Predictive processing and the representation wars: a victory for the eliminativist (via fictionalism)

Abstract

In this paper I argue that, by combining eliminativist and fictionalist approaches toward the sub-personal representational posits of predictive processing, we arrive at an empirically robust and yet metaphysically innocuous cognitive scientific framework. I begin the paper by providing a non-representational account of the five key posits of predictive processing ("prediction-signal", "error-signal", "prior", "likelihood", and "posterior probability"). Then, I motivate a fictionalist approach toward the remaining indispensable representational posits of predictive processing, and explain how representation can play an epistemologically indispensable role within predictive processing explanations without thereby requiring that representation metaphysically exists. Finally, I outline four consequences of accepting this approach and explain why they are beneficial: (1) we arrive at a victory for metaphysical eliminativism in the 'representation wars'; (2) my account fits with extant empirical practice; (3) my account provides guidance for future research; and, (4) my account provides the beginnings of a response to Mark Sprevak's IBE problem for fictionalist approaches toward sub-personal representation.

Distinguishing volumetric content from perceptual presence within a predictive processing framework

“…The role of sensation for both views is always as a corrective after the experience itself, which is profoundly dynamic. Of course, these expectations are cashed out within the PPF as “models” or “hypotheses”, which is a move sensorimotor enactivists are likely to resist, but that is a question for another day (see Downey 2018 ).…”

Self-Knowledge in a Predictive Processing Framework

“…7 This is similar to treating mathematical objects as real, even though it is not fully clear what their ontological status actually is. That predictive processing leads to this fictionalist view of representation in general has been recently argued by Downey (2017). Moreover, there is good reason to opt for the fictionalist stance.…”

Mechanistic unity of the predictive mind

“…This might mean that the contents thus ascribed to an agent are not observer-independent, semantic properties of the agent’s internal states, causally shaping its behavior. Because FEP is so liberal about how prior beliefs are realized, ascriptions of prior beliefs may have a merely fictional, “as if” status (see Downey, 2018). Alternatively, the intentional commitments of FEP might be construed realistically, assuming a realist view that is relaxed with respect to commitments about internal mechanisms (see, e.g., Dennett, 1991, Schwitzgebel, 2002).…”