[ Seminar ] Durham–OIST Joint Seminar

Toward partnership between Durham University and OIST, we are organizing a joint hybrid seminar.

Talk 1: Naohiro Yamauchi

Neural Computation Unit, OIST

Title: Neural implementation of Bayesian inference and control as inference in the primary somatosensory and primary motor cortices addressed by a variable resistance lever push-pull task

Abstract: While decision-making and motor control under noise are often modeled via Bayesian inference and reinforcement learning, cortical layer-specific representations of underlying variables remain unclear.

Using a roving oddball paradigm with alternating lever resistance, we found that S1 activity increased for unexpectedly light resistance but was suppressed for unexpectedly heavy resistance. In a separate lever-pulling task, M1 showed a higher proportion of activated neurons during high-reward trials, reflecting action values.

Furthermore, we developed a motorize lever task where mice integrated prior probabilities and sensory evidence. Calcium imaging in S1 showed that prior information is more robustly decoded from layer 5 than layer 2/3. To clarify the key variables, we are developing a computational model that estimates the direction of least resistance and controls motor output. Our preliminary results qualitatively reproduce psychometric curves and “change of mind” behavior, offering a framework to understand the implementation of the duality of inference and control.

Talk 2: Dr. Marco Bocchio

Department of Psychology and Learning and Memory Processes (LAMP) Centre,
Durham University

Title: Circuit mechanisms shaping cell assemblies recruitment in the hippocampus

Abstract: Consistently co‑active neurons are often conceptualised as cell assemblies and are thought to be fundamental building blocks of learning and memory, particularly in the hippocampus. However, how specific assemblies are recruited within local circuits remains poorly understood. In this mini‑talk, I will present in vivo calcium imaging data from the mouse hippocampus together with computational modelling to explore the circuit mechanisms underlying cell assembly recruitment. The results highlight a previously underappreciated role for inhibitory interneurons and specific inhibitory circuit motifs in determining assembly recruitment. Together, these findings suggest that inhibitory circuitry is a key driver of structured population activity in the hippocampus, with important implications for how memory‑related representations emerge and are controlled at the circuit level.

Zoom Information

https://oist.zoom.us/j/92359181486?pwd=BUQMJFvSbgusrlFFd3SmjwcoolWe5A.1

Meeting ID: 923 5918 1486
Passcode: 451179

Host: Kenji Doya
Neural Computation Unit
Contact: ncus@oist.jp