Fig. 3
From: Symmetry breaking organizes the brain’s resting state manifold
Network dynamics. (A) The network input I modulates the probability of a noise-driven transition between the down- and up-state by increasing the basin on the attraction of the up-state (blue area) in the phase-plane of firing rate r and membrane potential v with nullclines \(\dot{r} = 0\) and \(\dot{v} = 0\) shown in red and green respectively. (B) Example of a cascade—coordinated increase in activity translating to a delayed correlated peak in the BOLD signal. Below we compare the network dynamics in and outside the working point, and the empirical data. In both empirical data and the working point (\(G=0.54\)), the BOLD co-activations follow the neuronal cascades of firing rate r (simulated) and EEG (empirical) (C), and show distinct spatial profiles which are recurrent in time (D): edge time series on the (top panel) captures the spatial profiles of the co-activations, the similarity across time is captured by the \(dFC_e\) matrix (middle panel), and the distributions of correlation between co-activation events (CA) and non-events (nCA) is compared (bottom panel).
