APICAL DENDRITES OF DEEP-LAYER PREFRONTAL CORTEX NEURONS DRIVE BEHAVIORAL FLEXIBILITY

Learning opportunities arise when rewards are unexpectedly obtained or withheld. The prefrontal cortex is thought to guide goal-directed behavior based on these prediction errors in order to maximize behavioral outcomes. However, it remains poorly understood how the layered neural circuitry of the cortex integrates past experience with current behavioral outcome to produce optimal decision strategies. We performed dual Neuropixels recordings, capturing hundreds of neurons in the dorsomedial prefrontal cortex (dmPFC) and the anterolateral motor cortex (ALM) while mice performed a head-fixed naturalistic foraging task. Animals were trained to infer stochastic, hidden reward contingencies from trial outcome history. This design allowed us to derive precise estimates of the animal's reward expectations on a trial-by-trial basis - a variable that is difficult to access in most behavioral paradigms. We find that dmPFC and ALM encode a rich representation of past experience in a layer- and area-specific manner. Notably, the apical dendrites of deep-layer pyramidal neurons in dmPFC, but not ALM, contribute to behavioral flexibility.