Abstract
Although the neuronal spike counts across repeats of single condition often vary considerably, it has been considered suitable to describe the responses according to mean and variance because the distributions have been considered unimodal. Recently, Wiener and Richmond (2003) pointed out that even unimodal distributions (V1 neuronal responses) are not consistent with simple, single Poisson processes, and could be fit nicely as mixtures of a few Poisson processes, raising the question of whether such fits to the firing rate distribution arose as a computational tool, or as a consequence of some state change in the neuronal responses. We have recorded single neuronal responses from anterior insula, anterior cingulate cortex, and ventral striatum that we conceptualize as part of a system related to the balance between work and reward. The raster plots of those responses seem to have trials falling into 2 or 3 groups based on the spike counts in single task condition. Here, we report that more than 95 percent of those neuronal responses were well fit by mixtures of 3 or fewer Poisson processes. Moreover, substantial number of them showed changes in the ratio of each Poisson process in relation to the task states (progress through trials toward reward in this case). These mode changes of the neurons might be related to behavioral or cognitive factors. [Jpn J Physiol 55 Suppl:S182 (2005)]
