A neuronal model of predictive coding accounting for the mismatch negativity

Area of Science:

• Neuroscience
• Computational Neuroscience
• Auditory Perception

Background:

• Mismatch negativity (MMN) is linked to prediction and deviance detection.
• A detailed neuronal model for MMN mechanisms and computational basis is lacking.

Purpose of the Study:

• To propose a detailed neuronal model of auditory cortex based on predictive coding.
• To account for critical features of MMN using this model.

Main Methods:

• Developed a spiking neuronal network model of auditory cortex.
• Incorporated predictive coding principles with excitatory and inhibitory neurons.
• Utilized a spike-timing dependent learning rule based on NMDA receptor transmission.

Main Results:

• The model replicates key MMN properties: frequency-dependent responses, reaction to unexpected repeats, sound omission responses, and NMDA antagonist sensitivity.
• The model shows MMN does not consider global sequence context.
• Validated model predictions with a new magnetoencephalography experiment.

Conclusions:

• The proposed model successfully explains major MMN empirical properties.
• MMN generation is attributed to active cortical prediction, challenging passive habituation theories.
• The model provides a neurobiological basis for MMN and suggests future research directions.