Inferring brain-wide interactions using data-constrained recurrent neural network models

Behavior arises from the coordinated activity across anatomically and functionally distinct brain regions. Modern experimental tools allow unprecedented access to large neural populations spanning many interacting regions brain-wide. Yet, understanding such large-scale datasets necessitates robust, scalable computational models to extract meaningful features of inter-region communication and principled theories to interpret those features. Here, we introduce current-based decomposition (CURBD), an approach for inferring brain-wide interactions using data-constrained recurrent neural network models that autonomously produce dynamics consistent with experimentally obtained neural data. CURBD leverages the functional interactions inferred from such models to reveal directional currents between multiple brain regions simultaneously. We first show that CURBD accurately isolates inter-region currents in simulated, ground-truth networks with known connectivity and dynamics. We then apply CURBD to multi-region neural recordings obtained from many species-larval zebrafish, mice, macaques, and humans-to demonstrate the widespread applicability of CURBD in untangling brain-wide interactions and inter-area communication principles underlying behavior.