Temporal network approach to unraveling collective neuron firings

As interest escalates in the area of temporal networks, the possibility arises that such tools might help unravel the complex dynamics of potentially correlated spatiotemporal events in complex systems. Here, we present such an analysis for a dataset comprising neuronal spike trains from retinal ganglion cells in a salamander. The neuron firing events are detected experimentally using a flat array of extracellular electrodes at micrometre scale separations, with spikes from the ganglion cell layer then being sorted into single-unit spike trains.We develop a temporal network analysis of this ensemble of spike trains that allows us to explore potential causality between firings. We compare our results to a randomized system in order to deduce a statistical Z-score and find examples of both event amplification and inhibition. Although the application to neuron firings is of direct interest for understanding brain function, the network approach that we present can in principle be applied to any set of timelines detailing the occurrence of events in particular regions, sectors or entities, including the search for causality between events in human activities such as crime.