Speaker: Prof. Dr. Benjamin Lindner, Bernstein Center for Computational Neuroscience & Department of Physics, Humboldt University (Berlin)
Abstract: In the brain, nerve cells (neurons) are connected in large recurrent networks that process and transmit information. It is both interesting and important to understand the autonomous (stimulation-free) dynamics of such networks. Remarkably, even if single neurons obey completely deterministic dynamics (for instance, in a model network of randomly connected integrate-and-fire neurons), the nonlinear interactions within the network may result in an highly irregular, asynchronous, and apparently random activity that strongly resembles the spiking in many brain areas in vivo. This activity can be described as a stochastic process and is subject to a self-consistency condition: each neuron is subject to but also contributes to the network noise. The statistics and, in particular, the temporal correlations of the input fluctuations are thus related to those of the output fluctuations of a neuron. In my talk I review different models in which pronounced temporal correlations of network activity are observed and outline different analytical and numerical approaches to the self-consistency problem.
Host: Dr. Peter Thomas, Associate Professor CWRU Department of Mathematics, Applied Mathematics, and Statistics