Predictive Coding and Conceptual Thought

* Daniel Williams, ‘Predictive Coding and Thought’, Synthese (2018),
* Andy Clark, ‘Beyond the “Bayesian Blur”: Predictive Processing and the Nature of Subjective Experience’, Journal of Consciousness Studies, 25 (2018), 71-87.

So regular readers may have noticed that I’ve been taken by aspects of the Predictive Processing / Free Energy school of thought in cognitive science and philosophy. I posted about Andy Clark’s book Surfing Uncertainty here, I used it as a reference point in one of my posts about John Skelton’s poetry, and I wrote here about a conference in which various contributors to the field came to Cambridge and strutted their stuff. And that’s not all.
      Now one of the organizers of that conference has shown his philosophical hand in a very interesting paper. Predictive Processing seems to work very well as a model for perception and action, but it has been hailed as a comprehensive account of cognition that can include conceptual thought as well. Williams argues that the case is as yet incomplete in relation to two particular features of our thought: its generality (‘the fact that we can think and flexibly reason about phenomena at any level of spatial and temporal scale and abstraction’) and its compositionality (‘the specific way in which concepts productively combine to yield our thoughts’). And this seems fair enough, as does Williams’s analysis of the challenges facing the Predictive Processing model if it aspires to account for Everything.
      I really like, a lot, this kind of interface between a philosophical appeal to the characteristics of the ways we consciously think, and a scientific appeal to evidence of phenomena in brain scans and behavioural experiments. There can be ambitious pressure from either side, wary scruples likewise. From the semi-outside, enjoying paradoxes as I do, I like to think of my mind as a-machine-that-isn’t-a-machine. Whether or not you are more committed than I am to resolving such an impasse, I think this essay is well worth reading, and I am interested to see what response there may be.
      And also out now… Andy Clark himself tackling an interesting objection to the Predictive Processing approach, which is that our experience of perception is that it is not a matter of algorithms and probabilities: things are either there for us, or they aren’t, or so it seems. He finds a way round, via the slogan that perception is a ‘slave to action’. The field’s on the move; let’s keep up!

E-mail me at rtrl100[at]

Leave a Reply

Your email address will not be published. Required fields are marked *