Avalanches in directed complex networks of neurons connected by stochastic synapses

TL;DR

This paper investigates how stochastic synapses in directed neural networks lead to phase transitions and avalanches, with activity patterns characterized by exponential size and duration distributions near criticality.

Contribution

It introduces a model of directed neural networks with stochastic synapses and analyzes avalanche behavior and phase transitions within this framework.

Findings

Avalanches occur after random neuron excitation near phase transition.

Avalanche size and duration follow exponential distributions.

Only a small fraction of neurons are active during long avalanches.

Abstract

Directed complex network of two-state model neurons linked by synapses which can be blocked or activated stochastically in time undergoes phase transition between the quiescent phase with zero activity and the active one with persistent activity. Avalanches of activity appear in the quiescent phase after occasional excitation of randomly chosen neuron. Size and duration distributions of long avalanches which we observe near phase transition point are exponential. Only a small fraction of all neurons are active at any moment of a long avalanche.