Modeling and Simulation Group Meeting Old

Abstract: Spiking neuronal networks have been used to model brain dynamics for a long time, yet they are hard to analyze due to the curse of dimension. Many previous model reduction methods focused on coarse-graining weakly-coupled networks, hence hard to capture highly-correlated dynamics between neurons. In this talk, I will introduce a sequence of reduced models to capture the brain dynamics called Gamma oscillations, which exhibit highly spatiotemporal correlations. We find that Gamma oscillation can be mostly represented by a two-dimensional manifold in the space of reduced models, and the layout of the manifold explains many key features of the dynamics.