An Information Theoretic Analysis of Neural Multiplexing

Abstract

How the brain encodes information in sequences of voltage spikes is an open question. Past literature suggests the importance of bursts, high-frequency spike events, as a key step towards answering this question. In particular, it was recently shown that neurons could use bursts to communicate two streams of information simultaneously, resulting in higher information rates than seen with other neural code theories. However, it is unknown how a neuron’s spiking statistics might affect communication via this new code. To investigate the influence of spike statistics, we study a bursting neuron model with the goal of estimating its information rate as a function of its spike statistics. To this end we extend a recently proposed method for estimating information rate. We find the information rate in our burst-multiplexing model is robust to changes in spike-train statistics, providing evidence for the utility of a burst-multiplexing code to diverse brain networks.