Avalanches of Activation and Spikes in Neuronal Complex Networks

TL;DR

This paper investigates how network topology influences neuronal activation avalanches, using a mean-field model based on hierarchical degrees, with applications to theoretical and real-world networks like C. elegans.

Contribution

It introduces a mean-field model based on hierarchical degrees to analyze activation avalanches and their timing in neuronal networks, linking topology to dynamics.

Findings

Hierarchical degrees determine avalanche intensity and timing.

Model accurately predicts activation onset in various networks.

Avalanches may be universal across different network sizes.

Abstract

As shown recently (arXiv:0801.3056), several types of neuronal complex networks involving non-linear integration-and-fire dynamics exhibit an abrupt activation along their transient regime. Interestingly, such an avalanche of activation has also been found to depend strongly on the topology of the networks: while the Erd\H{o}s-R\'eny, Barab\'asi-Albert, path-regular and path-transformed BA models exhibit well-defined avalanches; Watts-Strogatz and geographical structures present instead a gradual dispersion of activation amongst their nodes. The current work investigates such phenomena by considering a mean-field equivalent model of a network which is strongly founded on the concepts of concentric neighborhoods and degrees. It is shown that the hierarchical number of nodes and hierarchical degrees define the intensity and timing of the avalanches. This approach also allowed the identification of the beginning activation times during the transient dynamics, which is particularly important for community identification (arXiv:0801.4269, arXiv:0801.4684). The main concepts and results in this work are illustrated with respect to theoretical and real-world (\emph{C. elegans}) networks. Several results are reported, including the identification of secondary avalanches, the validation of the equivalent model, the identification of the possible universality of the avalanches for most networks (depending only on the network size), as well as the identification of the fact that different avalanches can be obtained by locating the activation source at different neurons of the \emph{C. elegans} network.