Constructing the hierarchy of predictive auditory sequences in the marmoset brain

Area of Science:

• Neuroscience
• Auditory Processing
• Predictive Coding

Background:

• The brain constantly predicts sensory input, comparing it to actual input to generate prediction errors.
• The hierarchical nature and neural mechanisms of predictive coding are not well understood.

Purpose of the Study:

• To investigate the hierarchical depth of predictive auditory processing.
• To explore the neural mechanisms underlying prediction errors and updates in auditory sequences.

Main Methods:

• Combined functional magnetic resonance imaging (fMRI) and electrocorticography (ECoG) in marmoset monkeys.
• Utilized an auditory local-global paradigm with stimuli at two hierarchical temporal levels.
• Examined neural responses to auditory mismatches and omissions.

Main Results:

• Identified a hierarchical gradient in auditory processing: midbrain/sensory regions for local (short-time) predictions, and anterior temporal/prefrontal areas for global (long-time) predictions.
• ECoG differentiated prediction-error signals (bottom-up γ oscillations) and prediction-update signals (top-down β oscillations).
• Omission responses confirmed hierarchical top-down prediction processing in auditory, temporal, and prefrontal areas.

Conclusions:

• Findings support the hierarchical predictive coding framework.
• Demonstrated how neural networks and spatiotemporal dynamics represent hierarchical auditory sequences in the marmoset brain.