Film-Philosophy

Journal | Salon | Portal (ISSN 1466-4615)

Vol. 5 No. 25, August 2001

 

 

Richard Allen

Looking at Motion Pictures (Revised) [1]

 

 

 

I

What is it we see when we look at a motion picture? This is a fundamental question addressed by all film theory. It derives from a much older question: what is it we see when we look at a picture? But the answer to this question depends in turn upon how we understand the activity of seeing itself. The philosophical understanding of what seeing is has been dominated by the causal theory of perception. The concept of a causal connection is central to understanding the natural world and it serves to characterize the physical connection between our sensory organs and what it is that we perceive. We can only see objects of a certain colour, shape, or size because these objects impinge causally on our senses. However, the causal theory of perception makes two further claims: first, that by acting causally upon our senses, objects cause us to have a visual experience; second, that the asserted causal connection does not simply describe our knowledge of the physical world but it is part of our 'ordinary notion of perceiving'. [2] The first claim assumes a sharp divide between 'the mental' and 'the physical' worlds in order to assert the existence of a causal connection between the two. The second claim, wishing to deny this sharp divide, asserts that the idea of a causal connection is built into the concepts we use to describe our interaction with the world, in this case of the concept of perception. It is the burden of Wittgenstein's later philosophy to contest both these claims. First, he argues there are not two worlds, 'mental' and 'physical', that are causally connected to each other, but one physical world that consists of causal connections. Secondly, he argues that a philosophical understanding of the mind consists in attaining a perspicuous overview of the concepts we use to characterize it and the way these concepts are grounded in human behaviour, not in the investigation of the causal preconditions for the application of those concepts. In other words, there is a sharp distinction between conceptual and empirical inquiry: grammar, in particular the grammar of our mental concepts like 'perception', is autonomous. The causal theory of perception provides a signal instance of the failure to observe this distinction.

 

I shall begin this essay by sketching some Wittgenstein-influenced arguments as to why the causal theory of perception is inadequate. However my main concern is to explore the ramifications for pictorial perception of understanding perception in terms of the causal theory. When 'our ordinary notion of perceiving' is characterized in terms of the existence of a causal connection between an object perceived and our sensory experience of that object the case of pictorial perception generates a paradox. For when we look at a representational painting, a photograph, or a film, arrayed before us is the two-dimensional disposition of pigment upon a canvas or light registered upon photographic paper or projected on a film screen organized into the shapes of recognizable objects. According to the causal theory, my 'visual experience' of a picture is caused by my two dimensional object array. Yet when we look at a painting, photograph, or film, and report upon our 'visual experience' we commonly report that we see not simply the disposition of pigment or array of light but what is depicted in the representation. This is especially true when we report on what we see in photographs, on television, and in the cinema. It seems that we see William Shatner in _Star Trek_, not simply see a representation of him, and consequently we see that he has grown older in _Star Trek: The Motion Picture_. However, we may also report seeing William Shatner when he is depicted in a painting, though we may feel that our experience of seeing is different in the case of paintings as opposed to photographs. To the extent that our ordinary concept of seeing permits us to speak of seeing what is depicted in a motion picture, our ordinary understanding of perception in the cinema conflicts with the causal theory of perception, for how can we report that we see William Shatner when our 'visual experience' is of a representation of William Shatner?

 

Visual theorists have sought to resolve the paradox of pictorial perception in a way that conforms to the causal theory of perception. As they have been applied to the problem of perceiving motion pictures, these theories are of at least four kinds: illusion theories, transparency theories, imagination theories, and recognition theories. According to illusion theories of pictorial perception we see what a picture depicts because a picture causes us to have a visual experience that is like the visual experience that we have when we see the object. We may be deceived by the illusion, in which case it is a cognitive or epistemic illusion, or the illusion may be merely sensory or perceptual. Theories of illusion were highly influential in the film theories of the 197Os and early 1980s that sought to explain the special power of movies to shape the imagination. Transparency theories are associated with the realist tradition of film theory: Bazin, Kracauer, and Cavell. Transparency theorists claim that the unique properties of the photographic image allow us, in some sense, to actually see the object when we look at a motion picture. The photographed object causes the image to be produced in the photograph which causes us to have a visual experience of the object photographed. The thing photographed is not absent but indirectly made present to us via its photographic reproduction. Imagination and recognition theories of pictorial perception are of more recent vintage and are associated with the turn toward 'cognitive' approaches to understanding motion picture perception within film theory. Imagination theorists of pictorial perception deny the fact that we see what is depicted in a motion picture, our 'visual experience' is simply one of a two-dimensional representation. If we are still inclined to speak of seeing what is depicted in a picture it is because although we do not actually see what a picture depicts we imagine seeing what a picture depicts. Recognition theorists, like imagination theorists, resolve the paradox of pictorial perception by denying that, strictly speaking, we see what is depicted in a picture. For recognition theorists, pictorial perception mobilizes a capacity to recognize objects already possessed by the spectator. Recognition theorists may also deny that we imagine that we see something and simply claim that what we see is the disposition of colours on a flat surface that cues our recognition of what the picture is of.

 

Each of these theories attempts to resolve the paradox of looking at pictures by rejecting one or other horn of the dilemma generated by conceiving pictorial perception in terms of the causal theory of perception. Either we see what a picture depicts because what we see is really the thing itself or an illusion of it, or we do not see the thing itself and instead we either imagine that we see what the picture is of or it affords us a recognition of what it depicts. I shall argue that none of these solutions are satisfactory: we require an understanding of seeing (motion) pictures that, contrary to imagination and recognition theorists, respects that fact that seeing what a (motion) picture is of is a genuine case of seeing, without committing ourselves to the idea that what we see is either the thing itself, as opposed to a representation of it, or else an illusion of the thing itself. Such an understanding is predicated upon a rejection of the causal theory of perception and hence of the paradox it gives rise to. When I see a representation of William Shatner in a painting or film I see William Shatner. To deny the role of the causal theory of perception in the explanation of pictorial perception is not, of course, to deny the existence of a physical, causal connection between the painting or film and the sensory organs of the viewer. It is simply to claim that this connection cannot explain or justify the content of our perceptual report. But how do I justify my own claim that (motion) picture perception is a genuine case of seeing? I do not offer an alternative theory here about what it is we see when we look at motion pictures. Instead, my strategy will be to diagnose what it is that impels the theorist of vision to insist there is something strange about our talk of seeing something that is depicted in a motion picture. Dispel the strangeness, or recognize how widespread this strangeness is, and the pressure is allayed to revise our customary ways of speaking on account of a theory.

 

Following Wittgenstein, I shall argue that the content of our perceptual reports is specified not by a theory but simply by what we report that we see when we make veridical reports using perceptual verbs. In the _Philosophical Investigations_ Wittgenstein discusses cases of visual perception that involve the seeing of an aspect, such as looking at Jastrow's duck-rabbit figure or seeing a family resemblance. [3] In aspect seeing any characterization of the physical properties of what is seen fails to unambiguously determine the nature of what it is that we see, for while the physical properties of what we see do not change we seem to see something different. For example, when previously we have seen the duck-rabbit figure as a rabbit, now we see it as a duck, but we cannot explain the change in what it is that we see by any physical change in the thing perceived. It is tempting to understand the switch from the duck to the rabbit as an interpretation of the figure, for we can choose to look at it one way or another once we recognize that the figure contains two aspects. However, when the aspect first dawns upon us we are not in a position to 'choose' what we see, our perception of the duck-rabbit as a duck is compelled upon us by the figure. And it is the way the figure compels us to see it under any given aspect (unless we switch) that renders the perception of an aspect like ordinary seeing. I shall argue that pictorial perception in general is a case of 'continuous aspect seeing': (motion) pictures allow us to continuously see what they depict where the difference in what we see between merely seeing a two dimensional array and seeing what is depicted by that array is not specifiable in terms of the physical properties of what we see. Once again, this is not to deny the existence of a causal relationship between an object with certain physical properties and the sensory organ of the observer, it is to deny that a specification of the physical properties of the object can serve to characterize what we really see, and hence support a denial that we really see what is depicted in picture.

 

My concern is to address the paradox of looking at pictures. However, contained within the problem of whether or not we can see what a picture depicts is a problem posed by a certain class of pictures: pictures of fictions. Fictional objects do not exist. How can something that is non-existent be depicted, and if it cannot be depicted how can we see it in a picture? The idea that we can perceive fictional objects is surely nonsensical: a 'metaphysical impossibility'. [4] My approach to the problem of perceiving fictions is simply this. Assuming that fictions can be depicted, then the arguments that I make about looking at pictures in general apply to pictures of fictions as well. That is, just as we might report that we saw Larry Hagman acting the part of J.R. being shot, we might also report, if we were avid _Dallas_ watchers, 'J.R. has been shot . . . I just saw it.' Similarly, assuming that what is fictional can be depicted, then we can see what is depicted in a painting, whether the painting is of a horse or a unicorn. However, I shall not argue the case here.

 

II

The causal theory of perception is motivated by two apparently common-sense arguments about visual perception. The first argument contrasts ordinary seeing with hallucinations and the perception of illusions. The dagger that Macbeth thinks he sees is an hallucination. But suppose that while Macbeth is hallucinating the dagger a real dagger was placed before him that corresponded to the hallucinated dagger in every conceivable respect. We would still not claim that Macbeth actually saw the dagger, and it seems natural to explain this by saying that the reason Macbeth did not see the dagger is that the dagger did not cause his visual experience. Or suppose, to consider another case, I was looking at a pillar and unknown to me a mirror was interposed in front of the pillar that reflected a numerically different pillar. It is certainly tempting to explain the fact that I see the second pillar (in the mirror) and not the first by the fact that the first pillar was not causally relevant to my perception. [5] Finally, suppose that it seems to me that I am perceiving a clock on the shelf but it turns out that my visual cortex is being stimulated in such a way so as to engender the impression that I am seeing a clock, then it seems natural to claim that I do not actually see the clock on the shelf because the clock on the shelf did not cause my experience of seeing the clock. It is tempting to think in these cases that perceiving something and hallucinating something or seeing an illusion have something in common, namely, a perceptual or visual experience. The difference between the case of genuine perception and hallucinations or illusions seems to lie in the fact that in the case of genuine perception my visual experience is caused by the presence of the object in my visual field.

 

As John Hyman points out in his carefully argued critique of the causal theory, these arguments fail to establish what they seem to. [6] These stories appear to illustrate different ways in which visual impressions may be caused: they may be caused by hallucinations or illusions or they may be caused by an uninterrupted perception of something. However, what they conclusively demonstrate are only the different ways in which someone can be causally prevented from seeing something. They do not by themselves show that the concept of perception contains the idea of causal connection between the environment and our experience of it. They merely illustrate the platitude that if I see something I am free from any causal constraints that would prevent me from seeing it. Nonetheless, these arguments do prepare the ground for us to think of our concept of perception as one that contains a causal relationship, for they allow us a way to think of perception as a distinctive type of subjective perceptual experience that is connected to something outside subjective experience, that is, to the world that surrounds us. If our 'visual experience' in the case of the hallucination is one that is caused, and the 'visual experience' we have when we hallucinate an object is identical to the 'visual experience' we have when we actually see something, then it is natural to conclude that the 'visual experience' we have when we see something is one that is caused by the presence of an object in our visual field.

 

This intuition gains support from a second kind of argument based on the idea that perception is a cognitive faculty, a way of finding out about the world around us that is responsive to that world. How can we think of perception as being dependent, in this way, upon things in the environment unless we think of perception as causally explained by things in the environment?

 

We think of perception as a way, indeed the basic way, of informing ourselves about the world of independently existing things: we assume, that is to say, the general reliability of our perceptual experiences; and that assumption is the same as the assumption of a general causal dependence of our perceptual experiences on the independently existing things we take them to be of. [7]

 

In the case of hallucination, we do not find anything out about our environment and this is why hallucination is not a genuine case of seeing. But to say that when we genuinely see something we are finding out something about the environment is to say, it seems, that something in the environment caused us to see it.

 

The general claim that perception is a way of informing ourselves about the world, and therefore that our concept of perception contains in it the idea of a causal connection, is credible only if it makes sense to conceive of perception in terms of a relation that obtains between a 'subjective' psychological event or perceptual experience and the object perceived. Once this psychological event is detached from the overall activity of seeing, then a conceptual space has been created between what is seen and the experiencing of seeing that allows for the specification of a causal connection. For a causal connection obtains only between two distinct existences, cause and effect, where it is logically possible for one to exist without the other. In this case, the causal connection lies between the object perceived and the putative perceptual experience. Only then can we speak of 'a general causal dependence of our perceptual experiences on the independently existing things we take them to be of'. But as I have suggested the comparison between seeing something and hallucination seems to establish just this point. The theory of what it is to see an illusion or have an hallucination of an object -- that it involves the same visual experience as the one involved in seeing the object -- seems to show how it is that our visual experience of seeing can be detached from the specification of what it is I actually see. The 'visual experience' is the same in both cases but only in one case -- the case in which the visual experience is causally attached to the object seen -- is a case of genuine seeing.

 

However, is this claim coherent? Does it makes sense to claim that the concept of 'visual experience' can in this way be detached from and then re-attached to the concept of seeing something? 'Is it not implausible', Hyman writes, 'that the experience of hearing something and the experience of seeing something are experiences once can have despite failing to hear or see anything at all -- even in total darkness and perfect silence?' [8] One cannot have the experience of a university education without having a university education, or of playing the piano without playing the piano, or have the experience of a toothache without having a toothache. Why is the case of visual perception different? Clearly from the third person perspective seeing something and hallucinating it do not describe the same experience. 'They are so different indeed', P. M. S. Hacker notes, 'that they do not even look alike, for no observer would mistake Macbeth's having an hallucination of a dagger for Macbeth's seeing a dagger.' [9] Furthermore, even the person who hallucinates or sees the illusion of a dagger will distinguish their experience from the case where they actually see a dagger or where someone else actually sees a dagger. They will report that they seem to see a dagger or that what they see looks like a dagger, rather than reporting that they actually see a dagger. So what is the visual experience in common between hallucination and perception that allows us to think of ordinary perception as containing a visual experience as its subjective part?

 

Suppose we consider the hallucination from standpoint of someone who, unlike Macbeth, doesn't realize that they are hallucinating a dagger. Surely it seems to this person that they are seeing something in exactly the same way as they would if they were actually seeing the dagger. Indeed from the subject's point of view there is no difference. But what exactly is it that is shared in these two cases? The person who believes he sees a dagger believes that he is having the same experience as a person who actually sees a dagger. In this sense we can speak of what someone believes they see as being, from the subject's point of view, indistinguishable from actually seeing something. However, for the person who experiences an illusion to *believe* that he is having the same experience does not mean that he has the same experience. For although the reports we make about what we see in general provide a sound basis for ascribing a certain kind of visual perception to us, these reports, like any other first-person avowal, may be defeated by (anomalous) circumstances. The defeating circum stances are, in this case, precisely that the person is not seeing a dagger but hallucinating it, and thus although he reports that he sees a dagger, he actually only seems to see it. While there is a clear distinction between hallucinating x and seeing x, the subject who is hallucinating is not in a position to draw it. Thus the absence of his ability to distinguish between hallucination and perception does not entail that the experiences he has are the same.

 

There is not something -- a visual experience -- in common between genuine cases of perception and cases where we merely seem to see something, therefore the comparison between perception and hallucination provides no basis for conceiving of perception as containing within it such a visual experience as a component part, one that is detachable as it were, from the concept of perceiving something. Since, as I have argued, the causal theory of perception depends upon the intelligibility of the idea that our visual experience of something can be detached from the exercise of our perceptual capacities in order to establish the idea of a causal relation between the visual experience and what is perceived, the failure to establish an experience in common between the genuine case of perception and hallucination renders the causal theory of perception itself unintelligible. The causal theory of perception requires that we can separate out a mental component of perception from the physical component to make room for an explanation of the causal connection between them. But there is no room to be had. Mental and physical are indissolubly linked in our perceptual reports.

 

III

The illusion thesis of pictorial representation is one that is associated with contemporary film theory, though its roots are ancient. The theory that pictures are cognitive or epistemic illusions that cause us to believe mistakenly that what we see is real has been thoroughly debunked and can be laid to rest. [10] But elsewhere I have tried to defend a weaker version of the illusion thesis: cinema and other forms of pictorial representations can function as a form of perceptual or sensory illusion that I call 'projective illusion'. I define projective illusion as a form of illusion that is akin to our experience of an illusion like the Muller-Lyer illusion, where we know that what we see is an illusion yet our senses are still deceived. However, it is also a weaker form of illusion than this kind of sensory deception, since in projective illusion, as I define it, I can bring to bear my knowledge of the fact that the representation is a depiction to prevent projective illusion taking hold, or to break the hold of the illusion entirely. [11]

 

This idea of projective illusion is both empirically false and conceptually confused. What is distinctive about an illusion is that it is a special form of representation that is configured in such a way as to confound our senses: our senses become unreliable guides to what lies before us. For example, in the Muller-Lyer illusion we seem to see two lines of unequal length, even though we know that the two lines are of the same length. However, representations do not, in general, drive such a wedge between perception and belief in this way; we do not customarily take a representation for something other than it is. I try to build into the theory of projective illusion a recognition of the fact that my knowledge that the representation is a depiction is sufficient to break the hold of the putative illusion. However, the kind of mistaken perception engendered by an illusion is not one that can be corrected by my knowledge: that is what makes it an illusion. I argue that the correction occurs in the manner that I can change my perception when I perceive an ambiguous figure like Jastrow's 'Is it a duck? Is it a rabbit'. However, the idea that the correction of our perception of projective illusion occurs in this way still presupposes that I see the image as an illusion in the first place. As Noel Carroll points out, if the illusion theory is wrong, then so is the explanation of how the spectator's experience of the illusion is countered. [12]

 

The confusion that characterizes my own theory is, I think, typical of illusion theories. However, it is important to recognize the feature of looking at motion pictures that illusion theories, however confusedly, are trying to explain. Illusion theories take seriously the fact that it is meaningful to speak of looking at what a motion pictures depicts. However, illusion theorists are mistakenly drawn to the counter-intuitive conclusion that pictures therefore must be understood as illusions, for they rely upon the assumption that what we specify in our perceptual reports must be the content of a visual experience that is caused by what it is that we see. Since we make our perceptual reports in the presence of a picture and not what is depicted in it, we must be under the spell of an illusion that causes us to have a visual experience of the object without the object being present.

 

I have argued, following Hyman, that the causal theory of perception mistakenly assumes (a) that the content of a putative 'visual experience' can be detached from the specification of what it is that I see, and (b) that the seeing of an illusion shares this content -- a visual experience -- with actual seeing. It is these mistaken ideas that motivate the analogy between pictures and illusions: in seeing a picture of an object I am said to have the same visual experience as when I see the object itself. However, as we have seen, the assumptions underlying the causal theory of perception are mistaken. The factor that lies in common between the seeing of an illusion of x and seeing x is simply the belief that one is seeing the same thing, and, in the cases where the illusion is merely sensory, not even this belief is shared. Clearly, in the case of seeing what is depicted in a picture I do not erroneously believe that I am seeing the object itself, but neither do I seem to see something in a way that confounds my knowledge that what I see is only an illusion. Once we abandon the causal theory of perception together with the explanatory role it bestows upon illusion for understanding the role of 'visual experience' in perception, the argument that seeing things depicted in a picture is like seeing an illusion and hence like seeing things themselves loses its appeal.

 

IV

Kendall Walton has provided a rigorous, defensible version of the central claim of the realist tradition in film theory, that what we see in a photograph or motion picture is not simply a representation of the object but the object itself. Bazin claimed that: 'No matter how fuzzy, distorted or discolored, no matter how lacking in documentary value the image may be, it shares, by the process of its becoming, the being of the model of which it is the reproduction; it is the model'. [13] This assertion is based upon a conflation of photographic representation with what the photograph is of. Walton revises Bazin's account in the following way. He argues that when we look at a photograph or a motion picture we may speak with justification of seeing an object photographed just in those cases where the object depicted in the photograph is something that existed in front of the camera when the photograph was taken (something that we can assume in the case of the standard photograph of a non-fictional object). However, we can see the object not because the representation is in some sense the object photographed, as Bazin claimed, but because the photograph allows us to see the object indirectly. The object photographed is, in this case, seen *through* the photograph in a manner that is akin to the way that we look through eyeglasses or a telescope at an object. [14] We thus really are looking at William Shatner when we see _Star Trek: The Motion Picture_, but he is not directly present to us, nor do we believe that he is present to us in the manner of an illusion.

 

Walton argues that the main reason we can see what is depicted in a photograph is because the photograph preserves a 'counterfactual dependence' between the photographic image and what the image depicts that is independent of the beliefs held by the photographer about what he or she is photographing. In other words, irrespective of what the photographer believes he is taking a picture of, the photograph will register what is in front of the camera. The 'counterfactual dependence' of the photograph on the thing photographed means that were the thing photographed to change (counter to fact), the picture would change in a corresponding fashion, regardless of the photographer's beliefs or intentions. The photograph contrasts in this respect to the painting. Regardless of what actually stands before the painter when he paints, what he paints and we see is what he intends to paint and for us to see. However, in the standard case, even if the photographer intends to photograph something else, what he photographs and hence what we see stood before the camera. Gregory Currie expresses the contrast this way: the counterfactual dependence in photography is 'natural' rather than 'intentional'. [15] Natural counterfactual dependence is not sufficient to establish transparency because we can imagine, for example, a light sensitive machine that printed out descriptions of what it records that would obviously fail to produce representational transparency. Thus Walton argues that visual transparency is only guaranteed by the fact that photographs, unlike verbal descriptions, preserve real similarity relationships between objects: the word 'house' is more likely to be confused with 'hearse' than with 'barn', but a photograph of a house can be confused with a barn just as a house can be mistaken for a barn. The range of discriminatory error is alike in the case of seeing photographs and seeing things.

 

Walton's argument about photography and cinema has the advantage that it can make sense of the fact that we do speak of seeing the object depicted without resorting to an illusion theory. However, it does so upon the basis of a theory that draws a sharp distinction between looking at photographic depictions of existing objects or persons, like William Shatner, and looking at photographs of fictional objects or persons, like Captain Kirk, or looking at a painting of either William Shatner or Captain Kirk. For Walton, when we look at a representational painting or a motion picture fiction we imagine seeing what the painting or the fiction represents. Walton is committed to the idea that there is a sharp or categorical difference between what we report when we exclaim 'There is William Shatner' when we look at a painted poster of _Star Trek: The Motion Picture_, and when we look at the film. In the first case, we imagine that we see what is depicted, in the second case, we are reporting upon what we actually see via its photographic reproduction. Yet, is my conviction that I can see what is depicted justified only in the case of looking at photographs or motion pictures and not in the case of representational paintings?

 

Walton's argument about natural counterfactual dependency is insufficient by itself to establish the sharp distinction that Walton requires between what we see when we look at a photograph and what we see when we look at a painting. It merely establishes an important difference between representational paintings and photographs. The further argument needed is one which establishes a relationship between the character of the natural counterfactual dependency that characterizes photography and the counterfactual dependency involved in ordinary seeing. Walton writes:

 

'Why is it that we see Lincoln when we look at photographs of him but not when we look at his painted portrait? The answer requires an account of seeing . . . I would subscribe to some variety of causal theory: to see something is to have visual experiences which are caused, in a certain manner, by what is seen. Lincoln (together with other circumstances) caused his photograph and, thus, the visual experiences of those who view it.' [16]

 

Reports about what we see are counterfactually dependent upon what is seen. If the animal caught in my headlights was a bear and not a deer, my report about what I saw (if veridical) would be that I saw a bear rather than a deer. However Walton is committed not simply to the counterfactual dependency of ordinary seeing, but to a characterization of that counterfactual dependency in causal terms. [17] Since, when we see something, we have a visual experience that is caused by what is seen, ordinary seeing exhibits the same kind of natural dependency that photographs exhibit. Furthermore, since ordinary seeing is transparent, then so too is our perception of what is depicted in a photograph.

 

Once the casual theory of perception is abandoned, the natural counterfactual dependency that characterizes photographs can be used to explain only the difference in relationship that a representational painting or drawing has to the scene painted, and a photograph to what is photographed: photographs are natural representations, paintings are intentional representations. This difference between paintings and photographs is crucial to understanding the evidential nature of photographs and films: photographs can reliably show us where things are located and how they looked at a certain time and place. But this difference cannot be conceptualized in terms of a difference between a case of genuine seeing and a case in which we do not, properly speaking, see the object depicted at all. Furthermore, the fact that photographs or films can count as evidence of what they depict does not entail that when we look at a photograph or film we actually see what the photograph of film depicts through the image, or that we are brought into direct perceptual contact with the object. Such a thesis could be justified only if it were correct to describe 'ordinary seeing' and the counterfactual dependency it entails in terms of the causal relationship that characterizes the dependence of the photograph upon the object photographed. But as we have seen there are good reasons for denying the causal theory of perception.

 

V

I have suggested that illusion and transparency theorists are right to take seriously the idea that we see what is depicted in a picture, but they mistakenly try to fit their explanation of what it is that we see within the framework of the causal theory of perception; it is this attempt that leads to error. Pictures are not like illusions and the transparency thesis draws too sharp a contrast between looking at representational paintings and looking at photographs. However, one may agree with this criticism of illusion and transparency theories without accepting that the reason for their failure is the fact they rely on a particular theory of perception. Instead, the reason for their failure might seem to lie in their mistaken assumption that when we look at what is depicted in a picture we actually see what it depicts.

 

As we have seen, other than in the case where we look at a standard photograph of a non-fictional object, Walton argues that looking at what is depicted in a picture is a form of imagined seeing rather than seeing, though he emphasizes that the activity in which we imagine seeing the object depicted is bound up with our actual seeing of the surface of the picture. Referring to Meindert Hobbema's picture 'Wooded Landscape with a Water Mill', Walton writes: 'Rather than merely imagining seeing a mill, as a result of actually seeing the canvas (as one may imagine seeing Emma upon reading a description of her appearance in _Madame Bovary_), one imagines one's seeing of the canvas to be a seeing of a mill and this imagining is an integral part of one's visual experience of the canvas.' [18] Walton's thesis of imagined seeing is advanced as a general thesis about seeing pictures; he offers the transparency thesis as a further thesis that pertains to looking at photographs alone.

 

Walton's account of imagined seeing is derived in part from Wittgenstein's contemplations (in the _Philosophical Investigations_ and elsewhere) on aspect seeing. However, Wittgenstein's reflections on aspect seeing are more complex and nuanced than Walton's extrapolation. Under the rubric of 'aspect-seeing' Wittgenstein investigates a range of cases that illuminate the diversity and complexity of the concept of seeing and in particular the borderline between seeing and imagined seeing. The relationship of 'seeing aspects' to imagination is illustrated by Wittgenstein in the example of a triangle that can be seen in a number of different ways: as a solid object, as the geometrical drawing of a triangle, as standing on its base, as hanging from its apex, as a mountain, as a wedge, as an arrow, as a pointer, as an overturned object which is meant to stand on the shorter side of the right angle, as a half-parallelogram, and as various other things. [19]

 

At least two considerations suggest that seeing the aspects of the triangular figure is a case of imagined seeing. First, discriminations between the different aspects are not specifiable in terms of the physical properties of what we see; we cannot discriminate one aspect from another on the basis of the physical properties of the object depicted. Secondly, it undoubtedly requires imagination to see the triangle as one thing and another, just as it requires imagination to see the shape of creatures in the clouds as Leonardo asked of an aspiring painter. Aspect seeing in this case is subject to the will in a way that ordinary seeing is not. [20] The first consideration is not conclusive. Despite the fact that we do not discriminate aspects on the basis of different sets of physical features, it is the physical features of the triangle that lead us to see it one way or another; we do not simply project properties onto the triangle at random. Yet it is hard to conceive of our perception of the different aspects as the perception of something different, for the aspect lacks the fixity or permanence that typically characterizes an object of sight. The 'perception' of the aspects of the triangle does seem to be a good candidate for imagined seeing. However, to concede that seeing the aspects of a triangle is a case of imagined seeing is not to concede that the thesis of imagined seeing provides a general explanation of what we see when we look at a picture, for it is only an unusual type of picture that elicits or encourages this kind of imaginative activity. Typically, a picture will endow a given aspect with a sense of permanence, that is, a given aspect becomes a property that the picture compels us to see: this is a picture of a wedge, this is a picture of an arrow. While we must be able to recognize what the picture is of in order to see picture of a wedge or an arrow, in these cases it takes no distinctive activity of the imagination to do so: our recognition of what the picture is of is immediate and the aspect is a permanent feature of the picture.

 

The affinity of looking at what is depicted in a picture with seeing, as opposed to an activity of the imagination or imagined seeing, can be illustrated by considering Jastrow's drawing 'Is it a duck? Is it a rabbit' and Wittgenstein's discussion of it. The duck-rabbit figure contrasts with the triangular figure for we are not free to interpret what it is that we see in different ways. Instead, we are compelled to see the figure one way or another. Imagination is not required to see the two aspects of the duck-rabbit figure, simply the capacity to recognize ducks and rabbits. Seeing the figure, say, as a duck, is like seeing a duck, for two main reasons. First, the perception is immediate and direct. Someone who has only seen the duck-rabbit figure as a duck, that is, who perceives the figure as an unambiguous picture, will not report when asked what it is that she sees: 'Now I am seeing the figure as a duck', as if seeing the figure as a duck was the product of some special kind of mental activity on her part. She will simply point to a duck or picture of a duck, or make duck noises in order to explain what it is that she sees in the same way that she would explain what it is that she sees when she sees an actual duck. It would make as little sense for her to say, 'Now I am seeing the figure as a duck', as it would for her to say at the sight of a knife and fork, 'Now I am seeing this as a knife and fork'. Wittgenstein writes: 'One doesn't 'take' what one knows as the cutlery at a meal for cutlery; any more than one ordinarily tries to move one's mouth as one eats or aims at moving it.' [21] Secondly, like a visual perception, seeing the duck aspect is a mental state with a measurable duration. For example, we might say, 'I saw the figure as a duck for exactly two minutes, but now I am seeing the figure as a rabbit and I shall do so for a further five minutes.'

 

The standard case of pictorial perception does not involve the distinctive visual experience of aspect dawning for a picture does not typically offer the possibility of two contrasting perceptions. The physical cues provided by the standard picture as to what we see in it are not ambiguous. Rather, representational paintings and photographs are like the case where we see the duck-rabbit figure as a duck or a rabbit in ignorance of the alternative 'interpretation' of the figure. In the case where we just see the duck-rabbit figure as a duck or as a rabbit, we do not see the 'duck' or 'rabbit' as an aspect of the figure at all. Here the perception of the duck or rabbit is simply like the case of standard pictorial perception. Representational paintings, photographs, and films manifest what Wittgenstein terms 'continuous aspect seeing' in which, as I have said, the aspect becomes a property of the picture that the picture compels us to see. Unlike cases of imagined seeing, no distinctive activity of the imagination is commonly required to recognize what a painting or photograph is a picture of. Given the affinities that continuous aspect seeing or looking at a pictorial representation bears to seeing in general, it is surely only the prejudicial adherence to a mistaken theory of perception that prevents visual theorists from acknowledging that seeing what is depicted in a picture is a form of seeing and not simply a form of imagined seeing. For only the causal theory of perception requires us to insist that what we really see when we look at a picture is only a two dimensional image and not what is depicted by that image. [22]

 

VI

Some visual theorists agree that the thesis of imagined seeing is not required to explain the activity of looking at pictures, but argue instead that the activity can be explained through the idea that when we look at pictures we deploy capacities to identify and recognize what they depict. The exercise of recognition capacities does not require that we actually see the thing that we recognize, merely that we are provided with a sufficient number of visual cues to correctly identify something. Pictorial representations can be understood to provide such cues without it being necessary to postulate that we see or imagine seeing what they depict. Noel Carroll argues against the thesis of imagined seeing on the grounds that it seems unnecessary to postulate the activity of imagining looking in at least one class of depictions: non-fictional depictions. What is in common between looking at non-fiction and fictional depictions is simply the fact that we recognize what they are by looking, rather than, say, by reading: 'Recognition', he writes, 'without the additional process of imagining seeing, is basic to analyzing depictive representation.' [23]

 

Gregory Currie has offered a detailed elaboration of the recognition thesis. Currie, like Carroll, proceeds from the assumption that 'cinematic images, like paintings, are representations, that we perceive representations of things when we see photographs'. [24] However, this is obviously only a starting-point for an analysis of what it is we see when we look at a picture, for we can easily imagine circumstances in which we look at a picture but we cannot see what it is the picture is of: our glasses are misty or the room is dark, we see a rectangular shape on the wall but we cannot see that it is a still life. The difference between seeing a picture and seeing what is depicted in it consists, for Currie, in recognizing what the picture is of. The apparent advantage of the concept of recognition is that it allows us to capture what is in common between seeing an object and seeing a picture of the object in a manner that is consistent with the causal theory of perception. When we see a horse, seeing the horse involves recognition that it is a horse we see; when we see a picture of a horse we recognize by looking at the picture that it is a horse depicted. What lies in common between seeing a horse and a picture of a horse is that both activities deploy horse-recognition capacities, but in the one case we see a horse, in the other case we merely look at the picture of a horse. But how is it that when we recognize that the picture is a picture of a horse by looking, we do not see a horse?

 

The specifics of Currie's account of how we draw upon our object recognition capacities when looking at pictures develops a theory of pictorial cognition proposed by Flint Schier. [25] Following Dennett, Fodor, and others, Currie argues that the brain is organized into a number of relatively autonomous subsystems that operate on a hierarchy of the complex to the primitive. Primitive subsystems of the brain act to categorize the 'visual input' as the object itself, whether or not what we see is a depiction of the object or the object itself. More complex subsystems then operate to correct this diagnosis if it is erroneous and allow us to recognize the depicted object as a depicted object, or alternatively, if what we see is the object, they confirm the diagnosis. He admits that his view of picture recognition constitutes a sort of illusionism about pictures: 'but this is an illusionism we can live with. It allows, exactly, that the person seeing can recognize a picture as representing a horse without him supposing he is actually looking at one'. [26]

 

Is this an illusionist theory and can we live with it? Presumably Currie does not want to maintain that the primitive subsystem of the brain actually sees the visual input, so the 'deception' here is scarcely an illusion. Furthermore, it seems inappropriate in this context to speak of recognition or deception at all, since the failure involved is more like that of a robot that lacks the capacity to discriminate between certain sensory inputs than the response of a human being. Can a robot display the shock of recognition, can it be aware of its mistake without quite knowing how to correct it? So why does Currie personify the parts of the brain as homunculi that see and are confused by what they see? It is, I think, because he realizes there is a need to explain what it is that we see when we recognize what is depicted in a picture, while at the same time holding to the assumption that what we see must be specifiable in terms of the physical properties of the physical object that lies in our visual field. Since we do not in this sense see the depicted object when we look at a picture of it, the only way of explaining how it is we do see what is depicted is the illusion theory. But the illusion theory is untenable as a theory of consciously looking at pictures, so it is buried in the subsystems of the brain, in the cognitive unconscious, so to speak, where it is hoped it can do no harm. But the illusion thesis cannot be buried there without entailing conceptual confusion. [27] If we subtract from Currie's discussion the misleading talk of homunculi in the brain doing the seeing and being confused, we are still bereft of an explanation of how seeing enters into our recognition of the depicted object.

 

Is it meaningful to speak of recognizing something by looking in a manner that does not imply that we are looking at the thing we recognize? Carroll's formulation, quoted earlier, certainly implies this. He suggests we can recognize by reading or by looking in such a way that the concept of recognition is detached from the concept of seeing what it is that we recognize. However, as Walton argues, the concept of recognition here is detached from the idea of seeing what it is that we recognize only by misconstruing the distinction between reading about something and seeing a picture of it. [28] Carroll asserts that when we look at pictures we recognize by looking rather than by reading, but when we read it is only by looking that we recognize the words on the page, so Carroll has failed to articulate what is distinctive about looking at pictures. Currie accounts for the difference between our experience of written and visual fictions in the following terms: visual fictions as opposed to written fictions involve 'perceptual imagining'; for example, 'I see displayed on the screen a man with a knife, and I imagine that there is a murderer.' However, when Currie writes, 'I see displayed on the screen a man with a knife', he means that my man-with-knife recognition capacities are triggered by the image. Since the recognition thesis already leaves us bereft of a distinction between seeing something and reading about it, the thesis of 'perceptual imagining' that presupposes the prior deployment of our recognition capacities -- what we recognize cues our imagination -- cannot restore the distinction that has been elided. [29] Looking at a picture, whether a painting, a photograph, or a film, we recognize what the picture is of because we see what it depicts.

 

VII

I have addressed what are, arguably, the four main kinds of theses that film theorists and philosophers have offered to explain what it is we see when we look at a motion picture: illusion theories, the transparency theory, the thesis of imagined seeing, and the recognition thesis. I have suggested that all these theories are individually flawed, but that their underlying problem is the assumption that they share in common: the causal theory of perception. By requiring that we separate our perceptual reports into a mental component and a physical component the causal theory of perception imposes a conceptual straight-jacket upon our understanding of what it is to look at a picture: what it is that we see when we look at something must be specifiable in terms of a physical object array that produces in us a visual sensation. This profoundly distorts our understanding of what it is we see when we see depictions because it entails that we when we look at pictures what we really see is a picture, photograph, or film and not what they depict. Each of the explanations I have examined thus far as to what it is to see a motion picture conforms to the straitjacket of the causal theory of perception either by producing by theorizing that we see an illusion of things or the things themselves or by denying that we really see what is depicted at all.

 

The insight afforded by the illusion theory is to take seriously the idea that we do see the object depicted when we see a motion picture. However, that insight is then distorted when it is construed as the claim that we have a visual experience that is identical to the visual experience we have when we actually see an object that is not a depiction. Walton is correct to recognize that there are differences in the way we look at paintings and motion pictures. However, the transparency thesis misconstrues the distinctive character of cinematic depiction by claiming that in photography and cinema we see the thing itself through the photographic representation. As a generalized account of seeing pictures, the thesis of imagined seeing acknowledges that there is more to seeing a picture than simply recognizing what it is of, but it mistakenly construes seeing what a picture is of as a mental activity that is added on to the activity of looking; but looking at what a picture is of does not require a further activity other than simply looking. Finally, the recognition thesis either fails to offer an explanation of the visual basis of pictorial recognition at all (Carroll) or it offers this only by recourse to an illusion theory (Currie).

 

But surely one of these theories, or a theory like them, must be right? Is it not superstitious to say that we can see what is depicted in a picture? It would be superstitious to claim that when we look at what a picture depicts something that is absent or non-existent is made present to us. But, of course, this is not happens when we see what is depicted in a picture, for we do not when looking at what is depicted cease thereby to see a depiction. But then how can I claim that we see what is depicted in a picture at all? It is because, as Wittgenstein points out, the verb 'to see' has many uses, only one of which is captured in reports about the deployment of things in space, and we speak of seeing what is depicted in paintings, photographs and films because we react to these things in the same spontaneous way that we react to the things themselves. As Hyman has written:

 

'when looking at a painting, the natural answer to the question 'What do you see?' is a description of the depicted scene, and not a description of the disposition of pigments. This is not simply because we have learned to assume that this is what the question is after. We can see what is depicted; but it is generally more difficult, and it may be very difficult indeed, to see how the pigments are disposed.' [30]

 

Photographs, like paintings, have a surface, but the elements of the photograph are not constituted out of marks inscribed upon that surface. The surface of a photograph registers or records what it depicts. Furthermore, the projected image has no surface other than the screen upon which it is projected. Features of the surface of the photograph, such as graininess, may enter into our perception of it but they are not constitutive of what a photograph depicts. The difference in the relationship between surface and image in a painting or drawing compared to a photograph or film suggests that we can revive Walton's transparency claim once it is shorn of the argument that what we see when we look at a photograph or motion picture is the object itself. A photograph is transparent because even though it has a surface, it is usually not constituted out of marks inscribed upon its surface. The projected moving image is transparent for it lacks a surface altogether, other than the screen upon which it is projected. Paintings, by contrast, tend to lack transparency, though paintings that mimic photographs appear transparent. This interpretation of transparency has the distinct advantage that it is indifferent to whether or not a photograph is mechanically or digitally produced, or whether a film is animated or live action. That is, a photograph or film remains transparent whether or not its causal origins lie in the registration of reflected light from an object.

 

New York University, USA

 

 

My special thanks to Malcolm Turvey for encouraging me, by his example, to think through the significance of Wittgenstein for film theory and for his comments on this paper. Thanks also to Berys Gaut, Paisley Livingston, Murray Smith, Steven Schneider, and Michael Zryd for their helpful comments.

 

 

Footnotes

 

1. This paper has been revised in order to clarify a confusion in the original version -- published in Richard Allen and Murray Smith, eds, _Film Theory and Philosophy_ (Oxford: Clarendon Press, 1997) -- between the causal theory of perception as an explication of the concept of perception and the role of causality in perception. The result of this confusion was that in the name of denying the causal theory of perception I at times seemed to deny the role of causality in perception altogther. This is all the more ironic, since this was precisely a distinction that I was keen to observe. I am grateful to Samuel Guttenplan for precisely identifying this confusion in his review of _Film Theory and Philosophy_ ('Analytic Philosophy and Film', _Film-Philosophy_, vol. 2 no. 36, November 1998 <http://www.film-philosophy.com/vol2-1998/n36guttenplan>; accessed 16 April 2001). While I have made significant changes throughout, I have preserved, as much as possible, the structure and content of the original essay, whose argument, I believe, still withstands scrutiny.

 

2. This characterization of the causal theory of perception is taken from John Hyman, 'The Causal Theory of Perception', _Philosophical Quarterly_, vol. 42 no. 168, 1992, p. 278.

 

3. The main source for Wittgenstein's discussion of seeing aspects is _Philosophical Investigations_, 2nd edition, trans. G. E. M. Anscombe (Oxford: Blackwell, 1958), Part II, xi. Further extensive discussion of the concept is contained in three volumes of Wittgenstein's notes on the philosophy of psychology: _Remarks on the Philosophy of Psychology_, vols 1 and 2 (Chicago: University of Chicago Press, 1980), and _Last Writings on the Philosophy of Psychology_, vol. 1 (Chicago: University of Chicago Press, 1982).

 

4. See, for example, Noel Carroll, _Theorizing the Moving Image_ (Cambridge: Cambridge University Press, 1996), p. 368.

 

5. This example and the next are taken from H. P. Grice, 'The Causal Theory of Perception', _Proceedings of the Aristotelian Society_, suppl. vol. 35, 1961, p. 142.

 

6. Hyman, 'The Causal Theory of Perception', p. 279. The argument that follows is largely indebted to Hyman's paper.

 

7. P. F. Strawson, 'Perception and its Objects', in G. F. MacDonald, ed., _Perception and Identity_ (London: Macmillan, 1979), p. 51.

 

8. Hyman, 'The Causal Theory of Perception', p. 283.

 

9. P. M. S. Hacker, _Appearance and Reality_ (Oxford: Blackwell, 1987), p. 234.

 

10. See Noel Carroll, _Mystifying Movies: Fads and Fallacies in Contemporary Film Theory_ (New York: Columbia University Press, 1988), pp. 89-146.

 

11. See Richard Allen, _Projecting Illusion: Film Spectatorship and the Impression of Reality_ (New York: Cambridge University Press, 1995), pp. 106-14.

 

12. Noel Carroll, _Theorizing the Moving Image_, p. 368. I attempt to circumvent this kind of objection to the theory by revising the concept of projective illusion to one in which we entertain in thought or imagine that we see the represented object. However, it is not appropriate to label the thesis of imagined seeing projective illusion for it is not an illusion theory at all. If I imagine seeing I do not see, just as if I imagine eating I do not actually eat. If imagined seeing is not seeing, then it is not the seeing of an illusion either.

 

13. Andre Bazin, 'The Ontology of the Photographic Image', _What is Cinema?_, vol. 1, trans. Hugh Gray (Berkeley and Los Angeles: University of California Press, 1967), p. 14.

 

14. Kendall Walton, 'Transparent Pictures: On the Nature of Photographic Realism', _Critical Inquiry_, vol. 11 no. 2, 1984, p. 252.

 

15. Gregory Currie, _Image and Mind: Film, Philosophy and Cognitive Science_ (New York: Cambridge University Press, 1995), p. 55.

 

16. Walton, 'Transparent Pictures', p. 261.

 

17. My interpretation of Walton's argument is consonant with that of Gregory Currie in _Image and Mind_, p. 63. Against Currie, Walton denies that he has settled on an account of what it is to see something; see 'On Pictures and Photographs: Objections Answered', in _Film Theory and Philosophy_, p. 69. Walton's dismissal of Currie's reconstruction of his argument flies in the face of the text I have quoted that demonstrates his clear commitment to a causal theory of perception. For an article that makes a similar diagnosis of Walton to the one I make here see Jonathan Friday, 'Transparency and the Photographic Image', _British Journal of Aesthetics_, vol. 30 no. 1, January 1996, pp. 30-42.

 

18. Kendall Walton, _Mimesis as Make Believe: On the Foundations of the Representational Arts_ (Cambridge, Mass.: Harvard University Press, 1990), p. 301. Walton derives his thesis of imagined seeing, in part, from Richard Wollheim's characterization of looking at paintings in terms of 'seeing-in'. Unlike Walton, Wollheim claims that 'seeing-in' is not a form of imagined seeing but a case of seeing. According to Wollheim, a picture is not an illusion because seeing what is depicted does not preclude our attention to the surface of the painting. The term 'seeing-in' is designed to capture what he terms the 'two-foldedness' of the experience of seeing what a picture depicts. Yet, in spite of his intention, Wollheim's account of 'seeing-in' becomes an illusion theory since his analysis of seeing is cast in terms of the causal theory of perception. He defines 'seeing-in' as the capacity to have 'perceptual experiences of things not present to the senses' -- _Art and Its Objects_ (New York: Cambridge University Press, 1980), 2nd edition, p. 217. According to Wollheim the picture of an object causes us to have a mental state that is like the mental state of those who actually see the object. While this experience of seeing something not present to the senses is meant to coincide with the experience of seeing the surface of the picture, Wollheim does not explain how this coincidence of illusion and veridical perception can be achieved. For criticism of Wollheim on this point see John Hyman, _The Imitation of Nature_ (Oxford: Blackwell, 1989), p. 217. It is understandable then, that Walton should interpret Wollheim's concept of 'seeing-in' as a form of imagined seeing: we imagine seeing something that is not present to the senses.

 

19. Ludwig Wittgenstein, _Philosophical Investigations_, Part II, xi, p. 201.

 

20. The role of the imagination in aspect seeing is emphasized by T. E. Wilkerson in 'Pictorial Representation: A Defense of the Aspect Theory', _Midwest Studies in Philosophy_, vol. 16, 1991, pp. 152-66. While this emphasis is valuable as an account of aspect seeing it distorts the role of the imagination in looking at pictures.

 

21. Wittgenstein, _Philosophical Investigations_, Part II, xi, p. 195.

 

22. My own thoughts on Wittgenstein and aspect-seeing are partly inspired by the imaginative and suggestive exploration of Wittgenstein's remarks by Malcolm Turvey in 'Seeing Theory: On Perception and Emotional Response in Current Film Theory', in _Film Theory and Philosophy_, pp. 441-53.

 

23. Noel Carroll, 'Critical Study: Kendall L. Walton, _Mimesis as Make-Believe_', _Philosophical Quarterly_, vol. 45 no. 178, 1995, p. 97.

 

24. Currie, _Image and Mind_, p. 78.

 

25. See Flint Shier, _Deeper Into Pictures_ (New York: Cambridge University Press, 1986), pp. 188-95.

 

26. Currie, _Image and Mind_, p. 86.

 

27. On the 'reckless application of human-being predicates to insufficiently humanlike objects' see Anthony Kenny, 'The Homonculus Fallacy', in John Hyman, ed., _Investigating Psychology_ (New York: Routledge, 1991), pp. 155-65.

 

28. See Walton's remarks in 'On Pictures and Photographs: Objections Answered', in _Film Theory and Philosophy_, pp. 65-66.

 

29. In 'On Pictures and Photographs', pp. 62-65, Walton offers further good reasons for thinking that Currie's thesis of 'perceptual imagining' should be rejected.

 

30. Hyman, _The Imitation of Nature_ (Oxford: Blackwell, 1989), pp. 42-3.

 

 

Copyright © Richard Allen 2001

 

Richard Allen, 'Looking at Motion Pictures (Revised)', _Film-Philosophy_, vol. 5 no. 25, August 2001 <http://www.film-philosophy.com/vol5-2001/n25allen>.

 

 

Save as Plain Text Document...Print...Read...Recycle

  

Join the Film-Philosophy salon,

and receive the journal articles via email as they are published. here

 

Film-Philosophy (ISSN 1466-4615)

PO Box 26161, London SW8 4WD, England

Contact: editor@film-philosophy.com

 

Back to the Film-Philosophy homepage