Self-sustained collective irregular dynamics

Ekkehard Ullner


The origin of the self-sustained brain activity in the resting state is not fully understood. We propose the collective irregular dynamics (CID) as a possible mechanism. Collective irregular dynamics (CID) is stochastic-like behavior on macroscopic scales.
We study two very different neuronal network types expressing CID. On the one hand we investigate fully coupled networks of spiking neurones with a dispersion of their natural frequencies and analyse the robustness of CID in a context where the microscopic dynamics cannot be chaotic in itself. In particular we test the robustness
against variations in the amount of disorder and in the shape of the phase response curve. On the other hand we discuss the emergence of CID in networks of spiking neurons with quasi-balanced activity. A detailed investigation of the thermodynamic limit for fixed density of connections shows that the asymptotic regime is characterized by self-sustained irregular, macroscopic (collective) dynamics. As long as the connectivity is massive, this regime is found in many different setups. The irregularity of the collective dynamics is justified by the power spectra of the neural activity, a fractal-dimension analysis and standard synchronisation measures.
The detailed simulations performed for different model setups and various parameter values confirm that irregular collective dynamics is very ubiquitous.