"What" and "when" predictions modulate auditory processing in a mutually congruent manner.

Area of Science:

• Neuroscience
• Cognitive Science
• Auditory Perception

Background:

• Predictive processing is vital for behavior, integrating stimulus content ('what') and timing ('when').
• Real-world stimuli, like music, involve multi-level predictions (e.g., notes, beats) that interact.
• It remains unclear if the brain integrates these predictions congruently or uses distinct neural mechanisms.

Purpose of the Study:

• To investigate if 'beat' and 'interval' timing predictions interact with 'what' predictions.
• To determine if integrating different timing predictions involves separable neural correlates.
• To explore the neural basis of integrating multi-level predictions in sensory streams.

Main Methods:

• Manipulated 'what' and 'when' predictions at local (interval) and global (beat) levels.
• Recorded neural activity using electroencephalogram (EEG) during a repetition detection task.
• Employed dynamic causal modeling for effective connectivity analysis.

Main Results:

• Temporal predictions (beat/interval) modulated responses to 'what' prediction violations.
• Modulations of 'what' predictions by timing were shared across EEG spatiotemporal distributions.
• Integration of 'what' and 'when' predictions increased connectivity between the superior temporal gyrus and fronto-parietal network.

Conclusions:

• The brain integrates diverse predictions with high congruence within a shared cortical network.
• This contrasts with theories suggesting separate mechanisms for beat-based and memory-based predictions.
• Findings highlight a unified neural basis for integrating multi-level predictions in sensory processing.