Jeffrey T. Dean Getting a Good View of Depiction Robert Hopkins Picture, Image, and Experience Cambridge: Cambridge University Press, 1998 ISBN 0521-58259-8 (hbk) 205 pp. '... it seems no accident that there are no spoken pictures, the way that a description can be spoken as well as written; and that pictures seem especially suited to representing the visible world. Thus in some way depiction seems bound to the visual, although this thought certainly needs clarifying' (14-15). [1] With Picture, Image and Experience, Robert Hopkins attempts to clarify the thought that depiction is bound to the visual, and does so in a concise but densely argued book about the nature of pictorial representation and mental visualization. Hopkins's style is clear and free from jargon, and while his arguments are at times somewhat too compressed to be fully persuasive, he handles a variety of difficult topics and divergent philosophical traditions with acumen and sensitivity. Of course, I am somewhat biased, since the kind of view that Hopkins defends -- namely, that both depiction and visualization are best explained, in crucial respects, by their relation to vision -- is one with which I am sympathetic. But even (or especially) those who demur owe Hopkins's book careful consideration, since it provides a sophisticated treatment of the recent history of debate on these matters; much of the book's considerable value lies in its cogent discussion of the influential views of others, such as Nelson Goodman, Richard Wollheim, Kendall Walton, Flint Schier, Christopher Peacocke, Wittgenstein, and Sartre. Some may be disappointed that outside of a few brief discussions of Goodman, Hopkins spends little time on views which favor some kind of conventionalist account of depiction. However, the lack of sustained attention to alternatives does not seem to me particularly remiss in this case, given that Hopkins's main goal is to provide us with reasons to believe that some kind of resemblance view of depiction (whether it be his own or some other permutation) may well have the resources to overcome what have appeared to many to be devastating objections and internal problems. [2] By far the greater part of Hopkins's book is spent discussing depiction, with discussion of visualization left to the final chapter. Due to its brevity, and to a further reason addressed below, Hopkins's treatment of visualization is somewhat less satisfying than his work on depiction. Because of this, and because the discussion of depiction is more directly relevant to the interests of Film-Philosophy, I spend the great majority of this review article on Hopkins's treatment of depiction, with only a few remarks on visualization at the close.
Hopkins begins chapter one by considering a painting (in this case, Picasso's Guernica), noting that: 'there are two aspects to the picture's nature. It is on the one hand a material object, on the other a representation. One way to present our problem is to ask how one thing can fill both these roles. How can a paint-covered surface represent other objects and scenes at all?' (7) Of course, as Hopkins notes, the same question can be asked about written linguistic representation, which also involves marks on a surface that represent other objects and events. But there are important differences (in case there were any doubt). For instance, 'a written description has to be read in a certain order, but the eye is free to roam over a picture without confusing the viewer. The colour of the marks rarely matter to what a description says, but often affects what a picture represents. The relative location of different bits of the picture dictates the spatial relations between the objects they stand for, but the same does not seem true for the words in the description, not at least in any very direct way' (7). By this Hopkins does not mean to suggest that pictures can represent only through depiction; this is clearly not the case. For example, the idea of unity may be represented in a painting by the depiction of a ring; both a ring and unity are represented in the painting, but only the former is represented via depiction. Thus Hopkins is interested in explaining what is normally taken as being unique to pictures, i.e. their ability to depict their objects. The most immediately obvious answer to this question -- namely, that pictures depict their objects by resembling them -- is widely believed to fail due to three central objections (I think there are only two objections here, but Hopkins presents them as three). The first is that while resemblance is a relation between two particulars, not all pictures represent particular objects, persons, or events. So, while it is at least prima facie plausible to say that a portrait of Bob Dole depicts Bob Dole by resembling him, it makes no sense at all to say that the mother weeping over her child in Guernica is depicted by her resemblance to a particular weeping mother, since no particular weeping mother is depicted. And even the prima facie plausibility of resemblance in the case of representations of particulars appears to evaporate in the face of the second objection to the resemblance view, which is that there are many more respects in which pictures differ from their objects than resemble them. A portrait of Bob Dole is flat, made of canvass and paint, encompassed in a frame, incapable of speech, and so on, while Bob Dole is none of these things. Indeed, for any point of resemblance between a depiction and its object, there is almost always a greater number of points of difference. And this is so even in 'realistic' or 'naturalistic' paintings and photographs, to say nothing of paintings like Picasso's Guernica. This brings us to the third objection (really an extension of the second), which is that many pictures which depict their objects do not even remotely resemble them, as is the case with the weeping mother in Guernica, as well as caricatures, stick-figure drawings, and so on. So, even if we could make out a way in which certain depictions can be said to resemble their objects (when such objects are particulars), this would hold true for only a subset of depictive representations, and not for all of them. Such objections have been enough for many, in concert with Nelson Goodman, to conclude that the resemblance view is hopeless. Hopkins, however, sets himself the task of not only rescuing the resemblance view of depiction from these well-known objections, but of giving a detailed explanation of why such a view has the appeal and
power it does. Indeed, part of what Hopkins finds troublesome with Goodman's alternative to the resemblance view -- aside from the fact that Goodman's conditions for depiction are insufficient to distinguish it from other kinds of (non-depictive) representation -- is that Goodman's view does little to explain the nature of depiction, and why we are so powerfully drawn to think of it as being closely related to vision. Hopkins's basic view is that depiction involves a special visual experience, one which, along with Richard Wollheim, he calls 'seeing-in'. In essence, Hopkins takes seeing-in to be a matter of experienced resemblance. Thus, for example, when I see a painting of a tractor, my seeing a tractor in the painting is a matter of experiencing (some aspect of) the painting as resembling a tractor. What this requires is not that a picture resemble its object, but that it is experienced as resembling its object (to say that a picture 'looks like' its object is ambiguous between these options; Hopkins's experienced resemblance view allows us to retain such terminology without falling into the difficulties generated by claims about actual resemblance). Notice that because seeing-in involves only the experience of resemblance between a depiction and its object, it is not subject to the first objection to the (actual) resemblance view. Just as the sound of rain can be heard to resemble the sound of applause (but not necessarily any particular applause), so a picture can be seen to resemble a woman (even if not any particular woman). Experienced resemblance, unlike actual resemblance, is not necessarily a two-place relation between particulars. Moreover, my experiencing one object as resembling another does not depend on their having a certain percentage of shared properties. Rather, it requires only that they share certain features salient to producing the experience of resemblance (see below for a discussion of what these features are). [3] On Hopkins's characterization, seeing-in includes four essential elements: i) a distinctive phenomenology; ii) an experience whose content includes the picture's object; iii) a way of seeing the picture; and iv) the integration of the thought of an absent object and seeing marks on a surface (15). According to Hopkins, these are among the most central general elements of depiction any account must preserve and explain. Hopkins rejects three influential positions which do preserve these elements. These include: 'illusionism' (which involves the claim that regardless of what belief states are or are not induced by our experience of depictive representations, the experience of seeing a depiction of X is phenomenologically indistinguishable from the experience of seeing X); Wollheim's own account of seeing-in; and Kendall Walton's account of pictorial representation. Hopkins argues that illusionism is simply false, and maintains that while the accounts offered by Wollheim and Walton are broadly promising, each suffers from deficiencies in their formulation and ability to explain how depiction is related to, but different from, both vision and visualization. Hopkins assessment of these views is brief, but lucid. And while I expect that Walton would be able to shore up the weakness revealed by Hopkins analysis, his doing so would probably leave him with a view much like the one defended by Hopkins. Indeed, the view Hopkins defends is very much like Walton's, though couched in somewhat different language, and certainly more fully developed (no surprise, perhaps, given that Walton's account of depiction is embedded in a more general account of representation).
Part of what leads Hopkins to reject the accounts offered by Goodman, Wollheim and Walton is his further specification of what any adequate account of depiction must explain. After motivating the inclusion of each item, Hopkins lists the explananda as follows: (X1) There is a significant minimum pictorial content. (X2) Everything is depicted from some point of view. (X3) Whatever can be depicted can be seen. (X4) Pictorial misrepresentation is possible, but has its limits. (X5) General competence with depiction and knowledge of the appearance of O (be it a particular a or merely a, but no particular, F-thing) suffice for the ability to interpret depiction of O. (X6) General competence with depiction and knowledge of the appearance of O are necessary for the ability to interpret depiction of O. (27-31) Hopkins argues that none of the above-mentioned accounts is successful in explaining each and every member of X1-X6. Thus any adequate experiential view of depiction will have to improve on these accounts. It is worth noting that Hopkins also rejects Flint Schier's influential view, which holds that what is essential to pictorial representation is that it engages the same capacities to recognize the depictions of objects as are required to recognize the objects themselves. The thrust of the objection is that Schier's view cannot adequately account for X1, and that this failure gives his view trouble with X3 and X4 as well. Because Hopkins does a nice job of motivating the importance of each of the six explananda, and of articulating their interconnectedness, I believe the failure of other views to adequately explain them does tell against these views. At the same time, it is open to others to disagree, and to argue, perhaps, that the inability to deal with the explananda is not due to problems with the other views, but with the nature or formulation of the explananda themselves. A defender of Schier's view, for example, might argue that the view can in fact explain X1; but she might also argue that even if Schier's view cannot explain X1, this is not due to a flaw in the view, but in X1 itself. Of course, it will have occurred to some by now that a number of important questions have yet to be answered. First, in virtue of what are we said to experience resemblance between a depiction and its object? And second, is there anything that determines whether a given experience of resemblance is warranted? The former is a question about what it is about a depiction P of an object O that encourages us to see O in P (where being able to see O in P is necessary for P to depict O). The latter is a question about what else is required for P to depict O, since clearly seeing O in P is not sufficient for depiction (e.g., my seeing a bunny rabbit in the cloud does not entail that the cloud depicts a bunny rabbit, since clouds do not depict anything); it is a question about a standard of correctness. The answer to the first question is 'outline shape'. Formally, as Hopkins notes, we can 'define an object's outline shape at a point as the solid angle it subtends at that point. Two items will resemble in outline shape to the extent that, at some point, on subtends a solid angle similar to that subtended, at some point, by the other' (55). We can think very roughly of this outline shape as a silhouette, but one which admits of differentiation within its boundaries (internal silhouettes, if you will). For example, imagine seeing George Washington through a
window at some distance. There is a particular outline shape that he has from your perspective. Now imagine tracing his figure and features onto the surface of the window. What you wind up with, presuming you have done your tracing accurately, is George Washington's outline shape, from your point of observation, on the window as well (if he is facing your right, perhaps the tracing will look roughly like the picture of George Washington on a dollar bill). Thus when we experience a picture as looking like its object, what we are experiencing is similarity in outline shape (that is, at the very least; we may also be experiencing similarities in color, shading, and so on, but these are not necessary similarities for depiction -- similarity in outline shape is). The answer to the second question, the question about a standard of correctness, is disjunctive. With respect to pictures, the answer is 'intention'. What determines whether your seeing the Spice Girls in a picture is warranted (assuming that you are able to see the Spice Girls in the picture, i.e. that there are aspects of the picture whose outline shape is similar to the outline shape of the Spice Girls) is a matter of whether whoever produced the picture intended for the Spice Girls to be seen therein. With respect to photographs, the standard is somewhat different, owing to the fact that cameras are often set up with the intention that they take a picture of something, without there being anything in particular they are intended to take the picture of. The basic claim is that a photograph depicts an object just in case the photograph is produced by a system intended to produce pictures causally related to the object so that the objects can be seen in the picture. Hopkins formal formulation of depiction is thus as follows: An item P depicts *a*/something F iff [if and only if] (1) *a*/something F can be seen in P and either (2i) (1) because someone intended that a/something F be seen there or (2c) P is the product of a system successfully intended to produce surfaces causally related to objects in such a way that those objects can be seen in those surfaces, and (1) because P is so related to *a*/something F. (77) And, of course, part of what separates Hopkins's view from the views of others is his characterization of what seeing-in amounts to: 'Something O is seen in a surface P iff P is experienced as resembling O in outline shape.' (77) The remainder of the book (save the final chapter on visualization) tackles the six explananda head-on, taking note along the way of how other experienced resemblance views (especially Christopher Peacocke's) fare in this regard. The discussion is clear and concise, though occasionally leaving the sense that too much time is spent on minnows when there are bigger fish to fry. The central topics here are the problem of pictorial misrepresentation (i.e., the problem of how, given the experienced
resemblance view, pictures can depict objects which they misrepresent) and the problem of indeterminacy and interpretation (i.e., of how it is, given the experienced resemblance view, a picture can depict an object when the marks on the surface are to some greater or lesser extent indeterminate with respect to that object). Although questions remain (as Hopkins himself acknowledges), the experienced resemblance view is shown to have surprisingly deep resources, both in terms of its explanatory power, and its ability to deflect or absorb objections without resorting to ad hoc apologia. I want to now close with a few comments on the last chapter of the book, which addresses the issue of visualization. Those with an interest in cognitive science and the philosophy of mind will be familiar with what is often called the 'imagery debate', a debate about whether there are, in addition to propositional mental representations, depictive mental representations as well. What is in question is not whether we experience mental imagery; it is widely agreed that we do. What is in question is whether such imagery is adequately explained by propositional mental representation alone, or whether it also relies on depictive mental representations -- where the former can be thought of as a list of (mental) sentences, and the latter an array of points in a space. [4] The questions raised and addressed in this debate, which includes participants not only from the philosophy of mind, but from psychology and cognitive science as well, are numerous and difficult, and there is no need to elaborate on them here. What is important for our purposes is that while Hopkins acknowledges that the work done in his book may have some relevance to these debates, when it comes to the central point of overlap between issues in theory of mind, cognitive science, and pictorial representation -- namely, mental visualization -- Hopkins remains mysteriously silent (save a single footnote on page 195 to the effect that he is not going to address the relation of his inquiry to these other disciplines). What is especially peculiar is that Hopkins believes that at least part of what is essential to visualizing has to be modeled on seeing, and he gives a number of nicely nuanced reasons for why this should be so (he also argues that there are, in addition, important respects in which our understanding of visualization must rest on our understanding of depiction). The discussion here is again subtle and perceptive, including compelling discussion of Sartre, Wittgenstein, and Christopher Peacocke. At the same time, he makes no mention at all of some of the recent work done on precisely this issue, for example that of Stephen Kosslyn and his associates in cognitive psychology. [5] Kosslyn uses compelling neurological evidence to make the case that visualization uses the same physiological resources as vision, that mental images are generated by the triggering of elements of our visual system. Hopkins says that he is only interested in philosophical explanations (by which he seems to mean primarily conceptual and phenomenological analysis), but does concede that 'understanding the psychological and neurological mechanisms involved in visualizing may indeed offer certain kinds of explanation for its properties' (195). Given that Hopkins also concedes that there is a rather severe limit to the extent to which philosophical inquiry (in his sense) can illuminate the nature of visualization (unlike the case of depiction), it is surprising that he does not say something more about the promising work on the subject in cognitive science. For if Kosslyn and others who share his view are right, then the connection between visualization and
vision goes significantly deeper than their phenomenology, and is rooted in their emergence from an overlapping set of neural functions and cognitive hardware (i.e. parts of the brain). Indeed, the phenomenological similarities between seeing and visualizing are explained by reference to their shared etiology. And this is surely a crucial part of any explanation of how vision and visualization are related. While I believe that failure to discuss any of the recent cognitive science literature on visualization is an important omission (and that, indeed, if I had any complaint about Hopkins's book it would have to be with respect to what is not included in it), I do not wish to imply anything other than great esteem and appreciation for the extent to which Hopkins has succeeded in defending and vitalizing an experienced resemblance view of depiction. This is no mean feat, and Hopkins is to be commended for it. University of Wisconsin-Madison, USA Footnotes 1. In this regard, Hopkins's view appears to lend support to the main contention of James Elkins's book, On Pictures and The Words That Fail Them (Cambridge University Press, 1998), recently reviewed here by Kathleen Johnson: ''To See or Not To See', That Is the Question', Film-Philosophy: Electronic Salon, vol. 3 no. 1, January 1999 <http://www.film-philosophy.com/vol3-1999/n1johnson>. 2. As will become clear in the discussion below, Hopkins in fact defends an experienced resemblance view, rather than a resemblance view simpliciter. 3. Of course, one can experience any object as resembling any other. The respect in which such an experience is warranted, relevant to depiction, is discussed below in relation to a standard of correctness for depiction. 4. As Stephen Kosslyn notes: The space in which the points appear need not be physical, such as this page, but can be like an array in a computer, which specifies spatial relation purely functionally... In a depictive representation, each part of an object is represented by a pattern of points, and the spatial relations among these patterns in the functional space
correspond to the spatial relations among the parts themselves. Depictive representations convey meaning via their resemblance to an object, with parts of the representation corresponding to parts of the object. In this case, a 'part' can be defined arbitrarily, cutting up the representation in any way; no matter how you cut it, the part will still correspond to a part of the object. Of course, when Kosslyn speaks of depictive representations here he is including both physical pictures and mental representations. See Stephen M. Kosslyn, Image and Brain: The Resolution of the Imagery Debate (Cambridge, Mass.: MIT Press, 1994), p. 5. 5. Stephen M. Kosslyn, Image and Brain, op. cit.