Description and Depiction

Citation preview

Description and Depiction Eddy M. Zemach Mind, New Series, Vol. 84, No. 336. (Oct., 1975), pp. 567-578. Stable URL: http://links.jstor.org/sici?sici=0026-4423%28197510%292%3A84%3A336%3C567%3ADAD%3E2.0.CO%3B2-P Mind is currently published by Oxford University Press.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/oup.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact [email protected].

http://www.jstor.org Tue Jul 31 23:44:16 2007

Description and Depiction EDDY M. ZEMACH

What is it that makes a realistic picture of an object, 0 , so different from a verbal description of O? T h e traditional answer to this question was that the picture resembles the object it depicts, while the verbal description is related to 0 by linguistic conventions only. Most contemporary aestheticians, however, reject this solution. First of all, they argue, resemblance is a symmetrical and transitive relation while depiction is neither; hence resemblance cannot be a necessary condition for depiction. Secondly, every object resembles every other in some respect. One may perhaps say that a picture resembles the object it depicts more than it resembles other objects, but this concept is vague; there is no way to weigh resemblance in one respect against resemblance in another and say which resemblance is more extensive. Moreover, it does not seem true that a picture resembles the object it depicts more than it resembles any other object; Two paintings seem to resemble each other more than they resemble the objects they depict. Thirdly, even a realistic picture uses the representational conventions of its art and period; someone unfamiliar with these conventions will not regard that picture as realistic. Thus, contemporary aestheticians, following E. Grombrich, mostly insist on the conventionality of representation and stress the amount of decoding necessary for 'reading' a realistic picture correctly (i.e., recognizing the object depicted in it). T h e obvious difficulty of these accounts is in showing how does depiction differ from description. I n this article I shall examine the solutions offered to this problem by two articulate representatives of this group, Nelson Goodman and Kendall Walton. I shall try to show that these solutions are unsatisfactory, and what is wrong with the entire conventionalistic approach to depiction. Having done that, I shall attempt to suggest in which direction a better solution can be found.

.~568

EDDY M . ZEMACH:

Nelson Goodman tends to our question twice in his Languages of Art,l once in the concluding section of the first chapter ('Depiction and Description') and again in the first section of the concluding chapter ('Pictures and Paragraphs'). His first answer is this: T o represent, a picture must . . . function in a system such that what is denoted depends solely upon the pictorial properties of the symbol. . . . A pictorial characterization says . . . what colors the picture has at what places. And the properties correctly ascribed to a picture by a pictorial characterization are its pictorial properties (pp. 41-42). T h e difference between a description of 0 and a depiction of 0 consists, then, in that unlike the first, the second denotes 0 by virtue of its pictorial properties only. But this cannot be right. The pictorial properties of a picture determine whether it is, e.g., a bulldog-picture or a man-picture. But Goodman has insisted that a bulldog-picture need not denote a bulldog; it may denote a man (i.e., be a picture of a man). Thus, to consider solely the pictorial properties of the picture D is to consider not what D denotes but what denotes it, i.e., what kind of picture it is. If, however, one takes into account the entire symbolic system in which D functions then one can, presumably, find out by considering its pictorial properties also which object or objects it denotes (if any). But then the pictorial properties of any inscription taken as functioning in a certain symbolic system determine, not only what kind of inscription it is, but also which objects it denotes (if any). That is, just as the pictorial properties of the inscription 'table' in the symbolic system of the English language determine which objects are denoted by it, so do the pictorial properties of a table-picture functioning in a certain symbolic system determine which objects are denoted by it (and these need not be tables). Thus, Goodman's first distinction between depiction and description fails. The second attempt is this. We are told that the symbols in the pictorial scheme are relatively replete . . . one dense scheme is more diagrammatic than a second if the character-constitutive aspects under the first are properly I

N. Goodman, Languages of Art, Bohbs-Merrill, 1968.

D E S C R I P T I O N AND D E P I C T I O N

$9

included among the character-constitutive aspects under the second. . . . Descriptions are distinguished from depictions not through being more arbitrary but through belonging to articulate rather than to dense schemes (pp. 230-231). Let us examine repleteness first. Take two indistinguishable realistic Napoleon-pictures and interpret the first of them, but not the second, in such a way that its weight (in grams) gives the weight of Napoleon (in pounds), the size of the blue areas in it stands for the extent of his historical influence, and the origin of the canvas denotes his birthplace. Both pictures belong to dense schemes, but, according to the above definition, the second picture is more diagrammatic than the first and thus, in this context, only the first is a representational painting. But this is plainly false: as we use the term 'representational', both pictures are equally representational paintings. Moreover, the inscription 'Napoleon' may be, relative to some system, equally dense and more replete than a realistic picture of Napoleon, and thus be more of a depiction than the picture. Goodman will not deny that; but then, in what sense is his theory relevant to our problem of distinguishing description from depiction? Secondly, is it true that syntactic density is what distinguishes a pictorial mark from a descriptive mark? Let us consider the language Anglish, which is just like English, except that every inscription in it is ambiguous between six characters: the typeword associated with that inscription in English and the first five words which follow the inscription of that English type-word in the Roget thesaurus. The inscriptions of Anglish are undifferentiated, and hence dense, throughout. Thus, the Anglish word 'table' is a picture! T o put it briefly, there are many things which are not realistic pictures although they can be used as dense, relatively replete symbols in a system with which we may be familiar. I suppose that Goodman will say that if this is the case then it is just too bad for the ordinary-language category of realistic pictures. But to say this is to give up the entire task of elucidating what makes realistic (i.e., representational) pictures (i.e., those objects that we call 'representational pictures') what they are. One final comment. Essential to Goodman's account of depiction is his concept of representation-as. What we would call 'a realistic picture of a man' is termed by Goodman 'a representation of a

man as a man'. This concept is defined thus: x represents an F as an F iff it represents an F and it is an F-picture.l But this definition (whether or not it is sufficient to capture what we mean by 'a realistic picture') is already more than Goodman can legitimately afford. T h e reader is supposed to take it for granted that the term 'F' occurs in 'x is an F-picture' and thus it becomes reasonable to say that if x also represents an F then x represents an F as an F. But if 'F-picture' contains 'F' as a meaningful part, 'F-picture' is a structured predicate, i.e., there is some function which takes us from Fs to F-pictures. But what can this function which relates, e.g., men to man-pictures be if not representation of, or is a picture of? In this case Goodman's definition of representation-as is a petitio principii, since the concept of an Fpicture would already presuppose the concept it was supposed to explain, that of a picture of an F (an object which represents Fs). Thus it seems that Goodman must regard 'F-picture' as one unbreakable predicate determining a certain kind of object, like 'chair' and 'table'. Under this interpretation, 'man' does not occur in 'man-picture' any more than 'hair' occurs in 'chair' and 'ale' in 'table'. But then it becomes completely mysterious why an object representing a man would be a representation of a man as a man if it is of the kind manpicture rather than of the kind table, or manacle! Goodman's claim is therefore either unreasonable or circular.

II Walton's solution2 is based upon the idea that some false sentences may be true in a certain fictional framework. He uses the symbol '*p" (MB)' to say that the sentence 'p' (e.g., 'Smerdyakov killed his father') is true in some make-believe framework (Dostoyevski's Brothers Karamaxov). What distinguishes 'p' being makebelievedly true of D from 'p' being fictionally true of D is that in the first case 'p' is true of D by virtue of the actual properties of D and not by stipulation only. If, e.g., D is a globe of mud which counts in this game as a pie then 'D is F' (when F is the property, has raisins in it) is make-believedly true only if D has some This is the corrected definition as found in N. Goodman, Problems and Projects, Bobbs-Merrill, 1972, p. 123. For the original definition, see Languages of Art, pp. 21-31.In the new definition, 'x represents y' stands for 'x denotes y and x is a dense and relatively replete symbol'. Kendall L. Walton, 'Pictures and Make-Believe', Philosophical Review, 2 82, 283-319 (1973). I

DESCRIPTION AND DEPICTION

57'

property G (e.g., has pebbles in it) such that it is by virtue of D having G that 'D is F' is make-believedly true. Both pictures and novels allow for some make-believe games. Let 0 be the object depicted in the picture and described in the novel. What is the difference between the depiction and the description? Walton attempts to give the necessary conditions for D's being a depiction. One condition is that members of some society S must have mastery of the rules which make it true, by virtue of D having certain properties, that "0 exists* (MB) and that "0 is F" (MB), and they must have internalized those rules sufficiently so that they can ascertain those make-believe facts on examining D without explicitly inferring them from the relevant properties of D and the rules. This condition (and several others which I shall not reiterate here) applies, however, to linguistic descriptions as well. Most of us can ascertain, by reading Brothers Karamazov, that "Smerdyakov existsk (MB) and that "Smerdyakov killed his fatherk (MB) without consciously going through a process of inferring those facts from the geometrical shapes of the inscriptions in the book and the rules of language (i.e., without using a chart of the alphabet and a dictionary). So far, no difference between description and depiction has been noted. Walton's main idea is, however, that the elusive distinction is to be found in the following feature of some make-believe games:

I suggest that representational pictures are distinguished from novels mainly by their role in a game of make-believe of a certain kind-a game which allows for our performing various make-believe visual actions, for our "seeing, looking at, staring at, noticing, recognizing, visually examining things" (MB). . . . What novel characters lack, if they do not happen to be pictured in illustrations, is the possibility of being objects of make-believe visual actions (p. 303). 'This idea is repeated several times (e.g., on pp. 309 and 318), and is given a precise formulation as conditions I C and ~d for D to be an F-depiction (Goodman's F-picture): There is in S a game of make-believe such that certain actions that members of S might perform count (would count) as their make-believedly performing various perceptual actions vis-a-vis 0 , such as . . . their "seeing 0" (MB), "looking at 0" (MB), "staring at 0" (MB), "gazing

572

E D D Y M . ZEMACH:

at OX (MB), *recognizing that 0 is a P* (MB), *examining 0 ' s features* (MB). These actions . . . are themselves instances of perceiving D in various ways (p. 313). I t seems to me, however, that this condition is empty. I t does not tell us anything about the kind of games that fulfil it, and, a forteriori, it does not distinguish novels from pictures. T h e reason is, that it is not clear what the crucial expressions '*seeing O* (MB)', '*looking at O* (MB)', etc., mean, since we cannot say which real actions count as *seeing OX (MB) or *looking at OX (MB). Let us remember that actions which may be correctly described as cases of "looking at 0 * (MB) need not be even remotely similar to cases of looking at 0 . According to Walton any action may count in some game or other as an action of "looking at 0 * (MB). But if there is no special feature or property by virtue of which some actions and not others are to be considered as actions of "looking at O* (MB), then it is not clear what does it mean to claim that an action of "looking at O* (MB) has taken place with respect to D. The impression that we have learned something by being told that an object, D, is used in a game of make-believe in the course of which actions of *looking at 0" (MB) are performed is wholly due to the fact that we seem to recognize the expression 'looking at 0'. But an action of looking at 0 is very much unlike an action of *looking at 0 * (MB). For one, it is certainly not the case that every action at all can count as an action of looking at 0 . Moreover, we also know what is it for one to make-believe as if one if looking at 0: ordinarily, we would say this if one behaves in a may which is very similar to ones' behaviour when one is looking at 0 , although in this case one is not looking at 0. But none of this is true of *looking at 0* (MB). Walton cannot say that actions of "looking at 0 * (MB) are similar to actions of looking at 0 , because this will take him back to the resemblance theory which he wishes to replace. But if we have no idea what is it for an action to be a case of "looking at OX (MB) we cannot distinguish the game, of which this action was supposed to be the distinguishing feature, from other games. Nor can we distinguish the objects used in this game (i.e., pictures) from other objects. The net import of Walton's definition is therefore only this: pictures are objects which may be involved in games, in the course of which people perform perceptual actions with respect to those objects.

DESCRIPTION AND D E P I C T I O N

573

But I do not know of any object which this description will not fit. One may perhaps think that "looking at 0" (MB) can be characterized internally, i.e., by its logical connections to other make-believe actions. For example, if there is some action A (say, raising one's hand) which counts in some game as *looking at 0" (MB) then another action B which counts in this game as *seeing 0" (MB) stands in some logical relation to A. Unfortunately, however, it is normally impossible to characterize a game by enumerating the game-actions and indicating the logical connections between them. One also needs to specify, at least generally, which real actions may count as embodiments of these game-actions. For example, suppose you are told that chess is a game in which the king must be defended if checked. Will this help you identify chess and distinguish it from other games? The information is completely useless as long as you are not told what real actions may count as checking or defending the king. I t may be argued that one minimal piece of information was indeed given, i.e., that chess is a game in which one may check and defend the king, and this distinguishes it from games in which one may not do so. But even this is not true. In China, they may play chess exactly as we do, but instead of saying 'check!' in the appropriate circumstances, the players say, 'long live the chairman!'. Thus a Chinaman may tell you that in chess you cannot check the king, although you can salute the chairman. Unless you know which moves saluting the chairman and checking the king consist in, you would believe that you have found a contradiction. Yet there is none here : both those who salute the chairman and those who check the king play chess. A characterization of games by game terms only is entirely uninformative; one must also know what real actions they involve. Walton can retort that he has met this requirement, since he has characterized depiction by saying that it involves not only *perceptual actions* (MB) but also perceptual actions. This condition, however, is fulfilled in almost any game and thus irrelevant to our problem, i.e., how to distinguish depiction from description. Reading a novel is also a perceptual action. Let us, for example, invent the following game: the action of reading any sentence about Alyosha Karamazov will count, in this game, as "looking at Alyosha* (MB). The action of reading sentences is a perceptual action, it will count as a "perceptual action" (MB), and

.

E D D Y M. ZEMACH: 574 the game in question fulfils all Walton's other requirements. Hence if we lay this game the novel Brothers Karamasov will become a painting. But this is certainly false. Indeed, the make-believe statements that would be true in this game will be quite different from those which are true in the game of looking at a picture of A1yosha.l E.g., in reading the novel we may Yearn that Alyosha has three brothers, and then that he wishes to be a priest, and then that he has black hair* (MB); this will not be true of the game of looking at a picture of Alyosha. How can one learn, by "looking at Alyosha" (MB) that he has three brothers? This seems strange, and renders my counterexample very suspicious. Walton, however, has nowhere limited his discussion to those games in which "A" (MB) may result in "B" (MB) only if A may result in B. But even if this limitation is now imposed, Brothers Karamasov can still be regarded, according to his definition, as a picture. All we have to do is limit the above game to those sentences (even one will do!) in which a description of Alyosha's external appearance is given. As long as there are some sentences (e.g., 'this is a black square') with respect to which this game can be played, we have sufficient evidence that Walton's definition will not do. T h e phrase 'this is a black square' is not a picture of a black square; it is a description. Any theory which has as a consequence that if I now decide to play a certain game this sentence would become a picture of a black square is plainly wrong.

I would like to suggest that what is distinctive about depiction is neither the syntactic density of the system in which an F is represented, nor the fact that, in the framework of some makebelieve game, it is true to say 'we look at an F' when we look at an F-picture. Rather, it seems to me that what is distinctive about representational pictures is this. We are able to imagine, when we look at a realistic picture, that the observation conditions which prevail are such, that a real object of the kind depicted in that picture would look to us exactly as its depiction does in fact look to us. Of course we know that these conditions do not really prevail. For example, we know that we look at the picture from a distance of a few yards and not from the considerable distance I

This objection was kindly pointed out to me by Professor Walton.

D E S C R I P T I O N AND D E P I C T I O N

575

needed to make right the small size of the object we see, had it really been a man and not a man-picture. Yet we seem to have the skill of unconsciously and almost automatically bringing ourselves to look at the picture as if those other conditions do really exist, and thus see things as if we are under those conditions. T h e conventionalists-Gombrich, Goodman, Wollheim, Walton, et. a1.-are indeed right in criticizing the resemblance theory; a picture does not resemble, tout court, the object it depicts. But these aestheticians are wrong to believe that this is the end of the matter, and the traditional theory of resemblance can be dismissed. They wrongly assume that since a conventional system of notation is involved a depiction cannot look very similar to the object depicted. Thus the conventionalists are right in insisting that one has to 'read' a painting in much the same way that one reads a written text. As Goodman points out, even a trompe l'oeil painting looks like the object it depicts only for a short time and under rather abnormal circumstances. What I would like to argue, however, is that this is no reason to reject the resemblance theory in toto. On the contrary, since it is an undeniable fact that a realistic picture does look to us very much like the object depicted in it, one should ask how these abnormal circumstances are in some sense brought about, or what serves as a substitute for them. I therefore agree with the conventionalists that looking at a realistic painting and recognizing the object depicted in it requires a familiarity with certain conventions and an ability to 'read' the picture according to the key of interpretation assumed. I n this, depiction is very much like description. But while this is the entire stoiy of reading descriptions, it is only the beginning of the process of seeing a depiction as such. My view is, then, that reading a description is a one-step process, while seeing a depiction as such involves two steps. Once the picture is correctly 'read' and identified as, e.g., a tree-depicting picture, the trained observer who knows that under certain circumstances (which do not actually obtain at that time) a tree would look to one exactly like this, can actually imagine, that those observation conditions do prevail. The visual adjustment is almost instantaneous and is normally not noticed; yet I would like to maintain that, invariably, it does occur whenever we see a picture of something as a picture of that thing. I n some cases, however, it is easier to discern the two steps involved. This is the case when one, looking at a picture, fails at first to make any sense of it; then,

.

- ~7~

E D D Y M. Z E M A C H :

when one is told what it is that the picture represents, the details fall into place, and then, all of a sudden, one can see the object right there, in the picture. An excellent demonstration of this suggestion is supplied by many of the much discussed, but slightly understood, cases of seeing-as. T h e puzzling thing about seeing something as an A and then seeing it as a B is that, as Wittgenstein says, 'I see that it has not changed; and yet I see it differently' (Philosophical Investigations, 11, xi). How can an object look quite differently to me, when the aspects change, although I cannot point out a single item in it which has changed? The suggestion I wish to make in this article offers a simple solution to the difficulty in some cases of seeing-as. Take Wittgenstein's own example, the triangle which 'can be seen as a triangular hole, as a solid, as a geometrical drawing, as standing on its base, as hanging from its apex; as a mountain, as a wedge. . .' (ibid.). I t seems to me that what supplies the missing differentiating factor between these cases of seeing-as is the assumed location of the observer. T o see the triangle as a hole in the ground I have to position myself (i.e., to imagine that I am located) at a point high above the centre of the triangular area. I t then looks precisely as a hole in the ground would look to me were I in fact located at this position. T o see it as a mountain I must place myself (make-believedly) at a distant point to the side of the base of the triangle looking at its peak sidewise from below. T o see it as a geometrical drawing I place myself at a dimension which it does not occupy, etc. Thus we can explain why it is that, in these 'changes of aspects', no detail of what is seen is seen as having changed: what have changed is not the object observed but the location of the observer. Another application of this principle can be found in the famous cases of the Necker Cube and the Reversible Staircase. I n order to see a Necker Cube one way, the observer has to 'place his eyes' (i.e., imagine that he sees what he sees while being located) slightly to the left and below the centre of the cube; to see it the other way he has to imagine that he is located slightly to the right and above the centre of the cube. The same trick will reverse the staircase inside out; no head motion is necessary. An interesting case is that of non-realistic depiction, e.g., cubistic paintings. There certainly are some highly complex conditions of observation (involving mirrors, refracted light, etc.) under which a person will look to us as a cubistic portrait of his

DESCRIPTION AND D E P I C T I O N

577

does look to us. But these circu~nstancesare extraordinary and almost never obtain. We have no experience in seeing persons under such conditions, and hence we cannot bring ourselves into imagining that they obtain, i.e., we cannot make the necessary visual adjustments, bringing into play the relevant shape and size constancies. This is why we cannot regard such paintings as realistic. We recognize what is portrayed in them because they retain some of the features of realistic pictures, but we know that they are 'wrong'. 'Nobody looks like that!' For the last seventy years enthusiastic aestheticians (including Goodman: cf. Languages of Art, p. 33) have assured us that once we become used to the new style in painting, Picasso's women will look to us just as natural and life-like as Vermeer's. Alas, the miracle has not happened. We are thoroughly familiar with, and admire, Picasso, but we do not think that his women are realistic or life-like. The conventionalistic view seems to imply that Picasso actually sees things very differently than we do, and eventually we shall come around to see things as he does. I believe that this is ridiculous. Picasso's women do not look like actual women simply because Picasso does not wish to depict women as they look to him. Great art need not be realistic. Yet the conventionalists are not entirely wrong. It is logically possible that we may acquire the skill of imagining that the conditions under which people would look to us as they do in a Picasso painting actually obtain. But what ought to occur for this to happen is not merely that we become familiar with a certain symbolic system, but a radical change in the physical conditions of normal perception. The conventionalists are also right in claiming that the realistic shades into the diagrammatic. But again this is so not because there can be gradual additions to the syntactic repleteness of the symbolic system involved. The reason is, rather, that we can stretch our ability of imagining that certain observation conditions exist, but only so much, and no further; it does not stretch indefinitely. Although for every object 0 and every painting D there are some observation conditions C such that under C, 0 looks to us as D looks to us under normal conditions, we cannot bring ourselves to imagine that C exists unless we have in the past seen things under C and are used to the fact that things like 0 sometimes look that way, i.e., like D. The more uncommon and contrived C is, the less realistic (and more diagrammatic) would D look.

- 578

EDDY M. ZEMACH: D E S C R I P T I O N AND D E P I C T I O N

Note, that on the theory of depiction here advocated all three principles offered by the three theories criticized above are involved in seeing D as a depiction of 0. ( I ) an identification of 0 by 'reading' D according to a conventional system of symbol interpretation, (2) a game based on our ability to act perceptually on assumptions known to us to be false (i.e., on my view, the game of looking at D as if one is positioned at a certain distance and angle from D), and finally (3) seeing a resemblahce between D and 0: when one performs (I) and (2) a man-picture would look to one very much like a real man. HEBREW UNIVERSITY O F JERUSALEM