When seeing is not believing: A mechanistic basis for predictive divergence.

Area of Science:

• Cognitive Science
• Neuroscience
• Philosophy of Mind

Background:

• Visual illusions demonstrate incongruence between perception and belief.
• Modular theories explain this, but Predictive Processing (PP) framework faces challenges.
• Existing PP models posit a unified inferential hierarchy for perception and cognition.

Purpose of the Study:

• To reconcile visual illusions and incongruent perception-belief states within the Predictive Processing framework.
• To propose a mechanism within PP that accounts for stable divergences between perception and belief.
• To maintain a hierarchical and unitary view of cognition despite perceptual conflicts.

Main Methods:

• Reviewing recent insights into the neural implementation of Predictive Processing.
• Analyzing the roles of top-down and bottom-up information processing streams.
• Proposing a model based on the interaction of context-dependent and context-independent prior information.

Main Results:

• Prior information is approximated by both top-down (flexible) and bottom-up (inflexible) processing streams.
• A stable divergence between perception and belief arises when higher-level flexible priors conflict with lower-level inflexible priors.
• This mechanism allows PP to account for incongruent percepts and beliefs.

Conclusions:

• Predictive Processing can accommodate stable perception-belief incongruence.
• The proposed model integrates conflicting perceptual and cognitive states within a hierarchical framework.
• This advances our understanding of how the brain reconciles competing information.