A rate-independent measure of irregularity for event series and its application to neural spiking activity

Although higher-order statistics of neuronal firing have been characterized in neuroscience, many analyses ignore the nonstationarity of the background firing rate. We discuss how to measure the irregularity of interspike intervals in a rate-independent manner. Under the framework of semiparametric statistical models, we develop an estimator of firing irregularity which remains after the effects of rate modulations are removed. We found that firing irregularity is robust and reproducible in neurons in olfactory cortex irrespective of the rate modulation during the task period. As the level of irregularity varies among neurons, we classified neurons in olfactory cortex by using the proposed measure as a feature.