A principled approach to community detection in interareal cortical networks

Structural connectivity between cortical areas, as revealed by tract-tracing is in the form of highly dense, weighted, directed, and spatially embedded complex networks. Extracting the community structure of these networks and aligning them with brain function is challenging, as most methods use local density measures, best suited for sparse graphs. Here we introduce a principled approach, based on distinguishability of connectivity profiles using the Hellinger distance, which is relatable to function. Applying it to tract-tracing data in the macaque, we show that the cortex at the interareal level is organized into a hierarchy of link-communities alongside with a node-community hierarchy. We find that the ½-Rényi divergence of connection profiles, a non-linear transform of the Hellinger metric, follows a Weibull-like distribution and scales linearly with the interareal distances, a quantitative expression between functional organization and cortical geometry. We discuss the relationship with the extensively studied SLN-based hierarchy.