Neurocomputational Underpinnings of Expected Surprise

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Psychiatry

Background:

• Predictive coding theories explain perception but face a paradox regarding surprise responses.
• This paradox involves how prediction error signals change with predictability and context.
• Understanding this is crucial for conditions like psychosis and cognitive inflexibility.

Purpose of the Study:

• To resolve the paradox of prediction error signaling in perception.
• To investigate the role of precision weighting in modulating brain responses to surprise.
• To explore the neurophysiological underpinnings of context-sensitive perception.

Main Methods:

• Employed a novel experimental paradigm with unexpected and expected surprising sounds.
• Combined electroencephalography (EEG) and magnetoencephalography (MEG) for high temporal and spatial resolution.
• Utilized trial-by-trial modeling and dynamic causal modeling (DCM) to analyze brain responses.

Main Results:

• Brain responses reflect precision-weighted prediction error, resolving the paradox.
• Demonstrated adaptive learning of surprise, with increased integration of past information for expected surprises.
• Identified precision encoding via neuronal excitability in the auditory hierarchy, even during unattended tasks.

Conclusions:

• Perceptual learning is context-sensitive, adapting to sound predictability.
• Precision weighting links to neuronal self-inhibition, offering insights into healthy and pathological perception.
• Findings have direct implications for understanding and potentially treating psychiatric disorders like schizophrenia and autism.