Model-Agnostic Neural Mean Field With The Refractory SoftPlus Transfer Function.

animal models, so the problem of modeling the behavior of large-scale neuronal systems is more relevant than ever. The statistical physics concept of a mean-field model offers a tractable approach by modeling the behavior of a single representative neuron and extending this to the population. However, most mean-field models work only in the limit of weak interactions between neurons, where synaptic input behaves more like a diffusion process than the sum of discrete synaptic events. This paper introduces a data-driven mean-field model, estimated by curve-fitting a simple transfer function, which works with larger interaction strengths. The resulting model can predict population firing rates and bifurcations of equilibria, as well as providing a simple dynamical model that can be the basis for further analysis.