Cognitive Neurobiology Talk

20.03.2026

Towards Linking Neural Mechanism and Behaviour

Abstract:
A central goal in neuroscience is to link neural mechanisms to behavior. Progress requires both a principled understanding of neural computation and experimental paradigms that capture behavior under flexible conditions. In this talk, we present approaches that address both challenges. To gain deeper insight into neural computation, we study recurrent neural networks (RNNs), a powerful yet often difficult-to-interpret model. We introduce a method that links distinct functional modules to specific subspaces of the network connectivity. Using this approach, we show that only a low-dimensional subspace of the typically high-dimensional connectivity is functionally relevant, whereas large subspaces of connectivity can be removed without affecting task-related dynamics or performance. These insights simplify further analysis and provide a deeper understanding of how the observed dynamics emerge from the underlying network connectivity. In the second part, we explore the methodological challenge of studying behavior in socially housed, freely moving non-human primates. Using an in-cage training system that allows rhesus macaques to participate in experiments within their home enclosure, we systematically compare different training conditions. Our results show that lower reward values promote more balanced participation among group members without compromising task performance. These results highlight the importance of careful environmental and reward design to optimize engagement and minimize social tension in social settings. Overall, these studies advance two essential components for bridging neural mechanisms and behavior: mechanistic insight into neural computation and refined behavioral methodology.

Speaker:
Dr. Renate Krause (Institute of Neuroinformatics, University of Zurich and ETH Zurich)

Host:
Cognitive Neurobiology (CNB) Department (RUB)

Venue:
MB 7/159

Starting time:
4:00-5:30 PM