The role of fluctuations in coarse-grained descriptions of neuronal networks (Q357514): Difference between revisions

The paper reviews recent results of the authors on the coarse-graining of neural network dynamics and addresses both synaptic-inputs as well as network-connectivity fluctuations. First, an all-to-all coupled excitatory I\&F neural network with the dynamics described by a set of differential equations in the membrane potential and excitatory conductance for each neuron is introduced. The Boltzmann equation governing the evolution of the probability density is deduced and the limiting case in which neither synaptic nor connectivity fluctuations are present is described. The mean-field equation is explicitly solved. Further, there is a presentation of the extension of kinetic theory to the case neural network takes synaptic-input fluctuations into account. The Boltzmann equation is simplified to become a (2+1)-dimensional nonlinear advection-diffusion equation, and, then, via the maximum entropy principle, the two dimension dynamics \((v, g)\) is reduced to the dynamics in one dimension \(v\). The (1+1)-dimensional kinetic equations governing the coupled excitatory neural networks are obtained. Recent work of authors on the bifurcations of steady-state gain curves described by the Fokker-Planck equation as well as interconnections between the kinetic-theoretic, Fokker-Planck and mean-field models are presented. Follow a discussion on how can be extended the I\&F network model from an all-to-all coupled to a heterogeneous coupling architecture and how can be derived the corresponding generalizations of Boltzmann equation and the mean-field system. Neuronal firing rates as well as the statistics of these firing rates are presented.