Abstract

Over the past twenty years network approaches have proved invaluable in helping interpret the large, increasingly complex, multivariate datasets in neuroscience. This shift towards systems-level thinking has shed light on the organization of macro-scale brain networks derived from neuroimaging, as well as micro-scale neural circuits, and even molecular-scale networks of protein interaction or gene co-expression in the brain. In each case, the network-level features of these systems has proved to be intimately linked to their function. This enabled, for example, network analyses to pinpoint key nodes or meso-scale structures in the network. In this talk I argue that moving beyond network structure to incorporate dynamical processes supported by these networks unlocks deeper, often more mechanistic insights. I will illustrate this point with several strands of work, including (i) control theoretic approaches in C. elegans neural networks and in the human brain, and (ii) a reservoir computing framework to link the organization of macro and micro-scale brain connectivity to cognitive function.