A modularity-optimizing community detection algorithm was applied to this network, and strong community structure was detected at all ages examined, consistent with the spring embedded layouts. Interestingly, the modularity value attained by the algorithm did not change (qualitatively) over age, indicating that the “quality” of community structure was similar (and high) in children, adolescents, and adults. Importantly, however, the composition of modules did change over development. In children, modules grouped nodes largely by anatomical location, whereas modules in adults grouped nodes almost perfectly by their functional role (default, fronto-parietal, cingulo-opercular, and cerebellar error categories).
