Modeling of the spike trains of electrically stimulated auditory nerve fiber models with random point process

Abstract

It is investigated whether or not the spike trains of auditory nerve fiber (ANF) models in response to pulsatile electric stimuli can be modeled as an inhomogeneous Poisson process using computer simulations. Inhomogeneous Poisson process is one of the random point processes for modeling highly localized events, like the spike firings observed in neurophysiology experiments, and the temporal processes can be characterized by the parameterized intensity function. In this study, the intensity function is assumed to be expressed as a series of von Mises distribution with a periodic time interval, so as to capture quasi-periodic spike firings of ANF models in response to sinusoidally modulated pulsatile electric stimuli. The results of computer simulations suggest that the maximum likelihood estimates of the parameters for von Mises distribution and of the spike firing rates can represent the characteristics of the intensity function in inhomogeneous Poisson process.