Predictive Processing, Implicit and Explicit

Abstract

This chapter employs the implicit/explicit distinction to shed light on the theoretical commitments of the predictive processing (PP) account of cognition. I propose that depending on the details, PP may be realized in two different ways. One way involves explicit representational structures that participate in implicit inferential processes, that is, processes that approximately accord to a (Bayes-)rational rule without the rule itself being represented in the system. The second way in which PP may be realized involves ‘representations’ implicitly embodied in processing biases that do not undergo inferential updating. This distinction at least partially maps onto the ‘intellectualist’ and ‘radical’ readings of PP, respectively. The upshot is that that ‘intellectuals’ and ‘radicals’ about PP may both get part of the story right, as there is some theoretical and empirical traction to the claim that the brain (if it engages in PP at all) employs both aforementioned variants of PP.