Can predictive processing explain self-deception?

Abstract

The prediction error minimization framework (PEM) denotes a family of views that aim at providing a unified theory of perception, cognition, and action. In this paper, I discuss some of the theoretical limitations of PEM. It appears that PEM cannot provide a satisfactory explanation of motivated reasoning, as instantiated in phenomena such as self-deception, because its cognitive ontology does not have a separate category for motivational states such as desires. However, it might be thought that this objection confuses levels of explanation. Selfdeception is a personal level phenomenon, while PEM offers subpersonal explanations of psychological abilities. Thus, the paper examines how subpersonal explanations couched in the PEM framework can be thought of as related to personal level explanations underlying selfdeception. In this regard, three views on the relation between personal and subpersonal explanations are investigated: the autonomist, the functionalist, and the co-evolutionary perspective. I argue that, depending on which view of the relation between the personal and subpersonal is adopted, the PEM paradigm faces a dilemma: either its explanatory ambitions should be reduced to the subpersonal domain, or it cannot provide a satisfactory account of motivated reasoning as instantiated in self-deception.

The role of Marr’s Levels of Explanation in Cognitive Sciences

“…Dosadašnjoj raspravi bi se moglo prigovoriti da pretpostavlja valjanost paradigme iz kognitivnih (neuro)znanosti koja je tek u povojima i za koju nije jasno da može zahvatiti sve važne aspekte ljudske psihe. 31 Na primjer, neki ukazuju na to da prediktivno kodiranje nije analitički rješivo, u smislu da nije jasno da je to algoritam koji može u razumnom vremenu obaviti izračune koji bi omogućili adaptivno ponašanje i rješavanje kognitivnih zadataka s kojima se organizmi svakodnevno suočavaju. 32 Budući da PSE svodi djelovanje i percipiranje na minimizaciju slobodne energije, drugi prigovaraju onda da nije u mogućnosti napraviti jasnu razliku između motivacijskih stanja, poput želja, i spoznajnih stanja, poput vjerovanja.…”

Testable or bust: theoretical lessons for predictive processing

“…By means of complete class theorem, PP may even necessitate needless reconceptualizations of relatively well-understood phenomena (Jurjako, 2022). Let us consider self-deception, which is a motivationally biased process of forming beliefs, in which they evolve in the direction opposite to what is provided by evidence.…”

“…Let us consider self-deception, which is a motivationally biased process of forming beliefs, in which they evolve in the direction opposite to what is provided by evidence. PP takes such cases as instances of idiosyncratic rationalities (given a unique history of internal model development), doing away with a palpable phenomenal distinction between beliefs and desires just because it does not differentiate between them at the algorithmic level (Jurjako, 2022;Williams, 2021). A variety of forms of motivated and unmotivated reasoning is aggregated under a homogenous category of optimality.…”