Decision and Behavior
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 performance. Jonathan Pillow (Princeton) discussed flexible descriptive models for tracking time-varying policies in animals. He introduced dynamic Generalized Linear Models (Sidetrack) and hidden Markov models (GLM-HMMs) that capture day-to-day and trial-to-trial fluctuations in choice behavior, including abrupt switches between “engaged” and “disengaged” states. These models provide new insights into how animals’ strategies evolve under learning. Finally, Kenji Doya (OIST) highlighted the importance of unifying reinforcement learning with Bayesian inference, exploring how cortical-basal ganglia networks might implement model-based and model-free strategies. He also described Japan’s Brain/MINDS 2.0 and Digital Brain initiatives, aiming to integrate multimodal data and computational principles into cohesive “digital brains.”
Experience dependent changes of sensory representation in the olfactory cortex
Sensory representations are typically thought as neuronal activity patterns that encode physical attributes of the outside world. However, increasing evidence is showing that as animals learned the association between a sensory stimulus and its behavioral relevance, stimulus representation in sensory cortical areas can change. In this seminar I will present recent experiments from our lab showing that the activity in the olfactory piriform cortex (PC) of mice encodes not only odor information, but also non-olfactory variables associated with the behavioral task. By developing an associative olfactory learning task, in which animals learn to associate a particular context with an odor and a reward, we were able to record the activity of multiple neurons as the animal runs in a virtual reality corridor. By analyzing the population activity dynamics using Principal Components Analysis, we find different population trajectories evolving through time that can discriminate aspects of different trial types. By using Generalized Linear Models we further dissected the contribution of different sensory and non-sensory variables to the modulation of PC activity. Interestingly, the experiments show that variables related to both sensory and non-sensory aspects of the task (e.g., odor, context, reward, licking, sniffing rate and running speed) differently modulate PC activity, suggesting that the PC adapt odor processing depending on experience and behavior.