Continuing through our summer reading, we’ve just looked at Tim Bayne’s “Perception & the Reach of Phenomenal Content”, forthcoming in Phil. Q., where Tim attempts to defend the idea that so-called “higher-level” properties figure in the phenomenal content of perceptual experience. A few remarks/worries about the issue and Tim’s argument.
THE ISSUE: Conservatives think that perceptual experiences only represent so-called ‘lower-level’ properties such as, in the case of vision, colour, shape, spatial, topological, and mereological properties, orientation, etc., whilst liberals insist that perceptual experiences also represent so-called ‘higher-level’ properties including (still in the case of vision) object types such as being a tomato, being a eucalyptus tree, a kangaroo, a Toyota, etc. Liberals don’t just want perceptual representation tout court: they want perceptual representation of ‘higher-level’ properties with some sort of phenomenal saliency. That’s in line with the sort of ‘contrast’ arguments typically advanced for liberalism, attempting to show that whether or not a ‘higher-level’ property is represented in experience makes a phenomenal difference to it. (As Tim rightly points out, the terminology is unfortunate, since the difference between lower-level and higher-level really corresponds to different levels of processing in the visual system, not to different kinds of properties as such.)
Tim says that the issue is vague, somewhat messy, and may well collapse. In any case, reading through Tim’s paper, it strikes me that, currently at least, the action should pretty clearly take place between the following 2 claims:
Conservatives: whenever two experiences allegedly differ in content in that one represents a ‘higher-level’ property so that they are phenomenally different as a result, what is really the case is that (i) neither represents a higher-level property, but (ii) there is a representational difference in the representation of lower-level properties, which (iii) explains the phenomenal difference.
This embodies the conservatives’ best re-description strategy against the liberals’ contrast arguments: grant the phenomenological difference but explain it away as a difference in the representation of lower-level properties. In contrast,
Liberals: two experiences can differ phenomenally in that one represents a ‘higher-level’ property in a phenomenally salient way, even though there is no difference in the ‘lower-level’ properties represented.
Unless liberals argue for this strong view, their contrast arguments, it seems to me, simply won’t have much bite against the conservatives’ strategy, and we’ll end up in a stalemate. And the weaker liberal position (status-quo liberals), flagged by Tim, according to which the representation of higher-level properties might supervene on the representation of lower-level properties, is of little help in this respect.
PERCEPTION & THOUGHT: Tim doesn’t like the conservatives’ re-description strategy much. He asks why, in all the relevant cases, there should always be ‘low-level’ representational/phenomenal differences. Of course, there doesn’t have to, but insofar as the conservative can propose plausible alternative explanations involving such ‘low-level’ differences, that’s all it takes to block the liberals’ contrast arguments.
Nor does he like another re-description strategy (advanced by Michael Tye and others), according to which perceptual experiences only represent ‘lower-level’ properties, and ‘higher-level’ properties only figure in doxastic states based on experiences: whether a given ‘higher-level’ property figure in such states will depend on whether the subject recognizes the property on the basis of her experience of ‘other lower-level’ properties. Against this, Tim argues that an associative agnosiac (of which more below) could believe that the thing in front of her is a pipe, and still lack the experience of a pipe as of a pipe, which shows, he suggests, that recognition of ‘higher-level’ properties isn’t doxastic (he also mentions the fact that recognition resists cognitive penetration). He further argues that such a belief could even satisfy Tye’s requirement that recognition be based on experience, if the agnosiac’s belief is caused by a neurological condition causing her to believe everything she sees is a pipe. This is meant to show, I take it, that (a) Tye’s account of recognition as doxastic is no good because recognition is perceptual, and hence that (b) ‘higher-level’ properties can be represented in experience.
Regarding (a), I suspect Tim’s response fails to consider Tye’s suggestion seriously enough. To say that perceptual recognition is a doxastic state based on experience isn’t to simply identify perceptual recognition with beliefs, or to take the basing relation as merely causal. Perceptual recognition, rather, is a process by which a certain kind of transition from an experience to a belief takes place, where a belief with a certain content that can represent ‘higher-level’ properties is formed as the output of such a process. As such, there’s no reason to think it should be cognitively penetrable in exactly the same way beliefs can be revised on the basis of other beliefs: one may be able to revise one’s recognitional beliefs on the basis of further information, but not the process by which these beliefs are produced. And though recognition is causal, it isn’t just that For one thing, the basing relation must clearly be sensitive to the content of the experience (obviously not the case in Tim’s example), and, to some extent, reliably so: if all perceptual identifications a subject comes up with are completely inaccurate in relation to the content of her experiences, this suggests the subject isn’t sensitive in the right way to those contents. But that’s also lacking in Tim’s example.
Finally, it doesn’t help liberals to suggest, as Tim seems to, that recognitional beliefs, like thoughts, have phenomenology (‘cognitive phenomenology’). For it provides the conservatives another weapon: they can grant that there are further phenomenological differences when ‘higher-level’ properties are allegedly represented, but these differences owe simply to the cognitive phenomenology of recognitional beliefs, which accompany a perceptual experience but are distinct from it.
THE ARGUMENT FROM ASSOCIATIVE AGNOSIA: Tim claims that traditional contrast arguments for liberalism haven’t fared all that well and offers a new one, based on the phenomenon of associative agnosia. He relies on the traditional description of associative agnosia, according to which patients with this condition (a) perceive shape perfectly well but (b) are unable to recognize and identify what they perceive: the impairment is one of ‘category perception’, not one of ‘form perception’. The argument goes: (1) associative agnosiacs have experiences phenomenally different from ours, based on the fact that they can’t recognize types of objects, and (2) what is missing from their experiences, and explains the phenomenal difference, is the representation of ‘higher-level’ properties, which suggests that (3) our experiences do represent such properties.
Some worries:
i) I’m not as well versed as Tim in the empirical literature on this topic, but I would have thought that the traditional description of associative agnosia is contentious, at best, given Martha Farah’s rather compelling case for scepticism in the 2nd edition of her book (2004). One traditional reason for thinking that associative agnosiacs perceive shapes perfectly well is their ability to draw very good copies of what they see. But, Farah points out, one little mentioned fact is that their drawing is very slow and piecemeal. She suggests that, in fact, associative agnosiacs do suffer from a perceptual impairment involving global spatial perception, based on the fact that, typically, (a) they also fail to recognize meaningless shapes , (b) their mistakes usually involves types of objects similar in shape, (c) they cannot discriminate geometrically impossible objects from possible ones, and (d) have difficulties perceiving fragmented shapes. If she’s right, there’s some reason to think that associative agnosia may in fact involve a deficit in the representation of low-level (thought complex) spatial properties.
ii) Even if the traditional description of associative agnosia is correct, it’s far from clear it supports Tim’s case. The evidence primarily suggests that associative agnosiacs can’t recognize what they see. This is entirely consistent with the suggestion that the representational content of their experiences is exactly the same as ours, it’s just that their faculty of recognition is impaired, thus precluding them from forming perceptual beliefs about types of objects.
iii) Tim grants that we can’t really know what the phenomenology of their experiences is like. Yet, oddly, he insists that it is “extremely plausible” that their experiences are phenomenally different from ours. I find this puzzling. I’d like to know what makes this plausible–as opposed to agnosticism. And if standard contrast arguments based on allegedly introspectible phenomenal differences don’t fare all that well, as Tim points out, it’s a bit mysterious to me how this new contrast argument based on some phenomenal difference we have no access to is supposed to fare any better.