Talk Title: Sensory cortical computations for learning

Talk Abstract: 

Learning about sensory cues from the environment is essential for animal survival. The sensory cortex is thought to subserve this learning by increasing the saliency of behaviorally relevant cues at the timescale of behavioral improvements. However, reports suggest that the acquisition of task knowledge may occur faster than performance improvements, challenging the stimulus-related plasticity model of sensory cortical plasticity. To investigate the relationship between learning and sensory cortical representations, I trained mice on a novel behavioral paradigm that dissociated the rapid acquisition of task contingencies (learning) from its slower expression (performance). I combined longitudinal two-photon imaging in the auditory cortex to track thousands of neurons across weeks, cutting-edge tensor decomposition algorithms to reveal latent dynamics at multiple timescales, and behaviorally closed-loop optogenetics. Rather than enhancement or expansion of cue representations, we found that the sensory cortex manifests higher-order computations, reward prediction and action suppression, that separably drive rapid learning and slower performance improvements, respectively. These results reshape our understanding of the fundamental role of the sensory cortex. At the end of my seminar, I will briefly discuss my future directions focusing on investigating multi-region neural dynamics that underlie memory-guided behaviors.

This talk will not be live streamed