A subsequent study by Fair et al. (Fair et al., 2009) used spring embedding layouts to examine the development of a network composed of 34 published default, task control, and error-responsive ROIs (Dosenbach et al., 2007; Dosenbach et al., 2006; Fair et al., 2007a; Fox et al., 2005). Nodes, edges, and subjects were defined as before. Figure 5 shows layouts of the graphs at several ages. Two striking patterns are noted. First, nodes that are anatomically proximal tend to have strong correlations in children, which weaken over development (the light-blue-rimmed frontal lobe nodes of Figure 5A). Second, nodes that are functionally related in adults are not strongly associated in children, but come to associate strongly over development (e.g. the red-filled DMN nodes of Figure 5B). The patterns of edges clearly change across ages, and it is evident that the network is clumped in certain areas where nodes are densely connected, indicating that the network possesses community structure.
