Topic: Hidden Markov models

This webinar addressed computational perspectives on how animals and humans make decisions, spanning normative, descriptive, and mechanistic models. Sam Gershman (Harvard) presented a capacity-limited reinforcement learning framework in which policies are compressed under an information bottleneck constraint. This approach predicts pervasive perseveration, stimulus‐independent “default” actions, and trade-offs between complexity and reward. Such policy compression reconciles observed action stochasticity and response time patterns with an optimal balance between learning capacity and performan

Understanding how internally driven cognitive processes (COPs) give rise to behavior in different species is one of the key challenges in systems neuroscience. Traditionally, neuroscience has aimed to isolate individual COPs by simplifying sensory stimuli and behavioral readouts. Yet in natural behavior, several COPs can co-occur, often with overlapping neural representations. The dynamics of such superimposed COPs are yet to be understood, as well as the extent to which they generalize across species. To tackle this question, we developed a naturalistic Virtual Reality setup in which two spec

The hippocampus and the medial prefrontal cortex (mPFC) participate in episodic and working memory, respectively. In trace conditioning, animals learn to associate a sensory stimulus (CS) and a reward (US) separated by a delay (trace period). Although the neural activity in both areas can be linked to memory representation formation, the neural mechanisms involved in learning the temporal association between CS and US are not known. In this work, we use auditory trace conditioning in head-fixed mice with simultaneous 128-channel silicon probe recordings from the hippocampal CA1 subfield and th