Goal-directed learning involves creating an internal model linking relevant stimuli, possible actions, and their consequences. These models must be flexible to accommodate new associations and unexpected stimuli encountered in the environment. Perirhinal cortex (Prh) is a structure heavily interconnected with sensory neocortex and is part of the medial temporal lobe which is implicated in learning and memory. Here, the functional changes at this interface between incoming external information and internal representations of the world were investigated as mice learned a complex sensory task. First, a system was developed to enable unbiased, high-throughput training of mice on such a task via automated training in the home cage environment. This system was then combined with chemogenetics to confirm the involvement of Prh in abstract sensory learning. Chronic two-photon calcium imaging, population analysis, and computational modeling showed evidence of sensory prediction error signals as well as stable, yet flexible, stimulus-outcome associations in Prh. This also revealed that the population encodes expected outcomes which are linked to these associations via cholinergic signaling, demonstrated by acetylcholine imaging and perturbation. These findings suggest Prh participates in updating an internal model of the task by evaluating sensory stimuli and linking those sensations with subsequent events.
展开▼
