Perceptual Inference: A Matter of Predictions and Errors

Area of Science:

• Neuroscience
• Cognitive Science
• Perception Research

Background:

• Predictive coding theories propose the brain constantly generates predictions about sensory input.
• Deviations from these predictions generate prediction errors, crucial for updating internal models.

Purpose of the Study:

• To investigate the neural mechanisms underlying predictive coding in the human brain.
• To determine if distinct neuronal populations encode perceptual predictions and prediction errors.

Main Methods:

• Utilized functional magnetic resonance imaging (fMRI) to monitor brain activity.
• Analyzed neural responses in the inferotemporal cortex during a visual perception task.

Main Results:

• Identified distinct neural populations within the inferotemporal cortex.
• One population selectively responded to predicted stimuli.
• Another population specifically encoded prediction errors.

Conclusions:

• The findings provide neural evidence for the segregation of prediction and prediction error signals.
• Supports the role of the inferotemporal cortex in implementing predictive coding frameworks.
• Suggests a neural basis for how the brain learns and adapts based on sensory evidence.