Can Hierarchical Predictive Coding Explain Binocular Rivalry?

Explanation of binocular rivalry is taken as a classic illustration of hierarchical predictive coding's explanatory power. I revisit the account and show that it cannot explain the role of reward in binocular rivalry. I show that a modified version of the predictive processing approach may account for the role of reward by recasting it as a form of optimism bias. Accepting this account, however, is at odds with the epistemic commitments favored by proponents of hierarchical predictive coding.