The Perception of Pictures - Part 3 - Art Without Illusion?


[Hans Memling - The Last Judgment]

Gibson may end up with a problem of how he can account for pictures of things not perceivable, but I want to leave that problem to stew and turn to some insights Gibson has -- insights which point to strong reasons to prefer his theory of perception and picture perception.

In "Pictures, Perspective, and Perception" [PPP], Gibson spends much time discussing the fidelity of a picture. In this discussion, he uses a 2-component definition of fidelity. He eventually discards the first component, so I'll look at that only briefly. Gibson originally subscribes to a version of what he later calls 'The Point-Projection Theory of Pictorial Information.' This essentially says that the fidelity of a picture to the object depicted depends upon the identity of the two sheaves of light rays. Without going into the technical explanation of this, I will only note that this theory presupposes an explanation in terms of geometric optics (
rays of light), and that Gibson, in preferring descriptions in terms of his own ecological optics, rightly changes his tune.

A better definition of fidelity will not be in terms of geometric optics and absolute values of points of color, but in the picture's ability to convey the right sort of information. This is a picture's functional fidelity to the thing depicted. “A picture’s functional fidelity to the scene represented is simply the degree to which the variables to which the eye is sensitive are the same in one array as the other” [223]. These variables are the familiar invariants of the optic array, things like contours (defined as 'abrupt transitions of intensity in light).

Concluding this discussion of fidelity, Gibson notes that the development of perspectival rendering is an
advancement in depiction which facilitates visual education – it allows for faithful representing of solid objects and “permits the vicarious experiencing of an absent thing or the mediated perception of a distant place” [224]. He goes on to describe this faithful depiction as important knowledge, citing anatomical drawings and scientific records as the sorts of pictures from which we can get knowledge at second-hand when first-hand knowledge is not possible.

This may seem like a widely applicable definition of pictures, but we should note that Gibson seems constrained to privilege faithful or veridical depictions over non-veridical depictions insofar as he sees visual perception as communicating more or less useful information. Mediated perception will be useful so long as it conveys ‘truthful’ information – reading ‘truthful’ as ‘information that could be found by the percipient if she could actually experience first-hand.’ This implicit tie to 'real' experience is understandable for Gibson, but will not allow him to provide a satisfactory account of the variety and strangeness of the illusions that are an essential component of pictures.

I want to stress that this is not Gibson's final definition of a picture, nor is it his final discussion of depiction, but on this account, we should see what he can claim and what he cannot claim.

Gibson can claim that pictures 'work' by mediating visual perception -- that when the individual cannot go and see Mont Sainte-Victoire (for a variety of reasons, including the interesting puzzle of how one actually 'sees' mountains to begin with), they can still pick up information about that mountain by looking at a picture of it. To be sure, the information available will be already constrained in an important way (the artist made a series of selections in depicting the visual experience), but for Gibson, these constraints are themselves valuable and available for information pickup (a percipient like Berenson may be a skilled selector of the constraints Da Vinci uses when depicting).

Gibson can also claim that we understand a picture when we receive the information conveyed -- and that this information does not depend upon the beholder knowing some language of pictures or symbols, but on the information carried by the invariant structure of light. This doesn't mean that upon looking at the picture above, any beholder will know it as a picture of the Last Judgment, nor as Hans Memling's particular depiction, but they will be able to perceive the low-level invariants which all percipients learn through normal experience in the world. For Gibson, that's a fact of how the perceptual system develops.



[Albrecht Durer - Rhinoceros]

What Gibson cannot claim is that he has successfully described pictures at all. As Ernst Gombrich elegantly shows in Art and Illusion, the very process of depicting is rife with tricks, conventions, illusions and learned developments. Pictures themselves are never actual imitations, and they often erroneously depict the thing they were trying to depict. Durer depicted the rhinoceros with armored plating, the Dutch flower painters depicted spring tulips in the same bouquets as summer roses and peonies, Parmagianino depicted a strangely 'stretched' Madonna and Child, and so on. Even depictions which 'preserve figure' and attempt to imitate reality are founded on a framework of illusion and trickery in order to accomplish their imitation.

Gibson also cannot claim to give us any way of understanding any of the modern art trends away from figure and toward what is called 'abstraction.'

Gibson has a way around these two problems, but does he solve them at the expense of another aspect of his theory? I turn to his effort to resolve some of these problems next.