David Hansel, theoretical neuroscientist, Senior Researcher at CNRS, visiting professor at The Hebrew University; visiting scientist at Janelia Research Campus (HHMI)
601 Pao Yue-Kong Library
Strong temporal irregularity and right-skewed, long-tailed distributions of firing rates are distinctive features of cortical spiking activity. Both features are quite puzzling upon consideration of the large number of synaptic inputs a cortical cell receives and the weak correlations among these inputs. A minimal theoretical framework accounting naturally for these features – the balance hypothesis – was proposed by van Vreeswijk and Sompolinsky in two seminal papers in the 1990s. In fact, we showed recently that log-normal distributions of mean firing rates as reported in cortex in-vivo, emerge naturally in the balanced regime (Roxin et al.. 2011). I will summarize the physics and the general phenomenology of the balanced regime as well as its functional consequences for the selectivity of neuronal responses in primary visual cortex (Hansel and van Vreeswijk, 2012, Pattadkal et al., in press). I will then present very recent results regarding the emergence of spatial correlations in the dynamics of recurrent networks operating in the balanced regime (Darshan et al., Nature Comm., 2017; Darshan, van Vreeswijk & Hansel, Phys. Rev. X, in press). In conclusion I will very briefly discuss some possible directions to develop further the theory of the balanced regime and to test experimentally the balance hypothesis.
Ph.D in theoretical physics (Ecole Normale Supèrieure, Paris), Theoretical Neuroscientist; Senior Researcher at CNRS, co-head with Dr. Carole Levenes (experimentalist) of the Theory/experiment team “Cerebral dynamics, Plasticity and Learning” at CNRS-Paris Descartes University; Visiting professor at The Hebrew University; visiting scientist at Janelia Research Campus (HHMI).
Topics of research: Cortical dynamics; Network mechanisms of visual processing; network mechanisms of working memory; Dynamics of the Basal Gangli network.