Optimal Burstiness in Populations of Spiking Neurons Facilitates Decoding of Decreases in Tonic Firing

Abstract

Abstract A stimulus can be encoded in a population of spiking neurons through any change in the statistics of the joint spike pattern, yet we commonly summarize single-trial population activity by the summed spike rate across cells: the population peristimulus time histogram (pPSTH). For neurons with a low baseline spike rate that encode a stimulus with a rate increase, this simplified representation works well, but for populations with high baseline rates and heterogeneous response patterns, the pPSTH can obscure the response. We introduce a different representation of the population spike pattern, which we call an “information train,” that is well suited to conditions of sparse responses, especially those that involve decreases rather than increases in firing. We use this tool to study populations with varying levels of burstiness in their spiking statistics to determine how burstiness affects the representation of spike decreases (firing “gaps”). Our simulated populations of spiking neurons varied in size, baseline rate, burst statistics, and correlation. Using the information train decoder, we find that there is an optimal level of burstiness for gap detection that is robust to several other parameters of the population. We consider this theoretical result in the context of experimental data from different types of retinal ganglion cells and determine that the baseline spike statistics of a recently identified type support nearly optimal detection of both the onset and strength of a contrast step.