Increased firing rates monotonically expand neuronal coding bandwidth

Abstract

Sensory information is often processed by populations of neurons that vary with respect to their levels of spiking activity. It appears plausible that neurons that fire more often are capable of following faster changes in the input signal than those with lower firing rates. In this study, we test this intuitive assumption by systematically investigating how the spiking activity of a neuron affects information transmission. To this end, we employ the coherence function as a spectral measure of information transmission and study the relation between the width of the coherence function, which we call coding bandwidth, and the firing rate. The reexamination of experimental data from a previous study on weakly-electric fish and the simulation of biologically inspired neuron models fitted to real neurons reveal indeed a significant correlation between firing rate and coding bandwidth. In addition, known analytical expressions for the coherence function for the stochastic leaky integrate-and-fire model demonstrate a linear relation for high firing rates. However, in contrast to intuition, the width of the coherence function becomes independent of the firing rate and is completely set by the membrane time constant for very low firing rates.