Signaling events in brain networks unfold over multiple topological scales. Areas may exchange information over local circuits, primarily encompassing direct neighbours and areas with similar functions. Alternatively, areas may exchange information over global circuits, encompassing more distant neighbours with increasingly dissimilar functions. In the present report, we study communication in cortico-cortical networks by characterizing a region’s structural embedding over a continuous range of topological scales. We find that the centrality of a brain region varies across scales and that connection diversity determines scale preference, with less diverse unimodal regions showing preference for local communication and more diverse multimodal regions showing preferences for global communication. These preferences manifest as scale-specific structure-function relationships, with unimodal areas showing optimal coupling at local scales and multimodal regions showing optimal coupling at global scales. Altogether, the present findings reveal how functional hierarchies emerge from hidden but highly structured multiscale connection patterns.
bioRxiv Subject Collection: Neuroscience