrom-a-certain-angle-and-distance? It is not clear what this might mean. And the
angle and distance from which (in some sense) she sees Rinaldo, in the picture,

16. Wollheim, Painting as an Art, p. 101; cf. On Pictorial Representation, p. 225.

17. Wollheim, Painting as an Art, chap. 3.
18. Wollheim, Painting as an Art, p. 102.
19. Wollheim, Painting as an Art, p. 103.
20. Robert Hopkins (among others) has argued for the plausible thesis that depiction
is necessarily from a point of view (Explaining Depiction, p. 428; Picture, Image and
Experience, p. 36), while noting that there can be signicant indeterminacies in a pictures
perspective. (Ambiguities also, I would add.) Dominic Lopes observes rightly that a picture need not represent things from a single point of view (Understanding Pictures [New
York: Oxford University Press, 1996], p. 120). I think a picture might conceivably lack
a point of view entirely but that this virtually never happens.
21. My point here is essentially the same as that of my roller coaster example, in Pictures and Hobby Horses: Make-Believe Beyond Childhood, [chap. 5, this volume].

D E P I C T I O N , P E R C E P T I O N , A N D I M A G I N AT I O N


is in any case not her actual perspective or point of viewnot in anything like
the sense of perspective that applies to ordinary vision. Wollheim insists that the
viewer actually sees Rinaldo, but it is unclear how this seeing-in can actually be
from a certain perspective, apart from the ordinary perspective one actually has
on the picture itself.
The solution is staring us in the face: on looking at the picture one imagines
seeing Rinaldo from a certain (approximate) angle and distance.22 This, curiously,
is what Wollheim seems to be saying about the special case of pictures that possess a spectator in the picture. The viewer identies with this personage, and so
imagines seeing what he sees from his perspective. Why cannot the viewer imagine seeing the depicted objects from a given perspective without having such a
spectator to identify with? Otherwise, the obvious fact that most or all pictures
depict things from a perspective or point of view remains mysterious.

The upshot of these several worries is that Wollheims characterization of seeingin, of the experience of picture perception, is seriously incomplete. One way to
indicate what is missing, while sidestepping any misunderstandings that may
arise from different conceptions of the imagination, is to observe that there is no
place, or no obvious place, in his account of the content of the viewers experience
for anything like the thought or idea or impression or awareness or conception or
notion of an ordinary seeing of an ordinary woman.23 Instead, he has perceivers
seeing objects of a different kind, in a special way. Perhaps if pressed, he would
acknowledge some such thought or idea or impression. That would be a big
step toward my way of understanding his notion of seeing-in, whether or not he
agreed to speak of imagining.
He does say that the recognitional aspect of seeing-in, when one sees a boy in
a stained wall, for instance, is capable of being described as analogous to the
experience of seeing a boy face-to-face. But he insists that the two sorts of experience are phenomenologically incommensurate, and that it is a confusion to ask
how experientially like or unlike the one is to the other. We get lost once we
start comparing the phenomenology of our perception of the boy when we see
him in the wall . . . with that of our perception of [the] boy seen face-to-face.24

22. Walton, Mimesis as Make-Believe, pp. 337348.

23. The thought, if we call it that, need not be involved in a sense that entails that the
person articulate it, or say to himself, I see a woman. But the naturalness of describing
ones experience in this way, ones readiness to do so, suggests that the thought, in an
unarticulated form, is already there. Robert Hopkins appears to recognize part of what is
needed. When you see a horse in a picture, he says, the thought (or some such) of a horse
enters your experience of the picture (Picture, Image and Experience, p. 16).
24. Wollheim, Painting as an Art, pp. 4647. Cf. On Pictorial Representation, p. 221.



Insofar as I understand this, I think I agree. I take it to mean, roughly, that the
two experiences differ not in degree, but in kind, that it is wrong or misleading to describe the experience of seeing-in or its recognitional aspect as an experience somewhat like that of face-to-face seeing (as some resemblance theories
of depiction might have it). Wollheim constantly refers to seeing-in as a distinct
kind of perception, or a visual experience with a distinctive phenomenology.
Agreeing with this does not require excluding the thought or impression
of face-to-face seeing of a boy from the phenomenology of seeing a boy in the
marks. Rather than being somewhat like engaging in face-to-face seeing, seeing-in
is a visual experience that involves (as I choose to put it) imaginingmerely imaginingdoing so. And what is imagined is not just somewhat like face-to-face
seeing, but the real thing. The difference between imagining seeing and actually
doing so is, I take it, a difference in kind. Malcolm Budd argues that if the
two phenomenologies are incommensurate, the recognitional aspect of seeing-in
cannot be understood on the analogy of face-to-face seeing. Hence, the recognitional aspect of seeing-in . . . is revealed as having no nature of its own.25 Understanding the recognitional aspect to involve imagining seeing makes sense of the
claim that it is both analogous to and incommensurate with face-to-face seeing.
Wollheim urges that there is an important causal trafc between seeing-in
and seeing face-to-face. Children learn to recognize many familiar and unfamiliar
objects through rst seeing them in the pages of books.26 This is no surprise
on the imagining seeing account. It is a familiar fact of experience, conrmed
by empirical research, that doing things in imagination can often improve ones
ability to do them in fact.27 Imagining (visualizing) a face from a verbal description may help me to recognize it in the esh.28
Jerrold Levinson has, inadvertently, provided support for my claim about the
lacuna in Wollheims notion of seeing-in. He claims to agree with Wollheim
that seeing-in is generally prior to, and not to be analyzed in terms of, imagined

25. Malcolm Budd, On Looking at a Picture, in Psychoanalysis, Mind and Art: Perspectives on Richard Wollheim, ed. Jim Hopkins and Anthony Savile (Oxford: Blackwell,
1992), p. 271.
26. Wollheim, Painting as an Art, p. 47.
27. Cf., for example, the studies described by Lien B. Pham and Shelley E. Taylor,
From Thought to Action: Effects of Process- Versus Outcome-Based Mental Simulations on Performance, Personality and Social Psychology Bulletin 25 (1999): 250260;
Roger N. Shepard, The Mental Image, American Psychologist 33 (1978): 125137; and
Shelley E. Taylor, Lien B. Pham, Inna D. Rivkin, and David A. Armor, Harnessing the
Imagination: Mental Simulation, Self-Regulation, and Coping, American Psychologist 53
(1998): 429439. Shepard takes some of his experiments to provide evidence that the
very same mechanisms are operative in imagery as in perception (The Mental Image,
p. 134). Thanks to Gregory Walton for these references.
28. Wollheim claimed that the make-believe theory has grave difculties in accounting for this phenomenon (Paintings as an Art, p. 360, n.8).

D E P I C T I O N , P E R C E P T I O N , A N D I M A G I N AT I O N


seeing.29 But his discussion (sketchy though it is) suggests that he is thinking of
this experience very differently from the way Wollheim does, and points strongly
in the direction of the kind of imagining seeing I take to be central.30
He makes a stab at clarifying the recognitional aspect of seeing-in: In looking at a picture of a woman, he proposes, it seems to you as if you are seeing
a woman (alternatively, you have an impression of seeing a woman), in virtue of
attending visually to portions of the canvas. The core of seeing-in . . . is a kind
of as-if seeing that is both occasioned by and inextricably bound up with such
registering.31 To say that it seems as if something is the case often implies,
perhaps even entails, that it is not the case; Levinsons stab thus appears to conict with Wollheims insistence that to see a woman in a picture really is to see a
woman, though in a special manner.
The kind of seeing of a woman Levinson has in mind, the seeing it seems as
if one engages in, is surely ordinary face-to-face seeing of an ordinary woman.
Levinson thus brings on board exactly what I claim, most fundamentally, to be
missing from Wollheims account of seeing-in, the (unarticulated) thought or
impression or idea of seeing an ordinary woman in an ordinary manner.
Why does Levinson deny that seeing a woman in a picture involves imagining
seeing a woman face-to-face? In one discussion he simply declares that he nds it
odd to say this,32 but a footnote to his essay on Wollheim reveals more: On my
conception of it, imagining is necessarily active or contributory. . . . By contrast, seeming to one as if . . . is passive or receptive, not something one brings about and actively
sustains, but something that . . . simply occurs. Seeing X in Y is something that
happens to one.33 Since I have always insisted that the kind of imagining central
to my theory can be and frequently is nondeliberate, something that happens to
us (often as a result of prompting by a picture or other prop), his disagreement
with my account turns out to be verbal rather than substantive. Understanding
as if seeing as imagining seeing (in my sense), it is not hard to construe Levinsons
suggestion that the viewer imagines seeing a woman in virtue of attending

29. Jerrold Levinson, Wollheim on Pictorial Representation, Journal of Aesthetics and

Art Criticism 56 (1998): 227.
30. I will not examine the most obvious difference between Levinson and Wollheim:
Levinson rejects the necessity of twofoldness, which Wollheim has always taken to be
at the very heart of his conception of seeing-in.
31. Levinson, Wollheim on Pictorial Representation, p. 229, emphasis in original.
32. Jerrold Levinson, Making Believe, in The Pleasures of Aesthetics (Ithaca, N.Y.:
Cornell University Press, 1996), p. 294.
33. Levinson, Wollheim on Pictorial Representation, p. 232, n.3, emphasis in
original. Wollheim expressly allows for imaginings being involuntary or passive rather
than active, and so does not share Levinsons conception of imagining. Cf. Wollheim,
Imagination and Identication, in On Art and the Mind (Cambridge, Mass.: Harvard
University Press, 1974), pp. 69 ff.



visually to portions of the canvas as the claim that the viewer imagines his
seeing of the canvas to be a seeing of a woman. (Other construals are possible
as well.)
Levinson does offer another reason for resisting analyzing seeing-in in terms of
imagining seeing, but it backres. If all seeing-in involves imagined seeing, he
claims, we lose a resource for explaining some of the special character, whether
of immediacy, intimacy, absorbingness, or emotional impact, of some pictures as
opposed to others.34 I have identied enormous resources available to my theory
for making distinctions of these kinds within the class of things serving as props
in visual games of make-believefor accounting for differences of realism, in
several senses, among depictions, and for understanding different styles of depiction. The visual games in which pictures are props vary greatly in richness and
vivacity. They are more or less indeterminate, in various respects. The principles
of make-believe may or may not be linked to resemblances of one sort or another,
and they may be internalized to different degrees. Some pictures restrict viewers
participation in the game, in one dimension or another. And so on.35 Levinson
would forgo all of these resources for the sake of a simple crude contrast between
inducing or not inducing perceivers to imagine seeing the object represented.
I claim support for my account of depiction, also, in a recent discussion by
Catherine Abell and Gregory Currie. Pictures aid the simulated seeing of their
objects, they propose. A depiction is an input to the simulation of seeing something. And they speak of simulations as involving pretend perceptions.36

I turn now to objections Wollheim has raised to my theory of depiction.
In Painting as an Art, Wollheim claimed that my make-believe theory holds
that there is a conventional link between the appearance of the picture and what
we are led to make-believedly see and so fails to ground what a painting represents in the kind of visual experience that the representation will cause in suitable spectators.37 My reply is that (what I call) principles of generation or principles

34. Levinson, Wollheim on Pictorial Representation, p. 227.

35. See Walton, Mimesis as Make-Believe, pp. 293352.
36. Catherine Abell and Gregory Currie, Internal and External Pictures, Philosophical
Psychology 12 (1999): 440441. I do not know how Currie will square this suggestion with
his skeptical remarks in Image and Mind: Film, Philosophy and Cognitive Science (Cambridge:
Cambridge University Press, 1995), which I discuss in On Pictures and Photographs:
Objections Answered, in Film Theory and Philosophy, ed. Richard Allen and Murray
Smith (Oxford: Oxford University Press, 1997), pp. 6075 [chap. 7, this volume].
37. Wollheim, Painting as an Art, pp. 77 and 361, n.21. I responded to this objection in Mimesis as Make-Believe, pp. 301302. Cf. my exchange with Wollheim in Philosophy and Phenomenological Research 51 (1991), Symposium on Mimesis as Make-Believe,

D E P I C T I O N , P E R C E P T I O N , A N D I M A G I N AT I O N


of make-believe are not in general conventional in any robust sense. In Mimesis

as Make-Believe I warned against characterizing them thus.38 Also, although
the principles specify what imaginings are appropriate, given the nature of the
props, appreciators do not usually consult them, or even have them in mind, in
deciding what to imagine. Indeed, they usually do not decide what to imagine,
but simply nd themselves imagining in a certain manner, prompted by properties of the work before them.39
The role of the principles of make-believe in my account is exactly analogous
to that of Wollheims own standard of correctness in his. Neither compromises
the visual nature of depiction:
While a standard of correctness applies to the seeing appropriate to
representations, it is not necessary that a given spectator should, in order to see
a given representation appropriately, actually draw upon, rather than merely
conform to, that standard of correctness. He does not, in other words, in seeing
what the picture represents, have to do so through rst recognizing that this
is or was the artists intention. On the contrary he mayand art historians
frequently doinfer the correct way of seeing the representation from the way
he actually sees it . . . and, for a spectator reasonably condent that he possesses
the relevant skills and information, this is perfectly legitimate.40

An entirely different objection, which Wollheim has advanced more recently,

focuses on my claim that on viewing a picture of a re engine, for instance, one
imagines ones actual perceiving of the picture to be a perceiving of a re engine.
His argument consists of rhetorical questions:
My difculty . . . is how to understand the core project, or imagining one
perceptual experience to be another. For if we succeed, in what way does the
original experience retain its content? For, what is left of the experience of seeing
the surface when I successfully imagine it to be some other experience? However,

pp. 401406 and 423427; and Walton, Seeing-In and Seeing Fictionally [chap. 8, this
volume]. More recently, Wollheim described me as the major contemporary advocate of
the theory that we relate to the content of pictures through the imagination rather than
perceptually (A Passionate Sightseer, review of Michael Podro, Depiction, in Times Literary Supplement 23 [April 1999], p. 20). I decline the honor, having designed my theory
to establish and explain the fundamentally perceptual nature of pictorial representation.
38. Walton, Mimesis as Make-Believe, pp. 38 and 4041.
39. Walton, Mimesis as Make-Believe, pp. 1316, 23, 68, 139, 185186, 216, 311,
351352, and especially 217, 301302.
40. Richard Wollheim, Art and Its Objects, 2nd ed. (Cambridge: Cambridge University Press, 1980), p. 207. My principles are not tied essentially to artists intentions, as
his standards of correctness are.



if I do continue to see the surface, or this experience retains its content, how have
I succeeded in imagining it, the experience, to be an experience of seeing a face?41

He sees no difculty, in general, in imagining one action or experience to be

a different one. One may move ones hands in a jerky and irregular fashion,
imagining this to be an action of conducting an orchestra. What he claims to
nd problematic is imagining of a perceptual experience of one kind that it is a
perceptual experience of a different kind.
This does not seem to me to be a problem at all. Why should imagining a
perceptual experience to have one content while recognizing that it actually has
a different one be any more difcult than imagining an object to have properties
different from those one realizes it really possessesimagining a glob of mud
to be a pie, for instance? I listen to a Glenn Gould recording of Bachs Art of the
Fugue. My actual perceptual experience is of sounds emanating from a speaker
in my living room, but I imagine my experience to be of a live performance in a
concert hall.42 Attending a performance of Die Zauberte, I hear the sounds produced by the utist in the pit orchestra, imagining my experience to be of sounds
produced by Papageno with his crude wooden instrument.
Patrick Maynard reminded me that Scottie, in Hitchcocks Vertigo, dresses up
Judy precisely in order to enjoy a vivid imaginative experience of perceiving
the now deceased woman he knew as Madeleine. Surely Scotties actual experience remains one of perceiving the dressed-up Judy; and surely he imagines this
experience to be one of perceiving Madeleine. Never mind that Judy turns out to
be Scotties Madeleine. Scotties imaginative project as he conceived it, believing
Judy and Madeleine to be different persons, is perfectly coherent.
It is surely coherent, also, to suggest that in interacting with her husband or
her boss a person might, perhaps unconsciously, imagine herself to be interacting
with her father or her mother. Of course this suggestion includes the hypothesis
that she imagines her perceptions of the husband or boss to be perceptions of her
father or mother.
In all of these cases, not only is the actual object of a persons perceptual experience in fact different from what she imagines it to be and not only does she know
this to be so, it is likely that the actual intentional content of her experience is
different from what she imagines it to be, that is, the original experience retains
its content even as she imagines it to have a different content. The sounds
produced by the utist performing Die Zauberte seem to the listeners to be just
that, while they imagine themselves to be hearing Papagenos playing.
41. Wollheim, On Pictorial Representation, p. 224.
42. If the recording is of an actual concert performance (as many of Goulds recordings are not), in hearing (directly) the sounds from the speakers, I am hearing, indirectly,
the actual performance. I imagine my experience to be one of hearing the performance
directly. (Cf. my Transparent Pictures [chap. 6, this volume].)

D E P I C T I O N , P E R C E P T I O N , A N D I M A G I N AT I O N


Wollheim notes that imagining one experience to be another is something

more experiential than simply imagining that one experience is the other.43
This seems right (or rather it seems right that a more experiential imagining is
involved in picture perception). It is not easy to say what experiential means
here. But the examples of imagining one experience to be another, presented
above, seem to me to be appropriately experiential.
If there is some sort of incompatibility between the two intentional contents,
why is this not a problem for Wollheims own notion of seeing-in? Seeing-in is
an experience characterized by what he calls twofoldness: one sees the marked
picture surface, and one sees the subject of the picture. These are not two independent experiences, he insists, but two aspects of a single one. It is hard to know
what this means, and Wollheim offers little explanation. But he clearly says
that we have a single perceptual experience involving two different intentional
contents. Why doesnt he think the one content interferes with the other? How
can the perception of the surface retain its content if one succeeds in making
the subject of the painting the content of ones perceptual experience? Well,
the experience has two different aspects. But what does this mean? Wollheim
rejects the duck-rabbit analogy, precisely on the grounds that it suggests an
incompatibility; one cannot presumably see the gure as a duck and as a rabbit
simultaneously. I propose that my theory goes some way toward showing how
two different intentional contents can be combined. The experience is a perception of the pictorial surface imagined to be a perception of a re engine, or of
whatever is depicted.

More needs to be said about this experience. I have not fully specied, either here
or previously, the nature of the imaginings involved or how they are related to
ones actual seeing of the picture surface. But I hope to have shown that imaginings (or whatever one chooses to call them) along the lines I have suggested are
an essential ingredient of picture perception, and that Wollheims worries about
supposing this to be so are unfounded.44
43. Wollheim, On Pictorial Representation, p. 225.
44. I am indebted to Malcolm Budd, David Hills, Jerrold Levinson, Patrick Maynard,
and Richard Wollheim for discussions of the ideas in this essay.

This page intentionally left blank

What Do You See and How Long Do You See It?

How often have we experienced a day when the light was so crystalline, when every object within our vision was etched with such
clarity, that we longed for the moment to be preserved forever?
The purpose of art since the days of the cave dwellers has been
to arrest the passage of time in order that the moment may be
contemplated at leisure.
James Borcoman, Magicians of Light: Photographs from
the Collection of the National Gallery of Canada
What the camera does . . . is to x the appearance of [an]
event. It removes its appearance from the ow of appearances
and it preserves it, not perhaps for ever but for as long as the
lm exists. . . . The camera saves a set of appearances from the
otherwise inevitable supercession of further appearances. It holds
them unchanging. . . . [The ratio of the life of a photograph to
the instant appearance it preserves may be 20,000,000,000:1.]
Perhaps that can serve as a reminder of the violence of the ssion
whereby appearances are separated by the camera from their
John Berger, About Looking

Picturesstill pictureshave been around forever. That they are still was hardly
noticed, probably, during the millennia before photography. For until approximately then pictures of no other kind, nothing anyone might want to call motion



pictures, were available.1 The rst photographic pictures were stills, of course,
although they were made from moving imagesthe images of the camera obscura,
for instance, which move with the motion of the objects reecting light through
the pinhole. The problem for picture makers, for the inventors of photography, was
how to x the image, how to preserve it. And xingthe only kind of xing they
could envision at rst, no doubtsacriced the images motion. They arrested the
image, froze it, in order to preserve it, thus creating a still picture. The pictures
produced by this new method are remarkable in several respects. They are nonetheless entirely unambiguous members of the ancient category of still pictures. Later
inventors gured out how to preserve images without arresting their motion, and
produced pictures of a radically new kind, moving pictures. These constitute an
exciting new medium, but they also call attention to the still-ness of still pictures,
and put us in a better position to appreciate what is special about them.
Still pictures are perfectly capable of depicting motion; this capacity was not new
with motion pictures. But it is by moving or changing themselves that moving
pictures depict movement or change. What is remarkable about stills in contrast
to their dynamic cousins is their ability to depict motion while remaining static.
Much has been written about the portrayal of three-dimensional objects on a at
surfacewhich both still and moving pictures doabout the development of
perspective, modeling techniques, and so on. An equally astonishing but infrequently examined trick is that of depicting movement or change by means of
unmoving marks on an inert picture surface.2 This is my topic in section 2 of
this essay. Three very different but equally convincing photographic examples,
gures 10.110.3, illustrate several of the many devices picture makers use in

This essay originated as a talk for The Light Symposium in St. Johns, Newfoundland, in 2002, but has since gone through many transformations, with the assistance of
audiences on that rst and many other occasions. A version of section 3 was published
separately as Landscape and Still Life: Static Representations of Static Scenes, in Rivista
dEstetica (1995), and reprinted in Photography and Philosophy: Essays on the Pencil of Nature,
edited by Scott Walden (Oxford: Blackwell, forthcoming). My warm thanks, for invaluable suggestions, to Alon Chasid, Herb Clark, Stephen Davies, Stacie Friend, Robert
Gordon, Daniel Herwitz, David Hills, Thomas Hofweber, Daniel Jacobson, Eileen John,
John Kulvicki, Peter Ludlow, Patrick Maynard, Mike Martin, Bence Nanay, Ian Proops,
Barbara Savedoff, Murray Smith, Scott Walden, Alicyn Warren, Jessica Wilson, Stephen
Yablo, and Gideon Yaffe.
1. The rst motion pictures, plausibly so-called, were not photographic, however,
but what one saw using devices like Joseph Plateaus phenakistoscope (1832).
2. There has been some examination of this topic. See Ernst Gombrich, Standards of Truth: The Arrested Image and the Moving Eye, Critical Inquiry 7, no. 2
(1980): 237273; Ernst Gombrich, Moment and Movement in Art, in The Image
and the Eye: Further Studies in the Psychology of Pictorial Representation, pp. 4062; James
Cutting, Representing Motion in a Static Image: Constraints and Parallels in Art,
Science, and Popular Culture, Perception 31 (2002): 11651193; and Scott McCloud,
Understanding Comics: The Invisible Art (New York: Harper, 1993), p. 95. See also

Figure 10.1 Ron Chapman, Bicycles, 1981. Photograph.




Figure 10.2 Erwin von Dessauer, Children on the Beach, ca. 1933. Gelatin silver
print, 28 25.3 cm. ML/F 1983/182.

accomplishing this trick.3 But I will be interested primarily in the experiences

viewers enjoy when they look at still depictions of motion or change, however
the depiction is accomplished.
There would seem to be no trick at all in making motionless marks depict
motionless scenes, as in gure 10.4, just as there is no mystery about how sculptures, themselves solid chunks of stone or steel or clay, manage to portray solid
three dimensional objects, or how at paintings can portray at surfaces (as

various psychological studies of motion perception by Jennifer J. Freyd and collaborators, e.g., The Mental Representation of Movement when Static Stimuli Are Viewed,
Perception and Psychophysics 33, no. 6 (1983): 575581; and Five Hunches about Perceptual Processes and Dynamic Representations, in Attention and Performance 14, ed.
David Meyer and Sylvan Kornblum (Cambridge, Mass.: MIT Press, 1993), pp. 99119.
Robin Le Poidevin develops an account of how static images depict change and movement, using a notion of depiction different from mine, in The Images of Time: An Essay on
Temporal Representation (Oxford: Oxford University Press, forthcoming). See also Robin Le
Poidevin, Time and the Static Image, Philosophy 72, no. 280 (1997): 175188.
3. For an interesting discussion of some important techniques, and their strengths and
weaknesses for various purposes, see Cutting, Representing Motion.

Figure 10.3 Udo Ernst Block, Ewiger Kreislauf. Photo courtesy of the artist.




Figure 10.4 Kendall L. Walton, Mt. Geryon. Kendall L. Walton. Photo courtesy
of the artist.

Jasper Johnss paintings of targets and ags do). But the portrayal of static scenes
in the medium of still pictures, which I will examine in section 3, is much more
problematic than one would expect, and viewers experiences of stasis-depicting
still pictures are, if anything, even more puzzling and more fascinating than
their experiences of motion-depicting ones.
Most of my discussion of both topics applies equally to photographic pictures
and to hand-made ones (paintings, drawings, etc.), and I will use examples of
both kinds. There are some extra wrinkles in the case of photographs, however,
and photographs will be especially useful when it comes to comparing still pictures with moving ones.

Still and Moving Pictures

What are still pictures? How do they differ from moving ones? The answer is
not as easy as it seems. Moving pictures do not necessarily move. Andy Warhols
Empire (1964)a lm consisting of a single eight-hour, dawn-to-dusk shot of
the Empire State Buildingis nearly motionless, and may be entirely so during
periods when nothing in the scene before the camera changed.4 And still pictures
need not be still: paintings fade over time, a picture drawn in sand erodes, an
4. Arthur Danto once wrote that moving pictures are just that: pictures which move, not
just (or even necessarily at all) pictures of moving things (Moving Pictures, Quarterly



artist touches up his canvas or a restorer cleans it, pigeons deposit droppings
on it. Still pictures can depict either motion or stasis, as we noted, and moving
pictures can do so as well.5 The essential difference between still and motion
pictures consists neither in temporal properties of the images themselves nor
in their representational content, but in the relation between the two, the relation between changes or lack of them in pictures, and changes or lack of them
in picture worlds.
What is it about images of the two kinds by virtue of which they depict movement or stasis? In the case of motion pictures, temporal properties of the images
do the job.6 Absent camera mobility, movement of the image of a horse on a
movie screen depicts the horse as running; when the image stops, it depicts the
horse as stopped. With carefully arranged camera movement, a stationary horse
image on a changing background might depict the horse in motion, or a changing image might depict it as stationary. But it remains true that temporal properties of the screen image are responsible for the depiction of the horses temporal
Still pictures work differently. Whether they depict movement or stasis
depends not on what happens to the image over time but on features of the
image that are present all at once. Blurred images, or marks depicting unstable
congurations of objects serve to portray motion or change, as in our recent
examples; other all-at-once properties such as motion lines and multiple images
do the job in other cases, as we shall see. This is why stills can depict motion
or change without moving or changing themselves. If a still picturethe conguration of marks on its surfacedoes change, the process of change does not
have the representational signicance it would have in lm. Fading paint or
restoration work or pigeon droppings may alter a pictures depictive content;
what the picture depicts after the change may be different from what it depicted
before. But a change in what something depicts is not necessarily a depiction of
change. If a prankster or exceptionally clever pigeons transform the surface of
Mt. Geryon (gure 10.4) so that it depicts a monster instead of a mountain, it
has not thereby depicted the transformation of a mountain into a monster, nor

Review of Film Studies 2 [1979]: 15; italics in original). He can hardly have meant the former claim, for he introduces, as a hypothetical example, eight hours of Warhol-inspired
lm footage of the title page of War and Peace.
5. Both can depict the absence of visible motion or change. Motion pictures are no
better than still ones at depicting imperceptible motion or change. See the section below
titled Static Depictions of Static Scenes: Mt. Geryon.
6. As Gregory Currie and Gideon Yaffe put it, lms represent time by means of time.
See Currie, Image and Mind: Film, Philosophy and Cognitive Science (Cambridge: Cambridge
University Press, 1995); and Yaffe, Time in the Movies, in Midwest Studies in Philosophy:
Meaning in the Arts, ed. Peter A. French and Howard K. Wettstein (Oxford: Blackwell,



has it depicted a mountain being replaced by a monster. We just have a picture

which, at one time, depicts a mountain and later a monster; the mountain and
the monster need not be understood to belong to the same ctional world.7 And
at any or all particular moments in a changing images history, it may depict
things as unchanging.8
What features of still pictures, of the Mt. Geryon photograph for instance, are
responsible for depicting things as unchanging? Most obvious is the fact that
the marks on the photographic surface are such as to depict something we would
expect not to changenothing like the kids on the beach with their limbs in the
air. Also, the image is not blurred, as the bicycle picture is, nor does the picture
contain anything like motion lines or multiple images. Again, each of these features (or absences thereof ) is present in the picture all at once.
Freeze frames in lmstill pictures, in effect, inserted in the midst of moving
onesdemonstrate dramatically the difference between still and moving pictures. For a moment, until she realizes that it is a freeze frame, the viewer may
read the frozen image as portraying a frozen scenean athlete or dancer stuck in
midair, for instance. Once it is evident that the image is a still picture, once she
understands it as such, all-at-once features of the unmoving image may induce
her to see the athlete or dancer as in motion.
These examples suggest a general account of the difference between still and
moving pictures: A picture is a still one if temporal properties of the image
are representationally inert, if what happens or doesnt happen to the image
over time has no bearing on its representational content. Motion pictures are
pictures whose temporal properties do contribute to their representational

Picture Perception
The issues concerning temporality that I am interested in cannot be addressed or
even formulated very well without saying something about what pictures are and
what it is to perceive pictures as such. I will treat these issues in the framework of
an account of picture perception that I have developed elsewhere, and utilize its
resources in resolving them.9 This is not a neutral decision. The problems look

7. We may prefer to say that, if the marks on the surface of a picture change, it
becomes a numerically different picture, especially if the changes alter the pictures representational content. If this is so, it is even more obvious that the changes effected by the
prankster or pigeons do not depict change. We have a succession of two pictures on the
same surface, the rst depicting a mountain and the second a monster.
8. As undergoing no visible changes, anyway.
9. Kendall L. Walton, Mimesis as Make-Believe: On the Foundations of the Representational
Arts (Cambridge, Mass.: Harvard University Press, 1990), chap. 8. See also On Pictures
and Photographs: Objections Answered, chap. 7, this volume; and Depiction, Perception,



Figure 10.5 Peter Paul Rubens, An Autumn Landscape with a View of Het Steen
in the Early Morning. Reproduced by permission of the National Gallery.

somewhat different against the background of other theories, and on somevery

implausible ones in my opinionthey would disappear almost entirely.10
Here are the bare outlines of the account of picture perception that I favor.
I will say nothing now in its defense. When a person views Rubenss An Autumn
Landscape with a View of Het Steen in the Early Morning (gure 10.5),
(a) she sees the picture, the pattern of colored marks on the picture surface;
(b) she imagines seeing the depicted scene: trees and elds, a horse cart and a
hunter, clouds in the background, and buildings on the left;
(c) she imagines her actual perceiving of the picture surface to be a perceiving
of the trees and elds, clouds, horse cart, and so on.

These imaginings are rarely undertaken deliberately, of course. The viewer

doesnt decide to imagine seeing trees and elds, and so on, set out to imagine
them, and then do it. The marks on the canvas are such as to trigger the imaginings more or less automatically. She might, however, choose which part of the
picture to look at when, knowing that this will affect which particular imaginings she will nd herself engaged in as a result, in partly foreseeable ways.
and Imagination, chap. 9, this volume. My claims in the present essay require only that,
in perceiving pictures, the imaginatively colored perceptual experiences that I think constitute pictorial perception are ones we usually or typically enjoy; we neednt agree that
they are sufcient or even necessary for such perception.
10. On Nelson Goodmans account of pictorial representation, there would seem to
be no room for the puzzles I address here. See Goodman, Languages of Art (Indianapolis:
Bobbs-Merrill, 1968).



(a), (b), and (c) are necessary for seeing the painting as a picture of trees and
elds and buildings, and so on, for seeing them in the visual design (as Richard
Wollheim would put it). But the viewer is likely to engage in further, more
detailed imaginings as well, as her eyes scan the picture surface. She may imagine
surveying the scene for an extended period of time, noticing rst this, then that,
looking for this or that, and so forth. A typical viewer leisurely contemplating
Rubenss canvas in a typical manner might describe her experience as follows:
Looking out on a vast landscape, I see a hunter and his dog in the shadows near
me, as a farm wagon passes by, then catch sight of a small footbridge crossing
a stream, and beyond that a group of grazing cattle. I look to see what the
hunters quarry might be, and notice several birds in the eld. Examining them
more carefully I identify them as partridges. I pick out a blackberry bush in
the foreground. I watch the clouds drifting lazily across the sky, and several
birds circling above. Finally I catch sight of a village on the horizon, the town
of Malines.

These words could describe a persons experience as he looks out on an actual

rural scene. The viewer of the painting uses them to report her imaginative experience, what she imagines seeing as she scans the canvas. It is as though she is
surveying an actual landscape for several or many minutes, for as long as she
continues to look at the painting.
All of the above applies to photographs as well as paintings, and also to motion
pictures, not to mention other depictive representations such as sculptures and
theater. It may seem pretty innocuous, but qualms will set in soon enough.


The techniques that picture makers use to depict motion and change are even
more varied than our three initial examples (gures 10.110.3) suggest.11 Scott
McCloud has demonstrated some of the tricks of the cartoonists tradeseveral
varieties of motion lines and multiple images (gure 10.6).
Multiple images are used occasionally in other media as well, in this photograph by Harold Edgerton (gure 10.7), for instance, and in a famous painted
example (gure 10.8).
The multiple image technique has been tried even in sculpture (gure 10.9).
Some of the devices our several examples employ are less obvious than others.
The overall diagonal composition of Children on the Beach (gure 10.2)

11. The various devices well illustrate Patrick Maynards notion of tools in picture
makers toolkit. See Maynard, Drawing Distinctions (Ithaca, N.Y.: Cornell University
Press, 2005).

Figure 10.6 Scott McCloud, Drama, 1994. Courtesy of the artist.


Figure 10.7 Harold E. Edgerton, Gussie Moran Tennis Multiash, 1949. The
Harold E. Edgerton Foundation, 2004. Courtesy of Palm Press, Inc.


Figure 10.8 Marcel Duchamp, Nude Descending a Staircase, 1912. 2004 Artist Rights
Society (ARS), New York /ADAGP, Paris/Succession Marcel Duchamp.

Figure 10.9 Kendall L. Walton, photograph of Terry Allens Shaking Man, 1993.
Kendall L. Walton. Photo courtesy of the artist.




Figure 10.10 George Booth. Dog and Vase (New Yorker, 8 September 2003). The
New Yorker Collection 2003. From All rights reserved.

enhances its sense of motion, reinforcing the unstable attitudes of the childrens
bodies. Nude Descending a Staircase (gure 10.8) works especially well partly
because stepson a staircase, and those taken by a personare discrete stages,
corresponding to the more or less discrete images on the canvas. I would expect
multiple images to be less successful in portraying bicycles rolling smoothly
around a corner or the smooth motions of a tennis serve; blurred images are likely
to serve better in these cases.
Pictures can represent movement without depicting it. The distinction is nicely
illustrated by gure 10.10. Viewers see the dog scratching itself and the vase
jiggling, and infer that the vase will fall, probably on the dog. The picture
represents this entire unfortunate sequence of eventseverything that occurs in
the world of the picture. But it depicts, in my sense, only what viewers (properly)
imagine seeingthe scratching and the jiggling, not the vases falling. It is the
depictive content of pictures that I am primarily interested in, although what a
picture merely represents sometimes affects what it depicts.12

A Puzzle
Each of the motion-depicting still pictures before us depicts a momentary state of
affairs, a very short time slice of a changing world. That is what we imagine seeing

12. I distinguished between depiction and nondepictive representation in Mimesis as

Make-Believe, p. 297. The line between seeing and inferring, and, hence, that between
depicting and mere representing, is not sharp, especially if we allow for nonconscious
inferences, and there is room for disagreement about where best to draw it.



when we look at it. Yet we may observe the picture, we may continue seeing
the momentary state of affairs, for ve minutes or an hour or all day. How can
this be? How can we observe or even imagine observing a eeting moment of
reality for an indenitely extended period of time? Does the viewer imagine
her prolonged visual experience of Children on the Beach or Bicycles to be
a mere glimpse of the kids running toward the surf, or the bicycles rounding
the corner? Most (not all) motion-depicting still pictures present this puzzle.
Depicting objects in a particular conguration but as undergoing change, they
depict the conguration as lasting only a moment; yet the picture is there to be
seen for as long as we care to look.
Pictures may represent much longer sequences of events than they depict, of
course. It is ctional in Children on the Beach that the left foot of the guy on
the right will very shortly touch the sand, that the kids came from somewhere
farther up on the beach, and that they will soon be in the surf. But we dont see
these earlier and later states of affairs; we infer them from the one moment of
the race to the water that we do see. The puzzle consists of the fact that we can
continue looking at it indenitely.
By a moment I do not mean a mathematical instant, an extensionless point
of time. That would make it hard to understand how viewers imagine seeing
things moveor for that matter, how they see things, in imagination, at all. The
momentary event depicted in Bicycles, what viewers imagine seeing, arguably
lasts for one-fteenth of a second, the duration of the lms exposure when the
photograph was taken. Other depicted moments may be somewhat longer or
shorter, although there will usually be no denite answer as to exactly how long
they are. Motion lines and multiple images, like the blurs in Bicycles, extend
the space moving objects are depicted as traversing, and so the duration of the
time they spend traversing it. Figures 10.7 and 10.8 depict events lasting several
secondsthe entire execution of a tennis serve, and several steps of a persons
progression down a stairway.
I caution against regarding Nude Descending, for instance, as a combination of
several different (overlapping) pictures within a single frame, each depicting
a single short momentary state of affairs. That would ignore the importance
of attending to the several images all at once, and it would risk assimilating
Nude Descending to very different uses of multiple images in renaissance narrative paintings and story-telling scrolls. Masolinos fresco Healing of Cripple
and the Raising of Tabitha (gure 10.11) depicts St. Peter twiceonce healing a cripple and again raising Tabitha from the dead, two distinct moments
in St. Peters lifebut it doesnt depict his moving from one task to the
other. Observing the two images together does not induce viewers to imagine
seeing movement. On viewing Nude Descending, by contrast, one does see,
in imagination, the person taking several steps down the stairs. (I at least
nd this an apt way of describing my experience. If the reader does not, he
still must recognize and describe somehow the enormous difference between



Figure 10.11 Masolino da Panicale, Healing of Cripple and the Raising of Tabitha,
14241425). Photo credit: Scala / Art Resource, New York.

ones experiences of Nude Descending and of the twice-depicted St. Peter in

Masolinos fresco.)
Figure 10.12 is more complicated. It represents an extended series of events,
but I will leave to the reader the intriguing task of guring out what it depicts,
what the viewer imagines seeing. (Keep in mind that reading the text in the balloons can affect ones imaginative visual experience.)
The duration of depicted events varies from case to case, as we have seen. Nevertheless, their duration, in all the motion-depicting pictures we have looked at,
is likely to be far shorter than the time one spends contemplating the picture.

Figure 10.12 Scott McCloud, Smile, 1994. Courtesy of the artist.



What is depicted, what the viewer imagines seeing, has a more or less denite,
limited duration, yet there is no theoretical limit to how long one can look at
the picture. This paradox (if that is what it is) doesnt often bother viewers when
they look at still pictures. It is easy enough to nurture a sense of wonder once
we start thinking about our experiences, but enduring depictions of momentary
states of affairs do not ordinarily strike us as strange or puzzling or mysterious.
It is surprising that they dont; this is part of what needs to be explained.

A. Mt. Geryon
The puzzle demands that we examine more carefully the imaginative experiences
motion that depicting pictures induce in viewers. This is best done against a
contrasting background. So lets take a preliminary look at gure 10.4, which
does not depict motion and seems not to generate the puzzle.
Lingering before this photographfor ve minutes, lets sayI might describe
my experience in a narrative like the one I placed in the mouth of a hypothetical spectator of Rubenss Autumn Landscape, a narrative worthy of reporting an
extended visual examination of the Tasmanian mountain scene in the esh:
Looking toward the forbidding hulk of Mt. Geryon, I notice that it is almost
entirely without vegetation. My eye is drawn to the jagged ridge line. Then
I focus on the vertical ssures, looking for a climbing route to the top. Eventually,
I notice the mountains far in the background.

So I imagine spending ve minutes looking at the scene; I imagine my ve

minute visual experience of the picture to be a ve minute visual experience
of the mountain. If this is right then, surely, I imagine what I see of the mountain to persist for ve minutes; I imagine seeing a certain ve minute stretch of
its history, a ve-minute-thick time slice of it. Moreover, I see, in imagination,
the observed slice of the mountain lasting for ve minutes; it appears to me to
remain for as long as I see it. I imagine what I see of the mountain to appear to
me to persist for ve minutes. So each of the following is ve minutes long (cf.
row A in table 10.1):
(1) The duration of my (actual) seeing of the picture, which I imagine to be a
seeing of the mountain.
(2) What I imagine to be the duration of my seeing of the mountain.
(3) What I imagine to be the apparent duration of the time slice of the
mountain that I see.
(4) What I imagine to be the duration of the time slice of the mountain that
I see.

Table 10.1
What the viewer imagines to be:

(1) The duration of

the viewers seeing(s)
of the picture, which
she imagines to be of
the scene/event

(2) the duration of the

viewers seeing(s) of the

(3) the apparent duration

of the seen time slice of the

(4) the (actual) duration of

the seen time slice of the

A Mt. Geryon?; Film of

Mt. Geryon

5 minutes

5 minutes

5 minutes

5 minutes

B Empty Box

5 minutes

5 minutes

5 minutes


C Beach and Bicycles:

1st try.

5 minutes

5 minutes



D (slow motion) Beach

and Bicycles: 2nd try

5 minutes




E Beach and Bicycles:

3rd try. Mt. Geryon?





F (fast motion) Gussie

Moran; Nude Descending


a long moment:
3 seconds

a long moment:
3 seconds

a long moment:
3 seconds



There is no puzzle or paradox or mystery here, if this is right. I will assume for
now, that it is, although we will reconsider in section 3.
B. Empty Box
In Ryoji Akiyamas Empty Box on Its Way to a Reclamation Area
(gure 10.13), the camera freezes the action, as we say, in a way that it doesnt
in our previous examples of motion depictions. Do I imagine the box stuck in
midair? In any case arguably it appears to me, in my imaginative experience, to
be stuck in midair; I seem to see it stuck therefor the ve minutes that I spend
observing it. The box looks much like the sphere in a surrealistic photograph by
Jerry Uelsmann (gure 10.14).
But I gure out that the box in Akiyamas photograph is actually falling rather
quickly. (The title helps, and that inuences my imaginative experience.) Once I
do, I imagine what I see to be moving, not remaining in the position I see it in
for more than a moment. It still appears to me (in imagination) to be stuck there
for ve minutes, however, if I observe the picture for that long. So columns (3)
and (4) in table 10.1 come apart. What I imagine to be the (actual) duration of
the time slice of the box that I see is only a moment. But what I imagine to be
its apparent duration will still be ve minutes (row B).
The picture represents but doesnt depict the boxs motion, since I dont (in
imagination) see the box move. But the box is unlike the falling vase in Dog and
Vase in one respect: I do imagine seeing the box as it falls; I imagine that it is
moving while I am seeing it, while it appears to me to be suspended indenitely
in midair. I imagine suffering from a peculiar visual illusion, an illusion of a
kind that I am not susceptible to in any ordinary, real life circumstances. (This
explains the weirdness of the picture.)
If column (4) is momentary, how can (2) be ve minutes? If (in my imagination) the time slice of the box that I see lasts only a moment, how can I be seeing
it for ve minutes? We will address this question shortly.
C. Beach and Bicycles: First Try
What about our more ordinary puzzle cases? Unlike Empty Box, Children on
the Beach and Bicycles offer a vivid sense of motion; the objects in these
pictures dont seem frozen in the way the box does. We dont have to gure out
that the children and bicycles are in motion, and imagine accordingly; we see
them moving (in imagination, of course).13 So, in my imaginative experience,
it appears to me that what I see of the bicyclists and the kids, the position I see
13. The region [of the brain] specialized for visual motion processing is activated by
implied motion from static images. Sarah-Jayne Blakemore and Jean Decety, summarizing ndings of Jennifer J. Freyd (Five Hunches), in From the Perception of Action to
the Understanding of Intention, Nature Reviews/Neuroscience 2 (2001): 562.

Figure 10.13 Ryoji Akiyama, Empty Box on Its Way to a Reclamation Area, 1969.
Gelatin silver print. Courtesy of the artist.




Figure 10.14 Jerry N. Uelsmann, Untitled, 1979. Jerry N. Uelsmann.

Photo courtesy of the artist.

them in, lasts only a moment. (3) changes to momentary.14 But my imaginative
experience extends for ve minutes, the time I spend looking at the picture (1).
For ve minutes I imagine seeing the children or the bicyclists in a certain position, and in motion! This seems to mean that, in my imagination, the duration
of my seeing of the kids or the bicyclists is ve minutes (2). So I imagine spending
ve minutes observing something that lasts only a moment (row C).
14. Arthur Danto writes that we see things move when we watch moving pictures,
but we merely see that things are in motion, where this involves inference, in the case of
still representations such as Michelangelos David (Moving Pictures, p. 17). This leaves
unexplained the difference between stills like Empty Box and ones like Children on
the Beach and Bicycles. (David, I take it, is more like the latter than the former.)



This is strange, but not because there is anything incoherent, in general, about
the idea that perceiving an event might last longer than the event perceived.
This can happen in real life, let alone in imagination. Ones visual impression
of a ash of light may outlast the ash, by a fraction of a second; arguably one
continues seeing the ash after it stops existing.15 One can hear a shot (almost)
when it happens, when the sound arrives at ones ears, and also hear it a little
later, when an echo arrives.
One possibility is that what viewers of Children on the Beach and Bicycles
imagine is like the ash of light, except that they imagine the impression of
the event continuing for a full ve minutes. And unlike the ash of light, they
can, in imagination, continue seeing the momentary position of the bicyclists or
the running children indenitely, as long as they care to look at the picture. If we
are willing to say that dogs can smell a persons presence at a certain place (not
just the smell of the person, something the person leaves behind when she moves
on), they can presumably continue smelling the person not only long after she
has left the scene, but more or less indenitelyas long as the smell remains.
But this is smelling, not seeing.
I have no knockdown argument against this option. But it strikes me as
implausiblemainly because it seems to me that what we imagine experiencing
when we look at pictures like these is seeing of a very ordinary kind. One doesnt,
ordinarily, continue seeing a momentary slice of moving bicycles or running kids
for as long as one likes.16 In any case, there are other ways of understanding the
experience of these two photographs: another implausible one, and one (my third
try) which has viewers imagining only garden-variety visual experiences.
D. Beach and Bicycles: Second Try
Could column (2) be momentary? Could it be that what I spend ve minutes
imagining is an instantaneous act of seeing, seeing for an instant something that
lasts but an instant? This is not incoherent. There is no logical or conceptual
reason why the duration of ones imagining of an experience must be the same as
the duration of the experience imagined. After all, lots of other features of imaginings are different from features of the imagined experiences. (An imagining

15. Thanks to Alex Byrne.

16. I claim that in looking at a photograph one actually sees, indirectly, what was
photographed. A person who observes Children on the Beach, for ve minutes or
indenitely, sees an instantaneous event for ve minutes or indenitely. (Thanks to Alon
Chasid.) But this is seeing-via-photography, not direct seeing. Seeing-via-a-photograph
is a special mode of perception that is unusual in other respects as well; with the help of
photographs, one can also see the past and around corners. But what one imagines as one
observes a photograph is not, it seems to me, the indirect seeing-via-the-photograph that
one actually engages in but is, instead, direct, naked-eye seeing.



which occurs in a coffee shop in Ann Arbor may be of an experience in the

jungles of Borneo.)
Zenon Pylyshyn considers an example of the opposite kind: an imagining
which supposedly is much shorter in duration than the experience imagined:
In discussing how he imaged his music, Mozart claimed: Nor do I hear in the
imagination, the parts successively, but I hear them, as it were, all at once . . . 17
Mozarts imagining is supposedly instantaneous, and what he imagines in that
instant is having an extended auditory experience. Pylyshyn seems to think this
is incoherent:
If what [Mozart] claimed to be doing was that he imagined witnessing the
real event of, say, . . . hearing his Symphony Number 40 being played . . . and
if he imagined that it was actually happening before him in real time and in
complete detailincluding the most minute ourishes of the horns and the
trills of the ute and oboe, all in the correct temporal relations and durations
he would have taken nearly 22 minutes for the task. If he had taken less time, it
would signify only that Mozart had not been doing exactly what he said he was
said he was doing, that is he would not have been imagining that he witnessed
the actual event in which every note was played at its proper duration.18

But why suppose that it must take 22 minutes to imagine listening for 22
minutes to a musical performance? I understand that dream researchers have
established, by means of studies of rapid eye movements, that a dream of a
lengthy experience may occur in a very short period of time. There might be
omissions in the dream, but even with some omissions, surely the imagined
duration of the events that are included may exceed the few moments occupied
by the dreaming. (Also, the order of dreaming can be the reverse of the order
of the dreamed events. The actual sound of an alarm clock is dreamed to be
the ringing of a school bell, guaranteeing a tardy slip after a series of mishaps
prevented one from getting to school on time. Presumably the dreaming of the
earlier events didnt occur before the hearing of the alarm and the dreaming
17. Zenon Pylyshyn, Scanning Visual Mental Images: The First Phase of the Debate,
in The Philosophy of Mind: Classical Problems/Contemporary Issues, ed. Brian Beakley and
Peter Ludlow (Cambridge, Mass.: MIT Press, 1992), p. 230. Whether Mozart actually
claimed this and what exactly he meant by it do not matter for our purposes.
18. Pylyshyn Scanning Visual Mental Images, p. 230. Imagining that he witnessed
the event is not at all problematic. Pylyshyn must mean imagining witnessing (hearing)
the performance.
Mozarts claim, as quoted, is that he hears the parts of a piece simultaneously (possibly in a mathematical instant, although the simultaneity of the imaginings doesnt entail
that they are instantaneous). What Pylyshyn appears to think impossible is not just this
but also imagining hearing the parts of the piece successively, imagining this in a period
of time shorter than 22 minutes. I am interested now only in this weaker possibility.



of the school bellhow would the dreamer have known the sound would
Consider experiencing slow motion in lm. Here, the duration of imagining
may be longer than the duration of the imagined experience. Suppose that a shot
of a two-second eventa man running across a streetis slowed to six seconds.
Viewers might imagine the man running very slowly, taking six seconds to cross
the street, and they might imagine their observing of his crossing the street to
occupy the entire six seconds that they spend watching the slow-motion shot.
But it is likely, it seems to me, that viewers who are used to the device of slow
motion sometimes imagine seeing a normally paced event and seeing it in the
time it takes for it to happen, even though it takes them longer to imagine this.
The duration of the imagining is greater than the duration of the imagined seeing. A viewer may imagine her six-second watching of the slow-motion shot
to be a two-second observation of the mans two-second dash across the street.
There is perhaps some support for this suggestion in the fact that some viewers
seem not even to notice that an event is portrayed in slow motion, and dont
remember the portrayed event as having occurred slowly.20
Should we regard Children on the Beach and Bicycles as extreme instances,
limiting cases, of slow motionas very, very slow-motion depictions, indenitely long depictions, of momentary events? If I am right about slow-motion
sequences in lm, then, one may spend ve minutes imagining seeing, for a
moment, a momentary event. That would make (2) momentary (row D).
This does not work. Still pictures are not instances of slow motion. Indeed,
we will see that some of them are more like instances of fast motion! When I look
at Bicycles or Children on the Beach for ve minutes it doesnt take me the
entire ve minutes to imagine engaging in a momentary act of seeing, to imagine seeing the momentary event that is depicted. I did it in the rst minute,
the rst moment, of my ve-minute observation of the picture. But it takes all
six seconds of watching the slow-motion shot to imagine seeing the entire twosecond crossing of the street. After three seconds, I have imagined seeing only
half the event; I will have followed the man only to the middle of the street.
E. Beach and Bicycles: Third Try
Here is a better account of my experience of Bicycles and Children on the
Beach: At each individual moment during my ve-minute observation of the
photograph, I see (in imagination) the momentary occurrence that the picture depicts, a short time slice of the moving bicycles or the running kids, and
see it in a moment. At each moment I imagine my visual experience of the picture, at that moment, to be a glimpse of this momentary event. I do not imagine
19. Thanks to Ian Proops.
20. Thanks to Alicyn Warren.



anything of my ve-minute visual experience of the picture as a whole; only the

momentary parts of this experience are objects of my imaginings. So column
(1) is momentaryit being understood that there are lots of different moments
strung together, all of them adding up to ve minutes. Do I imagine seeing the
event again and again? No. I imagine, again and again, seeing it just once.
Compare the following:
(1) If I attend two different performances of Hamlet, on each of the two occasions I will imagine Hamlet killing Polonius. But it is hardly likely that I should
imagine Polonius dying twice at the hand of Hamlet.
(2) At the climax of Michelangelo Antonionis Zabriske Point (1970), a huge
explosion destroys the desert retreat of an evil real estate tycoon. The explosion
is shown again and again in the lm (from different perspectives and sometimes
in slow motion). The viewer repeatedly imagines seeing the explosion, but surely
she does not imagine its occurring repeatedly, or imagine seeing it repeatedly.
She imagine, repeatedly, observing a single explosion at a single moment.
The experience of looking at Bicycles or Children on the Beach is similar
to these experiences, except that the multiple successive imaginings are not discrete but continuous (separated only by other moments at which one also imagines
seeing running children or moving bicycles). I dont imagine my seeing of the
children or bicycles to occupy multiple moments, nor do I imagine this seeing to
continue for the ve minutes that my succession of imaginings occupy. The following, then, are all momentary: my seeings of the picture which I imagine to be seeings
of bicycles or kids, my imagined seeings of the bicycles or kids, and what I imagine to
be the apparent and the actual time slices of bicycles and kids that I see (row E).
If this is right, what am I doing when I scan the picture, my eyes roving from
one part of it to another, during a ve-minute viewing? I am scanning the picture.
I am not, in imagination, surveying the scene, examining one part of it, then
another, noticing rst this (e.g., the kid on the left), then that (the small kid on
the right, the surf in front of them, their reections in the wet sand), looking for
one or another feature (a toy boat in the water) and nding or not nding it. The
duration of the imagined seeing, of any of the momentary imagined seeings, is
too short for this to be possible. We can put this negative point by saying that
I do not, in imagination, watch the kids as they run toward the water, although
I do see them running.
In focusing on different parts of the picture, I am arranging for my imaginings
at various moments to be imagining the seeing of various particular features of
the momentary event. First I imagine focusing on the kid on the left, as the four
of them race to the surf; then I imagine seeing this instantaneous event with my
eyes focused on the kid on the right. (Compare television reruns of a play of the
day. First I watch to see whether the receiver catches the pass, then I watch the
same play to see whether his feet are in bounds when he catches it.)



F. Fast Motion
Lets look again at still depictions of relatively thick time slices of objects in
motion: Edgertons Gussie Moran Tennis Multiash and Duchamps Nude
Descending a Staircase (gures 10.7 and 10.8). The seen-in-imagination time
slice is three seconds long, lets say, in each of these cases. Suppose I look at the
picture for ve minutes. As in the previous examples, I dont imagine watching this three second event for ve minutes, nor do I imagine my ve-minute
visual experience of the picture to be a ve-minute observation of the womans
descent or Gussie Morans tennis serve. It doesnt take me even three seconds
to imagine seeing the event; I do it in an instant, in whatever tiny bit of time
it takes for the picture and its content to register (one-tenth second?). I might
blink my eyes open and shut, and between the blinks imagine observing the
three-second event.21 This seems to be approximately what, according to Pylyshyn, Mozart claimed to have done when he imagined an entire symphony in an
instant, except that perhaps Mozart did the imagining just once, in one moment,
and then stopped, whereas I look at the picture for ve minutes. At each moment
of my ve-minute perceiving of it, I imagine observing the three-second descent
or the three-second tennis serve, and imagine my momentary perception of the
picture to be a three-second observation of this three-second event. So column
(1) is momentary, and (4) is three seconds. This makes for an instance not of
slow motion but of fast motionthe depicted event is longer than the time
of depiction.
Assuming that (2) and (3) are also 3 seconds, we have row F. We might think
of these examples as special cases of E, except that the moments in column
(4), as well as (2) and (3), last three seconds, while the column (1) moment is
shorter. Is there a limit to how long column (4) moments might be? What
about a Nude Descendinglike picture in which multiple images portray a person
descending all the way to the bottom of the stairs, then going outside, across the
street, and into a coffee shopcall it Nude Descending a Staircase and Going Out
for Coffee. I doubt that I would, in the blink of an eye, imagine seeing all of this
when I observe this picture, although it might take no longer than this for me to
notice that all of it is depicted. More likely, I think, I would imagine observing
the womans extended trek as my eye follows the images on the picture surface,
focusing on different clusters of them in order. (The duration of this series of
imaginings might be the same as, or longer or shorter than, the duration of the
imagined trek.) This experience of Nude Descending a Staircase and Going Out for
Coffee would be like that of an experience of Masolinos fresco in some respects,
and like that of a motion picture depiction of the womans excursion to the coffee
shop in others.

21. Thanks to Eileen John.



Second Thoughts: Children on the Beach

If what I have said so far is right, viewers experience still photographs in very
different ways, depending (with some exceptions) on whether the photographs
portray objects in motion or static states of affairs. Mt. Geryon, I suggested, is
experienced in mode A (in the manner indicated in row A in table 10.1) and Children on the Beach in mode E. Skepticism is in order on both counts, though for
very different reasons. There is more than a little to be said for the hypothesis that
we experience Mt. Geryon in mode E. And an A-style experience of Children on
the Beach is not out of the question. (It goes without saying that viewers need not
be clearly aware of the nature of their experience, or able to articulate the content
of the imaginings it involves.) I will save my second thoughts about Mt. Geryon
for section 3. Lets look more carefully, now, at Children on the Beach.
There can be no doubt thatusually, normallyviewers of Children on the
Beach do imagine seeing the kids running. Given that, their experience is bound
to be in mode E. We can hardly (in imagination) spend ve minutes watching
the children running for ve minutes, since at each moment of our experience we
imagine seeing the kids in the same position. But this is probably not the whole
story. I suspect that at some level or in some manner, we imagine seeing the kids
frozen in a certain, admittedly unnatural, position, and experience the photograph
in mode A. (Or perhaps in mode Bperhaps we imagine its merely appearing to
us that the kids are frozen?) We certainly can imagine this without much effort, if
we choose to. And we can imaginatively survey the frozen scene, scan it, in the way
one might survey the Tasmanian mountain-scape. I suspect that we sometimes slip
into experiencing the photograph in this way without quite realizing what we are
doing.22 Or possibly we feel a tendency to imagine this, or are vaguely aware of
such a tendency, or of the ease with which we could so imagine. I dont rule out the
possibility that the best explanation of our responses to the photograph will have it
that we experience it (implicitly) in mode A (or B?) and mode E simultaneously.
It is entirely obvious, I take it, that the picture is not properly understood to
depict the children frozen in one position, or even as appearing to be motionless.
This means that A- and B-style experiences of it are unauthorized, involving
what I have called an unofcial game of make-believe. They may occur nonetheless, and may play an important part in our appreciation of the picture.23

22. Thanks to Daniel Herwitz.

23. Walton, Mimesis as Make-Believe, pp. 405411. Although I recognized, in Mimesis,
the possibility of unofcial games of make-believe and their importance in what we say
about works of ction, as well as claims of existence and nonexistence, I did not then realize how important these games are in our appreciation of representational works. Also,
in Mimesis I thought of unofcial make-believe as being primarily (as I later expressed
it) prop oriented (Metaphor and Prop Oriented Make-Believe, European Journal of Philosophy 1, no. 1 [1993]: 3957). But the unofcial games engaged in by appreciators are
often largely content- rather than prop-oriented.



Complicated multifaceted experiences of this kind, ones that include imaginings which conict with our judgments about what is depicted, are not uncommon. Barbara Savedoff observes that Walker Evanss photograph Torn Movie
Poster (1930)
is grotesque at least in part because the two dimensionality of the photograph
allows us to read the torn picture as a torn woman. . . . We see the woman as torn,
but we also see the wall coming throughasserting the fact that, after all, what
we see is a photograph of a poster peeling off a wall. . . . The photograph leads to
a double vision: we know we are looking at a photograph of a poster, but we see
it as a photograph of a woman.24

Photographs and other pictures whose frames cut off part of a persons face or
body are sometimes (not always) disturbing. The explanation of this reaction
may be that we imagine, implicitly, the person as deformed or cut off, or are
vaguely aware of a tendency so to imagine, even though it is clear that what the
picture depicts, and what we primarily and most explicitly imagine, is an ordinary whole person of whom we see only part.
That experiences of works of art are complicated and subtle is certainly not
news. But our examination of several kinds of still photographs has identied
and claried a number of their likely ingredients. These include (a) the imaginings that we actually engage ineither deliberately or spontaneously, perhaps
inadvertently, and either consciously or tacitly; (b) our impressions or judgments,
explicit or not, about which imaginings are prescribed or called for or appropriate, given the nature of the work and the traditions it is embedded in; and (c)
our impressions of what imaginings are possible or natural or easily engaged in,
whether or not we actually engage in them, and whether or not we take them
to be prescribed. The lesson to be learned from our consideration of Children
on the Beach is that questions about (a)what imaginings we actually engage
inmay admit of no simple answer, let alone an introspectively accessible one.
The same is true of (b) and (c), insofar as these are linked to (a).
These second thoughts about Children on the Beach do not apply equally,
if at all, to all motion depicting still pictures. It seems to me that viewers are
much less likely to imagine the bicyclists in Bicycles frozen in one position, than the boys in Children on the Beach. This difference is a result of
the very different devices the two pictures use for depicting motion, no doubt,
the fact that the bicyclists are blurred and the children are not. Gussie Moran
and Nude Descending a Staircase are different again. The unauthorized imaginings
that might creep into ones experience of them are, respectively, imagining a

24. Barbara Savedoff, Transforming Images: How Photography Complicates the Picture
(Ithaca, N.Y.: Cornell University Press, 2000), p. 53.



Figure 10.15 Kendall L. Walton, Mississagi River Rapids, 2006. Kendall L. Walton.
Photo courtesy of the artist.

many-armed tennis player, possibly frozen in a single position, and a parade of

staircase descenders.
Another variety of motion-depicting stills, which we have not yet considered, might be experienced in mode A without the dissonance Children on the
Beach is likely to involve.
The motion depicted in gure 10.15, raging rapids forming standing waves,
is not such as to render the appearance of the scene much different as time
passes. So imagining a ve-minute observation of the picture to be a ve-minute
observation of a ve-minute temporal chunk of the rapids does not require imagining the river to be, or appear to be, frozen. One can unproblematically imagine
continuing to see the water in motion, watching it moving, for ve minutes or
indenitely. Similar examples are common, including depictions of rising steam
or smoke, and of turning wheels if they are sufciently blurred. The motion lines
in Dog and Vase (gure 10.10) are such as to allow viewers to imagine watching
the dog scratching and the vase jiggling for an extended period of time, although
after awhile this imagining will conict with the fact that ctionally the scratching and jiggling continue only until the vase falls.
If Mt. Geryon is properly experienced in mode A, there would seem to be
no reason to deny that Mississagi River Rapids and other similar motiondepicting pictures are as well.




Static Depictions of Static Scenes: Mt. Geryon
I turn now to still pictures of static scenes and viewers experiences of them
which I predicted may be even more interesting than their experiences of motiondepicting stills.
Mt. Geryon (gure 10.4) will be my primary example, but its credentials as
a depiction of stasis need clarication. The photograph depicts the mountain as
severely eroded. Arguably it is ctional also, by implication, that it is currently
in the process of erodingvery slowly, even on a bright sunny day. Viewers dont
imagine seeing the erosion occur, so the picture represents but doesnt depict
the process. But surely it does not depict the erosion as not occurring; if it did,
its depictive content would conict with its nondepictive representational content. So the photograph does not depict total stasis or absolute immobility. It
does, presumably, depict the mountain as undergoing no visible changes, however (none visible from a certain perspective under certain conditions). I will
count this as constituting a depiction of stasis, and this is what I will mean in
what follows when I speak of depicting stasis or motionlessness. (It wont matter
for our purposes exactly how this is spelled out.) We can hardly expect viewers
to imagine seeing the absence of imperceptible change; so we shouldnt expect
pictures to depict things as not changing or moving at all. But if any pictures do,
what I say about Mt. Geryon will apply to them as well.
Motion pictures of static scenes can (in principle, anyway) be visually indistinguishable from still pictures of the same scenes. Consider a lmed portrayal of the
Tasmanian mountain scene, a ve-minute unedited shot made by a xed camera,
with no panning or zooming. The motion picture image is likely to be motionless
for as long as it lasts, and indistinguishable from the still photograph.25 If what
I have said so far is right, we might expect ones experience of the lm to be no different from that of looking at the still photograph for ve minutes. Observing the
lm image, one imagines seeing Mt. Geryon and imagines ones actual perceiving
of the image to be a perceiving of the mountain. And the lm viewer surely imagines continuing to observe the mountain scene for the ve minutes that the shot
lastsfocusing rst on one part of the scene, then another, looking for a climbing route, and so on. The extended narrative mentioned earlier is equally apt as
a description of a viewers experience of either picture. We experience the lm in
mode A, just as (as I tentatively suggested) we do the still photograph.

25. Peter Huttons [lm] New York Near Sleep for Saskia (1972) consists of a series of
beautifully photographed, deserted, mostly urban spaces. . . . The most remarkable feature
of these images, other than their beauty, is their lack of movement. So still are these
images that we might mistake some of them for photographs James Peterson, Dreams
of Chaos, Visions of Order: Understanding the American Avant-Garde Cinema (Detroit, Mich.:
Wayne State University Press, 1994), p. 34. Thanks to Murray Smith.



Something has gone wrong. For even if we are unable to tell the pictures
apart, our experiences of themour visual experiences of themare likely to
be very different (assuming that we realize that we are seeing a still picture
in one case and a lm, a motion picture, in the other). The two images of Mt.
Geryon do not appear different, but they appear differently. They afford different
visual experiences without giving the impression of being different.26 A plausible explanation of the difference is that the still photograph is experienced in
mode E, whereas the motion picture is experienced in mode A. It may be that
still pictures in generalstasis-depicting as well as motion-depicting onescall
for E style experiences, and motion pictures call for A style experiences. This
hypothesis ts nicely with what I said earlier about how still pictures differ, in
general, from motion pictures.

Experiencing Stasis-Depicting Stills

Viewers imagine of their perceiving of a picture that it is a perceiving of the
depicted object or event. Normally, of course, it is their perception of the
particular features of the picture by virtue of which the object or event is depicted
that they imagine to be their perceiving of the object or event. It is my perception of the marks on the surface of Rubenss canvas that depict a cart, those that
prescribe imagining a cart, that I imagine to be my perception of a cart.
The Mt. Geryon lm depicts the mountain as persisting for ve minutes, as
remaining motionless for ve minutes. It does so by virtue of the persistence of
the static image for the ve-minute duration of the shot. So, naturally, viewers
imagine their perception of the persisting image to be an observation of the
mountains persistence for a period of ve minutes.
The still photograph depicts the mountain scene as motionless, unchanging.
Does it depict the mountain as persisting unchanged for an extended period of
timefor ve minutes, or indenitely, or more than a moment? If so, all-at-once
features of the image, features viewers perceive in a moment, are responsible.
And we would expect viewers to imagine their perception of these features to
be their perceiving of the mountains extended persistence. This hardly seems
likely (though it is not inconceivable). Do viewers, in their rst glimpse of the
picture, imagine watching the mountain remain for ve minutes (or longer),
imagine observing a ve-minute time slice of its history? Surely it is more likely
that, in their initial sighting of the picture, they imagine seeing merely that the
mountain will last, or inferring that it will from what they see of it. What about
the next moment of the viewers experience after the rst one, and the nextthe
remainder of her ve-minute observation of the picture? Does she, at each of these
26. On this distinction, see Style and the Products and Processes of Art, chap. 12,
this volume; and Robert Hopkins, Aesthetics, Experience, and Discrimination, Journal
of Aesthetics and Art Criticism 63, no. 2 (2005): 119133.



moments, imagine spending ve minutes (or longer) watching ve minutes of

the mountains history? Does she imagine this again and again, as she continues
to look at the picture? The obvious suggestion is that her experience is in mode
E. At each moment she imagines catching a momentary glimpse of a momentary
state of the mountain, seeing that it is, at that moment, a massive, stable object,
momentarily static, and that it wont or is unlikely to change later. She does not,
in imagination, continue watching the mountain for an extended period of time,
and she does not, in imagination, observe a ve minute stretch of its history.
If this is right, if the viewer (properly) imagines seeing only a moment of the
mountains history, that is all of it that the picture depicts. It may nevertheless
represent the mountain as lasting longer; it may make it ctional that the mountain persists for ve minutes or indenitely. Some may propose that it is ctional
only that the mountain is likely to remain for ve minutes or indenitely, not
that it denitely will, nor for that matter, that it denitely wont. I prefer to say
that, ctionally, the mountain denitely does remain static more or less indenitely (but not foreversee below), and that features of the still image present
at each moment, including the very rst moment of the pictures existence, make
this ctional. This accords with a common pattern in our understanding of many
works of ction: If it is ctional that p is probable, it is frequently understood to
be ctional that p, absent any indication to the contrary.27 But no matter. Both
positions are entirely compatible with the idea that the picture depicts only a
momentary state of affairs, that viewers have E style experiences of it, and do not
imagine observing the mountain for more than a moment.
Now for some ip ops: We should not rush to endorse the hypothesis that viewers experience pictures like Mt. Geryon in mode E rather than mode A, or that
this is the natural or appropriate way of experiencing them. If it is right, extended
narrative accounts of ones experience of still pictures, such as those I mentioned
earlier, are problematic, as they appear to describe temporally extended imaginative surveys of depicted scenes.28 Such accounts certainly seem to be perfectly
natural, and they seem to report experiences that are perfectly proper. Viewers of
Mt. Geryon obviously can, if they choose, imagine catching much more than
a glimpse of the mountain. As their eyes roam over the picture surface, they can
imagine noticing rst this and then that, looking for something and nding or
not nding it, and so on. The photograph is visually indistinguishable from our
hypothetical stasis-depicting motion picture, and one can simply experience it as
though that is what it isas though it is a lm shot that continues indenitely, or
as long as one cares to look at it. Even if the viewer is well aware that the image is
a still picture, that its persistence is not what makes it ctional that the mountain
27. Cf. Walton, Mimesis as Make-Believe, 164. That the mountain is unchanging for an
extended period of time is an implied ctional truth (section 4.2).
28. These narratives might conceivably be construed in some less straightforward
way, on which they do not describe imaginative experiences of this kind.



persists, she might imagine her observation of the former to be an observation of

the latter. I have no doubt that viewers of stasis depicting stills do (sometimes?
often? to a considerable extent?) engage in such imaginings. And the idea that it
is somehow illegitimate to do so is not attractive.
But simply denying the hypothesis and insisting that viewers do (properly)
imagine observing a stretch of the scenes history lasting perhaps as long as they
look at the picture, would also leave us in an uncomfortable position. We already
know that it would complicate the task of explaining the difference between
experiences of still- and motion-picture look-alikes. But the discomfort doesnt
end there.
Mt. Geryon surely does not represent, let alone depict, the mountain as
remaining forever the same. Signs of severe past erosion are evident, as we noted.
It is ctional, presumably, that the erosion will continue and that eventually
the mountain will look very different from how it does now. Indeed, we might
reasonably understand it to be ctionalpart of the pictures representational
contentthat the appearance of the scene will be different very shortly, when the
sun sets or moves across the sky. If a viewer should observe the picture for hours
or months or years, will she have to imagine suffering from an illusion that the
mountain and its appearance are always the same?
Moreover, it is awkward to suppose that motion-depicting still pictures
(except, presumably, ones like gure 10.15 [Mississagi River Rapids]) and
stasis-depicting ones are experienced in radically different ways, the former in
mode E and the latter in mode A. The awkwardness is especially evident when
both motion-depicting and stasis-depicting images are included within the same
frame, as they are in gure 10.16. Does the viewer (in imagination) catch but a
glimpse of the moving horses, as they kick up a cloud of dust, but observe the
equestrian statues continuously for as long as he cares to gaze at the picture? And
what are we to say about the live horses temporarily at rest?
What shall we conclude? Given the conicts built into the medium of still
pictures, we should not expect any simple account of viewers imaginative experiences to be fully adequate. Perhaps the most we can say, in general, is that
people experience stasis depicting stills in a complicated combination of ways,
alternating imperceptibly between, or subtly intertwining, imagined glimpses
and prolonged visual surveys of depicted scenes.
Viewers are not usually aware of these complexities or bothered by the conicts. This is partly because, like so much of our mental lives, the imaginings
our experiences of pictures involve, or some aspects of them, are implicit and
not readily open to introspectionevident only, if at all, when we pay unusual
attention or draw inferences from what we say and do. Viewers rarely if ever
even have the alternatives we have considered clearly in mind. Also, some of
the questions I recently left hanging, questions highlighting the complexities
or confronting the conicts, are silly in the (slightly) technical sense I introduced in Mimesis as Make-Believe: they do not appropriately arise in the course



Figure 10.16 Jacques Philippe Joseph de Saint-Quentin (b. 1738), The Entrance to the
Tuileries from the Place Louis XV in Paris, c. 1775. Oil on canvas, 46 61 cm. Musee
des Beaux-Arts et dArcheologie, Besancon, France / Lauros / Giraudon / The Bridgeman
Art Library Nationality / copyright status: French / out of copyright.

of appreciation, appreciators are advised to avoid them, and they are likely not
to have denite answers.29 This goes for questions about the nature of viewers
experiences as well as questions about what pictures represent or depict.
What a picture depicts is (roughly) a matter of what viewers properly imagine
seeing, when they experience it. If as I suspect there are no especially relevant
propriety considerations affecting the kinds of experiences I have described,
there will be complications and uncertainties about pictures depictive contents
corresponding to those characterizing viewers imaginative experiences. We will
have to be satised with the recognition that pictures like Mt. Geryon are
ambiguous between depicting a momentary event and a longer period of the history of a scene, and with the realization that this is not a clean ambiguity; each
of the readings interferes, potentially at least, with the other.
Some may nd these conclusions frustratingly messy and inconclusive. I think
they reect facts about pictures and our experiences of them which are messy and
hard to pin downfacts which, partly for that very reason, may contribute to the
immense fascination that many pictures have for us.

29. Walton, Mimesis as Make-Believe, 174183.



Berger, J. 1980. Uses of Photography. In About Looking. New York: Pantheon,
pp. 4863.
Blakemore, S.-J., and J. Decety. 2001. From the Perception of Action to the Understanding of Intention. Nature Reviews/Neuroscience 2:561567.
Borcoman, J. 1993. Magicians of Light: Photographs from the Collection of the National Gallery
of Canada. Ottawa: National Gallery of Canada.
Currie, G. 1995. Image and Mind: Film, Philosophy and Cognitive Science. Cambridge:
Cambridge University Press.
Cutting, J. E. 2002. Representing Motion in a Static Image: Constraints and Parallels
in Art, Science, and Popular Culture. Perception 31:11651193.
Danto, A. C. 1979. Moving Pictures. Quarterly Review of Film Studies 2:121.
Freyd, J. J. 1983. The Mental Representation of Movement When Static Stimuli Are
Viewed. Perception and Psychophysics 33(6): 575581.
. 1992. Dynamic Representations Guiding Adaptive Behavior. In Time,
Action and Cognition: Towards Bridging the Gap, edited by F. Macar, V. Pouthas, and
W. J. Friedman, pp. 309323. Dordrecht: Kluwer Academic Publishers.
. 1993. Five Hunches about Perceptual Processes and Dynamic Representations. In Attention and Performance 14, edited by D. Meyer and S. Kornblum,
pp. 99119. Cambridge, Mass.: MIT Press.
Gombrich, E. H. 1980. Standards of Truth: The Arrested Image and the Moving Eye.
Critical Inquiry 7 (2): 237273.
. 1994. Moment and Movement in Art. In The Image and the Eye: Further Studies
in the Psychology of Pictorial Representation. London: Phaidon.
Goodman, N. 1968. Languages of Art. Indianapolis: Bobbs-Merrill.
Hopkins, R. 2005. Aesthetics, Experience, and Discrimination. Journal of Aesthetics and
Art Criticism 63(2): 119133.
Maynard, P. 2005. Drawing Distinctions. Ithaca, N.Y.: Cornell University Press.
McCloud, S. 1993. Understanding Comics: The Invisible Art. New York: Harper.
Peterson, J. 1994. Dreams of Chaos, Visions of Order: Understanding the American AvantGarde Cinema. Detroit, Mich.: Wayne State University Press.
Pylyshyn, Z. W. 1992. Scanning Visual Mental Images: The First Phase of the Debate.
In The Philosophy of Mind: Classical Problems/Contemporary Issues, edited by B. Beakley,
and P. Ludlow, pp. 229240. Cambridge, Mass.: MIT Press.
Savedoff, B. 2000. Transforming Images: How Photography Complicates the Picture. Ithaca,
N.Y.: Cornell University Press.
Walton, K. L. 1979. Style and the Products and Processes of Art (this volume).
. 1990. Mimesis as Make-Believe: On the Foundations of the Representational Arts.
Cambridge, Mass.: Harvard University Press.
. 1993. Metaphor and Prop Oriented Make-Believe. European Journal of
Philosophy 1(1): 3957.
. 1997. On Pictures and Photographs: Objections Answered [Reprinted, this
. 2002. Depiction, Perception, and Imagination: Responses to Richard Wollheim [Reprinted, this volume].
Yaffe, G. 2003. Time in the Movies. In Midwest Studies in Philosophy: Meaning in the
Arts, edited by P. A. French and H. Wettstein, pp. 115138. Oxford: Blackwell.


This page intentionally left blank


False judgments enter art history if we judge from the impression

which pictures of different epochs, placed side by side, make on
us. . . . They speak a different language.
Heinrich Wlfin, Principles of Art History

Paintings and sculptures are to be looked at; sonatas and songs are to be heard.
What is important about such works of art is what can be seen or heard in
them.1 This apparent truism has inspired attempts by aesthetic theorists to
purge from criticism of works of art supposedly extraneous excursions into matters not available to inspection of the works and to focus attention narrowly on
the works themselves. Circumstances connected with a works origin, in particular, are frequently held to have no essential bearing on an assessment of its
aesthetic nature. Thus, critics are advised to ignore how and when a work was
created, the artists intentions in creating it, his philosophical views, psychological state and personal life, the artistic traditions and intellectual atmosphere
of his society, and so forth. Once produced, it is argued, a work must stand or
fall on its own; it must be judged for what it is, regardless of how it came to
be as it is.
Arguments for this position need not involve the claim that how and in what
circumstances a work comes about is not of aesthetic interest or importance.
1. We should all agree, I think, . . . that any quality that cannot even in principle be
heard in it [a musical composition] does not belong to it as music. Monroe Beardsley, Aesthetics: Problems in the Philosophy of Criticism (New York: Harcourt Brace, 1958), pp. 3132.




One might consider an artists action of producing a work to be aesthetically

interesting, an aesthetic object in its own right, while vehemently denying its
relevance to an aesthetic investigation of the work. Robert Rauschenberg once
carefully obliterated a drawing by de Kooning, titled the bare canvas Erased de
Kooning Drawing, framed it, and exhibited it.2 His doing this might be taken
as symbolic or expressive (of an attitude toward art, or toward life in general,
or whatever) in an aesthetically signicant manner, and yet thought to have
no bearing whatever on the aesthetic nature of the nished product. The issue
I am here concerned with is how far critical questions about works of art can be
separated from questions about their histories.3
One who wants to make a sharp separation here may regard the basic facts of
art along the following lines. Works of art are simply objects with various properties, of which we are primarily interested in perceptual onesvisual properties of paintings, audible properties of music, and so forth.4 A works perceptual
properties include aesthetic as well as nonaesthetic onesthe sense of mystery and tension of a painting as well as its dark coloring and diagonal composition; the energy, exuberance, and coherence of a sonata, as well as its meters,
rhythms, pitches, timbres; the balance and serenity of a Gothic cathedral as well
as its dimensions, lines, and symmetries.5 Aesthetic properties are features or
characteristics of works of art just as much as nonaesthetic ones are.6 They are
in the works, to be seen, heard, or otherwise perceived there. Seeing a paintings
sense of mystery or hearing a sonatas coherence might require looking or listening longer or harder than does perceiving colors and shapes, rhythms and
pitches; it may even require special training or a special kind of sensitivity. But
these qualities must be discoverable simply be examining the works themselves

2. See Calvin Tompkins, The Bride and the Bachelors (New York: Viking Press, 1965),
pp. 210211.
3. Monroe Beardsley argues for a relatively strict separation (Aesthetics, pp. 1734).
Some of the strongest recent attempts to enforce this separation are to be found in discussions of the so-called intentional fallacy, beginning with William Wimsatt and
Beardsley, The Intentional Fallacy, Sewanee Review 54 (1946), which has been widely
cited and reprinted. Despite the name of the fallacy these discussions are not limited to
consideration of the relevance of artists intentions.
4. The aesthetic properties of works of literature are not happily called perceptual.
For reasons connected with this it is sometimes awkward to treat literature together with
the visual arts and music. (The notion of perceiving a work in a category, to be introduced
shortly, is not straightforwardly applicable to literary works.) Hence in this paper I will
concentrate on visual and musical works, though I believe that the central points I make
concerning them hold, with suitable modications, for novels, plays, and poems as well.
5. Frank Sibley distinguishes between aesthetic and nonaesthetic terms and
concepts in Aesthetic Concepts, Philosophical Review 68 (1959).
6. Cf. Paul Ziff, Art and the Object of Art, in Ziff, Philosophic Turnings (Ithaca,
N.Y.: Cornell University Press, 1966), pp. 1216 (originally published in Mind, N. S.
LX [1951]).



if they are discoverable at all. It is never even partly in virtue of the circumstances
of a works origin that it has a sense of mystery or is coherent or serene. Such circumstances sometimes provide hints concerning what to look for in a work, what
we might reasonably expect to nd by examining it. But these hints are always
theoretically dispensable; a works aesthetic properties must in principle be
ascertainable without their help. Surely (it seems) a Rembrandt portrait does not
have (or lack) a sense of mystery in virtue of the fact that Rembrandt intended it
to have (or to lack) that quality, any more than a contractors intention to make
a roof leakproof makes it so; nor is the portrait mysterious in virtue of any other
facts about what Rembrandt thought or how he went about painting the portrait
or what his society happened to be like. Such circumstances are important to the
result only insofar as they had an effect on the pattern of paint splotches that
became attached to the canvas, and the canvas can be examined without in any
way considering how the splotches got there. It would not matter in the least to
the aesthetic properties of the portrait if the paint had been applied to the canvas
not by Rembrandt at all, but by a chimpanzee or a cyclone in a paint shop.
The view sketched above can easily seem very persuasive. But the tendency of
critics to discuss the histories of works of art in the course of justifying aesthetic
judgments about them has been remarkably persistent. This is partly because
hints derived from facts about a works history, however dispensable they may be
in principle, are often crucially important in practice. (One might not think
to listen for a recurring series of intervals in a piece of music, until he learns that
the composer meant the work to be structured around it.) No doubt it is partly
due also to genuine confusions on the part of critics. But I will argue that certain
facts about the origins of works of art have an essential role in criticism, that aesthetic judgments rest on them in an absolutely fundamental way. For this reason,
and for another as well, the view that works of art should be judged simply by
what can be perceived in them is seriously misleading. Nevertheless there is
something right in the idea that what matters aesthetically about a painting or a
sonata is just how it looks or sounds.


I will continue to call tension, mystery, energy, coherence, balance, serenity, sentimentality, pallidness, disunity, grotesqueness, and so forth, as well as colors
and shapes, pitches and timbres properties of works of art, though property is
to be construed broadly enough not to beg any important questions. I will also,
following Sibley, call properties of the former sort aesthetic properties, but
purely for reasons of convenience I will include in this category representational and resemblance properties, which Sibley excludesfor example, the
property of representing Napoleon, that of depicting an old man stooping over a
re, that of resembling, or merely suggesting, a human face, claws (the petals of



Van Goghs sunowers), or (in music) footsteps or conversation. It is not essential

for my purposes to delimit with any exactness the class of aesthetic properties (if
indeed any such delimitation is possible), for I am more interested in discussing
particular examples of such properties than in making generalizations about the
class as a whole. It will be obvious, however, that what I say about the examples
I deal with is also applicable to a great many other properties we would want to
call aesthetic.
Sibley points out that a works aesthetic properties depend on its nonaesthetic
properties; the former are emergent or Gestalt properties based on the latter.7
I take this to be true of all the examples of aesthetic properties we will be dealing with, including representational and resemblance ones. It is because of the
conguration of colors and shapes on a painting, perhaps in particular its dark
colors and diagonal composition, that it has a sense of mystery and tension, if it
does. The colors and shapes of a portrait are responsible for its resembling an old
man and its depicting an old man. The coherence or unity of a piece of music (for
example, Beethovens Fifth Symphony) may be largely due to the frequent recurrence of a rhythmic motive, and the regular meter of a song plus the absence of
harmonic modulation and of large intervals in the voice part may make it serene
or peaceful.
Moreover, a work seems or appears to us to have certain aesthetic properties
because we observe in it, or it appears to us to have, certain nonaesthetic features
(though it may not be necessary to notice consciously all the relevant nonaesthetic features). A painting depicting an old man may not look like an old man
to someone who is color-blind, or when it is seen from an extreme angle or in
bad lighting conditions which distort or obscure its colors or shapes. Beethovens
Fifth Symphony performed in such a sloppy manner that many occurrences of
the four-note rhythmic motive do not sound similar may seem incoherent or
I will argue, however, that a works aesthetic properties depend not only on its
nonaesthetic ones but also on which of its nonaesthetic properties are standard,
which variable, and which contra-standard, in senses to be explained. I will
approach this thesis by way of the psychological point that what aesthetic properties a work seems to us to have depends not only on what nonaesthetic features
we perceive in it but also on which of them are standard, which variable, and
which contra-standard for us (in a sense also to be explained).
It is necessary to introduce rst a distinction between standard, variable, and
contra-standard properties relative to perceptually distinguishable categories
of works of art. A category is perceptually distinguishable if membership in it
is determined solely by features of works that can be perceived in them when
they are experienced in the normal manner. The categories of painting, cubist

7. Aesthetic and Nonaesthetic, Philosophical Review 72 (1965).



painting, Gothic architecture, classical sonatas, painting in the style of Czanne,

music in the style of late Beethoven, and most other media, genre, styles, and
forms can be construced as perceptually distinguishable. If we do construe them
this way we must, for example, regard whether a piece of music was written in
the eighteenth century as irrelevant to whether it belongs to the category of classical sonatas, and we must take whether or not a work was produced by Czanne
or Beethoven to have nothing essential to do with whether or not it is in the style
of Czanne or late Beethoven. The category of etchings as normally understood
is not perceptually distinguishable in the requisite sense, for to be an etching
is, I take it, to have been produced in a particular manner. But the category of
apparent etchings, works that look like etchings from the quality of their lines,
whether or not they are etchings, is perceptually distinguishable.8
A feature of a work of art is standard with respect to a (perceptually distinguishable) category just in case it is among those in virtue of which works in that
category belong to that categorythat is, just in case the absence of that feature
would disqualify, or tend to disqualify, a work from that category. A feature is
variable with respect to a category just in case it has nothing to do with works
belonging to that category; the possession or lack of the feature is irrelevant to
whether a work qualies for the category. Finally, a contra-standard feature with
respect to a category is the absence of a standard feature with respect to that
categorythat is, a feature whose presence tends to disqualify works as members of the category. Needless to say, it will not be clear in all cases whether
a feature of a work is standard, variable, or contra-standard relative to a given
category, since the criteria for classifying works of art are far from precise. But
clear examples are abundant. The atness of a painting and the motionlessness
of its markings are standard, and its particular shapes and colors are variable,
relative to the category of painting. A protruding three-dimensional object or
an electrically driven twitching of the canvas would be contra-standard relative
to this category. The straight lines in stick-gure drawings and squarish shapes
in cubist paintings are standard with respect to those categories respectively,
though they are variable with respect to the categories of drawing and painting.
The exposition-development-recapitulation form of a classical sonata is standard,
and its thematic material is variable, relative to the category of sonatas.
In order to explain what I mean by features being standard, variable, or contrastandard for a person on a particular occasion, I must introduce the notion of perceiving a work in, or as belonging to, a certain (perceptually distinguishable) category.

8. A category will not count as perceptually distinguishable in my sense if, in order

to determine perceptually whether something belongs to it, it is necessary (in some or
all cases) to determine, on the basis of nonperceptual considerations, which categories it
is correctly perceived in. This prevents the category of serene things, for example, from
being perceptually distinguishable.



To perceive a work in a certain category is to perceive the Gestalt of that category in the work. This needs some explanation. People familiar with Brahmsian
musicmusic in the style of Brahms (notably, works of Johannes Brahms)or
impressionist paintings can frequently recognize members of these categories
by recognizing the Brahmsian or impressionist Gestalt qualities. Such recognition is dependent on perception of particular features that are standard relative
to these categories, but it is not a matter of inferring from the presence of such
features that a work is Brahmsian or impressionist. One may not notice many of
the relevant features, and he may be very vague about which ones are relevant.
If I recognize a work as Brahmsian by rst noting its lush textures, its basically
traditional harmonic and formal structure, its superimposition and alternation of
duple and triple meters, and so forth, and recalling that these characteristics are
typical of Brahmsian works, I have not recognized it by hearing the Brahmsian
Gestalt. To do that is simply to recognize it by its Brahmsian sound, without necessarily paying attention to the features (cues) responsible for it. Similarly, recognizing an impressionist painting by its impressionist Gestalt, is recognizing
the impressionist look about it, which we are familiar with from other impressionist paintings; not applying a rule we have learned for recognizing it from its
To perceive a Gestalt quality in a workthat is, to perceive it in a certain
categoryis not, or not merely, to recognize that Gestalt quality. Recognition is a
momentary occurrence, whereas perceiving a quality is a continuous state which
may last for a short or long time. (For the same reason, seeing the ambiguous
duck-rabbit gure as a duck is not, or not merely, recognizing a property of it.)
We perceive the Brahmsian or impressionist Gestalt in a work when, and as long
as, it sounds Brahmsian or looks impressionist to us. This involves perceiving (not
necessarily being aware of ) features standard relative to that category. But it is
not just this, nor this plus the intellectual realization that these features make the
work Brahmsian, or impressionist. These features are perceived combined into a
single Gestalt quality.
We can of course perceive a work in several or many different categories at once.
A Brahms sonata might be heard simultaneously as a piece of music, a sonata, a
romantic work, and a Brahmsian work. Some pairs of categories, however, seem
to be such that one cannot perceive a work as belonging to both at once, much
as one cannot see the duck-rabbit both as a duck and as a rabbit simultaneously.
One cannot see a photographic image simultaneously as a still photograph and as
(part of ) a lm, nor can one see something both in the category of paintings and
at the same time in the category (to be explained shortly) of guernicas.
It will be useful to point out some of the causes of our perceiving works in
certain categories. (a) In which categories we perceive a work depends in part, of
course, on what other works we are familiar with. The more works of a certain
sort we have experienced, the more likely it is that we will perceive a particular
work in that category. (b) What we have heard critics and others say about works



we have experienced, how they have categorized them, and what resemblances
they have pointed out to us are also important. If no one has ever explained to
me what is distinctive about Schuberts style (as opposed to the styles of, say,
Schumann, Mendelssohn, Beethoven, Brahms, Hugo Wolf ), or even pointed
out that there is such a distinctive style, I may never have learned to hear the
Schubertian Gestalt quality, even if I have heard many of Schuberts works, and
so I may not hear his works as Schubertian. (c) How we are introduced to the
particular work in question may be involved. If a Czanne painting is exhibited
in a collection of French Impressionist works, or if before seeing it we are told
that it is French Impressionist, we are more likely to see it as French Impressionist than we would be if it is exhibited in a random collection and we are not told
anything about it beforehand.
I will say that a feature of a work is standard for a particular person on a particular occasion when, and only when, it is standard relative to some category
in which he perceives it, and is not contra-standard relative to any category in
which he perceives it. A feature is variable for a person just when it is variable
relative to all of the categories in which he perceives it. And a feature is contrastandard for a person just when it is contra-standard relative to any of the categories in which he perceives it.9


I turn now to my psychological thesis that what aesthetic properties a work seems
to have, what aesthetic effect it has on us, how it strikes us aesthetically often
depends (in part) on which of its features are standard, which variable, and which
contra-standard for us. I offer a series of examples in support of this thesis.
(1) Representational and resemblance properties provide perhaps the most
obvious illustration of this thesis. Many works of art look like or resemble other
objectspeople, buildings, mountains, bowls of fruit, and so forth. Rembrandts
Titus Reading looks like a boy, and in particular like Rembrandts son; Picassos
Les Demoiselles dAvignon looks like ve women, four standing and one sitting
9. I am ignoring some considerations that might be important at a later stage of
investigation. In particular, I think it would be important at some point to distinguish
between different degrees or levels of standardness, variableness, and contra-standardness
for a person; to speak, e.g., of features being more or less standard for him. At least two
distinct sorts of grounds for such differences of degree should be recognized. (1) Distinctions between perceiving a work in a certain category to a greater and lesser extent
should be allowed for, with corresponding differences of degree in the standardness for
the perceiver of properties relative to that category. (2) A feature which is standard relative to more, and/or more specic, categories in which a person perceives the work should
thereby count as more standard for him. Thus, if we see something as a painting and also
as a French Impressionist painting, features standard relative to both categories are more
standard for us than features standard relative only to the latter.



(though not especially like any particular women). A portrait may even be said to
be a perfect likeness of the sitter, or to capture his image exactly.
An important consideration in determining whether a work depicts or represents
a particular object, or an object of a certain sort (for example, Rembrandts son,
or simply a boy), in the sense of being a picture, sculpture, or whatever of it10
is whether the work resembles that object, or objects of that kind. A signicant
degree of resemblance is, I suggest, a necessary condition in most contexts for
such representation or depiction,11 though the resemblance need not be obvious
at rst glance. If we are unable to see a similarity between a painting purportedly
of a woman and women, I think we would have to suppose either that there is
such a similarity which we have not yet discovered (as one might fail to see a face
in a maze of lines), or that it simply is not a picture of a woman. Resemblance is
of course not a sufcient condition for representation, since a portrait (containing
only one gure) might resemble both the sitter and his twin brother equally but
is not a portrait of both of them. (The title might determine which of them it
It takes only a touch of perversity, however, to nd much of our talk about
resemblances between works of art and other things preposterous. Paintings and
people are very different sorts of things. Paintings are pieces of canvas supporting
splotches of paint, while people are live, three-dimensional, esh-and-blood animals. Moreover, except rarely and under special conditions of observation paintings and people look very different. Paintings look like pieces of canvas (or anyway
at surfaces) covered with paint and people look like esh-and-blood animals.

10. This excludes, e.g., the sense of represent in which a picture might represent
justice or courage, and probably other senses as well.
11. This does not hold for the special case of photography. A photograph is a photograph of a woman no matter what it looks like, I take it, if a woman was in front of the
lens when it was produced.
[Categories of Art was written before I developed my make-believe account of pictorial representation. I no longer understand resemblance to have as much to do with
depiction as I did then (unless resemblance is understood in a very special sense). See
the several essays on depiction in this volume, and especially Mimesis as Make-Believe, pp.
298, 302315, 350. As for photographs, I should have distinguished between what a
photograph is a picture of, and what it is a photograph of. See Postscript C to Transparent
Pictures, this volume.]
12. Nelson Goodman denies that resemblance is necessary for representationand
obviously not merely because of isolated or marginal examples of non-resembling representations (p. 5). I cannot treat his arguments here, but rather than reject en masse the
common-sense beliefs that pictures do resemble signicantly what they depict and that
they depict what they do partly because of such resemblances, if Goodman advocates
rejecting them, I prefer to recognize a sense of resemblance in which these beliefs are
true. My disagreement with him is perhaps less sharp than it appears since, as will be
evident, I am quite willing to grant that the relevant resemblances are conventional.
See Goodman, Languages of Art (Indianapolis: Bobbs-Merill, 1968), p. 39, n. 31.



There is practically no danger of confusing them. How, then, can anyone seriously hold that a portrait resembles the sitter to any signicant extent, let alone
that it is a perfect likeness of him? Yet it remains true that many paintings strike
us as resembling people, sometimes very much or even exactlydespite the fact
that they look so very different!
To resolve this paradox we must recognize that the resemblances we perceive
between, for example, portraits and people, those that are relevant in determining what works of art depict or represent, are resemblances of a somewhat special
sort, tied up with the categories in which we perceive such works. The properties
of a work which are standard for us are ordinarily irrelevant to what we take it to
look like or resemble in the relevant sense, and hence to what we take it to depict
or represent. The properties of a portrait which make it so different from, so easily distinguishable from, a personsuch as its atness and its painted lookare
standard for us. Hence these properties just do not count with regard to what (or
whom) it looks like. It is only the properties which are variable for us, the colors
and shapes on the works surface, that make it look to us like what it does. And
these are the ones which are relevant in determining what (if anything) the work
Other examples will reinforce this point. A marble bust of a Roman emperor
seems to us to resemble a man with, say, an aquiline nose, a wrinkled brow,
and an expression of grim determination, and we take it to represent a man
with, or as having, those characteristics. But why dont we say that it resembles
and represents a perpetually motionless man, of uniform (marble) color, who is
severed at the chest? It is similar to such a man, it seems, and much more so
than to a normally colored, mobile, and whole man. But we are not struck by
the former similarity when we see the bust, obvious though it is on reection.
The busts uniform color, motionlessness, and abrupt ending at the chest are
standard properties relative to the category of busts, and since we see it as a bust
they are standard for us. Similarly, black-and-white drawings do not look to us
like colorless scenes and we do not take them to depict things as being colorless, nor do we regard stick-gure drawings as resembling and depicting only
very thin people. A cubist work might look like a person with a cubical head to
someone not familiar with the cubist style. But the standardness of such cubical
shapes for people who see it as a cubist work prevents them from making that
The shapes of a painting or a still photograph of a high jumper in action are
motionless, but these pictures do not look to us like a high jumper frozen in midair.

13. The connection between features variable for us and what the work looks like is
by no means a straightforward or simple one, however. It may involve rules which are
more or less conventional (e.g., the laws of perspective). See E. H. Gombrich, Art and
Illusion (New York: Pantheon Books, 1960) and Nelson Goodman, Languages of Art.



Indeed, depending on features of the pictures which are variable for us (the exact
positions of the gures, swirling brush strokes in the painting, slight blurrings of
the photographic image) the athlete may seem in a frenzy of activity; the pictures
may convey a vivid sense of movement. But if static images exactly like those of the
two pictures occur in a motion picture, and we see it as a motion picture, they probably would strike us as resembling a static athlete. This is because the immobility
of the images is standard relative to the category of still pictures and variable relative to that of motion pictures. (Since we are so familiar with still pictures it might
be difcult to see the static images as motion pictures for very long, rather than as
[lmed] still pictures. But we could not help seeing them that way if we had no
acquaintance at all with the medium of still pictures.) My point here is brought out
by the tremendous aesthetic difference we are likely to experience between a lm
of a dancer moving very slowly and a still picture of him, even if objectively the
two images are very nearly identical. We might well nd the former studied, calm,
deliberate, laborious, and the latter dynamic, energetic, owing, or frenzied.
In general, then, what we regard a work as resembling, and as representing,
depends on the properties of the work which are variable, and not on those which
are standard for us.14 The latter properties serve to determine what kind of a
representation the work is, rather than what it represents or resembles. We take
them for granted, as it were, in representations of that kind. This principle helps
to explain also how clouds can look like elephants, how diatonic orchestral music
can suggest a conversation or a person crying or laughing, and how a twelveyear-old boy can look like his middle-aged father.
We can now see how a portrait can be an exact likeness of the sitter, despite
the huge differences between the two. The differences, in so far as they involve
properties standard for us, simply do not count against likeness, and hence not
against exact likeness. Similarly, a boy not only can resemble his father but can
be his spitting image, despite the boys relative youthfulness. It is clear that
the notions of resemblance and exact resemblance that we are concerned with are
not even cousins of the notion of perceptual indistinguishability.
(2) The importance of the distinction between standard and variable properties
is by no means limited to cases involving representation or resemblance. Imagine
a society that does not have an established medium of painting but does produce
a kind of work of art called guernicas. Guernicas are like versions of Picassos
Guernica done in various bas-relief dimensions. All of them are surfaces with the
colors and shapes of Picassos Guernica, but the surfaces are molded to protrude
from the wall like relief maps of different kinds of terrain. Some guernicas have
14. There is at least one group of exceptions to this. Obviously features of a work
which are standard for us because they are standard relative to some representational category which we see it ine.g., the category of nudes, still lifes, or landscapesdo help
determine what the work looks like to us and what we take it to depict.



rolling surfaces, others are sharp and jagged, still others contain several relatively
at planes at various angles to each other, and so forth. If members of this society
should come across Picassos Guernica, they would count it as a guernicaa perfectly at onerather than as a painting. Its atness is variable and the gures
on its surface are standard relative to the category of guernicas. Thus the atness,
which is standard for us, would be variable for members of the other society,
and the gures on the surface, which are variable for us, would be standard for
them. This would make for a profound difference between our aesthetic reaction
to Guernica and theirs. It seems violent, dynamic, vital, disturbing to us. But
I imagine it would strike them as cold, stark, lifeless, or serene and restful, or
perhaps bland, dull, boringbut in any case not violent, dynamic, and vital. We do
not pay attention to or take note of Guernicas atness; this is a feature we take for
granted in paintings. But for the other society, this is Guernicas most striking and
noteworthy characteristicwhat is expressive about it. Conversely, Guernicas color
patches, which we nd noteworthy and expressive, are insignicant to them.
It is important to notice that this difference in aesthetic response is not due
solely to the fact that we are much more familiar with at works of art than they
are and that they are more familiar with Guernicas colors and shapes. Someone
equally familiar with paintings and guernicas might, I think, see Picassos Guernica as a painting on some occasions and as a guernica on others. On the former
occasions it will probably look dynamic, violent, and so forth to him, and on
the latter cold, serene, bland, or lifeless. Whether he sees the work in a museum
of paintings or a museum of guernicas, or whether he has been told that it is a
painting or a guernica, may inuence how he sees it. But I think he might be
able to shift at will from one way of seeing it to the other, somewhat as one shifts
between seeing the duck-rabbit as a duck and seeing it as a rabbit.
This example and the previous ones might give the impression that in general
only features of a work that are variable for us are aesthetically importantthat
these are the expressive, aesthetically active properties, as far as we are concerned,
whereas features standard for us are aesthetically inert. But this notion is quite mistaken, as the following examples will demonstrate. Properties standard for us are not
aesthetically lifeless, though the life that they have, the aesthetic effect they have on
us, is typically very different from what it would be if they were variable for us.
(3) Because of the very fact that features standard for us do not seem striking
or noteworthy, that they are somehow expected or taken for granted, they can
contribute to a work a sense of order, inevitability, stability, correctness. This is
perhaps most notably true of large-scale structural properties in the time arts.
The exposition-development-re-capitulation form (including the typical key and
thematic relationships) of the rst movements of classical sonatas, symphonies,
and string quartets is standard with respect to the category of works in sonataallegro form, and standard for listeners, including most of us, who hear them as
belonging to that category. So proceeding along the lines of sonata-allegro form



seems right to us; to our ears that is how sonatas are supposed to behave. We feel
that we know where we are and where we are going throughout the workmore
so, I suggest, than we would if we were not familiar with sonata-allegro form, if
following the strictures of that form were variable rather than standard for us.15
Properties standard for us do not always have this sort of unifying effect, however. The fact that a piano sonata contains only piano sounds, or uses the Western
system of harmony throughout, does not make it seem unied to us. The reason, I think, is that these properties are too standard for us in a sense that needs
explicating (cf. note 9). Nevertheless, sonata form is unifying partly because it is
standard rather than variable for us.
(4) That a work (or part of it) has a certain determinate characteristic (of size,
for example, or speed, or length, or volume) is often variable relative to a particular category, when it is nevertheless standard for that category that the variable characteristic falls within a certain range. In such cases the aesthetic effect
of the determinate variable property may be colored by the standard limits of
the range. Hence these limits function as an aesthetic catalyst, even if not as an
active ingredient.
Piano music is frequently marked sostenuto, cantabile, legato, or lyrical. But how
can the pianist possibly carry out such instructions? Piano tones diminish in
volume drastically immediately after the key is struck, becoming inaudible relatively promptly, and there is no way the player can prevent this. If a singer or violinist should produce sounds even approaching a pianos in suddenness of demise,
they would be nervewrackingly sharp and percussiveanything but cantabile or
lyrical! Yet piano music can be cantabile, legato, or lyrical nevertheless; sometimes
it is extraordinarily so (for example, a good performance of the Adagio Cantabile
movement of Beethovens Pathtique sonata). What makes this possible is the
very fact that the drastic diminution of piano tones cannot be prevented, and
hence never is. It is a standard feature for piano music. A pianist can, however,
by a variety of devices, control a tones rate of diminution and length within the
limits dictated by the nature of the instrument.16 Piano tones may thus be more
15. The presence of clichs in a work sometimes allows it to contain drastically disorderly elements without becoming chaotic or incoherent. See Anton Ehrenzweig, The
Hidden Order of Art (London: Weidenfeld and Nicolson, 1967), pp. 114116.
16. The timing of the release of the key affects the tones length. Use of the sustaining
pedal can lessen slightly a tones diminuendo by reinforcing its overtones with sympathetic vibrations from other strings. The rate of diminuendo is affected somewhat more
drastically by the force with which the key is struck. The more forcefully it is struck the
greater is the tones relative diminuendo. (Obviously the rate of diminuendo cannot be
controlled in this way independently of the tones initial volume.) The successive tones
of a melody can be made to overlap so that each tones sharp attack is partially obscured
by the lingering end of the preceding tone. A melodic tone may also be reinforced after
it begins by sympathetic vibrations from harmonically related accompanying gures,
contributed by the composer.



or less sustained within these limits, and how sustained they are, how quickly or
slowly they diminish and how long they last, within the range of possibilities,
is variable for piano music. A piano passage that sounds lyrical or cantabile to us
is one in which the individual tones are relatively sustained, given the capabilities of the instrument. Such a passage sounds lyrical only because piano music is
limited as it is, and we hear it as piano music; that is, the limitations are standard
properties for us. The character of the passage is determined not merely by the
absolute nature of the sounds, but by that in relation to the standard property of
what piano tones can be like.17
This principle helps to explain the lack of energy and brilliance that we sometimes nd even in very fast passages of electronic music. The energy and brilliance of a fast violin or piano passage derives not merely from the absolute speed
of the music (together with accents, rhythmic characteristics, and so forth), but
from the fact that it is fast for that particular medium. In electronic music different pitches can succeed one another at any frequency up to and including that at
which they are no longer separately distinguishable. Because of this it is difcult
to make electronic music sound fast (energetic, violent). For when we have heard
enough electronic music to be aware of the possibilities we do not feel that the
speed of a passage approaches a limit, no matter how fast it is.18
There are also visual correlates of these musical examples. A small elephant,
one which is smaller than most elephants with which we are familiar, might
impress us as charming, cute, delicate, or puny. This is not simply because of its
(absolute) size, but because it is small for an elephant. To people who are familiar
not with our elephants but with a race of mini-elephants, the same animal may
look massive, strong, dominant, threatening, lumbering, if it is large for a minielephant. The size of elephants is variable relative to the class of elephants, but
it varies only within a certain (not precisely speciable) range. It is a standard
property of elephants that they do fall within this range. How an elephantss size
affects us aesthetically depends, since we see it as an elephant, on whether it falls
in the upper, middle, or lower part of the range.
(5) Properties standard for a certain category which do not derive from physical limitations of the medium can be regarded as results of more or less conventional rules for producing works in the given category (for example, the rules
of sixteenth-century counterpoint, or those for twelve-tone music). These rules
may combine to create a dilemma for the artist which, if he is talented, he may
resolve ingeniously and gracefully. The result may be a work with an aesthetic
17. The musical media we know thus far derive their whole character and their usefulness as musical media precisely from their limitations. Roger Sessions, Problems
and Issues Facing the Composer Today, in Paul Henry Lang, Problems of Modern Music
(New York: W. W. Norton, 1960), p. 31.
18. One way to make electronic music sound fast would be to make it sound like some
traditional instrument, thereby trading on the limitations of that instrument.



character very different from what it would have had if it had not been for those
rules. Suppose that the rst movement of a sonata in G major modulates to Csharp major by the end of the development section. A rule of sonata form decrees
that it must return to G for the recapitulation. But the keys of G and C-sharp
are as unrelated as any two keys can be; it is difcult to modulate smoothly and
quickly from one to the other. Suppose also that while the sonata is in C-sharp
there are signs that, given other rules of sonata form, indicate that the recapitulation is imminent (motivic hints of the return, an emotional climax, a cadenza).
Listeners who hear it as a work in sonata form are likely to have a distinct feeling
of unease, tension, uncertainty, as the time for the recapitulation approaches. If
the composer with a stroke of ingenuity accomplishes the necessary modulation
quickly, efciently, and naturally, this will give them a feeling of reliefone
might say of deliverance. The movement to C-sharp, which may have seemed
alien and brashly adventurous at the time, will have proven to be quite appropriate, and the entire sequence will in retrospect have a sense of correctness and
perfection about it. Our impression of it is likely, I think, to be very much like
our impression of a beautiful or elegant proof in mathematics. (Indeed the
composers task in this example is not unlike that of producing such a proof.)
But suppose that the rule for sonatas were that the recapitulation must be
either in the original key or in the key one half-step below it. Thus in the example
above the recapitulation could have been in F-sharp major rather than G major.
This possibility removes the sense of tension from the occurrance of C-sharp
major in the development section, for a modulation from C-sharp to F-sharp is as
easy as any modulation is (since C-sharp is the dominant of F-sharp). Of course,
there would also be no special release of tension when the modulation to G is
effected, there being no tension to be released. In fact, that modulation probably
would be rather surprising, since the permissible modulation to F-sharp would
be much more natural.
Thus the effect that the sonata has on us depends on which of its properties are
dictated by rules, which ones are standard relative to the category of sonatas
and hence standard for us.
(6) I turn now to features which are contra-standard for usones which have
a tendency to disqualify a work from a category in which we nevertheless perceive it. We are likely to nd such features shocking, or disconcerting, or startling, or upsetting, just because they are contra-standard for us. Their presence
may be so obtrusive that they obscure the works variable properties. Threedimensional objects protruding from a canvas and movement in a sculpture are
contra-standard relative to the categories of painting and (traditional) sculpture
respectively. These features are contra-standard for us, and probably shocking, if
despite them we perceive the works possessing them in the mentioned categories. The monochromatic paintings of Yves Klein are disturbing to us (at least at
rst) for this reason: we see them as paintings, though they contain the feature



contra-standard for paintings of being one solid color.19 Notice that we nd other
similarly monochromatic surfaceswalls of living rooms, for examplenot in
the least disturbing, and indeed quite unnoteworthy.
If we are exposed frequently to works containing a certain kind of feature
which is contra-standard for us, we ordinarily adjust our categories to accommodate it, making it contra-standard for us no longer. The rst painting with a
three-dimensional object glued to it was no doubt shocking. But now that the
technique has become commonplace we are not shocked. This is because we no
longer see these works as paintings, but rather as members of either (a) a new
categorycollagesin which case the offending feature has become standard
rather than contra-standard for us, or (b) an expanded category which includes
paintings both with and without attached objects, in which case that feature is
variable for us.
But it is not just the rarity, unusualness, or unexpectedness of a feature that
makes it shocking. If a work differs too signicantly from the norms of a certain
category we do not perceive it in that category and hence the difference is not
contra-standard for us, even if we have not previously experienced works differing from that category in that way. A sculpture which is constantly and vigorously in motion would be so obviously and radically different from traditional
sculptures that we probably would not perceive it as one even if it is the rst
moving sculpture we have come across. We would either perceive it as a kinetic
sculpture, or simply remain confused. In contrast, a sculpted bust which is traditional in every respect except that one ear twitches slightly every thirty seconds
would be perceived as an ordinary sculpture. So the twitching ear would be contra-standard for us, and it would be considerably more unsettling than the much
greater movement of the other kinetic sculpture. Similarly, a very small colored
area of an otherwise entirely black-and-white drawing would be very disconcerting. But if enough additional color is added to it we will see it as a colored rather
than a black-and-white drawing, and the shock will vanish.
This point helps to explain a difference between the harmonic aberrations
of Wagners Tristan and Isolde, and those of Debussys Pellas et Mlisande and
Schoenbergs Pierrot Lunaire as well as Schoenbergs later twelve-tone works. The
latter are not merely more aberrant, less tonal, than Tristan. They differ from traditional tonal music in such respects and to such an extent that they are not
heard as tonal at all. Tristan, however, retains enough of the apparatus of tonality, despite its deviations, to be heard as a tonal work. For this reason its lesser
deviations are often the more shocking.20 Tristan plays on harmonic traditions

19. This example was suggested by Gran Hermern.

20. Cf. William W. Austin, Music in the 20th Century (New York: W. W. Norton,
1966), pp. 205206; and Eric Salzman, Twentieth-Century Music: An Introduction (Englewood Cliffs, N.J.: Prentice-Hall, 1967), pp. 5, 8, 19.



by selectively following and aunting them, while Pierrot Lunaire and the others
simply ignore them.
Shock then arises from features that are not just rare or unique, but ones that
are contra-standard relative to categories in which objects possessing them are perceived. But it must be emphasized that to be contra-standard relative to a certain
category is not merely to be rare or unique among things of that category. The melodic
line of Schuberts song, Im Walde, is probably unique; it probably does not occur in
any other songs, or other works of any sort. But it is not contra-standard relative
to the category of songs, because it does not tend to disqualify the work from that
category. Nor is it contra-standard relative to any other category to which we hear
the work as belonging. And clearly we do not nd this melodic line at all upsetting. What is important is not the rarity of a feature, but its connection with the
classication of the work. Features contra-standard for us are perceived as mists
in a category which the work strikes us as belonging to, as doing violence to such a
category. Being rare in a category is not the same thing as being a mist in it.
It should be clear from the above examples that how a work affects us aestheticallywhat aesthetic properties it seems to us to have and what ones we
are inclined to attribute to itdepends in a variety of important ways on which
of its features are standard, which variable, and which contra-standard for us.
Moreover, this is obviously not an isolated or exceptional phenomenon, but a
pervasive characteristic of aesthetic perception. I should emphasize that my purpose has not been to establish general principles about how each of the three sorts
of properties affects us. How any particular feature affects us depends also on
many variables I have not discussed. The important point is that in many cases
whether a feature is standard, variable, or contra-standard for us has a great deal
to do with what effect it has on us. We must now begin to assess the theoretical
consequences of this.


The fact that what aesthetic properties a thing seems to have may depend on
what categories it is perceived in raises a question about how to determine what
aesthetic properties it really does have. If Guernica appears dynamic when seen as a
painting, and not dynamic when seen as a guernica, is it dynamic or not? Can one
way of seeing it be ruled correct, and the other incorrect? One way of approaching this problem is to deny that the apparently conicting aesthetic judgments of
people who perceive a work in different categories actually do conict.21
21. I am ruling out the view that the notions of truth and falsity are not applicable to
aesthetic judgments, on the ground that it would force us to reject so much of our normal
discourse and common-sense intuitions about art that theoretical aesthetics, conceived
as attempting to understand the institution of art, would hardly have left a recognizable
subject matter to investigate. (See this chapters epigraph.)



Judgments that works of art have certain aesthetic properties, it might be

suggested, implicitly involve reference to some particular set of categories.
Thus our claim that Guernica is dynamic really amounts to the claim that it
is dynamic as a painting, or for people who see it as a painting. The judgment
that it is not dynamic made by people who see it as a guernica amounts simply
to the judgment that it is not dynamic as a guernica. Interpreted in these ways,
the two judgments are of course quite compatible. Terms such as large and
small provide a convenient model for this interpretation. An elephant might
be both small as an elephant and large as a mini-elephant, and hence it might be
called truly either large or small, depending on which category is implicitly
referred to.
I think that aesthetic judgments are in some contexts amenable to such category-relative interpretations, especially aesthetic judgments about natural
objects (clouds, mountains, sunsets) rather than works of art. (It will be evident
that the alternative account suggested below is not readily applicable to most
judgments about natural objects.) But most of our aesthetic judgments can be
forced into this mold only at the cost of distorting them beyond recognition.
My main objection is that category-relative interpretations do not allow aesthetic judgments to be mistaken often enough. It would certainly be natural to
consider a person who calls Guernica stark, cold, or dull, because he sees it as a
guernica, to be mistaken; he misunderstands the work because he is looking at it
in the wrong way. Similarly, one who asserts that a good performance of the Adagio Cantabile of Beethovens Pathtique is percussive, or that a Roman bust looks
like a unicolored, immobile man severed at the chest and depicts one as such, is
simply wrong, even if his judgment is a result of his perceiving the work in different categories from those in which we perceive it. Moreover, we do not accord
a status any more privileged to our own aesthetic judgments. We are likely to
regard cubist paintings, or Japanese gagaku music, as formless, incoherent, or
disturbing on our rst contact with these forms largely because, I suggest, we
would not be perceiving the works as cubist paintings, or as gagaku music. But
after becoming familiar with these kinds of art, we would probably retract our
previous judgments, admit that they were mistaken. It would be quite inappropriate to protest that what we meant previously was merely that the works
were formless or disturbing for the categories in which we then perceived them,
while admitting that they are not for the categories of cubist paintings, or gagaku
music. The conict between apparently incompatible aesthetic judgments made
while perceiving a work in different categories does not simply evaporate when
the difference of categories is pointed out, as does the conict between the claims
that an animal is large and that it is small, when it is made clear that the person making the rst claim regarded it as a mini-elephant and the one making
the second regarded it as an elephant. The latter judgments do not (necessarily)
reect a real disagreement about the size of the animal, but the former do reect
a real disagreement about the aesthetic nature of the work.



Thus it seems that, at least in some cases, it is correct to perceive a work in

certain categories and incorrect to perceive it in certain others; that is, our judgments of it when we perceive it in the former are likely to be true, and those
we make when perceiving it in the latter false. This provides us with absolute
senses of standard, variable, and contra-standard: features of a work are standard,
variable, or contra-standard absolutely just in case they are standard, variable, or
contra-standard, respectively, for people who perceive the work correctly. (Thus
an absolutely standard feature is standard relative to some category in which the
work is correctly perceived and contra-standard relative to none, an absolutely
variable feature is variable relative to all such categories, and an absolutely contra-standard feature is contra-standard relative to at least one such category.)
How is it to be determined in which categories a work is correctly perceived?
There is certainly no very precise or well-dened procedure to be followed. Different criteria are emphasized by different people and in different situations. But
there are several fairly denite considerations which typically gure in critical
discussions and which t our intuitions reasonably well. I suggest that the following circumstances count toward its being correct to perceive a work, W, in a
given category, C:
(i) The presence in W of a relatively large number of features standard with
respect to C. The correct way of perceiving a work is likely to be that in which
it has a minimum of contra-standard features for us. I take the relevance of this
consideration to be obvious. It cannot be correct to perceive Rembrandts Titus
Reading as a kinetic sculpture, if this is possible, just because that work has
too few of the features which make kinetic sculptures kinetic sculptures. But of
course this does not get us very far. Guernica, for example, qualies equally well
on this count for being perceived as a painting and as a guernica.
(ii) The fact that W is better, or more interesting or pleasing aesthetically,
or more worth experiencing when perceived in C than it is when perceived in
alternative ways. The correct way of perceiving a work is likely to be the way in
which it comes off best.
(iii) The fact that the artist who produced W intended or expected it to be
perceived in C, or thought of it as a C.
(iv) The fact that C is well established in and recognized by the society in
which W was produced. A category is well established in and recognized by a
society if the members of the society are familiar with works in that category,
consider a works membership in it a fact worth mentioning, exhibit works of
that category together, and so forththat is, roughly if that category gures
importantly in their way of classifying works of art. The categories of impressionist painting and Brahmsian music are well established and recognized in our
society; those of guernicas, paintings with diagonal composition containing green
crosses, and pieces of music containing between four and eight F-sharps and at



least seventeen quarter notes every eight bars are not. The categories in which a
work is correctly perceived, according to this condition, are generally the ones in
which the artists contemporaries did perceive or would have perceived it.
In certain cases I think the mechanical process by which a work was produced,
or (for example, in architecture) the nonperceptible physical characteristics or
internal structure of a work, is relevant. A work is probably correctly perceived
as an apparent etching22 rather than, say, an apparent woodcut or line drawing,
if it was produced by the etching process. The strengths of materials in a building or the presence of steel girders inside wooden or plaster columns counts (not
necessarily conclusively) toward the correctness of perceiving it in the category
of buildings with visual characteristics typical of buildings constructed in that
manner. I will not discuss these considerations further here.
What can be said in support of the relevance of conditions (ii), (iii), and (iv)?
In the examples mentioned above, the categories in which we consider a work
correctly perceived probably meet all of these conditions. I would suppose that
Guernica is better seen as a painting than it would be seen as a guernica (though
this would be hard to prove). In any case, Picasso certainly intended it to be seen
as a painting rather than a guernica, and the category of paintings is well established in his (that is, our) society, whereas that of guernicas is not. But this of
course does not show that (ii), (iii), and (iv) each is relevant. It tends to indicate
only that one or other of them, or some combination, is relevant.
The difculty of assessing each of the three conditions individually is complicated by the fact that by and large they can be expected to coincide, to yield
identical conclusions. Since an artist usually intends his works for his contemporaries he is likely to intend them to be perceived in categories established
in and recognized by his society. Moreover, it is reasonable to expect works to
come off better when perceived in the intended categories than when perceived
in others. An artist tries to produce works which are well worth experiencing
when perceived in the intended way and, unless we have reason to think he
is totally incompetent, there is some presumption that he succeeded at least
to some extent. But it is more or less a matter of chance whether the work
comes off well when perceived in some unintended way. The convergence of
the three conditions, however, at the same time diminishes the practical importance of justifying them individually, since in most cases we can decide how
to judge particular works of art without doing so. But the theoretical question
I will begin with (ii). If we are faced with a choice between two ways of perceiving a work, and the work is very much better perceived in one way than it is
perceived in the other, I think that, at least in the absence of contrary considerations, we would be strongly inclined to settle on the former way of perceiving
22. See p. 199.



it as the correct way. The process of trying to determine what is in a work consists
partly in casting around among otherwise plausible ways of perceiving it for one
in which the work is good. We feel we are coming to a correct understanding of a
work when we begin to like or enjoy it; we are nding what is really there when
it seems worth experiencing.
But if (ii) is relevant, it is quite clearly not the only relevant consideration.
Take any work of art we can agree is of fourth- or fth- or tenth-rate quality. It is
very possible that if this work were perceived in some farfetched set of categories
that someone might dream up, it would appear to be rst-rate, a masterpiece.
Finding such ad hoc categories obviously would require talent and ingenuity on
the order of that necessary to produce a masterpiece in the rst place. But we can
sketch how one might begin searching for them. (a) If the mediocre work suffers
from some disturbingly prominent feature that distracts from whatever merits
the work has, this feature might be toned down by choosing categories with
respect to which it is standard, rather than variable or contra-standard. When
the work is perceived in the new way the offending feature may be no more
distracting than the atness of a painting is to us. (b) If the work suffers from
an overabundance of clichs it might be livened up by choosing categories with
respect to which the clichs are variable or contra-standard rather than standard.
(c) If it needs ingenuity we might devise a set of rules in terms of which the work
nds itself in a dilemma from which it ingeniously escapes, and we might build
these rules into a set of categories. Surely, however, if there are categories waiting
to be discovered which would transform a mediocre work into a masterpiece, it
does not follow that the work really is a hitherto unrecognized masterpiece. The
fact that when perceived in such categories it would appear exciting, ingenious,
and so forth, rather than grating, clich-ridden, pedestrian, does not make it so.
It cannot be correct, I suggest, to perceive a work in categories which are totally
foreign to the artist and his society, even if it comes across as a masterpiece in
This brings us to the historical conditions (iii) and (iv). I see no way of avoiding the conclusion that one or the other of them at least is relevant in determining in what categories a work is correctly perceived. I consider both relevant,
but I will not argue here for the independent relevance of (iv). (iii) merits special
attention in light of the prevalence of disputes about the importance of artists
intentions. To test the relevance of (iii) we must consider a case in which
(iii) and (iv) diverge. One such instance occurred during the early days of the
twelve-tone movement in music. Schoenberg no doubt intended even his earliest
23. To say that it is incorrect (in my sense) to perceive a work in certain categories
is not necessarily to claim that one ought not to perceive it that way. I heartily recommend perceiving mediocre works in categories that make perceiving them worthwhile
whenever possible. The point is that one is not likely to judge the work correctly when he
perceives it incorrectly.



twelve-tone works to be heard as such. But this category was certainly not then
well established or recognized in his society: virtually none of his contemporaries
(except close associates such as Berg and Webern), even musically sophisticated
ones, would have (or could have) heard these works in that category. But it seems
to me that even the very rst twelve-tone compositions are correctly heard as
such, that the judgments one who hears them otherwise would make of them (for
example, that they are chaotic, formless) are mistaken. I think this would be so
even if Schoenberg had been working entirely alone, if none of his contemporaries
had any inkling of the twelve-tone system. No doubt the rst twelve-tone compositions are much better when heard in the category of twelve-tone works than
when they are heard in any other way people might be likely to hear them. But
as we have seen this cannot by itself account for the correctness of hearing them in
the former way. The only other feature of the situation which could be relevant,
so far as I can see, is Schoenbergs intention.
The above example is unusual in that Schoenberg was extraordinarily selfconscious about what he was doing, having explicitly formulated rulesthat is,
specied standard propertiesfor twelve-tone composition. Artists are not often
so self-conscious, even when producing revolutionary works of art. Their intentions as to which categories their works are to be perceived in are not nearly as
clear as Schoenbergs were, and often they change their minds during the process
of creation. In such cases (as well as ones in which the artists intentions are
unknown) the question of what categories a work is corectly perceived in is left
by default to condition (iv), together with (i) and (ii). But it seems to me that in
almost all cases at least one of the historical conditions, (iii) and (iv), is of crucial
My account of the rules governing decisions about what categories works are
correctly perceived in leaves a lot undone. There are bound to be a large number
of undecidable cases on my criteria. Artists intentions are frequently unclear,
variable, or undiscoverable. Many works belong to categories which are borderline
cases of being well established in the artists societies (perhaps, for example, the
categories of rococo musicfor instance, C. P. E. Bachof music in the style of
early Mozart, and of very thin metal sculpted gures of the kind that Giacometti made). Many works fall between well-established categories (for example,
between impressionist and cubist paintings), possessing some of the standard features relative to each, and so neither clearly qualify nor clearly fail to qualify on
the basis of condition (i) to be perceived in either. There is, in addition, the question of what relative weights to accord the various conditions when they conict.
It would be a mistake, however, to try to tighten up much further the rules
for deciding how works are correctly perceived. To do so would be simply to
legislate gratuitously, since the intuitions and precedents we have to go on are
highly variable and often confused. But it is important to notice just where
these intuitions and precedents are inconclusive, for doing so will expose the
sources of many critical disputes. One such dispute might well arise concerning



Giacomettis thin metal sculptures. To a critic who sees them simply as sculptures, or sculptures of people, they look frail, emaciated, wispy, or wiry. But that
is not how they would strike a critic who sees them in the category of thin metal
sculptures of that sort (just as stick gures do not strike us as wispy or emaciated). He would be impressed not by the thinness of the sculptures, but by the
expressive nature of the positions of their limbs, and so forth, so he would no
doubt attribute very different aesthetic properties to them. Which of the two
ways of seeing these works is correct is, I suspect, undecidable. It is not clear
whether enough such works have been made and have been regarded sufciently
often as constituting a category for that category to be deemed well established
in Giacomettis society. And I doubt whether any of the other conditions settle
the issue conclusively. So perhaps the dispute between the two critics is essentially unresolvable. The most that we can do is to point out just what sort of a
difference of perception underlies the dispute, and why it is unresolvable.
The occurrence of impasses like this is by no means something to be regretted.
Works may be fascinating precisely because of shifts between equally permissible ways of perceiving them. And the enormous richness of some works is due
in part to the variety of permissible, and worthwhile, ways of perceiving them.
But it should be emphasized that even when my criteria do not clearly specify
a single set of categories in which a work is correctly perceived, there are bound
to be possible ways of perceiving it (which we may or may not have thought of)
that they denitely rule out.
The question posed at the outset of this section was how to determine what
aesthetic properties a work has, given that which ones it seems to have depends
on what categories it is perceived in, on which of its properties are standard,
which variable, and which contra-standard for us. I have sketched in rough outline rules for deciding in what categories a work is correctly perceived (and hence
which of its features are absolutely standard, variable, and contra-standard). The
aesthetic properties it actually possesses are those that are to be found in it when
it is perceived correctly.24

24. This is a considerable oversimplication. If there are two equally correct ways of
perceiving a work, and it appears to have a certain aesthetic property perceived in one
but not the other of them, does it actually possess this property or not? There is no easy
general answer. Probably in some such cases the question is undecidable. But I think we
would sometimes be willing to say that a work is, e.g., touching or serene if it seems so
when perceived in one acceptable way (or, more hesitantly, that there is something very
touching, or serene, about it), while allowing that it does not seem touching or serene
when perceived in another way which we do not want to rule incorrect. In some cases
works have aesthetic properties (e.g., intriguing, subtle, alive, interesting, deep) which
are not apparent on perceiving it in any single acceptable way, but which depend on the
multiplicity of acceptable ways of perceiving it and relations between them. None of
these complications relieves the critic of the responsibility for determining in what way
or ways it is correct to perceive a work.



I return now to the issues raised in section 1. (I will adopt for the remainder
of this paper the simplifying assumption that there is only one correct way of
perceiving any work. Nothing important depends on this.) If a works aesthetic
properties are those that are to be found in it when it is perceived correctly,
and the correct way to perceive it is determined partly by historical facts about
the artists intention and/or his society, no examination of the work itself, however thorough, will by itself reveal those properties.25 If we are confronted by a
work about whose origins we know absolutely nothing (for example, one lifted
from the dust at an as yet unexcavated archaeological site on Mars), we would
simply not be in a position to judge it aesthetically. We could not possibly tell
by staring at it, no matter how intently and intelligently, whether it is coherent,
or serene, or dynamic, for by staring we cannot tell whether it is to be seen as a
sculpture, a guernica, or some other exotic or mundane kind of work of art. (We
could attribute aesthetic properties to it in the way we do to natural objects,
which of course does not involve consideration of historical facts about artists or
their societies. [Cf. p. 211.] But to do this would not be to treat the object as a
work of art.)
It should be emphasized that the relevant historical facts are not merely useful
aids to aesthetic judgment; they do not simply provide hints concerning what
might be found in the work. Rather they help to determine what aesthetic properties a work has; they, together with the works nonaesthetic features, make it
coherent, serene, or whatever. If the origin of a work which is coherent and serene
had been different in crucial respects, the work would not have had these qualities; we would not merely have lacked a means for discovering them. And of two
works which differ only in respect of their originsones which are perceptually
indistinguishableone might be coherent or serene, and the other not. Thus,
since artistss intentions are among the relevant historical considerations, the
intentional fallacy is not a fallacy at all. I have of course made no claims about
the relevance of artists intentions as to the aesthetic properties that their works
should have. I am willing to agree that whether an artist intended his work to
be coherent or serene has nothing essential to do with whether it is coherent or
serene. But this must not be allowed to seduce us into thinking that no intentions are relevant.
Aesthetic properties, then, are not to be found in works themselves in the
straightforward way that colors and shapes or pitches and rhythms are. But I do
not mean to deny that we perceive aesthetic properties in works of art. I see the
serenity of a painting and hear the coherence of a sonata, despite the fact that the
presence of these qualities in the works depends partly on circumstances of their
25. But this, plus a general knowledge of what sorts of works were produced when
and by whom, might.



origin which I cannot (now) perceive. Joness marital status is part of what makes
him a bachelor, if he is one, and we cannot tell his marital status just by looking
at him, though we can thus ascertain his sex. Hence, I suppose, his bachelorhood
is not a property we can be said to perceive in him. But the aesthetic properties of
a work do not depend on historical facts about it in anything like the way Joness
bachelorhood depends on his marital status. The point is not that the historical
facts function as grounds in any ordinary sense for aesthetic judgments. By themselves they do not, in general, count either for or against the presence of any particular aesthetic property. Nor are they part of a larger body of information (also
including data about the work derived from an examination of it) from which
conclusions about the works aesthetic properties are to be deduced or inferred.
We must learn to perceive the work in the correct categories, as determined in part
by the historical facts, and judge it by what we then perceive in it. The historical
facts help to determine whether a painting is coherent or serene only (as far as
my arguments go) by affecting what way of perceiving the painting must reveal
these qualities if they are truly attributable to the work.
We must not, however, expect to judge a work simply by setting ourselves to
perceive it correctly, once it is determined what the correct way of perceiving it
is. For one cannot, in general, perceive a work in a given set of categories simply
by setting himself to do it. I could not possibly, merely by an act of will, see
Guernica as a guernica rather than as a painting, nor could I hear a succession
of street sounds in any arbitrary category one might dream up, even if the category has been explained to me in detail. Indeed, I cannot even imagine except
in a rather vague way what it would be like, for example, to see Guernica as a
guernica. One cannot merely decide to respond appropriately to a workto be
shocked or unnerved or surprised by its (absolutely) contra-standard features, to
nd its standard features familiar or mundane, and to react to its variable features
in other waysonce one knows the correct categories. Perceiving a work in a
certain category or set of categories is a skill that must be acquired by training, and exposure to a great many other works of the category or categories in
question is ordinarily, I believe, an essential part of this training. (But an effort
of will may facilitate the training, and once the skill is acquired one may be
able to decide at will whether or not to perceive it in that or those categories.)
This has important consequences concerning how best to approach works of art
of kinds that are new to uscontemporary works in new idioms, works from
foreign cultures, or newly resurrected works from the ancient past. It is no use
just immersing ourselves in a particular work, even with the knowledge of what
categories it is correctly perceived in, for that alone will not enable us to perceive
it in those categories. We must become familiar with a considerable variety of
works of similar sorts.
When dealing with works of more familiar kinds it is not generally necessary
to undertake deliberately the task of training ourselves to perceive them in the
correct categories (except perhaps when those categories include relatively subtle



ones). But this is, I think, only because we have been trained unwittingly. Even
the ability to see paintings as paintings had to be acquired, it seems to me, by
repeated exposure to a great many paintings. The critic must thus go beyond the
work before him in order to judge it aesthetically, not only to discover what the
correct categories are, but also to be able to perceive it in them. The latter does
not require consideration of historical facts, or consideration of facts at all, but
it requires directing ones attention nonetheless to things other than the work
in question.
Probably no one would deny that some sort of perceptual training is necessary,
in many if not all instances, for apprehending a works serenity or coherence, or
other aesthetic properties. And of course it is not only aesthetic properties whose
apprehension by the senses requires training. But the kind of training required
in the aesthetic cases (and perhaps some others as well) has not been properly
appreciated. In order to learn how to recognize gulls of various kinds, or the sex
of chicks, or a certain persons handwriting, one must have gulls of those kinds,
or chicks of the two sexes, or examples of that persons handwriting pointed out
to him, practice recognizing them himself, and be corrected when he makes mistakes. But the training important for discovering the serenity or coherence of a
work of art that I have been discussing is not of this sort. Acquiring the ability
to perceive a serene or coherent work in the correct categories is not a matter of
having had serene or coherent things pointed out to one, or having practiced recognizing them. What is important is not (or not merely) experience with other
serene and coherent things, but experience with other things of the appropriate
Much of the argument in this paper has been directed against the seemingly
common-sense notion that aesthetic judgments about works of art are to be
based solely on what can be perceived in them, how they look or sound. That
notion is seriousl misleading, I claim, on two different counts. I do not deny that
paintings and sonatas are to be judged solely by what can be seen or heard in
themwhen they are perceived correctly. But examining a work with the senses
can by itself reveal neither how it is correct to perceive it, nor how to perceive
it that way.

Figure 12.1 Jerzy Kolacz, Spot: Man Drawing Thumb Print. Jerzy Kolacz 2007.
Courtesy of the artist. Originally published in New Yorker, 30 November 1992.


A curious fact about our concept of style is that we seem unable to make up our
minds about what sorts of things have styles. Works of artpaintings, plays,
buildings, sculptures, operasare said to be in one or another style, and so are
objects such as bathing suits, neckties, and automobiles. But we often think of
styles as ways of doing things, ways of performing actions. There are styles of
teaching, styles of travel, styles of chess playing, and styles of selling insurance.
Are styles attributes of objects, or of actions?
James Ackerman regards them as attributes of objects, of works of art:
In the study of the arts, worksnot institutions or peopleare the primary data;
in them we must nd certain characteristics that are more or less stable . . ., and
exible. . . . A distinguishable ensemble of such characteristics we call a style.1

But Ernst Gombrich, in his encyclopedia article, predicates styles of actions:

Style is any distinctive, and therefore recognizable, way in which an act is
performed or an artifact made or ought to be performed and made.2

Meyer Schapiro shifts back and forth:

[The section titles are new in this reprinting.]
1. James S. Ackerman, Style, in Art and Archaeology, ed. James S. Ackerman and Rhys
Carpenter (Englewood Cliffs, N.J.: Prentice-Hall, 1963), p. 164.
2. Ernst Gombrich, Style, in International Encyclopedia of the Social Sciences, ed. David
L. Sill, 18 vols. (New York: Macmillan, 19681979), 15:352361.




By style is meant the constant formand sometimes the constant elements,

qualities, and expressionin the art of an individual or group. The term is also
applied to the whole activity of an individual or society, as in speaking of a lifestyle or the style of a civilization.3

It is unlikely that what we have here is a simple case of ambiguity, of two distinct
meanings of the term style which call for independent analyses. The very fact that
Schapiro and others, in discussions of style, seem to confuse products and processes
suggests that there is an intimate connection between styles of objects and styles of
behavior. I would suggest that styles of works of art are to be understood in terms
of the notion of styles of action. Specically, attributing a style to a work involves,
somehow, the idea of the manner in which it was made, the act of creating it.
This suggestion is supported by the fact that one way of describing the style
of a work is to speak of what style it is done in, and also by the fact that we talk
about styles of painting, of writing, and so on, in contexts in which it would seem
that we are really concerned with styles of paintings, writings, and so on, the
products rather than the processes.
It is especially noteworthy that the notion of style seems peculiarly irrelevant
to objects that are not products of human action, even when our interest in these
objects is aesthetic. What is the style of a tulip, or an alpine meadow, or a
pristine lake in the high Sierras? Are the Grand Canyon and Yosemite Valley in
the same style or different ones? Sunsets in the tropics are very different from
sunsets in the Arizona desert, and Arizona sunsets in January differ from Arizona
sunsets in June. But are these differences stylistic? We might allow that natural
objects can, in unusual cases, have styles. A chorus of chirping birds might, just
conceivably, chirp in the style of Haydn. But the notion of style here is obviously
parasitic on that which is applied to manmade artifacts. It is beginning to look
as though human action has something to do with all style attributions.
The fact that styles are attributed primarily to artifacts rather than to natural
objects is not, I think, just an insignicant peculiarity of the English word style;
it reects a profound difference between how we understand and respond to
works of art, and how we understand and respond to natural objects.4 The difference is evident in the fact that a wide range of very important aesthetic qualities
of works of art are not to be found in natural objects. Poems and paintings are
sometimes witty, or morbid, or sophisticated, but it is hard to imagine what a
witty tulip, or a morbid mountain, or a sophisticated lake would be like. A sunset can hardly be sentimental, or unsentimental for that matter, even though a
3. Meyer Schapiro, Style, in Aesthetics Today, ed. Morris Philipson (Cleveland: World
Publishing Co., 1961), p. 81. Cf. also pp. 82, 8384.
4. Nelson Goodman steamrollers this difference, it seems to me, when he allows that sunrises might reasonably be said to have styles. Goodman, The Status of Style, Critical Inquiry
1 (1975): 808.



realistic painting of the sunset might be a paradigm of sentimentality. The lines

in a drawing may be sensitive, or bold, or carefree, but one hesitates to attribute
these qualities to similar lines in nature. Could the pounding of a surf be pompous or exuberant or passionate or bombastic or energetic, as a performance of
a Rachmaninoff prelude might be? It is rarely appropriate to describe natural
objects as ponderous, deliberate, neurotic, anguished, pretentious, profound,
amboyant, expressive, or reserved.
I would like to make two further observations about these qualities whose
ranges seem limited to works of art, or at least to artifacts. First, the predicates
corresponding to them serve also, and perhaps primarily, to describe human
actions or to attribute to people properties that are expressed in action. Second,
when these qualities are possessed by works of art they are, in many cases at least,
aspects of the styles of the works. Thus a work may be in a sentimental style,
or in a morbid, or bombastic, or amboyant style. We will see below that sentimentality, bombast, amboyance, and so on do not constitute styles, but they are
what I shall call style qualities.
Our observations so far should make us uncomfortable with what I shall call
the cobbler model of the institution of art. The cobbler model has a three-part
structure. There is the producer, the product, and the consumer, that is, the
cobber, who makes shoes, which are worn by consumers. The point of the process
consists in how well the shoes t the feet and the needs of the consumer; the
proof is in the shoes. The cobblers work is merely a means to this end. Once the
consumer has the shoes he has no reason to concern himself with the cobblers
act of making them. What is important is the nature of the shoes themselves.
Natural objects with the right properties would serve just as well as the cobblers
artifacts, and it makes no difference whether the wearer thinks that his shoes are
artifacts or that they grew on trees.
Applying this model to the institution of art, we have the artist who, perhaps
together with a collaborating performer, counts as the producer; the work of art
which counts as the product; and the appreciator in the role of consumer. The
artist, the work, and the appreciator are supposed to have functions analogous to
those of the cobbler, the shoes, and the wearer of the shoes, respectively, although
of course the kind of value that the work has for the appreciator is not the same
as that which shoes have for wearers.
I am sure it is evident already what sort of objection I have to the cobbler model
as applied to art. It focuses attention too exclusively on the work of art, the object
itself, and not enough on the action of making it. If, as I have suggested, the
notion of styles as characteristics of works essentially involves that of the acts of
producing the works, one would not expect that the appreciator who wants to
appreciate the work for its style can simply wrap himself in the work itself and give
no thought to the artists action. Nor would one expect that it makes no difference
whether the work is indeed a work, an artifact, rather than a natural object. But



shortcomings in the cobbler model are evident even apart from the notion of style.
I shall examine some of them now, to set the stage, and return to style later.
A relatively minor deviation from the cobbler model occurs when the object
of the appreciators interest is not something that the artist produces, but rather
the action of the artist itself. In dance and theater our attention is directed to the
actions of dancers and actors.
But we should distinguish between the movements of dancers or actors and their
(intentional) actions. It may be the movements, the events consisting of bodies
in motion, which are the objects of appreciators interest, and these movements
are produced by the performers when they perform the actions.
There are other cases, however, in which it is evident that actions, not mere
movements, are objects of appreciation. The clearest examples of this occur in
the avant-garde. For instance, the Museum of Modern Art once exhibited the
leftovers so to speak of an event in which artist John Latham took a copy of Art
and Culture by critic Clement Greenberg, and shredded and blended it into a
kind of book-shake which he and some friends cheerfully gulped down.5
It is obvious that what is interesting in this case is the action that Latham
performed, not just the movements of his body. If his behavior had been unintentional, it would not have had anything like the same sort of signicance.
Some actions that are of interest are actions of making or displaying objects.
Lathams action did include the making of the book-shake, the remnants of
which were put on display. But it seems clear that what Latham did, not what
he made, is the main object of interest. The same might be said of Marcel
Duchamps readymades and John Cages indeterminately composed music. One
could argue that the readymades and the indeterminate music are not themselves very interesting (although Cage disputes this). In any case, Duchamps
act of displaying the readymades and Cages act of using indeterminate means to
compose his music certainly are interesting. This is obvious from the enormous
volume of literature about these actions.
But are the actions in these cases aesthetically interesting? My opinion is that
nothing is to be gained by pressing either this question or the related question of
what qualies as art.6 But if, for the sake of argument, we assume that a specically aesthetic kind of interest is to be recognized, a reasonable case can be made
for saying that our interest in the actions I have described may very well be aesthetic. They are easily understood as symbolic or expressive of certain attitudes
about life, or society, or the art establishment in very much the way that actions
of characters in literature very often are. They are, in fact, strikingly similar to
actions of characters in the theater of the absurd. The activities of many avantgarde artists can be, and have been, regarded as a kind of theater.
5. Newsweek, 30 April 1973, p. 89.
6. See my review of George Dickies Art and the Aesthetic in Philosophical Review 86 (1977):



Sometimes when artists make objects it seems obvious that the object is of
very little signicance and that it is only the act of making it which should
occupy our attention. But strangely enough, the objects, as ordinary or trivial
as they seem, are often treated with much the same sort of reverence we accord
to the masterpieces of Rembrandt and Shakespeare and Beethoven. They are put
in museums to be gawked at, they are bought and sold for incredible sums, and
so forth. The artists themselves often do not try to make it clear that attention
should be paid to their actions rather than the products of their actions. They
speak and write and behave as though their works are meant to be masterpieces,
or at least objects of interest, in something like the traditional way.
We can understand valuing these objects as mementos of the signicant activities that led to their existence, much as we value things like Beethovens piano
and Rembrandts printmaking equipment. Some of the signicance of the actions
rubs off in this way on the objects.
But I do not think that this accounts fully for the reverential treatment that
things like Duchamps readymades and Cages indeterminate music sometimes
receive. Another explanation that seems to me especially interesting and important is this: if the act of producing the object is symbolic or expressive in one
way, the act of buying or displaying it or just observing it may be symbolic or
expressive in another. Attending a concert of Cages indeterminate music may be
a way of expressing ones agreement with the point one takes Cage to have been
expressing in producing the music; the listener may be symbolically thumbing
his nose at the art establishment, or debunking the masters, or afrming a
kind of Cagian zest for life. This explanation does not suggest that the product,
Cages music, for example, is valuable, any more than the bread and wine used in
communion are themselves valuable. But it does suggest that there is something
signicant and important about behaving as though the objects are valuable
performing the music, buying the tickets to hear it, and listening to it.
I would like to point out especially that on the explanation I have offered the
act of appreciating the object (or should we say the act of pretending to appreciate it?) is closely analogous to the act of making it. Both are ritualistic or symbolic afrmations of probably similar attitudes or points of view.
Why do artists often appear not to recognize that it is their actions, rather than
the products of their actions, which are of interest? The action of interest is in
many cases that of behaving as though one is creating and/or displaying a valuable aesthetic object of a traditional kind, while actually creating or displaying
something that is nothing of the sort. It is the shock and absurdity of the contrast
between the object and the way it is treated which is symbolically signicant,
which can be seen as, for example, deating the pomposity and rigidity of traditional attitudes about art and the worshipful attitude toward what is deemed to
be very special. Just think how much less effective Duchamps act of displaying
his readymades would have been if he had attached notations to the pedestals
explaining clearly that the objects are not meant to be of any particular interest



and that attention is to be focused instead on his act of displaying them. His
action could not have been regarded as one of presenting trivial, uninteresting,
everyday things as though they were masterpieces, and hence his action would not
have had the same intriguing and, to some, maddening symbolic signicance.
The cobbler model is misleading even when our interest is directed toward the
product of an artists actions, rather than the action itself. What matters in
the cobbler case is the value of the shoes for the wearer. There usually would
be little point in making shoes if they were not to be worn. But works of art are
not made exclusively for the sake of their appreciation by spectators, listeners, or
readers. One way to appreciate music is by playing it. And musicians frequently
do play for the fun of it, with no thought of an audience.
The point here is not just that playing music is enjoyable, for the cobbler may
well enjoy making shoes also. The enjoyment of playing music strikes me as very
much like that of listening to it; both activities deserve the label of aesthetic
experience if anything does. Playing and hearing music are simply different
ways of appreciating it.7 But the cobblers experience, by contrast, is not at all
like that of the wearer of the shoes. His enjoyment of the activity of making the
shoes has little in common with the value that the shoes have for the wearer.
It might be thought that the similarity of the musicians experience to that of
the listener is explained simply by the fact that the musician is a listener also;
he listens to the sounds as he makes them. But the player does not listen in the
same way that a listener does. He is too occupied with what he is doing. Sounds
that a musician delights in making may be ones that would drive him up the
wall if he heard them as an audience does. I prefer to explain the similarity in
the opposite wayby the supposition that appreciation by the audience involves
some sort of empathy with the act of making the sounds. My inclination is to
understand appreciation by performing, rather than appreciation by listening, as
primary, even though the latter is much more common, at least in the tradition
of Western art music.
It is revealing to look beyond Western art music to its roots. Audiences are
superuous in many folk music and folk dance traditions (including our own
current tradition of hymn singing).8 People just get together and sing or dance.

[7. I recall being curious and puzzled, while composing this essay, as to why the experience
of making music and that of listening to it should be as similar as they clearly seemed to be,
why regarding them as alternative means of appreciating music should be so natural, even
unavoidable. The recent discovery of mirror neurons, the fact that the same kinds of neural activity occurs when subjects either perform certain actions or observe others performing them,
promises an explanation. Cf. Vittorio Gallese, Luciano Fadiga, Leonardo Fogassi, and Giacomo
Rizzolati, Action Recognition in the Premotor Cortex, Brain 119, no. 2 (1996): 593609;
and Rizzolati et al., Premotor Cortex and the Recognition of Motor Actions, Cognitive Brain
Research 3 (1996): 131141.]
8. Cf. Roger Sessions, Questions about Music (New York: W. W. Norton, 1970), pp. 1417.



Anyone listening or watching is incidental. There is no temptation to say that in

these cases the basic function of the artist, the singer, or the dancer is to produce
something for others to contemplate and appreciate. Nor, when there is no audience, are participants to be understood as merely practicing or playing at the
craft of performing for an audience. There is no sense in which the ultimate aim
is to please (or edify or entertain) passive observers.
Indeed it may be misleading to regard the aesthetic experience of the musician or dancer as appreciation of something at all. There is no object of appreciation
which is independent of the act of appreciation itself (at least if the singing or
dancing is entirely improvised and spontaneousotherwise the abstract song or
dance, of which the particular performance is an instance, might be regarded as
the object of appreciation). One simply enjoys doing something. One doesnt perform the action and also observe and appreciate the sounds produced or the action
performed. One doesnt (necessarily) reect on what one does or the sounds one
makes in this kind of way.
We are a long way from the cobbler model here. Not only is there no consumer who appreciates an object made by the producer; there is nothing that
can very comfortably be called an object of appreciation at all.
Audiences are done away with in some avant-garde traditions also. Many happenings of the 1960s, for example, were done not for the benet of onlookers,
but solely for that of the participants. In this respect, as in many others, the
avant-garde is not nearly as revolutionary as some of its practitioners would have
us believe.
Folk traditions do not always remain pure. Folk music and dance are sometimes performed for audiences of tourists. Folk singers give concerts and go
on concert tours. No doubt the nature of the singing or dancing is changed
in subtle, and sometimes not so subtle, ways to appeal to audiences. But what
strikes me as intriguing and revealing is the fact that activities of singing and
dancing which originally are done for their own sakes without thought of an
audience should so often be such that with little or no alteration they appeal to
audiences. The reason for this, I think, is that the audience empathizes with
the actions of the participants and so gets something of the same thrill or satisfaction or enjoyment from listening or watching which the participants get from
singing or dancing.
Much of our ne art has of course grown out of folk art traditions. It would
not be surprising if such works retained certain elements of their ancestors, if, for
example, appreciation by audiences of staged concerts of music and dance should
involve a sense of, or empathy with, the actions of the artists.
It is no accident that I have been concentrating just now on the performing
arts, for it is in these arts that the inadequacies of the cobbler model are most
glaring. But what about arts such as painting and sculpture and written literature which involve the production of relatively permanent physical objects that
can be appreciated long after the artist has nished his work? Is the creative



activity of painters and sculptors typically what anyone would want to call an
aesthetic experience, and is it anything like the experiences spectators have on
confronting the nished works? Does the spectators appreciation of the works
involve empathy with or understanding of the actions of the maker? I shall have
a lot to say about this in what follows. But now I would like to point out that
many of us like to doodle, to draw pictures just for the fun of drawing them.
Usually we do not intend or expect that anyone will see our doodles, nor that we
ourselves will spend any signicant amount of time examining or contemplating them after they are nished, let alone put them on permanent display. The
point is in the process, just as it is in the case of jam sessions and folk singing
and dancing.


The inadequacies of the cobbler model should encourage the idea that the notion
of the style of a work of art is to be understood in terms of the notion of the manner in which it was made. We may begin with the suggestion that we see in a
work the action of producing it,9 and that the works style is a matter of what sort
of action is visible (or audible, or otherwise perceptible) in it.
The action we see in a work may not correspond to what the artist actually
did in creating it; our perception may not be veridical. So it would seem that what
the artist actually did, the style or manner of his actual behavior, is not what constitutes the style of the work. If we should discover that one of Daumiers drawings was not done by Daumier at all but rather by a machine run by a computer
or in some other way, we would (probably) still feel comfortable saying that it is
in the style of Daumier.
The idea I wish to pursue is that it is how a work appears to have been made,
what sort of action or actions it looks or sounds or seems as though the artist
performed in creating it, which is crucial to the works style.
I do not claim that appreciators and critics are not ordinarily in a position to
know much of anything about artists acts of creation. It may be perfectly obvious, for example, that an artist performed an action of inscribing thin wiggly
lines or applying gaudy colors; this is evident just from the facts that the work
has wiggly lines or gaudy colors, and that it was made by someone. And given

9. This suggestion, or something like it, is to be found in Richard Wollheim, Expression,

in On Art and Mind, ed. Wollheim (Cambridge, Mass.: Harvard University Press, 1974), pp.
84100; Schapiro, Style, pp. 81, 85; Wayne Booth, The Rhetoric of Fiction (Chicago: University of Chicago Press, 1961); Guy Sircello, Mind and Art (Princeton N.J.: Princeton University
Press, 1972); Denis Dutton, Artistic Crimes: The Problem of Forgery in the Arts, British
Journal of Aesthetics 19 (1979): 302314; and in an unpublished paper by Timothy W. Bartel.
Booths implied author is one kind of what I shall call the apparent artist.



that artists can be assumed to intend the prominent features of their works, it
may be obvious that the action of making thin wiggly lines or applying gaudy
colors was intentional. But, as will be clear from my discussion, it is probably
the fact that the work appears to be the result of an act of intentionally making
thin wiggly lines or applying gaudy colors that is important. The appearance has
its effect even if, for some reason, it does not correspond to reality. Sometimes it
would be rash to suppose that a work was actually made in the manner it appears
to have been; yet the appearance alone is important.10
It is important to distinguish the apparent manners in which works were created
from ctional ones.11 A literary work that has a narrator or dramatic speaker can
be regarded as establishing the ction that it was created in a certain manner or
by a certain sort of person. Thus Samuel Becketts Malone Dies makes it ctional
that the words of that work were scribbled in a notebook with a stubby pencil
by a neurotic named Malone as he lay on his deathbed. Perhaps it is also true
that the work seems to have been written by such a person in such circumstances.

10. Guy Sircello recognizes the importance of thinking about works of art in terms
of the acts of creating them in his provocative book Mind and Art. But he claims that it
is actual actions that artists perform (artistic acts) that are important. What artistic
acts have been performed is to be discovered, he holds, just by examining the works
themselves; external evidence for them is either irrelevant or at least no better than the
internal sort (pp. 2728). I agree that for many of what he calls artistic acts, examination of the work itself is a crucial part of verifying that the act was performed. One must
examine The Love Song of J. Alfred Prufrock to ascertain that in it Eliot portrayed the
hero compassionately. But if Eliot did not write the poem, he did not perform this act, no
matter what the poem is like; and if the poem was written by a computer or a monkey,
no one performed the act of portraying the character compassionately. The words of the
poem are, to be sure, good evidence that someone wrote it, but it is conceivable that other
evidence (external evidence) should show that no one wrote it. Moreover, it is hard to
deny that whether at least some of Sircellos artistic acts are performed depends on
the intentions of the artist (e.g., the acts of inveighing angrily against the institution of
imprisonmentp. 25). The work is not the most direct evidence possible of the artists
intentions. Sircellos observations are best served, it seems to me, by recognizing that it
is how works appear to have been made that matters (and that is important for expression)although as we shall see, this does not mean that we can ignore all external
evidence. In fact, Sircellos book is a rich source of examples illustrating the importance
of the apparent actions of artists. I should emphasize that I am not postulating phantom
acts, airy nothings existing mysteriously in works of art (p. 28), for I do not advocate
quantifying over apparent actions (nor over apparent artists). My occasional apparent references to apparent actions are to be understood as eliminable in favor of descriptions
of what appears to be the case.
11. Seymour Chatman claims that the (actual) author is known by extraliterary, hence
irrelevant, information, and holds that the key to a literary works style is its persona (The
Semantics of Style, in Introduction to Structuralism, ed. Michael Lane [ New York: Basic Books,
1970], pp. 136, 143). I am not sure whether by persona here he means a narrator, i.e., a
ctional character, or the apparent author.



Figure 12.2 Roy Lichtenstein, Little Big Painting, 1965. Oil and synthetic polymer
on canvas, 68" 80". Purchase, with funds from the Friends of the Whitney Museum of American Art. Photograph 1996: Whitney Museum of American Art,
New York.

[I no longer think there is much of a distinction to be drawn between ctional and

apparent artists. It now seems to me that, when recognition of an apparent artist is
important in understanding and appreciating a work, we should think of him or her as
being ctional, as what I have called a storytelling narrator, typically, in the case of literature. Butand this is importantthese ctional artists usually belong to a ctional
world distinct from that in which the ordinary characters of the work reside. See Kendall
L. Walton, Mimesis as Make-Believe: On the Foundations of the Representational Arts (Cambridge, Mass.: Harvard University Press, 1990), pp. 368372.]



What ctionally is the case is sometimes apparently the case as well. But Malone
Dies seems also (at a deeper levelcf. below, section 4) to have been written by
a brilliant and imaginative author who is not neurotic. It is not ctional that this
is so; there is no brilliant, imaginative, non-neurotic character in the world of
the novel who creates Malone.
Occasionally we nd ctional creators outside of literature. Lichtensteins
paintings of brushstrokes are examples. It is ctional that someone produced
his Little Big Painting (gure 12.2) by several bold but sloppy strokes of a paintbrush. There are also lms about the making of themselves (e.g., Bergmans
Hour of the Wolf ). And one might regard a performance of Mozarts Musical Joke
as establishing the pretense, making it ctional, that it is the handiwork of an
utterly untalented and unimaginative, though earnest, eighteenth-century composer and a group of incompetent performers.12
But these are unusual cases. Most nonliterary works do not depict or portray themselves as having been made in a certain manner; few of them have
ctional creators. Van Goghs Sunowers depicts sunowers, not its own genesis.
Hitchcocks lms are about crimes and their solutions, not about the lms themselves. Most music is not representational at all; that is, it does not generate any
ctional truths. But in the case of all such works something can be said about
how they seem to have been created, and we do want to attribute styles to them.
So style attributions, at least in these cases, are not based on ctions about their
creation, but rather, I think, on how they appear to have been created.
In no area of the arts are the activities by which things seem to have been produced more important than in music. The sounds of a musical performance seem
to listeners to have been made by actions of banging, scraping, blowing, singing,
and so on. And they sound as though these actions were performed vigorously or
gently, carefully or with abandon. Usually the sounds we hear are in fact made
in pretty much the manner they seem to have been. But let us consider just how
the sounds sound, the impression they give of how they were made (even when
we dont see the performers), regardless of whether the impression corresponds to
the reality. There can be no doubt that much of the emotional impact of music
depends on what activities sound to the listener as though they are going on. It is
with reference to these apparent activities that we describe melodies or passages
of music as tender, nervous, raging, owing, or energetic, and that we characterize musical performances as sprightly or bombastic or timid or ponderous.

12. Fictional creators do not necessarily mediate our access to the ctional worlds of works
in the way that narrators do. (Cf. my Points of View in Narrative and Depictive Representation, Nos 10 [1975]: 4961.) Hour of the Wolf, for example, generates many ctional truths
that are not implied by ctional truths about the creation of the lm. Fictional characters in
depictions who mediate our access to ctional worlds, when there are such (cf. ibid., p. 61),
are usually not ctional creators.



Many of us, I think, nd much electronically generated musicas compared

with traditional musicethereal, disembodied, unreal, not very expressive (at
least not expressive in the way traditional music is). This, I believe, is because
the sounds of electronic music usually do not give the listener much of a sense of
physical activities by which they were made; they do not sound as though they
resulted from any familiar mechanical actions such as scraping, banging, blowing, and so forth.
But we are likely to have the impression that the sounds of an electronic piece
were chosen by someone who is witty, or imaginative, or tiresome; the music
seems to have resulted from acts of choosing of certain kinds, even if we have
little sense of the physical means by which the sounds were made.
In literature obviously what is important is the nature of the choices or decisions the author apparently made about how the work was to be. His decisions
may seem to have been motivated by certain passions, or aimed at certain objectives. They may seem to be the decisions of someone who has an ax to grind,
or who has certain beliefs or attitudes or sensitivities. We can say that authors
apparently acted passionately, or imaginatively, or with certain intentions, in
writing their works.
It is hardly necessary to mention the many works in the plastic arts which give
vivid impressions of the physical behavior of the artist as well as perhaps more
ambiguous impressions of his motivations and personality. There are Van Goghs
paintings with their visible brushstrokes, for example, and Jackson Pollocks
canvases with their drips and splashes. (Van Goghs paintings, unlike the Lichtenstein mentioned earlier, are not paintings of brushstrokes, but rather paintings of such things as wheat elds and sunowers. It is not ctionalthat is,
true in the ctional worldthat the artist manipulated a brush in a certain
manner; instead, the painting looks as though he did.)
Some other artists attempt to cover their tracks in their works. Neither Leonardo
da Vincis paintings nor those of Phillip Pearlstein leave very obvious clues about
the artists physical activities of applying the paint to the canvas. But in both cases
the artists seem not to have acted in the manner that either van Gogh or Pollock
seem to have, and they do seem to have worked carefully, deliberately, precisely,
and often with certain motivations (including the desire to cover their tracks).
The Last Supper scene of Buuels Viridiana (gure 12.3) is a fascinating play
with shifts in the degree of control apparently exercised by the director. A group
of bums have taken over a mansion. In the midst of the hubbub thirteen of
them strike a pose which we recognize as that of Leonardos fresco. Suddenly the
directors hand is apparent in a way that it was not previously. It is obvious that
nearly every detail of these several frames was carefully arranged; the remarkable
correspondence to Leonardos work could not have been accidental. Then this
moment of contrivance gives way to a resumption of the previous chaos; we no
longer have the impression, or at least not nearly as vivid an impression, of the
directors studied control over the details of the occurrences on the screen.



Figure 12.3 Luis Buuel, Viridiana (Last Supper Scene). The Museum of Modern Art.
Film Stills Archive, 11 West 53rd Street, New York.

There can be no doubt about the importance of how works seem to have been
made. A passionate work is one that seems to have been made by someone acting
in passion; a pretentious work one that seems to have resulted from pretension.
Many other aesthetic qualities of works of artincluding those of being exuberant, playful, compulsive, sensitive, sentimental, deliberate, neurotic, serene, sardonic, sophisticated, bold, amboyant, morbidare possessed largely or entirely
by virtue of appearing to have been made by actions of certain kinds, either actions
involving certain sorts of physical movements or ones performed from certain
motives or with certain intentions or as a result of certain personality traits.
These aesthetic qualities are qualities that we identied earlier as being, in
many cases at least, qualities of style. So the character of a works style is linked
in a crucial way to how it appears to have been made. Tentatively, to be in a
amboyant, sentimental, or timid style is to appear to have been created in a
amboyant or sentimental or timid manner.
I shall not claim that aesthetic qualities of this kind are always aspects of style,
that for example every sentimental work is a work in a sentimental style. One
reason for my hesitation is this: It is not clear to me that there could be just
one work in a given style. If there is only one, the style hasnt yet been established; there is no such style. I have some inclination to hold that if only one of
the paintings we call impressionist had existed (Monets Water Lilies, let us say)
it would not be correct to speak of the impressionist style. Water Lilies would
be like an exotic animal born unexpectedly to ordinary parents: not the sole
example of a new species, but a mutant form of an old one. But I see no reason



why there could not be only one sentimental work. So a mutant work of art
might be sentimental and yet not be in a sentimental style.
Are all qualities of a works style based on the apparent activities of its artist? It
is not obvious that describing something as being in a classical or baroque style
is to make reference to how it appears to have been made. Wlfin attempts to
dene classical and baroque (or anyway, Classical and Baroque) in terms
of features such as linearity and painterliness, and closed and open form. But it
may be that these features are felt to be ingredients of a works style because they
contribute to how works appear to have been made.13 (The same may be true
of characteristics such as balance, symmetry, etc.) Wlfin himself suggests as
much at least with respect to closed and open form (tectonic and a-tectonic):
[W]e nd the classic epoch following the principle that given conditions rule
the personal will, that is, the whole is made to look as if this lling were just
made for this frame, and vice versa.14
[In Leonards Last Supper the gure of Christ] coincides so exactly with the
high light of the central door that an enhanced effecta kind of halois
thereby achieved for Him. Such a support of the gures by their environment
is, of course, equally desired by the baroque: what is not desired is that this
coincidence of forms should look obvious and intentional.15
When, in . . . the great Ecce Homo in the oblong form (etching) [Rembrandt]
constructs, with clear reference to Italian models, a symmetrical architecture
whose mighty breath bears the movement of the little gures, once again the
most interesting point is that, in spite of all, he is able to cast the semblance of
hazard over this tectonic composition.16

If the style of something depends on what actions seem to have been performed
in creating it, a necessary condition for somethings having a style is its seeming
to have been created by the deliberate performance of some action. This explains
why natural objects do not ordinarily have styles. Many natural objects do seem
to have come about in certain ways; they look as though they were the result of
certain sorts of physical events. Half Dome, in Yosemite, appears to have been
sliced off by something like an incredibly gigantic bread knife. To some, the
Grand Canyon may seem to have been formed by a devastating ood, although to

13. Painterliness, symmetry, etc., are probably style constituents rather than style qualities. Cf. p. 00.
14. Heinrich Wlfin, Principles of Art History, trans. M. D. Hottinger (New York: Dover,
1932), p. 131.
15. Wlfin, Principles of Art History, p. 133.
16. Wlfin, Principles of Art History, p. 134.



people with a more sophisticated understanding of geology it probably appears

to have been created in the way it actually wasby the constant trickle of a relatively miniscule river and its tributaries over thousands of years. The sense we
have of how natural objects were made often has a lot to do with our aesthetic
appreciation of them. But we rarely have the impression that natural objects
are the handiwork of sentient beings, that they are the results of deliberate,
intentional human actions. We do not have the sense of a personality reected in
sunsets, alpine meadows, etc. This is why natural objects do not have styles.


I have left untouched a lot of important questions about the relations of apparent artists to styles. But it is high time that we probed deeper into the notion of
apparent artists. I have pretended so far that what it is for something to appear
to have been made in a certain manner is unproblematic. But nothing is further
from the truth. In most of the remainder of this article I shall point out some of
the problems, and point toward some solutions.
A good place to start is with the relevance of the notion of apparent artists to
the (so-called) intentional fallacy and related matters.
Monroe Beardsley and others have done a great service by forcing us to examine critically references to artists and their intentions in the writings of critics,
and to ask whether such references are not best construed as sloppy ways of talking about the works themselves rather than the artists. Much of the confusion in
this area can be traced to inattention to the distinction between apparent artists
and actual ones. Sense can often be made of seemingly illicit appeals to artists
intentions by reformulating them as appeals to what intentions it looks as though
the artists had, judging from their works.
But even when what is relevant to criticism is merely with what intentions a
work appears to have been made, or how, in other respects, it seems to have come
about, it still may be crucial to consider the actual historical context in which it
was created. We need to consider carefully in what sense the property of appearing to have been made in a certain manner is to be located in the work which
does so appear.
Jorge Luis Borgess story Pierre Menard, Author of Don Quixote will serve to
introduce my point. In this story Pierre Menard, a quixotic twentieth-century
author, wrote part of Don Quixote, that is, he authored a text that is word for
word identical with part of Cervantess Don Quixote. It is important that Menard
neither copied Cervantess work nor thought himself into Cervantess shoes and
wrote it, again, from Cervantess perspective; instead he wrote it from his own
twentieth-century perspective. Borgess narrator comments:
It is a revelation to compare the Don Quixote of Menard with that of Cervantes.
The latter, for instance, wrote:



. . . truth, whose mother is history, who is the rival of time, depository of

deeds, witness of the past, example and lesson to the present, and warning to the
Written in the seventeenth century, written by the ingenious layman
Cervantes, this enumeration is a mere rhetorical eulogy of history. Menard, on
the other hand, writes:
. . . truth, whose mother is history, who is the rival of time, depository of deeds,
witness of the past, example and lesson to the present, and warning to the future.
History, mother of truth; the idea is astounding. Menard, a contemporary of
William James, does not dene history as an investigation of reality, but as its
origin. Historical truth, for him, is not what took place; it is what we think
took place. The nal clausesexample and lesson to the present, and warning to the
futureare shamelessly pragmatic.17

Presumably, Menard had pragmatism in mind when he wrote this passage,

and Cervantes did not have it in mind when he wrote his corresponding one.
But perhaps this is not what is important. Perhaps Menards work has overtones
of pragmatism because it seems to have been written by someone with pragmatism in mind, and Cervantess work does not have overtones of pragmatism
because it seems to have been written by someone who did not have pragmatism
in mind.
How can it be that the works seem different in this way, given that they consist
of exactly the same words? The answer is that how the works seem is a function
not just of their words, but of what century they were written in. Menards text,
understood as a twentieth-century work, seems to have been written with pragmatism in mind, while Cervantess text, understood as a seventeenth-century
work, seems otherwise. So even if a critic can ignore what the authors actually
thought in favor of what they seem to have thought, he cannot ignore the historical context of the works and bury himself in the text alone.
Let us make this a little more precise. How a text seems to a particular reader
depends not just on the text itself but also on what century the reader takes it
to have come from. I think we can agree that it is correct (proper, appropriate,
normal) to read Cervantess Don Quixote as a seventeenth-century work, since
it was in fact written in the seventeenth century; to construe it as a product of
the twentieth century would be to misconstrue it, to read things into it which
are not there.18 What justies the judgment that it does not have overtones
of pragmatism is the fact that, when read in the appropriate manner, it seems

17. Jorge Luis Borges, Ficciones (New York: Grove Press, 1962), p. 53; italics in original.
18. There is a lot more to be said about the notion of correct and incorrect ways of reading texts. For some ideas on an analogous problem, see my Categories of Art [chap.11, this



not to have been written with pragmatism in mind. Menards Quixote is properly read as a twentieth-century work, and it seems when so read to have been
authored with pragmatism in mind (since the author can be expected to have
known about the ideas of William James); thus Menards Quixote does have overtones of pragmatism.
This example illustrates the well-known fact that how things look or sound or
seem is conditioned by what we know or believe, and hence by the experiences
that formed our beliefs. One general principle about such conditioning is this:
what sort of action a particular object appears to have resulted from depends in
large measure on our beliefs about what sorts of objects generally result from what
sorts of actions, at least when these beliefs are sufciently internalized. We
know from previous experience what a surface is likely to look like if an opaque
liquid has been dripped and splashed on it, or if such a liquid has been brushed
on it or smeared on with ones ngers. It is because of our understanding of these
matters that Pollocks paintings appear to have been dripped and splashed, Van
Goghs appear to have been executed with a brush, and ngerpaintings appear
to have been ngerpainted. Most of us have a less clear conception of what sorts
of colored surfaces are likely to result from what printmaking techniques
etching, woodcut printing, and so onor from using a brush in the way
that Leonardo did. This is why most of us, on looking at prints of Leonardos
works, do not have an especially vivid or detailed sense of how these works
were made.
It is against our vast background of experience with sound-making events
that the sounds of a musical performance sound as though they were made by
actions of certain kinds. Imagine what it would be like to hear music against the
background of radically different experience. Suppose that on Mars the harder
something is hit or scraped or blown, the softer is the sound that results. Giving
a cymbal a mighty wallop produces a mere tinkle, and barely touching it brings
forth a deafening roar. No doubt what sounds to us to have resulted from violent
actions would, to Martians, sound as though they resulted from gentle ones, and
vice versa. It is intriguing to speculate about how the Martians responses to a
performanceof, say, a Schubert symphonywould differ from ours; clearly the
differences would be enormous.
What information informs correct or appropriate perception of works of art? We
cannot expect a denitive and complete answer to this question. But it is clear
that the correct perception of most, and probably all, works is informed by some
knowledge of the sort I have described, especially what is pervasive, common
knowledge in the culture in which a given work is produced. No one will deny,
I think, that an impossibly naive viewer who has no understanding at all of how
liquids behave and so has no sense of the drippings and splashings that went into
a Pollock painting misperceives it. (He fails to perceive the spontaneity and sense
of abandon which the painting possesses, for example.) Nor would anyone deny
that the Martians I mentioned mis-hear the Schubert symphony.



The simple fact that how things look is context-dependent in the manner
I have described gives rise to some intriguingly subtle and complex situations in
the arts. Let us consider, for instance, the frequent stylistic innovations designed
to combat artice and contrivance in art and to achieve instead a sense of naturalness. I have in mind the avoidance of regular meters and rhymes in poetry,
the avoidance of symmetry in the visual arts, and the avoidance of sequences and
other too obvious repetitions of thematic material in music. Many of us have
a distaste for what seems too perfect, too much under the control of the artist;
we nd a sense of randomness or accident refreshing. It is clear already that the
artice or contrivance that seems objectionable can be understood in terms of
apparent artists. One does not want the artists hand to be too obvious in his
work. Contrived works are ones that seem too much to have been made carefully,
deliberately, with attention to the details, leaving little to chance. Coincidences
in literature are often felt to constitute undesirable intrusions by the author into
his work. And many innovations in the direction of realism in painting can be
understood partly as attempts to paint things as they are or as they appear, rather
than as they are painted.19
There may be instances in which the aim is to achieve the effect of natural
objects, to produce works which do not look made at all, which have no apparent
artists. Perhaps this is John Cages objective (although it is arguable whether he
succeeded). But it is likely that most artists who introduced new styles in order
to escape contrivance wanted to produce works with apparent artists who are
more spontaneous, freer, less uptight about details, works that seem to have been
made by someone, but someone who was willing to allow things to take their
natural course without always interfering.
The rule of thumb that in pictures the main subject should not be exactly in
the center is based on the desire to avoid contrivance. But it is easy to see how
the rule can backre. If it is consistently followed, if painters and photographers
consistently put their main subjects just off center, then pictures in which that
is done may well come to seem contrivedthey will come to look as though the
artist carefully, deliberately, placed the main subject just off center so as to avoid
the appearance of contrivance! Artists might then move their subjects farther
from the center, just off just-off-center. But if this becomes the general practice,
or even if it is merely the obvious way to (try to) make a picture that doesnt
look contrived, it may come to seem contrived also. If placing the subject just off
just-off-center is precisely what needs to be done to avoid contrivance, a picture
19. Wlfin describes the Baroque as, in part, a reaction against the contrivance of Classicism. Principles of Art History, pp. 131134. But the ultimate in the attitude I have described is
found in the writings of John Cage, who advises the composer to give up the desire to control
sound, . . . and set about discovering means to let sounds be themselves rather than vehicles for
man-made theories or expressions of human sentiments. Silence (Middletown, Conn.: Wesleyan University Press, 1961), p. 10.



with the subject just off just-off-center may for that very reason appear to have
been carefully planned so as not to look contrived. It is a continuous game of
hide-and-seek that artists play with audiences. We can almost understand how
one might, in frustration, resort to determining the composition of pictures by
chance methods, in the manner of John Cage.
But if the position of the main subject in a picture is decided by chance, the
subject might end up smack in the middle, or just off center. And if it does, wont
it look contrived? Perhaps. But let us not forget that how things look is conditioned by what we know. Suppose that the artists use of indeterminacy is advertised, as John Cage advertises his; suppose that everyone can be expected to know
about it. Our realization that the main subject got where it did by chance may
well prevent it from looking contrived. Its placement may strike us as a marvelous
or at least surprising coincidence, rather than as an indication of insipid artice.
Or, especially if the subject is not centered precisely, it may just look natural.20
We are now in a position to say more about the link between styles of works
and their apparent artists. What I previously called style qualities, qualities that
works possess in virtue of their apparent artists, are not essential properties of
styles. A style that is amboyant or sentimental or timid in one work, may not
be so in another. Features that make works of one period seem to be the result of
amboyant or sentimental or timid actions may make works of another period
seem to be the result of actions of very different kinds. What once suggested an
imaginative artist later suggests a dull one; evidence of a bold creator becomes
evidence of a timid one; the expressive becomes contrived; inspiration degenerates to clich. This of course is because the manner in which certain features seem
to have been made depends not just on the features but also on the context. What
needs to be noticed is that when the features remain the same but their apparent
genesis changes, the style has not necessarily changed. So style identity is tied up
with the features, rather than with the apparent artist.21 If features that in some
works suggest bold artists, suggest timid artists in works of a later period, we
dont have to say that the works are in different styles; rather the same style that
was bold in one context is timid on the other. The style of Pierre Menards Don
Quixote is archaic and that of Cervantess Don Quixote is not; yet these works are in
the same style. The style became archaic with the change of context.
Flamboyance, pretentiousness, timidity, and other properties that are linked
similarly to apparent artists, are, let us say, expressive. Styles are to be identied
not with what is expressed but with what in the work does the expressing; style
is not expression but the means of expression. What constitutes being in a given
style is not having a certain expressive nature, but having certain features (thin
wiggly lines, painterliness, balance, etc.) which are expressive.

20. The discussion in section 5 is relevant to this point.

21. I will not try to say just what changes of features constitute a change of style.



This account locates styles of works rmly in the works themselves, as rmly
as properties of having thin wiggly lines and being painterly are located there.
But the connection with behavior remains. For which properties of a work constitute its style is at least partly a matter of which of its properties give an impression of the artists action in creating the work, which ones are responsible for how
the work appears to have been made.


Our task has only begun. It is clear already that the notion of the manner in
which a work appears to have been created is not unproblematic. But we have
barely scratched the surface of the complexity of this subject.
Let us look at works that seem to have conicting appearances, works that
seem to have been created by acts of one sort and also seem to have been created
by acts of an opposite sort. We can set aside the simple cases in which a works
appearance is merely ambiguousfor example, a novel that contains some indications that it was written with the intention of being taken as a mere adventure
story and some indications that it was meant as an allegory. Here the contrary
indications tend to cancel each other out. To the extent that the novel seems to
have been meant allegorically, it seems not to have been meant as just an adventure story, and vice versa. But in the cases I am interested in, one impression does
not in this way tend to negate the other; in fact, one may depend on the other.
In these cases the two contrary impressions are, we might say, on different levels,
and this is why they do not conict in the way they would otherwise.22
It might be said that although Pollocks canvases appear supercially to have
been made in a haphazard, spontaneous manner, in a more basic way they give the
impression of having been thoughtfully planned and carefully executed. One can
hardly deny that Mozarts Musical Joke seems to be the work of an incompetent
eighteenth-century composer; yet one might also detect behind that impression
an impression of Mozarts genius.
That we have to make a distinction of some sort between levels of appearances
is clear from the following pair of cases: (1) a work that on the surface seems to
have been meant merely to be funny, just as a joke, and yet seems to be intended
to make a serious point by means of its humora political cartoon, for example,
and (2) a work that on the surface seems to have been meant seriously, but in
which we see a play for laughs behind the earnest exteriora joke told with a
straight face, for example, or Mozarts Musical Joke. In each of these instances
there is both an impression of playfulness on the part of the artist and an impression of his seriousness. The two cases are distinguished by which impression
comes in at a deeper level.
22. I have discussed this kind of case briey in Points of View in Narrative and Depictive
Representation, Nous 10 (1977): 5152. pp. 5152.



A deeper level appearance is likely to be a more reliable indication of the

reality than a more supercial one is. If we are interested in inferring how Pollock actually went about making his paintings from how they look, we are best
advised to go by the impression of careful planning that they give, rather than by
their obvious but supercial appearance of having been dashed off haphazardly.
We can ignore the signs of incompetence in the Musical Joke insofar as we see
behind them evidence of a joking genius, if our aim is to determine what sort
of person the composer actually was. (This point might be expressed by saying
that Pollocks works really appear to have been made carefully, and only seem to
look as though they were done haphazardly. Likewise for the Musical Joke.)
But of course our interest in these works is not merely, if at all, to discover
the facts about their actual creation. We are interested in the appearances for
their own sakes. We cant ignore the more supercial appearances. The whole
point of the Musical Joke would be lost if we did not recognize a level on which
the composer seems to have been incompetent. Part of what is interesting about
Pollocks works is the ironic interplay between their sense of haphazardness and
the sense that they were done with great care; both impressions are crucial to
Moreover, in many cases of this kind we cannot expect to recognize the deeper
appearance unless we recognize the supercial one. The talent evident in the
Musical Joke is a satirical one, a talent for satirizing, specically, incompetent
composers (and performers). It is evident in the work from the fact that Mozart
did such a brilliant job of composing something that sounds as though it was
composed by an incompetent. The Musical Joke wouldnt seem a work of talent if
it were not apparently the work of an incompetent! Thus the deeper appearance
of talent depends on the supercial appearance of incompetence.
The deeper impression that a Pollock painting gives may be similarly dependent on the more supercial one. Pollocks Blue Poles may seem to have been
made haphazardly, but it may be the sort of haphazardly made painting which
seems to have been carefully designed to look haphazard. The impression of
haphazardness is striking when one rst sees the painting; the work does have
important featuresits dripped and splashed lookwhich are likely to be found
in haphazardly made works. But when we look more closely and consider what
kind of an apparently haphazardly produced work it is, we may decide that it is
the sort that is likely to have been made by an artist attempting, deliberately and
carefully, to make his work look haphazard. In terms of style, we might say that,
simply as a painting, it is in a haphazard style, but as a painting in a haphazard
style, it is in a controlled style. (Compare: Forced laughter is apparent gaiety that
seems to proceed from something other than joy.)
There can be more than two levels of dependent appearances in a work. A threelayered example is a funny story told with a straight face, in which the humor
serves a serious purpose. Ones dominant rst impression might be that the storyteller does not intend to be funny, but because of his ridiculously exaggerated



air of seriousness we realize that his story is an apparently serious one that was
meant to be funny. And on reection we conclude that the storyteller, in telling
his apparently serious story in a manner that made it seem to have been meant to
be funny, apparently intended to be making a serious point. This last impression
depends (partly) on the apparent frivolity, which in turn depends on the supercial appearance of sobriety.


The complexity of the structure of a works appearance illustrated by this example is already intimidating. But there is more to come. Robert Rauschenbergs
Factum I and Factum II (gures 12.4 and 12.5), idealized somewhat, will serve
to introduce an issue about the notion of apparent artists which has especially
important consequences for the concept of style. Factum I is a painting/collage,
part of which was done by dripping paint in the manner of Pollock. In Factum II
Rauschenberg tried meticulously to reproduce Factum I.23 I do not know exactly
what techniques he used in Factum II, but let us suppose that he used eyedroppers to deposit each drop of paint one by one in its proper place. And let us suppose that he did so skillfully enough so that a viewer could easily be fooled into
thinking that the work was made by more or less random dripping.
If a viewer of Factum II is told how it was actually made, what effect does this
new information have on how it appears to him to have been made? No doubt
it will still be true to say that it appears to have been dripped. But the viewer is
likely also to have a sense of the meticulous task of placing the drops of paint one
by one in their positions on the canvas with eyedroppers. The new information
might draw the viewers attention to subtle features of the canvas which he didnt
notice before and in virtue of which the work seems to have been eyedropped.
It would be reasonable to infer from those features, perhaps, that an eyedropper
was used. But let us suppose that this is not so, that there is nothing at all on the
canvas to suggest that it was eyedropped rather than dripped.
Nevertheless, the viewer now sees in the painting the artists careful manipulation of eyedroppers. He has a sense of what different sorts of eyedroppers and
what eyedropping techniques were employed in depositing the various blobs of
paint on the canvas.
One way to describe the situation is as follows: Before the viewer was told
that Rauschenberg used eyedroppers, the painting looked as though, if it were
made with eyedroppers, it was made with eyedroppers of certain kinds manipulated in certain ways, although this is an aspect of the paintings appearance that
the viewer wouldnt have noticed unless the possibility of the works having
been eyedropped happened to occur to him. Later, against the background of

23. Barbara Rose, American Art since 1900 (London: Thames and Hudson, 1956), p. 217.

Figure 12.4 Robert Rauschenberg, Factum I, 1957. Mixed media, 156 91 cm. The
Panza Collection, The Museum of Contemporary Art, Los Angeles. Art Robert
Rauschenberg/Licensed by VAGA, New York.

Figure 12.5 Robert Rauschenberg, Factum II, 1957. Combine painting: 61 3/8" 35
1/2". The Museum of Modern Art, New York. Digital Image The Museum of Modern Art/Licensed by SCALA /Art Resource, New York. Art Robert Rauschenberg/
Licensed by VAGA, New York.



Figure 12.6 Gary Gilmore, Iceskaters. Drawing.

the realization that eyedroppers were in fact used, the impression is no longer
conditional; now the work appears, simply, to have been made with eyedroppers,
eyedroppers of certain kinds manipulated in certain ways.
Similar examples are common. When we are told that a drawing of a delightfully serene winter scene (gure 12.6) was done by Gary Gilmore, we may see,
behind the calm lines, anger, a vicious disposition, a criminal mind, even if
nothing in the lines themselves would suggest to anyone with no special information about the artist that it was the work of someone who was angry or vicious
or criminal. Where we cannot nd overt anger in the drawing we see anger suppressed. It may, for instance, look to us as though the artist chose a pastoral setting in order to mask the madness in him. The noises of a house we believe to be
haunted seem sinisterthe result of sinister forcesand all the more so because
they sound so normal! It is as though the ghosts in residence are trying to hide
their evil doings from us. Someones laughter may sound forced if we have reason
to believe that he is not happy but would like to seem so, even if the forced quality could not be detected in the laughter alone.
These are cases in which our beliefs affect our perceptual experience. As such,
there is nothing problematic about them; as we saw earlier there is no getting around the fact that many of our beliefs do condition how things look to
us. But the examples I just cited raise special problems. The information that



Rauschenberg used eyedroppers on Factum II makes us see in the work something that is not there, it will be argued. In the Gilmore case we read back into
the drawings the anger that newspapers have convinced us Gilmore must or may
have had. But arent we deluded if we attribute the appearance of an angry artist
to the drawings themselves? The doubts about these cases arise from the fact that
the information that supposedly makes it appear that the work came about in
a certain manner is the information that the work did, or may well have, come
about in just that manner.
Perhaps we can agree that if a viewers experience of Gilmores drawings is
inuenced by the newspapers in the way described, the drawings do appear to
him to be the work of an angry man. For that is what the viewer would say, especially if he didnt realize what inuence the newspapers had on his perception,
and perhaps we will agree that a person cannot be mistaken about how things
appear to him. But this is not to concede that the drawings appear to be the
result of anger in a sense that would support our saying that they are angry works
or works in an angry style. To do so, we need to locate the appearance of anger
more solidly in the works. The crucial question is whether perception of the
drawings inuenced by the externally acquired information is to be regarded
as correct or appropriate.
Perception inuenced by externally acquired information is sometimes clearly
incorrect or inappropriate, especially if the information is idiosyncratic. Suppose
that in a painting by a close friend I see the sweat and tears and frustration
that went into it, but only because I remember the anguish he suffered as he
worked. Certainly the painting does not appear to be the result of sweat, tears,
and frustration, in a sense in which that means that it is a laborious painting or
in a laborious style. It may well be in a casual, happy-go-lucky style.
I am not prepared to argue that the situation is different in the Gilmore and
Rauschenberg examples. But there are other cases in which it is much more
plausible that information about how a work actually did come about belongs
to the background knowledge that informs correct perception of it, and that
because of this information the work appears to have come about in the way
that it actually did. This, indeed, is the lesson to be learned from Cervantes and
Pierre Menard. The information that makes Cervantess Don Quixote seem not to
have been written with pragmatism in mind, and so makes it incorrect to attribute to it overtones of pragmatism, is the information that it was written in the
seventeenth century, or anyway long before William James, and hence that it is
unlikely that its author would have had in mind the doctrines of pragmatism.
This information is common knowledge, not at all idiosyncratic. We expect it to
inform any normal persons reading of Cervantes. Reading in this informed way
is reading correctly.
A great many facts about the origins of many works of art are common knowledge, especially facts about the societies in which they originated. We know,
of various works, that they are the products of societies that were intensely



religious, or authoritarian, or anarchic. We know of some works that they were

made in periods of widespread despair or new hope; that they came from industrial societies or from agrarian ones; that they were made before, or after, the time
of Darwin or Freud or Einstein. So we have a great deal of common knowledge
concerning what interests and attitudes are at least likely to have motivated the
creation of many works, the intentions many artists could or could not reasonably be expected to have had, and so on. If this information colors our perception
of the works, so be it. If our realization that a work was produced in medieval
Europe, rather than in the United States in the 1960s, makes it seem to have
been meant as a glorication of Christianity, rather than as an ironic, satirical,
debunking of the faith, it is eminently arguable that the work is a glorication
and not a debunking of Christianity.
But we need to look more closely at the reasons some may have for thinking
that we are reading things into works that arent there in the cases before us. We
are used to inferring how things are from how they seem. In fact, the point of
saying that such and such appears to be the case is often to suggest that perhaps
such and such is the case. But inferences of this kind are illegitimate in our
examples. If a person knows or believes from what he has read in the newspapers that Gilmore was (or is likely to have been) an angry man, and if it is just
because of this that Gilmores drawings appear to him to be the work of an
angry man, obviously he cant use this appearance to support the judgment that
the artist was in fact angry.24 If he thinks that there is an appearance in the drawings which supports that judgment, he is indeed suffering from an illusion. This
point might be made by simply denying that the drawings do have the appearance of having been made in anger, even denying that they appear that way to
the person in question (and thus giving up the idea that he cannot be mistaken
about how things appear to him). It seems to him that the drawings appear to him
to have been made in anger, but actually they do nothe was carried away by
his imagination. Likewise, if Menards text, in contrast to Cervantess, seems to
have been written with pragmatism in mind, it does so in a sense that provides
not the slightest support for the claim that Menard, unlike Cervantes, did have
pragmatism in mind.
For the sake of contrast, let us recall the earlier kind of example in which
beliefs affect appearances. Pollocks canvases appear to have been dripped and
splashed on partly because of our understanding of how dripped and splashed
liquids generally behave. But of course the fact that they appear dripped and
splashed on does support the conclusion that they actually were. The dependence

24. This is oversimplied. How easily the belief that the artist was angry makes it appear
that he was, may be a legitimate indication of whether the artist was in fact angry, even if
without the belief a viewer would derive no impression at all of an angry artist from the



of the appearance on the belief, in this kind of case, does not tempt one to refuse
to recognize that the works do indeed have the appearances in question.
From the point of view of art criticism and appreciation, how works appear to
have come about is important for its own sake, and not as an indication of how
they did come about. So from these points of view there is no reason to refuse to
recognize appearances of a sort that are not indications of the reality.
But historianscultural historians especially but some art historians as
wellhave a different point of view. They do, sometimes, want to draw conclusions about the sources of works of art, about artists and their societies, from the
appearances of the works. So they must, on pain of circularity, be careful to separate out those aspects of the impressions works give of their sources which merely
reect what we already know or believe about them. Suppose a work gives the
impression of having been born in sentimentality, because we have reason to
think that the artist, living in the mushy era in which he did, was sentimental.
Is the work sentimental, or in a sentimental style? Very likely, for the critic. But
the historian cannot agree, if he wishes to use the style quality of the work to
throw light on whether the artist or his society was sentimental.
We have the basis here of a fundamental tension between two conceptions of
style qualities, corresponding to the different interests of critics and historians.
The historian (sometimes) looks through the arts at the culture, whereas the
critic looks at the arts against the background of the culture. For the critic the
quality of a works style is something to be appreciated; for the historian it is a
clue to the artist or his society. The critic, concentrating on the products of art,
can in certain circumstances allow prior conceptions of its processes full play
in his imagination, while the historian, who focuses on the processes, must in
similar circumstances ruthlessly exclude his conceptions of the processes from his
understanding of the products.
The tension between these two conceptions of style qualities does not call for
resolution. But it does need to be clearly recognized. Recognizing it will help us
to grasp better what is going on when critics and historians talk about style, and
perhaps also when they talk past each other.


Ackerman, James S., 221

in literature, 232
actions, as objects of (aesthetic) interest,
in music, 231232
in the visual arts, 232233
See also appearances
grudging, 13, 18
pleasure of, 1120, 21
levels of, 231, 240242
Abell, Catherine, 152
reality dependent, 235239, 242248
aesthetic pleasure, 1314
versus indications of reality, 241,
aesthetic properties (qualities), 196198,
202, 210211, 216219, 233
appreciation, 3, 1920, 2830, 38, 39,
perception of, 195197, 217219
5051, 56, 76, 223228, 235,
aesthetic puzzle, the, 2830, 39, 41,
241, 248
4345, 4851, 57
art for arts sake, 17
aesthetic value, v, viii, 121, 2326,
awe and wonder, 1718
2830, 38, 41, 4851, 212,
213214, 224
Bazin, Andr, 79, 8284, 85, 88, 95 n. 18
Akiyama, Ryoji, Empty Box on Its
Beardsley, Monroe, 195, 196 n. 3, 235
Way to a Reclamation Area,
Beckett, Samuel, Malone Dies, 229231
175, 176, 177, 178 n. 14
Beethoven, Ludwig van, Eroica, 49
Alberti, Leone Battista, 143144
Bell, Clive, 4
Allen, Terry, Shaking Man, 166, 170
Berger, John, 157
Antonioni, Michelangelo, Zabriske
Block, Udo Ernst, Ewiger Kreislauf,
Point, 182
158, 161, 166
Arnheim, Rudolf, 84 n. 7, 86 n. 10,
Blocker, H. Gene, 97 n. 19
95 n. 18
Bond, Douglas, 92, 94
apparent artists, 228248
Booth, George, Dog and Vase, 171,
and ctional artists, 229231
176, 189
levels of, 231, 240242
Borcorman, John, 157



Borges, Jorge Luis, Pierre Menard,

Author of Don Quixote, 235237,
239, 246, 247
bootstrapping, 1718
Brady, Mathew, 80, 115
Bubble Boy, 21
Budd, Malcolm, 150
Buuel, Luis, Viridiana, 232233
Cage, John, 13, 224, 225, 238, 239
camera obscura, 158
Caravaggio, Michelangelo, Bacchus, 70,
Carroll, Lewis, 123124
Carroll, Nol, vi, 48 n. 6, 112 n. 5,
117119, 124, 126131
Carruthers, Peter, 55 n. 22
Cassavetes, John, Shadows, 105
categories, perceptually distinguishable,
perceiving in, 204205
Cavell, Stanley, 83 n. 4, 85 n. 9
Chapman, Ron, Bicycles, 159, 164,
171, 175
Chatman, Seymour, 229 n. 11
Chardin, Jean-Simon, 16
Cialdini, Robert, 53 n. 16
Close, Chuck, 9091, 103 n. 29, 115
cobbler model, the, 223224, 226,
227, 236
cognitive values, 3, 13, 3133, 76
See also photography: as a source of
contra-standard properties
contra-standard absolutely, 212
contra-standard for a person, 201
contra-standard relative to a
category, 199
effects on perception, 209210
convention(s), 36, 207
See also depiction: depiction and
convention; values: arbitrary
correctness, 133 n. 2, 153, 210219,
236237, 246

See also imagining: prescriptions to

counterfactual dependence, 99101,
cubism, 198199, 203, 211, 215
Currie, Gregory, vi, 55, 112 n. 5,
117132, 152, 163 n. 6
and Ian Ravenscroft, 48 n. 4, 246
Cutting, James, 163 n. 3
dance, 204, 224, 226227
Danto, Arthur, 162163 n. 4, 178 n. 14
da Vinci, Leonardo, 232, 237
depiction, vi, 37, 6378, 113114,
117126, 133192, 197, 198,
201204, 228, 231, 238239
and convention, 74, 98, 134, 140
142, 152153, 202 n. 12, 203 n.
distinguished from representation,
171, 176, 189, 189, 190
moving pictures. See lm (motion
picture perception, 6375, 117126,
133142, 143155, 164166
and resemblance, 64, 74, 134, 150,
152, 197, 201204
still pictures, vi, 8990, 157195,
200, 203204
See also imagining: imagining seeing;
music: musical depictions
Dessauer, Erwin von, Children on the
Beach, 160, 164, 166167,
172, 175, 176182, 184185
Devereaux, Mary, 48 n. 6
dreams, 74, 123124, 137, 139,
Dretske, Fred, 100 n. 21
Duchamp, Marcel
Nude Descending a Staircase, 169, 171,
172, 175, 183, 185186
readymades, 224, 225226
Edgerton, Harold, Gussie Moran Tennis
Multiash, 166, 168, 183


Ehrenzweig, Anton, 206 n. 15

egocentric information, 128130
empathy, vii, 3031, 34, 43, 54, 7678,
120121, 226, 227
Escher, M. C., 42
Evans, Walker, Torn Movie Poster, 185
expression, 76, 81, 97n, 98, 113, 196,
203, 205, 216, 224, 225, 239
Feagin, Susan, 18 n. 12
ction, vii, viii, 2745
emotional responses to, vii, 7378,
See also ctionality, of propositions;
ctionality puzzle, the
ctionality, of propositions, 5657,
6667, 8889
ctionality puzzle, the, 3345, 48,
5152, 5658
lm (motion pictures), 84, 85 n. 9,
8990, 92, 105, 117, 118124,
157158, 162164, 166,
178 n. 14, 183, 187188, 189,
200, 204
freeze frames, 164, 204
point-of-view shots, 120121
slow and fast motion, 175, 181,
182, 183
folk singing and dancing, 6, 12, 226227
forgery, vii
frames, 185
Gaut, Berys, 48 n. 6
Gendler, Tamar Szab, 45 n. 24, 47 n. 3,
48 n. 4, 52, 54, 5657
Gernsheim, Helmut, 83 n. 3
Giacometti, Alberto, 215216
Gibbard, Allan, 14 n. 9, 45 n. 23
Goldman, Alvin, 87 n. 12, 95 n. 17
Gombrich, Ernst, vi, 6365, 67, 74,
84 n. 6, 221
Goya, Francisco, 80
good because bad, 1819, 21
Goodman, Nelson, 165 n. 10, 202 n. 12,
222 n. 4


Goosens, William K., 101 n. 23

Greenberg, Clement, 21
Grice, Paul, natural and nonnatural
meaning, 101103
guernicas, 204205, 210211, 212
Hanslick, Eduard, 4
happenings, 227
Hills, David, 38 n. 14
Hitchcock, Alfred
The Birds, 119120
Vertigo, 154
Hogarth, William, 42
Hollander, John, 10
Hopkins, Robert, 148 n. 20, 149 n. 23
Hume, David, 2728, 30, 3234,
41 n. 18, 45, 47
humor, 13, 14, 29, 40, 4345, 49, 50,
51, 241244
imagining, 3046, 5056, 7273,
7576, 118126, 136142,
143, 145155, 165191
imagining seeing, 37, 52, 7376, 78,
8990, 114, 117126, 137138,
146, 149152, 153, 165166,
impossibilities, 4243, 5556
prescriptions to imagine, 37, 42, 56,
185, 188
imaginative puzzle, the, 3033, 48,
5156, 5758
imaginative resistance, v, 2746, 4759
intentions, artists, vii, 19, 37, 38, 41,
49, 100 n. 22, 103, 126, 133 n.
2, 153, 195, n.196 n. 3, 212,
214215, 217, 225226, 235
intuition, 110
Jacobson, Daniel, 18 n. 12, 45 n. 24,
48 n. 6, 50 n. 10
John, Eileen, 49 n. 6
Johns, Jasper, 136, 138, 162
Johnson, Jonathan Eastman, The Old
Stagecoach, 65, 66, 72, 73



Kant, Immanuel, 15 n. 11
Kertsz, Andr, Distortion #157,
96, 98
Kieran, Matthew, 49 n. 6
Klee, Paul, 37
Klein, Yves, 208
Kracauer, Siegfried, 86 n. 10
Latham, John, 224
Le Poidevin, Robin, 160 n. 2
Levinson, Jerrold, 150152
Lewis, David, 31 n. 4, 87 n. 12, 95 n. 16,
101 n. 24
Lichtenstein, Roy, Little Big Painting,
230, 231
linguistic analysis, 110111
literature, 910, 13, 3637, 7576,
125, 146147, 196 n. 4, 238
See also narrators
Loch Ness Monster, 89
Lopes, Dominic, 110 n. 1, 148 n. 20
Manet, Eduard, La Prune, 144, 146
Martin, Edwin, vi n. 1
Masolino da Panicale, Healing of Cripple
and the Raising of Tabitha, 172,
173, 183
Matravers, Derek, 48 n. 4
childrens games of, 5, 6, 12,
6378, 139
unofcial games of, 184
See also imagining
Maynard, Patrick, 113 n. 7, 114 n. 8,
154, 166 n. 11
McCloud, Scott, 166167, 173
Mencken, H. L., 49
metaphor, vii, 33, 49
Metz, Christian, 82
mirror neurons, 226 n. 7
Mississagi River Rapids, 186, 190
Modell, F. B., 6970
moral value, v, 13, 2746, 4851, 5354
moral realism, 35 n. 9
Moran, Richard, 44 n. 22, 48 n. 4

motion, depiction of, vivii, 10,

157195, 203204
motion lines, 163, 164, 166, 172, 186
motion pictures. See lm (motion pictures)
Mozart, Wolfgang Amadeus, 180, 183
A Musical Joke, 21, 231, 240241
Mt. Geryon, 162, 174176, 184,
multiple images, 163, 164, 166167,
171, 172, 183
music, vii, 7, 9, 10, 16, 49, 197200,
204211, 226227, 231232,
electronic, 207, 232
gegaku, 211
legato piano playing, 206207, 211
musical depictions, 136, 139, 204
twelve-tone, 209210, 214215
sonata-allegro form, 199, 205206
See also folk singing and dancing;
performance; Mozart
narrators, 3637, 3839, 75, 229231
omniscient, 36
natural objects, 18, 211, 217, 222223,
234235, 238
Nichols, Shaun, 48 n. 4, 52
Nude Descending a Staircase and Going Out
for Coffee, 183
orientation, 3233, 53
Panofsky, Erwin, 83nn.34, 84, 122 n. 23
pantoum, 10
direct and indirect, 88, 8990,
179 n. 16
perceptual contact, 87 n. 13, 91,
105106, 109, 113
seeing the past, 87
See also depiction; imagining:
imagining seeing
perceptual properties, 196
performance, 7, 9, 10, 206207,
226227, 231232, 237


perspective, 82, 98, 109, 118122,

125126, 147149, 158
Peterson, James, on Peter Hutton,
New York Near Sleep for Saskia,
187 n. 25
Pearlstein, Phillip, 232
photography, vi, vii, 16, 51, 69, 79116,
117, 126132
as a source of knowledge, 79, 86, 88,
9295, 98100, 106, 109, 113,
115116, 131132
darkroom manipulation, 92,
104105, 115
digital, 114116
and privacy, 80
still. See pictures: still
Picasso, Pablo, Guernica, 204205,
210211, 212
Peirce, Charles S., 134
pictures. See depiction
Plato, 28, 49, 50
pleasure. See admiration: pleasure of
Pollock, Jackson, 232, 237, 240241,
242, 247
Poussin, Nicolas, Rinaldo and Armida,
Pylyshyn, Zenon, 180, 183
pyramids, Egyptian, 49, 50
Rauschenberg, Robert
Erased de Kooning Drawing, 196
Factum I and Factum II, 242245
Rawls, John, 7
realism, 9, 10, 7985, 9098, 105, 109,
113, 222223, 238
Reality Principle (of implication),
3435, 47 n. 1
Reid, Thomas, 2021
representation. See depiction
Richter, Mischa, 6768
Riefenstahl, Leni, Triumph of the Will, 13,
28, 29, 38, 48 n. 6, 50. 98
Robinson, Jenefer, 40
Rubens, Peter Paul, An Autumn
Landscape, 165166, 174, 188


Rushdie, Salman, 114

Savedoff, Barbara, 185
Savile, Anthony, v, 2326, 136137
Schacter, Carl, 21
Schoenberg, Arnold, 209210, 214215
Scruton, Roger, 104 n. 30
sculpture, 136, 138, 139, 148, 160,
166, 170, 208, 209, 212,
215216, 217
seeing. See depiction; perception;
imagining: imagining seeing
seeing-in, 52, 133142, 143147,
149152, 155
twofoldness, 134135, 138139,
151 n. 30, 155
seeing the unseen, 123124
Shapiro, Meyer, 221222
Shepard, Roger, 150 n. 27
Shepard tones, 5556
Shirley, Richard, Resonant, 89
Sibley, Frank, 51, 196198
silly questions, 190191
similarity and discriminability, 106109
simulation, vii
Sircello, Guy, 229 n. 10
slow motion. See lm: slow and fast
Snyder, Joel, and Neil Walsh Allen,
81 n. 1, 90n, 97, 98n, 102
Sparshot, Francis, 131
sports and competitive games, viii,
5, 67, 89, 10, 11, 1213,
1516, 31
stamp collecting, 8
standard properties
standard absolutely, 212
standard for a person, 201
standard relative to a category, 199
effects on perception, 201208
Staneld, Clarkson, On the Dogger Bank,
6667, 6869
stasis, depiction of, 162, 163, 184,



Steichen, Edward, 79, 80

Stock, Kathleen, 56 n. 24
style, vii, 221248
style qualities, 223, 233234, 239240,
sublime, 18
superrealism, 9093, 103 n. 29, 115
supervenience, 4042
Talbot, William Henry Fox, 95 n. 18
Tanner, Michael, 48 n. 4,
theater, 83, 84, 95, 136, 138, 139,
166, 224
theory construction, 110111
titles, 37, 68
transparency (of photographs), vi,
79116, 117, 126132, 179 n. 16
Tversky, Amos, and Daniel Kahneman,
53 n. 16
Uelsmann, Jerry N., 104105, 178
value(s), vii, viii, 359
arbitrary, 811, 16, 17
institution-bound, 58, 11, 17
of institutions, 48
intrinsic and instrumental, 5, 6, 8, 9,
11, 12

second-order, 1517
See also aesthetic value; cognitive
values; moral value
Van Gogh, Vincent, 232, 237
Sorrow, 7678
variable properties
variable absolutely, 212
variable relative to a category, 199
variable for a person, 201
effects on perception, 201208
Wagner, Richard, 209
Warhol, Andy, Empire, 162
Weatherson, Brian, 48, 51 n. 11, 57
Williams, Bernard, 5354
Wiseman, Frederic, Titticut Follies,
86, 87
Wlfin, Heinrich, 195, 234, 238 n. 19
Wollheim, Richard, vi, 133142,
143155, 166
Wolterstorff, Nicholas, 34 n. 8
Yablo, Stephen, 48 n. 4, 51 n. 11, 57
Yaffee, Gideon, 163 n. 6
Zemach, E. M., 85 n. 9, 87 n. 12,
110 n. 1