22
OTHER BOOKS NELSON GOODMAN The Sttucturl' of Appearance Fact, Fiction, and Forecast NELSON GOODMAN / ,/ LANGUAGES OF ART/ AN APPROACH TO A THEORY OF SYMBOLS THE BOBBS*MERRILL COMPANY. INC. A Subsidiary of Howard W. Sams & Co .• Inc. PUBLISHERS' INDIANAPOLIS,! NEW YORK' KANSAS CITY , ",.

Nelson Goodman Languages Art Symbols

Embed Size (px)

Citation preview

Page 1: Nelson Goodman Languages Art Symbols

OTHER BOOKS B~ NELSON GOODMAN

The Sttucturl' of Appearance

Fact, Fiction, and Forecast

NELSON GOODMAN

/

,/ LANGUAGES

OF ART/

AN APPROACHTO A THEORY OF SYMBOLS

THE BOBBS*MERRILL COMPANY. INC.

A Subsidiary of Howard W. Sams & Co.• Inc.

PUBLISHERS' INDIANAPOLIS,! NEW YORK' KANSAS CITY

,",.

Page 2: Nelson Goodman Languages Art Symbols

I

3

REALITY REMADE

1. Denotation

Art is not a copy of the realworld. One of the damn thingsis enough.*

Whether a picture ought to be a representation or not isa question much less crucial than might appear from cur-rent bitter among artists, critics,

nature of

'" Reported as oeCUl'ring in an ell3l1ly on Virginia Woolf. I have beenunable to locate.the source.

reJ>rellentatiion issome arts, such as painting, and infrequent in

others, such as music, threatens trouble for a unified aes­

thetics;and confu$ion over<how._~~~~L~~J~r~~_e.~~~til)I!_~ '1:'c J, ~~

.~;~~~~~~~::~~~h~:~~e~;;~~~~:~;;i!f;~~::~>-sion on the other is fatal to any general theory of symbols.

The most naive view of representation might perhaps beput somewhat like this; "A represents B if and only if Aappreciably resembles B", or "A represents B to the extentthat A resembles B". ~t!ges of ~W.L~$SQlj;cl­refinements, persist in most writing on representation~Yet_.._~-_.__..-.,,_._._~--~ .._--_."..__.~.__....~. -",-"._.-~,-----~_._~-~.",~~~_.,-.-'-._-----~~ ....-'Drawing from Paul IDee's Piidagogische Skizzenbuch

(Munich, 1925; 2nd American edition, New York:, Freder~ick A. Praeger. Inc., 1953), p. 41; reproduced here by per-mission of the publishers. .

Page 3: Nelson Goodman Languages Art Symbols

REALITY REMADE

1 What I am considering here ill pictorial representation, or depiction,and the comparable representation that may occur in other arts. Nat­ural objects may represe;lt in the same way: witness the man in themoon and the sheep-dog 'in the clouds. Some writers use "representa­tion" as the general term; for all varieties of 'what I call symbolizationor reference, and use "symbolic" for the verbal and other nonpictori"lsigns I call nonrepresentational. "Represent" and its derivatives havemany other uses, and while I shall mention some of these later, othersdo not concern us here at all. Among the latter, fo~ e"ample, are theuses according to which an ambassador represents a nation and makesrepresentations to !l foreign government.

51,1

2 I use "object" indifferently for anything a picture represents, whetheran apple or a battle. A quirk of language makes a represented objecta subject.

3 Not until the next chapter wHl denotation be distinguished from~~of reference.

DENOTATION

even if we construe "picture" broadly enough to cover allpairttings~th;·f~~~~ia~T~-;id;·~f·th;~;;:k·-in-otherways:

-A-·Constr;lii;; pcinti~'~i 'M~;ib~~~~~h"C'astl;f;'~-;;;~lik~ Iany other picture than it is like the Castle, yet it representsthe Castle and not another picture--not even the closestcopy. To add the requirement that B must not be a picturewould be desperate and futile; for a picture may representanother, and indeed each of the. once popular paintings ofart aUeriesrepresents many others.

The plain fact is that a ictur~tJ? r~nt_an ~.£.!;.:~

~~.~.,!...~2:~l~~J0rit, stand for it refer to ~nd that,no degree of,~ese.IIlblance•is suffi.t?~t. to estalJlish tb!i~.~- .~

uisite relationship of referenCE<. Nor is resemblance nece_~. . u__ n.. _~_·_ .-....• ----. """"''-Z:::''''_'''~W~~''''r~::::::_·' .

/!lacy for refere.nce; almost anything may stand for almost·•anything else. A picture that represents-like a passage J

that describes-an object refers to and, more particularly, idenotes;; i Denotation is the core of re resentation and isl~ - -tindependent of resemblance. i--~Iftl;;-;elation between picture and what it representsis thus assimilated to the -relati;;r:;:'b-;t';~;~-~p~;di~at~---;nd

.whatltappffes-;o:'wE; ~;:i~·t;~;~~;-th~~~h~;~-;;t;;:i;ti~~of-'-'-~"-.-""-'''- _., '., _.H.__ ,..._ hn:-"'''--~

representation as a special kind of denotation,n Hat does·plcio;:IaCd;~;;t~i~~'h~;;-·i;-~:;;;;:;;;-·;ith;-;ndhow does

it differ from, verbal or diagrammatic denotation! ~

.Eat imp~~~ibleanswer is_.!.h~a!..~~t:iE.Qlans~~!J.ileno §!l~.

I,l4

more error could bardly be compressed into so short aformula.

Some of the faults are obvious enough. An object resem­bles itself to the maximum degree but rarely represents

itself; resem!:l'!.f!£~l_.,:gt;l1.i1t~....t;~P!:~ll~!lli!tiQ1.lj_jLt.'~f\exive.Again,';iclike representation,.r~.§~mblf;lr!9l")Ssymm~t~ictB~

.".~ ._n_"~.';'.__ -. .,'''_·_---··· _.. - '-",.__ -', h"_ '_.,_•."_~

is as much like A as A is like B, but while a painting mayrepresent the Duke of Wellington, the Duke doesn't rep­resent the painting. FurthermQre, in many cases neitherone of a pair of very like objects represents the other:none of the automobiles off an assembly line is a picture ofany of the rest; and a: man is not normally a representationof another man, even his twin brother.~Jy....t resem.:....

.l:l.!~~...Jp.~..Y._4~lIr~E! .. js..,~nj:LJ?llffi.cie.nt.-...£Qnd},!i.2!!~r

.!~Rf..§.§~.ation.l

Just what correction to make in the formula is not soobvious. We may attempt less, and prefix the condition "IfA is a picture, . • .".: Of course, if we then construe "pic­~ as "~presentation", we resign a large part .of thequestion: namely, ~hat epustitutes a representati~ But

Page 4: Nelson Goodman Languages Art Symbols

REALITY REMADE

2. Imitation Iil

7

'IMITATION

1,2

eye comes always ancient to its \work, obsessed by iU; own past and by old and new insinu­ations of the ear, nose, tongue, fingers, heart, and brain. Itfunctions not as an instrument self-powered and alone, butas a dutiful member of a complex and capricious organism.Not only how but what it sees is regulated by need andprejudice.~ It selects, rejects, organizes, discriminates, as-

tanything-that is described or pictured. Of course, none of this implies,'I either that nothing is described or pictured. See further section 5, '1and note 19 below.

r,In~ lIIW1i"n (New York, Pantheon Books, 19(0), PI'. 297-298and elsewhere. on1he general matter of the relativity of vision, seealso such works as R; L. Gregory, Eye and Brain (New York, McGraw­Hill Book Co., 1966), and Marshall H. Segall, Donald Campbell, endMelville J. Herskovits, The InRuence of Culture on Visual Perception(Indianapolis and New¥ork, The Bobbs-Merrill Co., Inc., 1966).

6 For samples of psychological investigation of this point, see JeromeS. Bruner's "On Perceptual Readiness", Psychological Review, vol. 64(1957), pp. 123-152, and other articles there clted; also William P.

tbe more nearly 1 succeeded, the less would-=:....::.::=,e--._-'the resullt be a realistic picture.

I am to copy then, it seems, is one such aspect,,<'[J11'etll the, ways the object is or looks. But not, of courSe,

one of' these at random-not, for example, the DukeWellinE~ton as he looks to a drunk through a raindrop.

>J~a1ther, we may suppose, the way the object looks to the,11,orllli.al eye, at proper range, from a favorable angle, in

without instrumentation, unprejudiced by:,ai(Iel:ticlOs or animosities or interests, and unembellished by ,

interpretation. In short, the object is to be \seen under aseptic conditions by the free and l,

1,26

cient condition for re1!!:.esentatio~_J~i:9stthe feature that

_~istinguishe~eseJ1E:!yon JIQ:q1---A!'U1otation of '. other Ikinds. Is it perhaps th,e case that if A denotes B, then A \

'rep;;sents B just to the extent that A resembles B? I think,even this warered-do'"(1l and innocuous-looking version of., I

our initial formula betrays a grave misconception of the I

nature of representatio:n.

"To make a faithful' picture, come as close as possible to'copying the object just as it is." This simple-minded in­junction baffies me; for the object before me is a man, aswarm of atoms, a complex of cells, a fiddler, a friend, afool, and much more. 'If none of these constitute the ob,.ject as it is, what else might? If all are ways the object is,then none is the way the object is.4 1 cannot copy all thes~

4 In "The Way the World b", Review of Metaphysics, vol. 14 (1960),pp. 48-56, I have argued that the world Is as many ways as It canbe !11'J¥. described, seen, plc~red, etc.,. and that there is no, sucbthing as the way the world· is. Ryle takes a somewhat similar position(Dilemmas [Cambridge, England, Cambridge University Press, 1954),pp. 75-77) in comparing the relation between a table as a perceivedsolid object and the table as a swarm of atoms with the relationbetween a college library a,ccording to the catalogue and according torthe accountant. Some heve proposed that the way the world is couldbe arrived at by conjoining all the several ways. This overlooks thefact that conjunction itself 'is peculiar to certain systems; for example,1 1we cannot conjoin a paragraph and a picture. And any attempted\ ,combination of all the ways would be itself only one-and a.peculiady}indigestible one--<lf the wilys the world is. But what is the world thatiis in so many ways? To speak of ways the world is, or ways of de·!scribing or picturing the world, is ,to speak of world-descripnons orworld-pictures, and does not imply'there is a unique thing--<lr indeed 4

Page 5: Nelson Goodman Languages Art Symbols

REALITY REMADE

Brown, "Conceptions of Perceptual Defense", British Journal ofPsycho!ollY Monograph! Supplement XXXV (Cambridge, England,Cambndge University Press, 1961). .

7 On the emptiness of the notion of epistemological primacy and thefutility of the search fpr the absolute given, see my Structure of Ap­pearance (2nd edition; Indianapolis and New York, The Bobbs­Merrill Co., Inc., 1966-hereinafter .referred to as SA), pp. 132-145,and "Sense and Certainty", Philosophical Review, vol. 61 (1952), pp.160-167.

I\r I

\I

9

IMITATION

1,2

8 And this is no less true when the instrument we use is a camerarather than a pen or brush. The choice and handling of the instrumentparticipate in the construal. A photographer's work, like a painter's,can evince a personal style. Concerning the 'corrections' provided forin some cameras, see section 3 below.

the.. most neutral eye and the most biased are.rely sophisticated in different ways. The most ascetic (

.and the most prodigal, like the sober portrait andcaricature, differ not in how much but only in \

!lfJ.l¥<.'(Jle~y interpret. \copy theory of representation, then, is stopped at

start by inability to specify what is to be copied. Not

1an object the way it is, nor all the ways it is, nor the way. itlooks. to the mindless eye. Moreover, something is wrongwith the very notion of copying any of the ways an object!is, any aspect of it. For an aspect is not just the objectifrom-a-given-distance-and-angle-and-in-a-given-light; it isl

.the object as we look upon or conceive it, aversion' or'(construal of the object. In representing an object, we donot, copy such a construal or interpretation-we achieve it.8

In other words, nothing is ever represented either shorn/".of or in the fullnes~of its properties. A picture neverImerely represents x, \but rather represents x as a man or!represents". to be a mountain, or represents the fact· that·x \.

is a melon. What coul~ be meant by copying a fact would \be hard to, grasp eyed if there were any such. things as Ifacts; to ask me to copy x as a soandso is a little like asking \me to sell something as a gift; and to speak of copying I

.something to be a man is sheer nonsense. We shall pres- \ently have to look further into all this; but we hardly need \

!

1,28

sodates,. classifies, a..•.·nalyzes, construc.ts. It does not so Imuch mlrror as take and make; and what it takes and \makes it sees not bare, as items without attributes, but asthings, as food, as people, as enemies, as stars, as weapons.Nothing is seen nakedly or naked.

The myths of t~e innocent eye and of the absolutejgiven are unholy accomplices. Both derive from and fosterthe idea of knowing as a processing of raw material re­ceived from the senses, and of this raw:"material as beingdiscoverable either through purification rites or by me­thodical disinterpretation. But reception and interpretationare not separable operations; they are thoroughly inter~

dependent. The Kantian dictum echoes here: the innocenteye is blind and the virgin mind empty. Moreover, whathas been received and what has been done to it cannot bedistinguished within :the finished product. Content cannotbe extracted by peeling off layers of comment.7

All the same, an artist may often do well to strive forinnocence of eye. The effort sometimes rescues him fromthe tired patterns of everyday seeing, and r~sults in freshinsight. The opposite effort, to give fullest rein to a per­sonal reading, can be equally tonic-and for the same rea-

Page 6: Nelson Goodman Languages Art Symbols

IIIIII

l

11!,3

PERSPECTIVE

10 From "Pictures, Perspective, and Perception",· Dattdalut/ (Winter19(0), p. 227. Gibson d",es not appear to have explicitly retractedthese statements, though his interesting recent book, The Senses Con­sidered as Perceptual Systems (Boston, Houghton MifHin Co., 19(6).deals at length with related problems.

11 Substantially tbis argument btu, of courY, been advanced by manyother writer•• For an interesting discussion see D. Gioseffi, Pt>rsptJtiv4Artiticialis (Trieste, Univenita dega studi di Trieste, In.tituto dlStoria dell'Arte Antica e Modema, 1951). and a long review of thesaInt> by M. H. Pirenne in The Art Bullt>tin, vol. 41 (1959), pp. 213­217. I am indebted to ProfeS$Or Meyer Schapiro for this reference.

son writes: ", .. it does not seem reasonable to assert thatthe use of perspective in paintings is merely a· convention,to be used or discarded by the painter as he chooses, . . •When the artist transcribes what he sees upon a two­dimensional surface, he uses perspective geometry, ofnecessity." 10

Obviously the laws of the behavior of light are no moreconventional. tfum any other scientific laws. Now supposewe have a motionless, monochromatic object, reflectingligbt of medium intensity only, The argument runs l1:_A {\picture drawn in correct perspective will, under specifiedconditions, deliver to the eye a bundle of light rays match- \ing that delivered by the object itself. This matching is apurely objective matter, measurable by instruments. And· is~ch .~atching constitutes fidelity of repr~sentation~ for. \Mce 11gbt rays are all that the eye can receive from eitherpicture or object, identity in pattern of light rays mustconstitute identity of appearance. Of course, the rays Iyielded by the picture under the specified conditions 'match not only those yielded by the object in question ifrom a given distance and angle but also those yielded Qy i

I,3

and 251.

10

look further to see ho"" little is representation a matter ofimitation.

The case for the relativity of vision and of representa­tion has been so conclusively stated elsewhere that I amrelieved of the need to argue it at any length here.Gombrich, in particular, has amassed overwhelming evi­dence to show how the way we·see and depict dependsupon and varies with experience, practice, interests, andattitudes. But on one matter Gombrich and others some­times seem to me to take a position at odds with suchrelativity; and I must therefore discuss briefly the questionof the conventionality of perspective.

9 Art and 111u~ion, pp.

REALITY REMADE

3. Perspective

An artist may choose his means of rendering motion,intensity of light, quality of atmosphere, vibrancy ofcolor, but if he wants to represent space correctly, be\must-almost anyone will tell him--obey the laws of per- \spective. The adoption of perspective during the Renais­sance is widely acceptec!' as a long stride forward in real- 1istic depiction. The laws of perspective are supposed to·provide absolute standards of fidelity that override differ­ences in style of seeing and picturing. Gombrich derides''the idea that perspective is merely a convention and doesnot represent the world as it looks", and he declares "Onecannot insist enough; that the art of perspective aims at ac;orrect equation: It, wants the image to appear like theobject and the object like the image." 9 And James J. Gib-

Page 7: Nelson Goodman Languages Art Symbols

-1$

REALITY REMADE

131,3

14.But note that owing to the protuberance of the cornea; the eyewhen rotated, .even with the head lixed, can often see slightly aroundthe sides of an object.

PERSPECTIVE

fi:xed eye is almost as blind as the innqcent one. What can I:the matching of light rays delivered under conditions that

make norm.al vision impossible ha~e ~o do with fidelity ~f I..representatIon? To measure fidelity m terms of rays di-

I'

reeted at a dosed eye would be no more absurd. But thisobjection need not be stressed; perhaps enough eye motioncould be anowed for scanning but not for seeing aroundthe object.14 The basic trouble is that the specified condi-\tions of observation are grossly abnormal. What can be Ithe grounds for taking the matching of light rays delivered (under such extraordinary conditions as a measure of .fidel- )ity? Under no more artificial conditions, such as ·the mter- \position of suitably contrived lenses, a picture far out of Iperspective could also be made to yield the same pattern of 1light rays as the object. That with clever enough stage- J

managing we can wring out of a picture drawn in perspec- {'tive light rays that match those we can wring out of theobject represented is an odd and futile argument for the \fidelity of perspective.

Furthermore, the conditions of observation in questionare in, most cases not the same for picture and object. Bothare to be viewed through a peephole with one, transfixedeye; but the picture is to be viewed face on at a d~tance ofsix feet while the cathedral represented has to be looked at

. from, 'say, an angle of 45 0 to its fa~ade and at a distance oftwo hundred feet. Now not only the light rays receivedbut also the attendant conditions determine what and how

1,312

any of a multitude O.;f other objects from other distances}and angles.12 Identity in pattern of light rays, like resem- \blance of other kinds, is clearly no sufficient condition forIrepresentation. The claim is rather that such identity is a

"'''ri= of ''''lity,1 of 00=" pi<torial """,en"ti",,,where denotation is otherwise established. .

If at first glance the argument as stated seems simple and!persuasive, it becomE;s less so when we consider the condi­tions of observation that are prescribed. The picture mu~tl .be viewed through a peephole, face on, from a certamldistance, with one eye closed and the other motionless.!

The object also must be observe.d through a peePhOle,,\from a gh;en '(but not usually the same) angle and dis­tance, and with a single unmoving eye. Otherwise, the \light rays will not match.

Under these remarkable conditions, do we not haveultimately faithful representation? Hardly. Under theseconditions, what we are looking at tends to disat;lpearrather promptly. Experiment has shown that the eye can­not see normally without moving relative to what it sees 13;

apparently, scanning is necessary for normal vision. The

12 Cf. Gombrich's dl'scussion of 'gates' in Art and Illusion, pp. 250-251.

13 See L. A. Riggs, F. Ratliff, ]. C. Cornsweet, and T. Cornsweet, "TheDisappean;\l1ce of Steadily Fixated Visual Objects", Journi't of theOptical Society of America, vol. 43 (1953), PP' 495-501. More re­cently, the drastic and rapid changes in perception that occur duringfixation have been investigated in d"tail by R. M. Pritchard, W. Heron,and D. O. Hebb in "Vis~"l P"rception Approach"d by th" Method ofStabilized Images", Canadian Journal of Psychology, vol. 14 (1960),pp. 67-77. According to. this article, the image tends to regenerate,som"tlmes transformed into meaningful units not initially present.

Page 8: Nelson Goodman Languages Art Symbols

REALITY REMADE

we see; as psychologists are fond of saying. there is moreto vision than meets the eye; Just as a red light says "stop"on the highway ami "port" at sea, so the same stimulusgives rise to diffet'ent visual experience under different !

circumstances. Even where both the light rays and the \

~ome~taryextemal con~itions are the same, the preced- Img tram of vIsual, experIence, together with information I

gathered from all Sources, can make a vast difference in lwhat is seen. If not even the former conditions are· the {same, duplication of light rays is no more likely to result in Iidentical perception than is duplication of the conditions \if the light rays differ. j

Pictures are norrnally viewed framed against a back­ground by a person free to walk about and to move hiseyes. To paint a picture that will under these conditionsdeliver the same light rays as the object, viewed under anyconditions, would be pointless even if it were possible.Rather, the artist's.· task in representing an object beforehim is to decide what light rays, under gallery conditions,will succeed in rendering what he sees. This is not a matterof copying but of.• · conveying. It is more a matter of

~£!!~~h~nessl~~!_duplicati~_.in the sense thata likeness lost maphotograph may%e caught in a carica­ture. Translation of a sort, compensating for differences incircumstances, is involved. How this is best carried outdepends upon countless and variable factors, not leastamong them the particular habits of seeing and represent­ing that are ingrained in the viewers. Pictures in perspec­tive, like any others, have to be read; and the ability toread has to be acquired., The eye accustomed solely toOriental painting dpes not immediately understand a pic-

151,3

PERSPECTIVE

15 Adaptation to spectacles of various kinds has been the subject ofextensive experimentation. See, for example, J. E. Hochberg, "Effectsof Gestalt Revolution: The Cornell Symposium on Perception",Psychological Review, vol. 64 (1959), pp. 74-75; J. G. Taylor, TheBehavi()<'al B~s of Perception (New Haven, Yale University Press,1962), pp. 166-185; and Irvin Rock, The Nature of Perceptual Adap-tation (New York, Basic Books, Inc., 19(6). Anyone can readilyverify for himself how easy iti'" to klarn to read. pictures drawn inreversed or. otherwise transformed~(;ti~~~;;vWied-'perspeCtiveofien-o~cu~sin Ori"eiifil;"By:.antine, and mediaeval art; sometimesstandard and reversed perspective are even used in dlft'erent pam ofone picture--see, for example, Leonid Ouspensky and VladimirLossky, The Meaning of Icons (Boston, Boston Book and Art Shop,

. j1952), p. 42 (note 1), p. 200. Concerning t:he fact t:het one has to!learn to read pictures in standard perspective, Melville J. Herskovits Iwrites in Man and His Works (New York, Alfred A. Knopf, 1948), p. \381: "More t:hen one ethnographer has reported t:he experience ofshoWing a clear photograph of a house, a person, a familiar landscape

. to people living in a culture innocent of any knowledge of photography,and to have the picture held at all possihkl angles, or turned over foran inspection of its blank hack, as the native tried to interpret thismeaningless arrangement of varying shades of grey on a piece of paper'I··For even the cklarest photograph is only an interpretstion of what thecamera sees~JJ' I

ture in perspective. Yet with practice one can accom- J

modate smoothly to distorting spectacles or to pictures IIdrawn in warped or even reversed perspective.15 Andeven we who are most inured to perspective rendering do ,,!not always accept it as faithful representation: the photo-J

., graph of a man with his feet thrust forward looks disl, torted, and Pike's Peak dwindles dismally in a snapshot. A~ ,the saying goes) there is nothing like a camera to make :t 1

. !/

molehill out of a mountain. 1

So far, I have been playing along with the idea that \pictorial perspective obeys laws of geometrical optics, and

1,314

Page 9: Nelson Goodman Languages Art Symbols

RFfA:LITY REMADE

17

PERSPECTIVE

1,3

all parallels in the plane of the fa~ade project geometri­cally as parallels onto the parallel plane of the picture. Thesource of unending debate over perspective seems to liein confusion concerning the pertinent conditions of obser­vation. In Figure 1, an observer is on ground level with

. eye at a; at b,c is the fac;ade of a tower atop a building; atd,e is a picture of the tower fa<;ade, drawn in standardperspective and 'to. a scale sucb that at the indicated dis­tances picture and fa~ade subtend equal angles from Ii. Thenormal line of vision to the tower is the line a,f; lookingmuch higher or lower will leave part of the tower fa~ade

out of sight or blurred. Likewise, the normal line of visionto the picture is a,g. Now although picture and fac;ade are

.parallel, the line a,g is perpendicular to the picture, so thatvertical parallels in the picture will be projected to the eyeas parallel, while the line a,f is at an angle to the fa~ade sothat vertical paraIlels there will be projected to the eye asconverging upward We might try to make picture andfac;ade deliver matching bundles of light rays to the eye byeither (1) moving the picture upward to the position 11,i,or (2) tilting it to the position i,k, or (3) looking at thepicture from a but .at the tower from m, some stories up.

exceptions, besides Klee, are Erwin Panof.ky, "Die Perspektive alll'Symbolische Form' ", Vortrage del' BibJiothek Warburg (1924-1925),pp. 258ff; Rudolf Arnheim, Art and Visual Perception (Berkeley,University of California Press, 1954), e.g., pp. 92ff, 226ff, and el.e­where; and in an earlier day, one Arthur Par.ey, who was taken to·task for his heterodox views by Augustus de Morgan in Budget ofParadoxes (London, 1872), pp. 176-177. I am indebted to Mr. P. T.·Geach for this last reference. Interesting discussions of perspectivewill be found in The Birth and Rebirth of Pictorial Space, by JohnWhite (New York, Thoma. Yoseloff, 1958), Chapters Vin and XIII.

1,316

that a picture drawn according to the standard pictorialrules will, under the very abnormal conditions outlinedabove, deliver a bundle of light rays matching that deliv­ered by the scene portrayed. Only this assumption givesany plausibility at all to the argument from perspective;

but the assumption i•...5 plainly false. By the pictoriabules,1railroad tracks running outward from the eye are drawnconverging, but telefJhone poles (or the edges of a fa~ade) !'running upward from the eye are drawn parallel. By the I'laws of geometry' the poles should also be drawn con- \verging. But so drawn, they look as wrong as railroad )tracks drawn. parallel. Thus we have cameras with tilting (I

backs and elevating lens-boards to 'correct distortion'­that is, to make vertical parallels come out parallel in our \photographs; we dOi not likewise try to make the railroad !tracks come out parallel. The rules of pictorial perspective \no more follow from the laws of optics than would rulescalling for drawing jthe tracks parallel and the poles con- iverging. In diametric contradiction to what 'Gibson says,the artist who wants'to p,roduce a spatial representation thatthe present-day Western eye will accept as faithful mustdefy the 'laws of geometry'.

If all this seems! quite evident, and neatly clinched b¥,Klee,HI_ there is n~vertheless impressive weight of au­thority on the other sidep relying on the argument that

16 see. the frontispiece ..•. to this chapter. As Klee remarks, the. drawing Ilooks quite normal if· taken as repn,senting a floor but awry as rep­resenting a fa91de, eyen though in the two cases parallelll in the \object represented recede equally from the eye. \

>7' Indeed, this i. the 6rthodox position, taken not only by Pirenne,Gibson, and Gombrich, but by most writers on the subject. Some

Page 10: Nelson Goodman Languages Art Symbols

19

SCULPTURE

4. Sculpture

18 The optimal way of seeing the tower fll~llde may be by lookingstraight at it from m; but then the optimal WilY of seeing the railroad,tracks would be ,by looking down on them from directly above tbemidpoint of their length,

The troubles with the copy theory are sometimes at~1tributed solely to the impossibility of depicting reality~in- ".th~TOund on a fiat surface. But imitation is no better

In the first two cases, since the picture must be also nearerthe eye to subtend the same angle, the scale will be wrong

.for lateral (left~right) dimensions. What is more impor­tant, none of these three conditions of observation is any­where near normaL We do not usually hang pictures farabove eye level, or tilt them drastically bottom toward us,or elevate ourselves at will to look squarely at towers}8With eye and picture in normal position, the bundle of Ili.ght ".I'.a.,Y.Ii delivered to the eye by the picture drawn in tstandard perspective is very different from the bundle d~livered by the facade. \

This argument by itself is conclusive, but my case doesnot: rest upon it The more fundamental arguments ad­vanced earlier would apply with full force even had thechoice of official rules of perspective been less whimsicaland called for drawing as convergent all parallels receding

in a,ny direction. Briefly, the behavior of light sanctions jneither our usual rior any other way of rendering space;'and perspective provides no absolute or independentstandard of fidelity.

1,3

__...... b

REALITY REMADE

18

f

m~__~~c

Page 11: Nelson Goodman Languages Art Symbols

21

FICTIONS

hayebeen considering only the representationperson or group or thing or scene; but a

predicate, may denote severally the mem7class. A picture accompanying a definition

ij)narr is. often such a representation, not denotingjSOI:Xl~ ~ne eagle, say, or collectively the class of.,distributively eagles in general.representations have neither unique nor multiple I~,What, for example, do pictures of Pickwick or J

rnrepresent? They do not represent anything; freprese~tations with null denotation. Yet how canll

that a plcture. represents Pickwick, or a unicorn, Ialso say that it does not represent anything2 Since \

·,>,ner'e.is no Pickwick and no unicorn, what a picture of,<l~i(:kwick d' f . \'" an a picture 0 a unicorn represent is the same. 1

su~elY to be a picture of Pickwick and to be a pictureIa Unlcorn are not at all the same. I

The simple fact is that much as most pieces of furnitureare readily sorted out as desks, chairs, tables, etc., so mostpictures are readily sorted out as pictures of Pickwick, ofPegasus" of a unicorn, etc., without reference to anythingrepresented. What tends to mislead us is that such JOCU_'/tions as "pict\lre of" and "represents" have the appearance

.

.. ~.f mannerly two-place predicates and can sometimes be somterpreted. But "picture of Pickwick" and "represents a

.. unicorn" are better considered unbreakable one--placepredicates, 0: ~las&:terms, like "desk" and "table". We I

..caltlnc,t reach lnSlde, any of these and quantify over parts of

1,420

REALITY REMADE

gauge of realism: in sculpture than in painting. Whatistd'be portrayed in ia bro.nz;e bust is a mobile, many~facete<J; ..and fluctuating person, encountered in ever changing ligt1~

and against ~l1aneous backgrounds. Duplicatingt··form of the heal1 at a given instant is unlikely to yieldnotahly faithful irepresentatiop. The very fixation ofa momentary pliase embalms the person much as a Phot*'graph taken at 1:00 short an exposure freezes a fountain. orstops a racehorse. To portray faithfully is· to .convey a. .person known and distilled from a variety of experi~cThis elusive conceit is nothing that one can meaningftry to duplicate lor imitate in a static bronze on a pedeStalin a museum. The sculptor undertakes, rather, a subtle and,'intricate problem of translation. J

Even where t1;te object represented is something simplerand more stable than a person, duplication seldom coincideswith realistic rel?resentation. If in a tympanum over a tallGothic portal, E!ve's apple were the same size as a Wine-­sap, it would not look big enough to tempt Adam. Thedistant or colosSal sculpture has also to be shaped verydifferently from 'what it depicts in order to be realistic, inorder to 'look right'. And the ways of making it 'lookright' are not reducible to fixed and universal rules; forhow an object looks depends not only upon its orientation,dist~nce, and lighting, but upon all we know of it andupon our trainin;g, habits, and concerns.

One need ha~dly go further ,to see that the basic caseagainst imitation as a test of realism is conclusive for sculp-ture as well as f9r painting. ~.

Page 12: Nelson Goodman Languages Art Symbols

REALITY REMADE

19 The 'ubstance of this and the following two paragraphs is containedin my paper, "On Likenes. o(Mesning", Analysis, vol, 1 (1949), pp.

1-7, and discussed f(.rther in the sequel, "On Some Differ~l1ce. aboutMeaning", A"alysis,!vol, 13 (1953), pp. 90-96. See also the paralleltreatment of the p~oblem of statements 'about fictive entities' in"About", Mind, '1'01.:70 (19tH), esp. pp. 18-22. In a series of papersfrom 1939 on (many of them reworked and republished in From aLogical Point of View [Cambridge, Mass~ Harvard University Press,1953]), W. V. Quine had sharpened the distinction between syncate­gorematic and other expressions, and had· shown that careful obser­VanCe of this distinction could dispel many. philosophical problems.

I u;e the device &f hyphenation (e.g., in "man-picture") as an aidin technical discour\;e only, not as a reform of everyday usage, wherethe context normally prevents confusion and where the impetus tofallacious existential inference is less compulsive, if not less con­sequential, than in; philosophy. In what follows, "man-picture" wiltalways he an abbr!,viation for the longer and more usual "picturerepresenting a man1', t;aken a. an unbreakable one-place predicate that

23

FICTIONS

The difference between a man~pict~re and a picture of ahas a close parallel in the differE1Dce between a man­

.d~(;riJPtion (orman-term) and a description of (or termfor) a man. "Pickwick", "the Duke of Wellington", "the

:.man who conquered Napoleon", "a man", "a fat man",.. ''the man with three heads", are all man-descriptions, but

all describe a man. Some denote a particular man, some]'denote each o~, many men, and some denote nothing.20

although "Pickwick" and ''the three-headed man" i

,j!lnd "Pegasus" all have the same nul! extension, the second"piffers from the first in being, for example, a many-headed­.m,an-dl~scriptic,n..while the last differs from the other twoin being,a winged-horse-description.

The way pictures and descriptions are thus classified,into kinds, like most habitual ways of classifying, is farfrom sharp or stable, and resists codification. Borderlines

and blur, new categories are always coming into.plrornitler.,ce, and the canons of die classification· are

clear than the practice. But this is only to say that we

. need not apply to, all or only to pictures that represent an actual man.The same general principle will"govern use of all compqunda of the

.form "--picture", Thus, for example, I shall not use "Churchill-.a~ an abbreviation for "picture painted by Churchill" or for

"niirt.,ra. belonging tu Churchill". Note, furthermore, that a square­IS not necessarily a square picture but a square-representing-

Strictl~, we should speak here of utterances and inscriptions; formstances of the same term may differ in denotation. Indeed

. .classifying replica. together to constitute term. is only one and ~'. f~r from 'imple, way of .·classifying utterances and inscripti~ns into·klnds.• Slle further SA, pp. 359-363, and also Chapter IV below.

1,522

them. From the fact that P is a picture of or represents a (

unicorn we cannot.•..infer that there is somet~ing that P ~s apicture of· or represents. Furthermore, a pIcture of PIck­wick is a picture of a man, even though there is no man itrepresents. Saying that a picture represents asoandso is \thus highly ambigtlous as between saying what the picture Idenotes and saying what kind of picture· it is. Some confu~ >sion can be avoidEid if in the latter case we speak rather ofa 'Pickwick-repre:renting-picture' or a 'unicorn-r.epresenting­picture' or a 'man-representing-picture' or, for short, ofa'Pickwick-pictute' or 'unicorn-picture' or 'man-picture'.Obviously a picture cannot, barring equivocation, both ,(represent Pickwick and represent nothing. But a picture 'Imay be of a certain kind-be a Pickwick-picture or a man-lpicture--without irepresenting anything.19 I

Page 13: Nelson Goodman Languages Art Symbols

REALITY REMADE

may have some trouble in telling whether certain pictures(in common parlance) 'represent a unicorn', or in settingforth rules for de¢iding in every case whether a picture is,aman-picture. &act and general conditions under whichsomething is a soandso-picture or a·soandso-descriptionwould indeed be hard to formulate. We can cite examples:Van Gogh's Postman is a man-picture; and in English, "aman" is a man-description. And we may note, for instanc~, '(that to be a soandso-picture is to be a soandso-picture as awhole, so that a picture containing or contained in a man­picture ..... 001 "",I' '" •m.""'''u,", But to ._pt ,much more is .to become engulfed in a notorious philo..sophical morass;; and the frustrating, if fascinating, prob­lems involved are no part of our present task. All thatdirectly matters ,here, I repeat, is that pictures are indeedsorted with varying degrees of ease into man-pictures,unicorn-pictures,: Pickwick-pictures, winged-horse-pictures,etc., just as pieces of furniture are sorted into desks,tables, chairs, etp. And this fact is unaffected ,by the, diffi­culty, in either Jase, of framing definitions for the severalclasses or eliciting a general principle of classification.

The possible tobjection that we must first understandwhat a man or J unicorn is in order to know how to apply"man-picture" or "unicorn-picture" seems to me quiteperv.erted. We .. can learn to apply "corncob pipe" or"staghorn" without first understanding, or' knowing howto apply, "corn" or "cob" or "corncob" or "pipe" or

"stag" or "horn".•....•.•.. as separa.te terms. And we can learn, o.n \.the basis of samples, to apply "unicorn-picture" not onlywithout ever h~ving seen any unicorns but without ever

24 1,5

FICTIONS

having seen or heard the word "unicqrn" before. Indeed, !largely by learning what are unicorn-pictures and lUlicorn- _'.descriptions do we come to understand the word "unicorn"; ,

.and our ability to recognize a staghorn may help us to Iirecognize a stag when we see one. We may begin to under­".,..-~ .. term by learning how to apply either the term itself 1or some larger term containing it Acquiring any of these!

may aid in~acquiring, but does not imply po~eSSing,'\1of the others. Understanding a term is not a precon- 1

clition, and may often be a result, of learning how to apply Iterm and its compounds.21

I said that denotation is a necessary condition for I>rl~pJ:esental:ioi[l, and then encountered representations.with()ut denotation. But the explanation is now clear. A

'.. piiet1Jre must denote a man to represent him, but need not i-~..~"~ anything to be a man-representation. Incidentally, \

'copy theory of representation takes a further beating I"6~_' for where a representation does not represent any- \

there can be no question of resemblance to what it irepresents. . (

.of such examples as Pickwick-pictures and unicorn.

know how to apply all compounds of a term would entail··.·> .••~:nQ,,,·irlg how to apply at least some compounds of all other terms in

th'~.·lilngua~:e. We normally say we understand a term when we knowre.,so~lablly well how to apply it and enough of its more usual com.!'K>uI1ds. If for a given "-picture" compound we are in dOUbt about

to apply it in a rather high percentage of calleS, this is also truethe correlative "represents as a --" predicate. Of Course, under­

, a term is not exclusively a matter of knowing how to applyIts compo,;,nds; such other factors enter as knowing what

inf'el'E,nc.~s can be drawn from and to statements containing the term.

25

Page 14: Nelson Goodman Languages Art Symbols

\j

REALITY REMADE

pictures may suggest that representations with null, deIi~tation are comparatively rare. Quite the contrary;thl!'world of pictures ~eems with anonymous fictional, persCJ~!places, and things. The man in Rembrandt's Landse8p¢with a Huntsman)s presumably no actual person; he isjulltthe man in Rembrandt's etching. In other words, theet<:h­ing represents n~ man but is simply a man-picture,

"\ !more particularly a the-man-in-Rembrandt's-Land\ ' with-a.-Huntsman-picture. And even if an actual man'

depicted here, his identity matters as little as the, artisblood-type. Furthermore, the information needed totermine what if ,anything is denoted by a picture isalways accessible. We may, for example, he unable to·whether a given representation is multiple, like ,an, eapicture in the dic!tionary, or fictive, like a Pickwick-picBut where we: cannot determine whether a pi " ,,'denotes anything or not, we can only proceed as if itdi(i.not-that is, con,fule ourselves to considering what kinpicture it is. Thus cases of indeterminate denotation are;treated in the same way as cases of null denotation. '

But not only Mrhere the denotation is null, or indetenhi­nate does the classification of a picture need to be consiered. For the d~notation of a picture no more determ 'its kind, than the kind of picture determines the de ,ti0IJ. Not every, man-picture represents a man, andco~­

versely not every picture that represents a man is a matt'"

\

~:~:~e~ ~~_~n,l,.',,;~~: :~:e:::c:n::::ce: ;:~:g ;~,.~"."",.,t,.".•,.,..••J.i,'.that denote a man, between those that do and those that donot represent him as a man. ",.

26

REPRESENTATION-AS

'}()cution:"represents ... as" has two quite different?Ilay that a picture represents the Duke of WeI­~~aninfant, or as an adult, or as the victor atqisoften merely to say that the picture represen:ts

lfe/atagiven time or period-that it denotes a cer­,t).~,orshort, continuous or broken) temporal part, 's;e'of him. Here"lils .. ." combines, ",ith,the \

Duke of Wellingt;;~"'to form-a'd;;c~ipti;;n:'~f ')',9;1:1:ion of the whole extended individual:22 Such ap~oJlcan always be replaced by another like "the i

keoLWellington" or ''the Duke of Wellington !!'10cl;;asion of his victory at Waterloo". Thus these 1. po difficulty; .all that is being said is that the

feJ:>resents the object so described.~~cond use is illustrated when we say that a given;represents Winston Churchill as an infant, where \ttlI'cdoes not represent the infant Churchill but ),ep~esents the adult Churchill as an infant. Here, as ,/'

hen.we say that other pictures represent the adultHasan adult, the "as ., ." combines with and ,

, ~'-~--"'" "'-', d" ". '" ,'H • _,' _ ••{~

the yerb;and we have genuine cases of represen- i""~-"'~"";"-""""'""~'.,,'-,. --.-"',,,__• .,,,.-, '.,-,,_,.•, ._. . .,., '"H. • •. ,. _ .' . ". • ,

.~_,Such representation-as wants now to be distillJfrom and related to representation. I

l.,pictiL1re that represents a man denotes him; a picture

indel:!ted to Mr. H, P. Grice and Mr. J. O. Urmson for com­ingto clarification of this point. Som~times. th~ portionmay }Ie marked ofi' along other tfian temporal lines. On

of a temporal part, see SA, pp. 127-129.

21

Page 15: Nelson Goodman Languages Art Symbols

29

REPRESENTATION-AS

1,6

as a whole both denotes k and is a soandso-picture.24

Many of the modifiers that have had to be included here.may, however, be omitted as understood in what follows;for example, "is or contains a picture that as a whole bothdenotes Churchill and is an adult-picture" may be short­ened to "is an adult-picture denoting Churchill".

Everyday usage is often careless about the distinctionbetween reprt;lsentation and representation-as. Cases havealready been cited where in saying that a picture repre­sents a soandso we mean not that it denotes a soandso butthat it is a soandso-picture. In other cases, we may meanboth. If I tell you I have a picture of a certain black horse,and then I produce a snapshot in which he has comeout alight speck in the distance, you can hardly convict me of

. lying; but you may well feel that I misled you. You under­standably took me to mean a picture of the black horse ass~ch; and you therefore expected the picture not only todenote the horse in question but to be a black-horse­picture. Not inconceivably, saying a picture represents the /' ?

black horse might on other occasions mean that it repre­sents the horse as black (i.e., that it is a black-thing-picturedenoting the horse) or that it represents the black thing inquestion as a horse (i.e., that it is a horse-picture denotingthe black thing).

The ambiguities of ordinary use do not end there. To

. 24 This covers cases where k is represented as a soandso by either awhole pkture or part of it. As remarked in the latter part of note 19above, there are restrictions upon the admissible replacements tor"soandso" in this t1ennitional schema; an old or square picture or onebelonging to Churchill doe. not thereby represent him as old or squareor self-possessed.

1,628

REALITY REMADE

23 The contained picture may, nevertheless, denote given objects andbe a soandsa-picture as a result ot ,its incorporation in the context ofthe containing picture, just as "triangle" through occurrence in ','tri­angle and drums" may denote given musical instruments and be amusical-imtrument-description.

ithat represents a fiCti.. ·onal man.iS a ~an-pictur~; and a pic-/ture that represents a man as a man IS a man-pIcture denotiing him. Thus while, the first case concerns only what the\

PiCtur.e deJ1otes, and.... the second only what kind of pic~it is, the third concerns both the denotation and th'\classification.

More accurate formulation takes some care. What a pic­t~e is said·to repre~ent may be denoted by the picture asa. whole or by a part of it. Likewise, a picture may be asoandso-picture as a whole or merely through containing asoandso-picture.23 90nsider an ordinary portrait of theDuke and Duchess of Wellington. The picture (as awhole) denotes thj:l couple, and (in part) denotes theDuke. Furthermore, it is (as a whole) a two-person-pic­ture, and (in part)! a man-picture. The picture representsthe Duke and Duchess as two persons, and represents theDuke as a man. But althoughjt,t~presentsthe Duke, and isa two-person-pictuI)e, it obvi,it'usly does not represent theDuke as two persons; and although it represents two per­sons and is a man-picture, it does not represent the two asa man. For the picture.neither is nor contains any picturethat as a whole both represents the Duke and is a two-man­picture, or that as a whole bq},h represents two persons andis a man-picture. ' .-' .

In ~eneral, then".an object k is represented as.a SOandsO\by a picture p if and only if p is or contains a pIcture that

Page 16: Nelson Goodman Languages Art Symbols

REALITY REMADE

!/ !

I311,7

INVENTION

If representing is a matter of classifying objects rather 1than of imitating them, of characterizing rather than of [r,~..copying, it is not a matter of passive reporting. The object Idoes not sit as a docile model with its attributes neatly

so also are objects classified by or under various pictorial Ilabels. And the labels themselves, verbal or pictorial, are in ,turn classified under labels, verbal or nonverbal. Objectsare classified under "desk", "table", etc., and also underpictures representing them. Descriptions are classifiedunder "desk-description", "centaur-description", "Cicero­name",etc.; and pictures under "desk-picture", ''Pickwick­picture", etc. Tne labeling of labels does not depend upon \

.what they are labels for. Some, like "unicorn", apply tonothing; and as we have noted, not all pictw::e.IL of soldiers \are soldier-pictures. Thus with a picture;~ with;'y'oih~;'label, there are always two questions: what it represents(or describes) and the sort of representation (or descrip-,tion) it is. The first question asks what objects, if any, it \applies to as a label; and the second asks about which Iamong certain labels apply to it. In representing, a picture \at once picks out a class of objects and belongs to a certaindass or classes of pictures.2~

26 The picture does nut denote the cIa.s picked out, but denotes theJ' }no or one or .everal members of that cIa••• A picture of coun.. belongsto countIess cIa"es, but only certain of these (e.g., the class of square­

,pictures, the class of. Churchill-pictures) and not others (e.g., the.class of square pictures, the class of pictUres belonglXlg to Churchill)bave to do with what the picture.represents-as.

I,630

say that the adult Churchill is represented as an infant (orIas an adult) is to'say that the picture in question is an!infant-picture (or ~n adult-picture). But to say that Pick-Iwick is represented as a clown (or as Don. Quixote) can­not mean that the picture is a clown-picture (or Don­Quixote-picture) rflpresenting Pickwick; for there is. noPickwick. :Rather, what is being said is that the picture be­longs to a certain rather narrow class of pictures that maybe described as Pickwick-as-clown-pictures (or Pickwick­as-Don-Quixote-pictures).

Distinctions obscured in much informal discourse thusneed to be carefully marked for our purposes here. Being amatter of monadic classification, representation-as differsdrastically from dyadic denotative representation. If a pic­ture represents k as a (or the) soandso, then it denotes kand is a soandso-pitture. If k is identical with h, the picturealso denotes and represents h. And if k is a sumand­such, the picture lilso represents a (or the) suchandsuch,but not necessarily as .a (or the) suchandsuch. To repre­sent the first Duk~ of Wellington is to represent ArthurWellesley and also to represent a soldier, but not neces­sarily to represent' him as a soldier; for some pictures ofhim are civilian-pictures.

/·Representations,' then, are pictures that function in I~somewhat the sam~ way as descriptions.25 Just as objects j

are classified by means of, or under, various verbal labels, \

25 The read~r will aIr.•••.•• eady have notic.ed that. "d.~scriPtlO~" ~n t?ell.present text'" not confined to what are called defimte descnptiOtlll tn

logic but covers all predicates from proper names through purplepassages, Whether with singular, miiftiple,-;;~;;:Uifdeiiotation.

Page 17: Nelson Goodman Languages Art Symbols

REALITY REMADE

separated and thru# out for us to admire and portray. Itis one of countless Objects, and may be grouped with anyselection of them; ~nd for every such grouping there is anattribute of the obj~ct. To admit all classifications on equalfooting amounts to imaking no classification at alL Classifi­cation involves pre(erment; and application of a label (pic­torial, verbal, etc.) .as often effects as it records a classifica­tion. The 'natural' ikinds are simply those we are in thehabit of picking ol1t for and by labeling. Moreover, theobject itself is not (ready-made but results from a way oftaking the world. The making of a picture commonly par- ]ticipates in making' what is to be pictured. The object andits aspects .depend upon organization; and labels of all sortsare tools of organization.

Representation and description thus involve and areoften involved in organization. A label associates togethersuch objects as it' applies to, and is associated with the

1o.ther labels of a kind or kinds. Less directl:" it a~sociatesits referents with these other labels and With their refer­ents, and so on. Not all these associations have equal force; \their strength varies with their directness, with the speci­ficity of the classifications in question, and with the firm­ness of foothold these classifications and labelings havesecured. But in al~ these ways a representation or descrip­tion, by virtue of' how it classifies and is classified, maymake or mark connections, analyze objects, and organizethe world.

Representation /)1' description is apt, effective, illuminat­ing, subtle, intrig' ing, tiD the extent that the artist or""Titer grasps and significant relationships and devises

331,7

INVENTION

means for making them manifest. Discourse or depictionthat marks off familiar units and sotts them into standardsets under well-worn labels may sometimes be serviceableeven if humdrum. The marking off of new elements orclasses, or of familiar ones by labels of new kinds or bynew combinations of old labels, may provide new insight.Gombrich stresses Constable's metaphor: "Painting is ascience . . • of' which pictures are but the experiments." 27

In representation, the artist must make use of old habitswhen he wants to elicit novel objects and connections. Ifhis picture is recognized as almost but not quite referringto the commonplace furniture of the everyday world, or ifit" calls for and yet resists assignment to a usual kind of 'picture, it may bring out neglected likenesses and differ­ences, force unaccustomed associations, and in some mea­s)lre remake our world. And if the point of the picture isnot only successfully made but is also well-taken, if therealignments it directly and indirectly effects are interest­ing and important, the picture-like a crucial experiment-makes a genuine contribution to knowledge. To a com­pleint that his portrait of Gertrude Stein did not look likeher, Pieasso is said to have answered, "No matter; it will"

In sum, effective representation and description requireinvention. They are creative. They inform each other; andthey form, relate, and distinguish objects. That nature imi­tates art is too timid a dictum. Nature is a product of artand discourse.

27 From Con.table's fourth lecture at the Royal Institution in 1836;see C.R. Leslie, Memoirs of the Life of John Constable, ed. JonathanMayne (London, Phliidon Press, 1951), p. 323.

1,732

Page 18: Nelson Goodman Languages Art Symbols

8. Realism

I

. r

I

351,8

REALISM

eludes mistaking it for anything else; iand the appropriate Iconditions of observation (e.g., framed, against a uniform rbackground, etc.) are calculated to defeat deception. Decep- \tion enlists such mischief as a suggestive setting, or a peel'- !

hote that occludes frame and background. And deceptionunder. such nonstandard conditions is no test of realism;for with enough staging, even the most unrealistic picturecan deceive. Deception counts less as a measure of realismthan as evidence of magicianship, and is a highly atypicalmishap. In looking at the· most realistic picture, I seldom 1suppose that I can literally reach into the distance, slice the (tomato, or beat the drum. Rather, I recognize the images l

as signs for the .oilie.cts. and. c.haracte....r.1.·..S.tic.s. represented-I)'signs that .work instantly and unequ;v9c~ny,:..:without beingconfused with 'W"l:iaf-theYdenot~:~Of course, sometimeswhere deception does oceur--say by a painted window in Ia mural-we may indeed call the picture realistic; but such Icases provide no basis for the usual ordering of pictures iin general as more or less realistic. I

Thoughts along these lines have led to the suggestionl'that the most realistic picture is the one that provides the!greatest amount of pertinent information. But this hy-\pothesis can be quickly and completely refuted. ConsiderIa realistic picture, painted in ordinary perspective andnormal color, and a second picture just like the first except \that the perspective is reversed and each color is replaced Iby its complementary. The second picture, appropriately \interpreted, yields· exactly the same information as thel'first. And any number of other drastic but information­preserving transformations are possible. Obviously, real-

1,834

REt\LITY REMADE

This leaves unansW;ered the minor question what consti­t11tes realism of representation. Surely not, in view of the

foregoing, any sort Of,.•.••.•. resemblance to reality. Yet we do inifact compare represel{-tations with respect to their realismor naturalism or fidelity. If resemblance is not the cri­terion, what is?

One popular answer is that the test of fidelity is decep-:1tion, that a picture is realistic just to the extent .that it is alsuccessful illusion, ·le<i)ding the viewer to suppose that it is,jor has the characteristics of, what. it represents. The pro-.posed measure of realism, in other words, is the proba­bility of confusing the .representation with the repre-­sented. This is some iimprovement over the copy theory;'!for what counts here lis not how closely the picture dupliJcates an object but how far the picture and object, underlconditions· of observapon appropriate to each, give rise tolthe same responses land expectations. Furthermore, theltheory is not immeqiately confounded by the fact thatfictive representations differ in degree of realism; for eventhough there are no yentaurs, a realistic picture might de-­ceive me into taking it for a centaur.

Yet there are diffiFulties. What deceives depends uponwhat is observed, and what is observed varies with inter-,ests and habits. If the probability of confusion is 1, we no- -I

longer have represen.tation-we have identity. Moreover,the probability seldoth rises noticeably above zero for eventhe most guileful trompe-l'miI painting seen underordinatY/gallery conditions. Fpr seeing a picture as a picture pre-\

Page 19: Nelson Goodman Languages Art Symbols

REALITY REMADE

28 Cf. Descartes, Meditations on the First Philosophy, trans. E. S.Halqane and G. R. T. Ross (New York, Dover Puplications, Inc.,HiSS), vol. 1, p. 155; also !Berkeley, "Essay Towards a New Theory ofVision", in Works on Vision, ed. C. M. Turbayne (New York, TheBobbs-Merrill Co.; Inc., 1963), p. 42.

371,8

REALISM

21t Or conventional; but "conventional" is a dangerously ambiguousterm: witness the con1n\st between "very conventional" (as "veryordinary") and "highly conventional" or "highly conventionalized"(as "very artificial").

Realism is relative, determined by the system of repre­sentation standard for a given culture or person at a giventime. Newer or older or alien systems are accounted artifi­cial or unskilled. For a Fifth-Dynasty Egyptian the

. straightforward way of representing something is not thesame as for an eighteenth-century Japanese; and neitherway is the same as for an early twentieth-century Eng­lishman. Each would to some extent have to learn how toread a picture in either of the other styles. This relativityis obscured by our tendency to omit specifying a frame ofreference when it is our own. "Realism" thus often comes f

to be used as the name for a particular style or system of Irepresentation. Just as on this planet we usually think of 'objects as fixed if they are at a constant position in relation \to the earth, so in this period and place we usually think ofpaintings as literal or realistic if they are in a traditiona1 29

European style of representation. But such egocentric el­lipsis must not tempt us to infer that these objects (of anyothers) are absolutely fixed, or that such pictures (or anyothers) are absolutely realistic.

Shifts in standard can occur rather rapidly. The veryeffectiveness that may attend judicious departure from atraditional system of representation sometimes inclines usat least temporarily to install the newer mode as standard.We then speak of an artist's having achieved a new degreeof realism, or having found new means for the realisticrendering of (saY') light or motion. What happens here is

1,836

. istic and unrealistic pi~tures may be equally informative;informational yield is :q.o test of realism.

So far, we have not Ileeded to distinguish between fidel-ity and realism. The criteria considered earlier have beenas unsatisfactory for t1\.e one as for the other. But we canno longer equate them. The two pictures just describedIare equal~y correct, e~ually faithful to what they repre-/sent, prOVIde the same land hence equally true information; I .yet they are not equ.all.•.•.y.· realistic or literal. For a picture to \be faithful is simply for the object represented to have theproperties that the picture in effect ascribes to it. But such ~~fidelity or correctness ~r truth is not a sufficient condition I \for literalism or realism.

The alert absolutist will argue that for the second pic­ture but not the first we need a key. Rather, the differenceis that for the first the key is ready at hand. For properreading of the second picture, we have to discover rules ofinterpretation and apP,ly them deliberately. Reading of the,'first is by virtually automatic habit; practice has rendered!the symbols so transparent that we are not aware of any ieffort, of any alternatives,-or of making any interpretation \at a11.2ll Just here, I tpink, lies the touchstone of realism: 1not in quantity of inf?rmation but in how easily it issues. !And this depends upon how stereotyped the mode of 1representation is, upon how commonplace the labels and \their uses have become.

Page 20: Nelson Goodman Languages Art Symbols

REALITY :REMADE

fr

REALISM

1,8

Our addiction, in the face of overwhelming counter- 1\

evidence, to thinking of resemblance as the measure of Irealism is easily understood in these terms. Representa­tional customs, which govern realism, also tend to generateresemblance. That a picture looks like nature often meansonly that it looks the way nature is usually painted. Again,what will deceive me into supposing that an object of agiven kind is before me depends upon what I have noticedabout such objects, and this in turn is affected by the way ,I am used to seeing them depicted. Resemblance and de- fceptiveness, far from being constant and independent I

sources and criteria of re;presentationa1 practice are inIsome degree products of it,31 I

81 Neither here nor elsewhere have I argued that there is no constantrelation of resemblance; judgments of similarity in selected and·familiar respects are, even though rough and fallible, as objective and

..Cl1tegorical a. any that are made in describing the world. But judg- }ment. of complex overall resemblance are another matter. In the first!place, they depend upon the aspects or factors in terms of which thelobjects in question are compared; and this depends heavily on .con-}ceptual and perceptual habit. In the second place, even with thes"factors determined, similarities along the seve.al axes are not imimediately commensurate, and the degree of total resemblance will\dep;i>nd upon how the .everal factors a.." weighted. Normally, fO"1e"ample, nean"'ss in geographical location has little to do with ourjudgment of resemblance among buildings but much to do with ourjudgment of resemblance among building lots. The assessment of totalr-esembla~ce is subject to influences galore, and our representationalcustoms are not least among these. In sum. I have sought to show tha~'insofar as Tesemblance is a constant and objective relation, resemblanca between a picture and what it reps:esents does not coincide witrealism; and that ipsofar as resemblance does coincide with realism\the criteria of resemblance vary with change. in repre""ntationa

practice. ,I.::

J .2']", ,:-. ( 39

1,838

something like the 'di~covery' that not the earth but thesun is 'really fixed'. A(:ivantages of a new frame of refer~

ence, partly because ~f their novelty, encourage its en·thronement on some 6ccasions in place of the customaryframe. Nevertheless, w~ether an object is 'really fixed' or apicture is realistic dep~nds at anytime entirely upon whatframe or mode is then{standard. Realism is a matter not ofany constant or absolute relationship between a pictureand its object but of a relationship between the system ofrepresentation employ,ed in the picture and the standardsystem. Most of the tiple, of course, the traditional systemis taken as standard; ~nd the literal or realistic or natural­istic system of represehtation is simply the customary one.

Realistic representa~ion, in brief, depends not upon imi­tation or illusion or information but upon inculcation. Al­most any picture may represent almost anything; that is,given picture and obje;ct there. is usually a system of repre­sentation, a plan of correlation, under which the picture

re;presents the Object.3....••• o How correct the picture is Under{that system depends upon how accurate is the information

about the object that.•... is ,obtained by reading the picture \according to that system. But how literal Or realistic thepicture is depends upon how standard the system is. If!representation is a matter of choice and 'Correctness iii mat- (ter of in(ormation, realism is a matter of habit. 1

30 Indeed, there are usually many such systems. A picture that underone (unfamiliar) system is a correct but highly unrealistic representa­tion of an object may. under another (the standard) system be a \

realistic but very. incorrec..•.•.t repre.sentation of the same object. Only ifaccurate information is yielded under the standard system will thepicture represent the obj~ct both correctly and literally.

Page 21: Nelson Goodman Languages Art Symbols

REALITY REMADE

9. Depiction and qescription

Throughout, I have ,stressed the analogy between pic­torial representation' and verbal description because itseems to me both ective and suggestive. Reference toan object is a n condition for depiction or descrip-tion of it, but. no degree of resemblance is a necessary orsufficient condition fdr either. Both depiction and descrip­tion participate in tl:ie formation and characterization ofthe world; and they; interact with each other and withperception and kno~ledge. They are ways of classifyingby means of labels ~aving singular or multiple or nullreference. The labelS; pictorial or verbal, are themselvesclassified into kinds; ;:jnd the interpretation of fictive labels,and of depiction-as arid description-as, is in terms of suchkinds. Application an~ classification of a label are relativeto a system 32; and there are countless alternative systemsof representation anf! description. Such systems are theproducts of stipulatio'n and habituation in varying propor­

tions. The cho.ice am"png systems is free; but gi~en a SYj'tern, the questIon wh~ther a newly encountered object isdesk or a unicorn-pieture or is represented by a certa'painting is a question of the propriety, under that system!.of projecting thepredicat~"desk" or the predicate "unicornJ

; 1

picture" or the painting over the thing in question!-; 'l

I

32 To anticipate fuller eXplanation in Chapter V, a symbol system!(not necessarily formal): embraces both the symbols and their inter­pretation, and a languag~ is a symbol system of a particular kind. Aformal system is couched in a language and has stated primitives androutes of derivation.

41

DEPICTION AND DESCRIPTION

1,9

and the decision both is guided by apd guides usage forthat system.33

The temptation is to call a system of depiction a lan~ Iguage; but here I stop short. The question what distin~

guishes representational from linguistic systems needs Iclose examination. One might suppose that the criterion of Irealism can be made to serve here, too; that symbols grade \from the most realistic depictions through less and less I

realistic ones, t6 descriptions. This is surely not the case;the measure of realism is habituation, but descriptions do Inot become depictions by habituation. The most com-monplace nouns of English have not become pictures. I

To say that depiction is by pictures while description isby passages is not only to beg a good part of the questionbut also to overlook the fact that denotation by .a picture~_~2!..,~,l~~X~;c2!l~ti1:\1tedepicti~Q~~io-;'~;;;~pl~,ii'pi~:-)tutes in a commandeered museum are used by a briefing iofficer to stand for enemy emplacements, the pictures dol

1not thereby represent these emplacements. To represent, alpicture must function as a pictorial symbol; that is, func-\

\

33 'On the interaction between specific judgment and general policy,see my Fact, Fiction, and Forecast (2nd edition; Indianapolis andNew York, The Bobbs-Merrill Co., Inc., 1965-hereinafter referredto as FFF), pp. 63-64. The propriety of projecting a predicate might ~

be said to depend upon what similarities there are among the objects !in question; but with equ"l truth, similarities among the objects might Ibe said to depend upon what predicates are projected (d. note 31 1above, and FEF, Pl'. 82, 96-99, 119-120). Concerning the relationship

-between the 'language theory' of pictures outlined above and the much,,"discussed 'Pltllite'theory"of'''lilnguage, see "The Way 'the WorJd'fs;; \(cited in note 4 above), pp. 55-56. '

1,940

Page 22: Nelson Goodman Languages Art Symbols

REALITY REMADE

84 The specification that follows has many shortcomings, among themthe absence of provision! for the often three-dimensional nature ofpicture surfaces. But While a rough distinction between pictorial andother properties is useful rem and in sOme later contexts, nothing veryvital rests on its precise fiJrmulation.

with description under denotation. Representation is thusdisengaged from perverted ideas of it as an idiosyncraticphysical process like mirroring, and is recognized as asymbolic relationship that is relative and variable. Fur­thermore, representation is thus contrasted with non­denotative modes of reference. Some of these will be con­sidered in the following chapter. Only much later shall 1come back to the troublesome question of distinguishing

representatio~lsystems from languages.

43

DEPICTION AND DESCRIPTION

,

1,942

tion in a system such ,that what is denoted depends solely Iupon the pictorial prqperties of the symbot The pictorial I

.., '. Iproperties might be rqughly delimited by a loose tecursive)l -)specification.34 An el~mentary pictorial' charact;;'izaiioX;states what color a picture has at a given place on its face.Other pictorial characterizations in effect combine manysuch elementary ones iby conjunction, alternation, quantifi­cation, etc. Thus a pictorial characterization may name thecolors at several place's, or state that the color at one ·placelies within a certain ~ange, or state that the colors at twoplaces are complementary, and so on. Briefly, a PictOrial

tcharacterization says m•......' ore or less completely and more orless specifically what colors the picture has at what places. .And the properties c6rrectly ascribed to a picture by pic­torial characterization are its pictorial properties.

All this,. though, is! much too special. The formula caneasily be broadened! a little but resists generalization.Sculptures with den<}tation dependent upon such sculp­tural properties as shape do represent, but words with denota­tion dependent upon such verbal properties as spelling donot. We have not yet captured the crucial difference be­tween pictorial and iverbal properties, between nonlin­guistic and linguistic s~mbo1s or systems, that makes the dif­ference between repr~sentation in general and description.

What we have don~ so far is to subsume representation