Relaxation and self-sustained oscillations in the time elapsed neuron network model

TL;DR

This paper analyzes the dynamics of a neuron network model based on the distribution of times since last firing, identifying conditions for relaxation to steady states or self-sustained oscillations, with explicit solutions and numerical validation.

Contribution

It provides a quantitative regime analysis for relaxation and oscillations in the time elapsed neuron model, including explicit periodic solutions and a nonlinear boundary condition approach.

Findings

Relaxation to steady state occurs in low or no connectivity regimes.

Self-sustained oscillations emerge in moderate connectivity regimes.

Explicit families of periodic solutions are constructed without bifurcation analysis.

Abstract

The time elapsed model describes the firing activity of an homogeneous assembly of neurons thanks to the distribution of times elapsed since the last discharge. It gives a mathematical description of the probability density of neurons structured by this time. In an earlier work, based on generalized relative entropy methods, it is proved that for highly or weakly connected networks the model exhibits relaxation to the steady state and for moderately connected networks it is obtained numerical evidence of appearance of self-sustained periodic solutions. Here, we go further and, using the particular form of the model, we quantify the regime where relaxation to a stationary state occurs in terms of the network connectivity. To introduce our methodology, we first consider the case where the neurons are not connected and we give a new statement showing that total asynchronous firing of neurons appears asymptotically. In a second step, we consider the case with connections and give a low connectivity condition that still leads to asynchronous firing. Our low connectivity condition is somehow sharp because we can give an example, when this condition is not fulfilled, where synchronous rhythmic activity occurs. Indeed, we are able to build several explicit families of periodic solutions. Our construction is fully nonlinear and the resynchronization of the neural activity in the network does not follow from bifurcation analysis. It relies on an algebraically nonlinear boundary condition that occurs in the model.These analytic results are compared with numerical simulations under broader hypotheses and shown to be robust.