Relating firing rate and spike time irregularity in motor cortical neurons

Introduction Cortical neurons exhibit highly irregular inter-spike intervals (ISIs) [1]. Differences in irregularity could be in part due to imbalances of excitatory and inhibitory inputs to the neurons, which determines the statistics of the net input [2]. There is experimental evidence that the intrinsic irregularity of neurons in the awake monkey is constant [3]. However, changes in irregularity have also been reported [4,5] in different cortical areas and different behavioral tasks. The classical measure of spike time irregularity is the coefficient of variation (CV), a global measure defined as the dispersion of the ISIs. However, the CV largely overestimates the irregularity in the case of pronounced changes in firing rate. This led several researchers to propose alternative measures of irregularity that are local in time and therefore relatively independent of rate changes. To our knowledge, these measures have never been compared to each other. Here we compare four such measures: the local coefficient of variation CV2 [6], the local variation LV [3], the measure IR [4] and the measure SI [7]. The first question we address is which of these measures is the most efficient for analyzing experimental data in a time-resolved manner where the number of ISIs is limited. Second, we study the variation of the spike time irregularity of neurons recorded in the motor cortex of a monkey while performing a delayed center-out task.