One dimensional Fokker-Planck reduced dynamics of decision making models in computational neuroscience (Q2443504)

The considered neuronal model consist of two interacting families of neurons where the time evolution for the firing rates is described by a system of two stochastic differential equations (SDE). The activation function is the sigmoid function. The system has been studied for instance in [\textit{G. Deco} and \textit{D. Martí}, Biol. Cybern. 96, No. 5, 487--496 (2007; Zbl 1122.92008)]. Some values for parameters introduced in this article will be used in numerical simulations of the present paper. There is a discussion first on how the decision making process for the neuronal network can be alternatively studied by means of the evolution of a 2D Fokker-Planck equation. The aim of the present study is to get an approximation of the 2D Fokker-Planck equation by a 1D Fokker-Planck reduced dynamics. The one dimensional reduction of the initial stochastic differential system is done in three steps: first the deterministic part is treated , then the stochastic part and finally the one dimensional Fokker-Planck model is obtained. The solutions for the one dimensional reduced Fokker-Planck equation are compared -- numerical simulation -- to the solutions of the original 2D model and the results are largely discussed. At the end of the article, the time behavior of the solution of the 1D Fokker-Planck equation as well as the reaction time and performance are studied.