Wednesday, 23 January 2008

Making up the Mind 7: Getting ahead by prediction

According to authors such as Anthony Kenny and Peter Hacker, the homunculus fallacy is to be understood as the fallacy of attributing to parts of organisms properties which can only sensibly be said to belong to the whole organism. I have followed Hacker and called this the 'mereological fallacy', instead reserving the 'homunculus fallacy' to describe a different conceptual confusion which arises when neuropsychologists attempt to describe the mechanics of perception, and end up writing as if the brain is some kind of mediating mechanism which serves up deliverances for the consciousness of an inner homunculus.

Along with the entifying fallacy (that of creating entities out of acts), these fallacies clearly work together, supporting one another in an unholy conceptual trinity. However I believe that the mereological fallacy is considerably less serious than the homunculus fallacy as I have described it. When authors such as Chris Frith describe the brain as thinking or remembering or inferring, we can usually translate out from this idiom without much trouble, and just remind ourselves that we have to do with metaphor here. But the homunculus fallacy makes for whole explanatory endeavours that are deeply misguided.

Consider for example how chapter 4 - Getting ahead by Prediction - continues. For several pages Frith describes various different learning paradigms (Thorndike, Skinner, Pavlov etc.), and sometimes talks of the brain predicting or learning or remembering or knowing. But the examples ("what our brain knows as a result of learning by association"; "we can see the brain predicting in this way if we look directly at the activity in nerve cells") can always be effortlessly translated out into a non-intentionalistic idiom. Frith is talking about the differential responses of the nervous system which causally support our knowing or learning, and for a shorthand has simply ascribed the psychological functions to the causal supports. No harm comes of it.

But now consider (pp. 101ff) the following Helmholtzian concern Frith has regarding how 'I experience myself in the world':

Consider a very simple action like walking round the room... I move and the world stays exactly where it is. And this is very odd because every time I move, this movement produces dramatic changes in what my brain senses about the world. Even just moving my eyes has a dramatic effect. On my retina, and again on my visual cortex at the back of my brain, a picture of the world is projected. But if Imove my eyes this projection will change completely. As I move my eye from left to right across the fir tree in the garden, the projection of the fir tree moves from the right to the left side of my retina. This is a dramatic change of sensation. And it raises a problem for my brain - is there a change in sensation because my eye is moving or because the fir free is moving?

Now what would make it the case that 'this is very odd'? Surely we would not expect the world to (appear to?) move when we walk around the place. Such movement of the world might render impossible the very possibility of perception. The only answer that presents itself is this: imagine if we were stuck behind our retinas, trying to read the world off from images that formed there. If that were our unfortunate dilemma, then of course we would need all the help we could get, and more, from cognitive processes. We would need some to bind together the two retinal images, some to reinvert them, others to stabilise them, and so on. At the end of it we would not be in touch with the world, perhaps, but at least we'd have a picture of it that stayed reasonably stable and was (but how would we ever know?) fairly reliable in its depiction. Frith presents no evidence (and what would that look like?) that this (being epistemically trapped on the receiving ends of neural deliveries from sense organs) is our predicament. Rather, he seems to simply assume it. For this reason the chapter presents its findings as if it is a remarkable discovery. As if we all assumed naively to start with that a 'picture of the world' could be read off 'retinal images' (because that is just what perception amounts to), and then found that things really couldn't be that simple, and there would need to be a lot of cognitive processes intervening between the ultimate 'image for consciousness' (or however we would have to describe it) and the original 'sensory input'.

Frith's mistake seems to be one of conflating the personal epistemic level of our contact with the world itself, and the subpersonal neurological level of the constitutive mechanics of vision. The personal element is pushed inwards at the end of the causal chain of the subpersonal dynamics. But this won't do: these really are two different levels of explanation and understanding, not two adjacent domains that are causally chained.

None of this is to say that the possibility of perception is not founded in the possibility of the brain reacting differently to identical changes in retinal stimulation when these changes are either due to object movement or to eye movement. That is surely a necessary precondition for the possibility of perception. Perhaps because of his homuncular 'mind as recipient of neurophysiological inputs' picture, Frith seems to think that perception would be simply a matter of passively receiving sensory stimuli, if it were not for the counfounding influence of personal movement on the retinal image. Yet we know that stable (personal-level) visual experience is dependent on our being able to make thousands of micromovements - it is - at the subpersonal level - an active (even 'enactive') process constituted by dynamic recursive loops in and out of the body - both sensory and motoric (and it even becomes hard to adequately distinguish the two at the subpersonal level). (See Susan Hurley's Consciousness in Action for an excellent exploration of the relation between personal-level phenomena such as visual consciousness, and subpersonal dynamically 'loopy' (from organism to world and back again) phenomena such as sensory and motor behaviour.) Far from our own movements constituting some kind of problem for the otherwise happy receipt of information about the world which we can read straight off the retinae, they are rather the very stuff of perception itself.

Perhaps I am getting carried away attributing these views to Frith? Well, consider the following (p. 102):

Our brain can use this prediction to make us perceive the world as stable even though the image of the world is jumping around on our retina as we move our eyes. This illusion of stability is important for our survival. ... visual changes caused by our own movements are of no relevance ... By predicting these unimportant changes of sensation, the brain can suppress our response to them.

Yet there is no 'illusion of stability' - the world really is stable in such instances. And it is not as if we could perceive perfectly well if we were completely still if we had no predictive machinery. The making of such 'predictions' by the brain is part of the very possibility of our perceptual experience of the world itself, not a mere stabilising factor without which everything would appear to move all the time. Without these 'predictions' we could not be said to have visual experience at all. A material condition of the very possibility of genuine visual consciousness of the world has been read by Frith as if it were instead a mere correcting facilty for the stabilisation of visual images.