Nave, K.
University of Edinburgh
If the brain is a probabilistic model of our environment, then why does visual perception seem to deliver a determinate world of ordinary objects, rather than a blur of possibilities - and is this apparent mismatch a problem for predictive processing?
I argue that it is not. Firstly, because as Clark (in press) points out, the role of perception in serving action planning, and the fact that actions do not come by probabilistic-degree, explains why perception would be pushed to serve up a world of determinate, action-relevant opportunities. Secondly because, as emphasised by Husserl (1907) and more recently developed by Cohen, M. A., et al. (2016) in an account of inattentional blindness, the determinacy of perceptual world may not be as simple as first impressions suggest.
How can these responses be combined? I claim that both the impression of determinacy and the goal of successful action guidance require only that the brain produces a percept that is sufficiently (not maximally) determinate. Visual experience determinately specifies a single value at the levels of granularity necessary for action planning or linguistic report. The levels of predictions about object size, colour, and category. This produces the misapprehension that experience is determinate simpliciter. Yet determinates at these categorical levels are still further determinable in turn (Funkhouser, 2006). To represent an object as determinately brown (and no other colour) still leaves indeterminacy regarding the object's shade, and so forth: colour>brown>tan>Pantone 14-1315.
Experience would be maximally determinate only if the properties determinately represented were not further determinable at some finer grain - that is, determinate regarding the exact patterns of sensory stimulation predicted. As determinacy at this level is necessary neither for action guidance, nor the initial impression of determinate objects, we can do justice to these while still allowing a pervasive indeterminacy in perceptual experience.