Task learning reveals neural signatures of internal models in rodent prefrontal cortex
Cortical population activity may represent sampling from an internal model. Using data from rats learning rules on a maze task, we show that sampling rates of population activity patterns in prefrontal cortex converge between waking and sleep over learning. Sample rate changes were greatest for activity patterns predictive of correct choice and expressed at the choice point. Our results suggest inference-by-sampling is a general computational principle for cortex.
Neuroscience↗