Keynotes | K&Ts | GACs | Talks | Posters | Search

Poster D in Poster Session D: Wednesday, August 5, 2:00 – 3:45 pm, Kimmel Center, Shorin & Rosenthal Rooms

A probabilistic model for context-dependent interval discrimination in mouse

Ali Shamsnia¹, Mengze Chen², Hüseyin Özkan³, Farzaneh Najafi²; ¹Sabanci University, ²Georgia Institute of Technology, ³Sabancı University

Presenter: Farzaneh Najafi

Adaptive decision-making in biological systems requires generating and updating predictions about the world. A key form of these predictions concerns the "when" of events, a process referred to as temporal predictive processing. How the brain forms and updates these temporal predictions remains unknown. To address this, we developed a blockwise interval discrimination task for mice and modeled their adaptive decision-making behavior by integrating temporal perception with latent beliefs about the environmental context. Within this framework, subjective time is modeled via a noisy sensory pathway consistent with scalar variability, featuring an internal accumulation process based on a gamma distribution. A parallel contextual branch estimates the latent block state using recursive Bayesian inference. These streams dynamically integrate to form a decision variable that governs probabilistic action selection. Furthermore, the model incorporates an uncertainty-modulated learning rule where trial-by-trial plasticity is driven by the variance of the context belief. Estimated through gradient-based maximum likelihood optimization enabled by the reparameterization trick, this interpretable framework provides a biologically plausible model of how neural systems achieve temporal predictive processing under uncertainty.

Topic Area: Decision-Making, Cognitive Control & Event Cognition