What Can Trial-to-Trial Variability Tell Us? A Distribution-Based Approach to Spike Train Decoding in the Rat Hippocampus and Entorhinal Cortex

Is it possible to robustly encode information from a population of neurons that rarely provide information? Neurons within the rat hippocampus and entorhinal cortex (EC) have been shown to possess spatial receptive fields that modulate their firing properties under different sets of behavioral or experimental conditions, termed contexts. Recent studies have identified cells in these regions that show changes in trial-to-trial firing rate variability as a function of experimental context. How then, could changes in variability, not observable in a single trial be useful to information processing in the brain? We propose a scenario in which individual neurons with high trial-to-trial variability each provide substantive information over small subsets of trials, which are distributed so that a population of the cells could robustly encode context across all trials. In this chapter, we explore the nature of trial-to-trail variability, and seek to verify our hypothesis by developing a decoding algorithm that predicts context from spiking data using a model characterizing changes in the full distribution of firing rate structure across trials. We compare this algorithm to another decoding procedure that accounts for only changes in mean firing rate. We apply these decoding algorithms to and experimental spiking data, and show that information is combined from many sparsely encoding cells within CA1 and dcMEC ensembles to robustly encode context.