Figure 2.

Schematic depictions of cognitive control network functional properties (details on network measures, see Materials and Methods and Results). In each panel, a “toy” version of the control network is prominently depicted in the center (with a reduced number of regions, or nodes, and simplified within-network connections), and out-of-network exemplars are depicted as truncated and surrounding the control network of interest (DAN: dorsal attention network, LAN: language network, DMN: default mode network, and SMN: somatomotor network). Each of these surrounding networks also contains within-network regions and connections, but these were not depicted here for simplicity. A, Regions in the FPN acted as flexible coordinators. This entailed high GVC and low partition deviation across task states. From example task state one to task state two, FPN regions maintained their within-network connectivity (low deviation) and out-of-network connectivity changes were variable across states (high GVC; for details, see Results). B, Regions in the CON acted as flexible switchers. This entailed low global variability and high partition deviation. From example task state one to task state two, CON regions dropped their within-network connectivity (high deviation) and out-of-network connectivity changes were consistent across states (low GVC; for details, see Results).