 |
| Ema Sullivan-Bissett |
This month, Ema Sullivan-Bissett introduces chapter 11 of Jakob Hohwy’s The Predictive Mind (OUP, 2013). Ema is a Research Fellow on Lisa Bortolotti’s ERC-funded project PERFECT, in the Department of Philosophy at the University of Birmingham.
Chapter 11 - The Fragile Mirror of Nature
Presented by Ema Sullivan-Bissett
In chapter eleven Hohwy seeks to challenge background beliefs we hold about the nature of our body, and characterizes us as ‘fundamentally fragile prediction error minimizing machines’ (p. 224). The fragility of the perceptual system is what is to be explicated in this chapter.
Hohwy starts by setting up a contrast between his prediction error minimization view of perception, and the bottom-up view. The latter casts us as ‘passive receivers of sensory input’, which does not require of us that we interpret such input making use of relevant background beliefs. Such a view is, of course, ‘fundamentally in stark contrast to the prediction error minimization idea’ (p. 225).
We see that Hohwy’s preferred view is well placed to account for disorder (see Ch. 7: Precarious Prediction), helped along by the wrong kind of input, or by the system’s stemming prediction error at the expense of truth. With respect to the first, Hohwy demonstrates this thought with reference to Lars Scwabe and Olaf Blanke’s (2008) work on out-of-body experiences (pp. 225–7). With respect to the second, Hohwy refers to the results of experiments on the rubber hand illusion (pp. 230–6) (If you are interested in recent work on the rubber hand illusion, see these three posts on the Imperfect Cognitions blog: The Rubber Hand Illusion and Anomalous Experiences, The Rubber Hand Illusion Using Representation and Similarity (1), and (2)).
The fragility of perception comes down to two points: ‘the imperative to stem prediction error at as many levels as possible and the fact we are restricted to using only the internal resources of the brain in doing this’ (p. 227). Hohwy notes that this is a difficult truth to acknowledge.
Hohwy starts by setting up a contrast between his prediction error minimization view of perception, and the bottom-up view. The latter casts us as ‘passive receivers of sensory input’, which does not require of us that we interpret such input making use of relevant background beliefs. Such a view is, of course, ‘fundamentally in stark contrast to the prediction error minimization idea’ (p. 225).
We see that Hohwy’s preferred view is well placed to account for disorder (see Ch. 7: Precarious Prediction), helped along by the wrong kind of input, or by the system’s stemming prediction error at the expense of truth. With respect to the first, Hohwy demonstrates this thought with reference to Lars Scwabe and Olaf Blanke’s (2008) work on out-of-body experiences (pp. 225–7). With respect to the second, Hohwy refers to the results of experiments on the rubber hand illusion (pp. 230–6) (If you are interested in recent work on the rubber hand illusion, see these three posts on the Imperfect Cognitions blog: The Rubber Hand Illusion and Anomalous Experiences, The Rubber Hand Illusion Using Representation and Similarity (1), and (2)).
The fragility of perception comes down to two points: ‘the imperative to stem prediction error at as many levels as possible and the fact we are restricted to using only the internal resources of the brain in doing this’ (p. 227). Hohwy notes that this is a difficult truth to acknowledge.
Hohwy asks whether perception is direct and indirect and suggests that we should not force a choice between these options, since there is a ‘grain of truth’ in both options (p. 228). Prediction error minimization is in favour of indirectness insofar as predictions are generated by internal models, but it is also in favour of directness because we are anchored in the ‘causal order of things’ (p. 228). The mind is fragile because though the state of the world guides the hypotheses generated by our internal model, thus making them reliable, this reliability should not be expected across changes in conditions, since if conditions are not normal, the hypotheses generating mechanisms may go awry; ‘[t]he mechanism is not very good at sticking to its learned truths when it moves outside its narrow zone of comfort’ (p. 230).
Next Hohwy turns to the way we perceive our body, and asks whether the perceptual fragility he has spoken of could spread to our body sense and the role it plays in perceptual inference. He then discusses recent research which points to it being ‘disconcertingly easy to disturb our body perception’ (p. 230). The overall thought of this discussion is that the results from empirical work on disturbing bodily perception and perceptual inference can be explained by appeal to Bayesian inference. As applied to the rubber hand illusion: ‘it takes a dynamic, probabilistic journey to leave behind the prior belief that touch is always felt on hand-like objects, but on the other hand, given the right sequence of evidence, this prior belief can in fact be left behind’ p. 232).
One might think that the fragility Hohwy points to is inconsequential. One reason to think this is that most of us do not experience the shattering of our ‘fragile perceptual worlds’ (p. 235). In response to this Hohwy points out that various mental disorders involve this kind of shattering, disorders which afflict ‘a surprisingly large proportion of us’ (pp. 235–6). We have learned then of the high malleability of the perceptual body image. We are willing to accept weird hypotheses to explain away some sensory input.
In this chapter Hohwy has demonstrated the fragility of perceptual inference, which leads to a ‘disconcerting and comforting’ understanding of the perceptual mind–world relation. Disconcerting because we seem to be at the mercy of the way in which we make the best out of the sensory input from the world, which, as we have seen, can lead to us opting for very strange hypotheses to minimize prediction error. Comforting because we cannot just come up with any hypothesis to minimize prediction error, we are constrained to hypothesize in ways which fit the sensory input.
No comments:
Post a Comment