Email updates

Keep up to date with the latest news and content from BMC Neuroscience and BioMed Central.

This article is part of the supplement: Eighteenth Annual Computational Neuroscience Meeting: CNS*2009

Open Access Poster presentation

First-to-fire neurons induced by clustering in sparse networks

Olav Stetter1*, Anna Levina12 and Theo Geisel12

Author Affiliations

1 Department of Nonlinear Dynamics, Max-Planck-Institute for Dynamics and Self-Organization, D37073 Göttingen, Germany

2 Bernstein Center for Computational Neuroscience, D37073 Göttingen, Germany

For all author emails, please log on.

BMC Neuroscience 2009, 10(Suppl 1):P153  doi:10.1186/1471-2202-10-S1-P153


The electronic version of this article is the complete one and can be found online at: http://www.biomedcentral.com/1471-2202/10/S1/P153


Published:13 July 2009

© 2009 Stetter et al; licensee BioMed Central Ltd.

Poster presentation

Recent studies demonstrated the dependence of the temporal evolution of avalanches in neural networks on the specific parameters of the network like connection strength and network size [1]. The dependence of the averaged network activity on the strength of an external stimulus has also been shown experimentally [2]. Additionally, in these experiments, the temporal order of activation has been shown to be non-random. There exists a topological hierarchy [3] with a number of neurons that are more likely to take part in an early phase of synchronized network activity ("burst"). Their activity can in fact be used to predict the "following" network behavior [4]. An exact characterization of these "burst initiation zones," however, is missing.

Here we ask under which topological conditions neural networks display such non-random behavior. As a model system we use a directed sparsely connected random graph with integrate-and-fire neurons. We observe that neurons with a high in-degree tend to be active slightly earlier during an avalanche. In fact, we show analytically that the mean first firing time of a neuron can be approximated by being proportional to C1 log(C2/k + 1) with the in-degree of a node k and constants C1 and C2 depending on the parameters of the network. Since the degree distribution Pk (k) is known, we can calculate the variance in the time of the first spike.

We then show how the introduction of clustering leads to a large increase in this variance. To do this we extend our model by giving each node a 2D position and using a distance-dependent connection probability (proportional to r-4). The mean first firing time, however, will then not only depend on the in-degree k of a node, but also on its neighborhood. To take this into account we characterize the different topologies using the Katz status index. We demonstrate that, using an appropriate coefficient, this index is much more efficient in predicting first-to-fire neurons than the degree alone.

References

  1. Eurich C, Herrmann M, Ernst U: Finite-size effects of avalanche dynamics.

    Phys Rev E 2002, 66:066137-066152. Publisher Full Text OpenURL

  2. Breskin I, Soriano J, Moses E, Tlusty T: Percolation in living neural networks.

    Phys Rev Lett 2006, 97:188102. PubMed Abstract | Publisher Full Text OpenURL

  3. Eytan D, Marom S: Dynamics and effective topology underlying synchronization in networks of cortical neurons.

    J Neurosci 2006, 26:8465-8476. PubMed Abstract | Publisher Full Text OpenURL

  4. Eckmann J-P, Jacobi S, Marom S, Moses E, Zbinden C: Leader neurons in population bursts of 2D living neural networks.

    New J Phys 2008, 10:015011. Publisher Full Text OpenURL