Stimulus-Specific Prediction Error Neurons in Mouse Auditory Cortex.

Area of Science:

• Neuroscience
• Auditory Processing
• Predictive Coding

Background:

• Predictive processing relies on comparing expectations with sensory experiences.
• Neural responses to unexpected self-generated stimuli suggest sensory cortex neurons may encode prediction errors (PEs).
• It remains unclear if PEs are general or specific to stimulus dimensions, and if they are distinct from movement-related neural facilitation.

Purpose of the Study:

• To investigate whether auditory cortex neurons encode specific prediction errors (PEs) related to self-generated sounds.
• To determine if PEs are dimension-specific or general error signals.
• To identify the neural mechanisms underlying prediction error signaling in the auditory cortex.

Main Methods:

• Mice were trained to associate a motor behavior with an expected auditory outcome.
• Auditory cortex activity was recorded while mice heard expected sounds or sounds deviating in specific acoustic dimensions.
• Neural responses were analyzed to identify neurons encoding prediction errors.

Main Results:

• The auditory cortex demonstrated suppression of responses to self-generated sounds across multiple acoustic dimensions.
• A distinct population of auditory cortex neurons, typically unresponsive to passive sounds, was identified.
• These neurons specifically encoded prediction errors, responding to one or two distinct violations of expectation, not a generic error signal.
• Prediction error neurons were abundant only in mice with learned motor-sensory expectations.

Conclusions:

• Cortical predictions regarding self-generated sounds exhibit specificity across multiple simultaneous dimensions.
• Cortical prediction error neurons encode specific violations of learned expectations.
• These findings suggest that prediction error signals are calculated early in sensory processing within the auditory cortex.