Unveiling causal activity of complex networks

Department of Physics, Indiana University - Bloomington, IN 47405, USA

^{(a) (b)} The original version of this article was uploaded to the arXiv on March 17th, 2016 [].Current address: Departments of Neurobiology and Mathematics, University of Pittsburgh - Pittsburgh, PA 15260, USA; rwgarcia@pitt.edu

Received: 26 October 2016
Accepted: 23 August 2017

Abstract

We introduce a novel tool for analyzing complex network dynamics, allowing for cascades of causally-related events, which we call causal webs (c-webs), to be separated from other non-causally-related events. This tool shows that traditionally-conceived avalanches may contain mixtures of spatially-distinct but temporally-overlapping cascades of events, and dynamical disorder or noise. In contrast, c-webs separate these components, unveiling previously hidden features of the network and dynamics. We apply our method to mouse cortical data with resulting statistics which demonstrate for the first time that neuronal avalanches are not merely composed of causally-related events.