Balanced Dynamics in Strongly Coupled Networks

Abstract: In the modeling of a variety of models of large-scale systems of interacting agents, mathematical models consider that the coupling terms representing the interactions between agents are scaled according to the network size to prevent divergence of the coupling terms. In a previous paper of ours and follow-ups from colleagues, we have explored the behavior of a large-scale system of interacting agents with diffusive coupling. It was shown that the system converges to a Dirac mass, that is free to evolve according to the intrinsic dynamics. We generalize these results to more general coupling functions, and show that, under specific conditions, the divergence of the coupling term does not preclude the well-posedness of limit equations. Instead, the system carries solutions associated to a balanced, so that the net input where leading order, vanishes. We also explore convergence to the balanced state. Such balanced regimes were widely observed in neuroscience and associated with complex regulatory mechanisms. This results offer an alternative, minimalistic perspective on the balance of excitation/inhibition in the brain.