Curious Inferences: Reply to Sun and Firestone on the Dark Room Problem

Published on July 23, 2020

Sun and Firestone [1] presented a challenge to predictive processing (PP) accounts of brain function by reviving the Dark Room problem – the idea that if agents are mandated to minimise prediction error, the best thing for them to do is to seek out highly predictable environments where nothing changes, and stay there. They argued that standard responses to this challenge have the potential to render the PP account untestable and explanatorily empty. We disagree.

Read Full Article (External Site)