Modelling mouse response to dynamics environments via recurrent neural networks

Abstract

To survive in dynamic environments, the nervous system must be able to generate flexible behavior, seamlessly weaving together past experience with the present context to achieve future goals. This project aims at revealing the neural mechanisms by which a mouse brain engages in specific processing of auditory vs. visual stimuli based on task demands. We hypothesize that these types of context-dependent behaviors operate through a flexible coupling and decoupling of neural networks mediated by changes in neural dynamics based on gain modulation, similar to mechanisms that are engaged in brain state regulation. Combining data analysis of whole-brain single cell recordings in behaving mice and theoretical modeling based on recurrent neural networks, this project will elucidate how the cortex flexibly reconfigures its functional interactions to produce contextually-appropriate behavior.

  • In collaboration with prof. Luca Mazzucato, University of Oregon, Eugene
Avatar
LIPh
Laboratory of Interdisciplinary Physics

Related