From regions to connections and networks: new bridges between brain and behavior

•Brain-behavior relations have traditionally been studied with univariate approaches.•Recent studies have revealed the modular architecture of brain networks.•Task demands reconfigure brain networks and break modular boundaries.•Brain-behavior can be directly accessed with the use of multivariate methods.•Multivariate tools can usefully complement graph-based network approaches.

Connections and interactions among distributed brain areas are increasingly recognized as the basis for cognitive operations and a diverse repertoire of behaviors. Analytic advances have allowed for brain connectivity to be represented and quantified at multiple levels: from single connections to communities and networks. This review traces the trajectory of network neuroscience, focusing on how connectivity patterns can be related to cognition and behavior. As recent initiatives for open science provide access to imaging and phenotypic data with great detail and depth, we argue that approaches capable of directly modeling multivariate relationships between brain and behavior will become increasingly important in the field.