Large-Scale Cortical Networks for Hierarchical Prediction and Prediction Error in the Primate Brain

Area of Science:

• Neuroscience
• Computational Neuroscience
• Auditory Perception

Background:

• Predictive coding theory posits hierarchical cortical processing of sensory information.
• Distinguishing neural signals for predictions versus prediction errors is challenging due to their interdependence.

Purpose of the Study:

• To investigate the neural implementation of hierarchical predictive coding in the auditory system.
• To differentiate the brain networks and oscillatory dynamics underlying prediction errors and prediction updates.

Main Methods:

• Utilized high-density electrocorticography (ECoG) in monkeys.
• Employed an auditory "local-global" paradigm with controlled temporal regularities at two hierarchical levels.
• Decomposed broadband ECoG data to identify neural signals and their frequency bands.

Main Results:

• Identified lower-level prediction-error signals in early auditory cortex and higher-level signals in anterior temporal cortex.
• Detected a prediction-update signal originating from the prefrontal cortex and projecting to the temporal cortex.
• Prediction-error signals were associated with gamma (γ >40 Hz) oscillations, while prediction-update signals involved alpha/beta (α/β <30 Hz) oscillations.

Conclusions:

• Findings provide strong evidence for hierarchical predictive coding in the brain.
• Demonstrated dynamic implementation of predictive coding through distinct cortical areas and specific frequency bands.
• The study elucidates the neural mechanisms of sensory prediction and error processing.