Modeling neuronal ensemble firing activity through intermittent Chaos

While a large body of experimental works provided results about stimulus evoked activities in primary sensory cortices, the functional role and the dynamics of spontaneous activity (SA) have been less widely addressed. From the dynamical perspective, a major open problem is represented by the evolution of SA, most modeling works tacitly assuming SA activity to be essentially random. The alternative possibility that SA is, at least for a significant part, deterministic, although highly chaotic, did not gain much attention. We addressed the problem by a novel synthetic approach. We first classified the multi-unit spike patterns into a reasonable number of classes. Through the use of symbolic dynamics we described SA by characterizing its long range correlations and maximum residence times for a pattern class. We found that we could exploit the nonlinear dynamics of a logistic map tuned in the region of Type I Intermittency. We finally applied our analyses on recordings taken both from normal and neuropathic rats and we found a great variety of behaviors. Some data exhibited complex dynamics. Others were more regular with intermittent-like phases constituted by long reiteration of the same pattern class. By combining a random noisy component with a logistic map we could generate class sequences that faithfully reproduced the long range correlations and maximum residence times measured on the dataset.