RT Journal Article
SR Electronic
T1 Non-assortative community structure in resting and task-evoked functional brain networks
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 355016
DO 10.1101/355016
A1 Richard F. Betzel
A1 Maxwell A. Bertolero
A1 Danielle S. Bassett
YR 2018
UL http://biorxiv.org/content/early/2018/06/25/355016.abstract
AB Brain networks exhibit community structure that reconfigures during cognitively demanding tasks. Extant work has emphasized a single class of communities: those that are assortative, or internally dense and externally sparse. Other classes that may play key functional roles in brain function have largely been ignored, leading to an impoverished view in the best case and a mischaracterization in the worst case. Here, we leverage weighted stochastic blockmodeling, a community detection method capable of detecting diverse classes of communities, to study the community structure of functional brain networks while subjects either rest or perform cognitively demanding tasks. We find evidence that the resting brain is largely assortative, although higher order association areas exhibit non-assortative organization, forming cores and peripheries. Surprisingly, this assortative structure breaks down during tasks and is supplanted by core, periphery, and disassortative communities. Using measures derived from the community structure, we show that it is possible to classify an individual’s task state with an accuracy that is well above average. Finally, we show that inter-individual differences in the composition of assortative and non-assortative communities is correlated with subject performance on in-scanner cognitive tasks. These findings offer a new perspective on the community organization of functional brain networks and its relation to cognition.