A functional theory of bistable perception based on dynamical circular inference

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience

Background:

• Bistable perception involves the brain switching between interpretations of ambiguous stimuli.
• Existing models explain bistability dynamics but not its link to general perception.

Purpose of the Study:

• To propose and investigate circular inference as the mechanism underlying bistable perception.
• To understand how prior beliefs interact with sensory evidence in hierarchical neural systems.

Main Methods:

• Developed a computational model based on circular inference in hierarchical neural networks.
• Simulated the model's response to ambiguous stimuli and analyzed its dynamics.

Main Results:

• The circular inference model successfully replicates key aspects of bistable perception, including Levelt's laws.
• Demonstrated that circular inference can transform perceptual integrators into bistable attractors.
• Showed the model's ability to predict aberrant belief formation from bistable perception behavior.

Conclusions:

• Bistable perception arises from circular inference, where feedback loops in the cortical hierarchy influence perception.
• This framework links perceptual inference to belief updating and aberrant belief formation.
• Feedforward/feedback loops may underlie both normal perception and psychotic symptoms when dysregulated.