The mechanisms for synchrony and for stability of persistent activity in large neuronal networks

David Hansel1 and German Mato2

1Laboratoire de Neurophysique et Physiologie du Systeme Moteur
2Comision Nacional de Energia Atomica and CONICE

We study how synchrony emerges in highly connected heterogeneous networks consisting of two interacting populations of integrate-and-fire or conductance based neurons. The populations are one excitatory (E) and one inhibitory (I). We find that three generic mechanisms exist: 1) Synchrony of spikes through inhibition in the I population. 2) Synchrony of spikes through the mutual interactions between the two populations. 3) Synchrony of network bursts through recurrent excitation. These mechanisms can be controlled by choosing the strength and time constants of the interactions between the populations, by heterogeneities and noise. For a large class of integrate-and-fire networks this was shown by computing analytically the regions in parameter space where the different mechanisms occur. For conductance-based neuronal models, we have relied on numerical simulations.

The cellular and synaptic properties underlying stable persistent state (PS) of activity in neuronal circuits have been debated recently in the context of models of working memories. Using our analytical approach we have derived conditions for the stability of PS. We have shown that PS with level of activity of the order of 5-20 Hz can be stable in a broad range of parameters of a two population network of neurons interacting with AMPA and GABA synapses, provided the level of heterogeneities in the intrinsic properties of neurons is sufficient and the inhibitory interactions between the inhibitory neurons are sufficiently strong. This is in contrast with previous theoretical studies which have suggested that saturating NMDA synapses are required to stabilize cortical persistent activity with physiologically realistic firing rates.