Phase space analysis of networks based on biologically realistic parameters

Abstract

We study cortical network dynamics for a spatially embedded network model. It represents, in terms of spatial scale, a large piece of cortex allowing for long-range connections, resulting in a rather sparse connectivity. The spatial embedding also permits us to include distance-dependent conduction delays. We use two different types of conductance-based I&F neurons as excitatory and inhibitory units, as well as specific connection probabilities. In order to remain computationally tractable, we reduce neuron density, modelling part of the missing internal input via external poissonian spike trains. Compared to previous studies, we observe significant changes in the dynamical phase space: Altered activity patterns require another regularity measures than the coefficient of variation. Hence, we compare three different regularity measure on the basis of artificial inter-spike-interval distributions. We identify two types of mixed states, where different phases coexist in certain regions of the phase space. More notably, our boundary between high and low activity states depends predominantly on the relation between excitatory and inhibitory synaptic strength instead of the input rate.

Publication
Journal of Physiology-Paris

header

Nicole Voges
Nicole Voges
PostDoc in Computational Neuroscience

Neuroscientist & Data Analyst, Wennigsen, Lower Saxony, Germany.

Laurent U Perrinet
Laurent U Perrinet
Researcher in Computational Neuroscience

My research interests include Machine Learning and computational neuroscience applied to Vision.