A predictive coding account of bistable perception - a model-based fMRI study.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Psychology

Background:

• Bistable perception involves fluctuating subjective awareness between two interpretations of a constant, ambiguous stimulus.
• The underlying neurocomputational mechanisms driving spontaneous perceptual transitions remain poorly understood.
• Existing models often focus on mutual inhibition, adaptation, or noise, but a unified framework is lacking.

Purpose of the Study:

• To develop and validate a predictive coding model for understanding endogenous perceptual transitions in bistable vision.
• To investigate the role of prediction errors in mediating shifts in subjective perception.
• To correlate model predictions with behavioral data and neural activity from fMRI experiments.

Main Methods:

• A Bayesian predictive coding framework was employed to model perceptual transitions.
• Simulations were used to assess the model's ability to reproduce key characteristics of bistable perception.
• The model was fitted to behavioral data and analyzed alongside fMRI data from participants experiencing bistable perception.

Main Results:

• The predictive coding model successfully simulated temporal dynamics of perceptual bistability.
• Model parameters correlated with individual differences in the frequency of perceptual transitions.
• Prediction error signals derived from the model correlated with neural activity in the inferior frontal gyri and anterior insulae.

Conclusions:

• The predictive coding framework offers a viable explanation for endogenous perceptual transitions in bistable vision.
• Prediction error signaling in frontal cortical areas appears crucial for resolving perceptual ambiguity.
• This approach provides a unified theoretical and analytical framework for studying bistable perception.