BioND — Dynamics of Biological Networks

Self-organized criticality in neural networks

Information storage and processing in complex systems is known to be maximally efficient close to critical states at which the dynamics of the system changes qualitatively. This suggests that also parts of our brain should be in (or at least close to) such a critical state. Indeed, evidence for criticality is found in EEG, MEG and EcoG data from humans and in direct measurements of activity in slices of rat cortex and in-vitro cell cultures.

An important current question is, how does the brain manage to remain close to criticality, despite ongoing changes occurring throughout development and due to cell death, aging and injuries. Addressing this question we demonstrated that self-tuning to critical states can be explained by a previously proposed abstract mechanism that relies on the adaptive nature of neural networks. In a series of simulations we showed that combining two well established model ingredients (spike-time-dependent-plasticity of synapses and integrate-and-fire neurons) is sufficient to obtain a neural system that robustly self-tunes to a synchronization transition (Meisel and Gross 2009). While being relatively simple, this model reproduces several experimental results on neural activity and predicts a neural connectivity of the correct order of magnitude. Furthermore, it evolves toward a distribution of synaptic conductances which is in good agreement with experimental results. It thereby provides the first theoretical model providing an explanation for the observed conductance distribution.

Our present work on this research topic focuses on verifying further predictions of the model by the analysis of EEG and EcoG data from humans. Furthermore we work on analytical models revealing the precise mechanism of self-tuning and its benefit for cognitive functions.

Key Publications

Analytical investigation of self-organized criticality in neural networks
Felix Droste, Anne-Ly Do, and Thilo Gross
Journal of the Royal Society Interface 78, 20120558, 2013.
(abstract) (link to publisher) (arXiv) (download preprint)

Additional Publications

Failure of adaptive self-organized criticality during epileptic seizure attacks
Christian Meisel, Alexander Storch, Susanne Hallmeyer-Elgner, Ed Bullmore, and Thilo Gross
PLoS Computational Biology 84, e1002312-8, 2012.
(abstract) (link to publisher) (arXiv) (download preprint)

Adaptive self-organization in a realistic neural network model
Christian Meisel and Thilo Gross
Physical Review E 80, 061917-6, 2009.
(abstract) (link to publisher) (arXiv) (download preprint)

Adaptive coevolutionary networks: a review
Thilo Gross and Bernd Blasius
Journal of the Royal Society Interface 5(20), 259-271, 2008.
(abstract) (link to publisher) (arXiv) (download preprint)