Figure 6.

A graphical representation of the next-time Jacobian terms, $\left|\frac{\partial {\mu }_{i,t}}{\partial s_{j,t-1}}\right|$, averaged over time and subjects. The features were grouped by a multi-level community algorithm Blondel et al. (2008), using the Pearson correlation coefficient to define an undirected graph (see Equation 12). Corresponding ROIs are provided on the right, and the complete legend can be found in Figure 4. Grouping (and coloring) was done by constructing an undirected graph using the Pearson coefficients, clustering the vertices using a standard community-based hierarchical algorithm.