Probabilistic Representation in Human Visual Cortex Reflects Uncertainty in Serial Decisions.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Decisions rely on uncertain sensory information, yet the neural basis of uncertainty is poorly understood.
• Bayesian models propose sensory uncertainty is encoded as probability distribution width.
• Serial dependence is a behavioral bias linked to sensory integration.

Purpose of the Study:

• To investigate if cortical population activity encodes sensory uncertainty as probability distribution width.
• To test the role of sensory uncertainty in a known behavioral bias (serial dependence).
• To provide evidence for Bayesian theories of neural coding and perception.

Main Methods:

• Functional magnetic resonance imaging (fMRI) to measure visual cortex activity.
• Decoding probability distributions from neural population activity.
• Behavioral experiments measuring orientation judgments with serial dependence.

Main Results:

• Serial dependence effects align with statistically advantageous sensory integration, weighting uncertain information less.
• Decoded probability distributions from brain activity reflect sensory uncertainty used in decisions.
• Neural representation of uncertainty matches predictions of Bayesian models.

Conclusions:

• The brain encodes sensory uncertainty as the width of probability distributions in cortical activity.
• This neural representation supports rational decision-making under uncertainty.
• Findings provide critical evidence for Bayesian theories in perception and decision-making.