• Home
• Staff & Students
• Vacancies

Daniel Durstewitz

 

 

http://www.zi-mannheim.de/en/research/departments/psychiatrie/arbeitsgruppen-psychiatrie/compnw.html

 

Wednesday 20th March 2013

Time: 4pm

 

Basement Seminar Room

Alexandra House, 17 Queen Square, London, WC1N 3AR

 

 

Neuronal Dynamics During Working Memory and Decision Making Reconstructed from Single-Trial Multiple-Unit Recordings

 

 

Speaker:
Daniel Durstewitz, Bernstein Center for Computational Neuroscience, Central Institute of Mental Health, Mannheim, Heidelberg University

A common idea in computational neuroscience is that cognitive processes are implemented in terms of the neuronal dynamics, e.g. transitions among attractor-like states. Such ideas are particularly popular in the study of higher-level (goal-directed) cognitive processing, like the active maintenance of goal-related memory items and their use in guiding choices. This is because these processes rely only partly on external sensory input but rather are assumed to evolve from internal recurrent dynamics. With the recent progress in multiple single-unit (MSU) recording techniques, ideas regarding the computational dynamics underlying higher-level cognition could be addressed more directly, potentially with single-trial resolution. In my talk I will discuss major results from our statistical analyses of MSU recordings obtained during an ecologically valid multiple-item working memory and decision making task in rodents that involves foraging on a radial arm maze.
Using tools from multivariate statistics, statistical learning, and nonlinear dynamics, we attempted to reconstruct neuronal processing and dynamics from single or only a few trials. The results suggest that task processing is characterized by transitions through unique neural population states with attractor-like properties, which break down under certain behavioral or pharmacological conditions, and that neural trajectories contain information about visited maze arms that is carried through delay phases of tens of seconds. The dynamical systems picture emerging from this work will be discussed.