Regensburg 2016 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 25: Networks: From Topology to Dynamics (joint session BP / SOE / DY)
SOE 25.3: Vortrag
Donnerstag, 10. März 2016, 17:15–17:30, H43
Global stability reveals critical components in the structure of multi-scale neural networks — •Jannis Schuecker1,4, Maximilian Schmidt1,4, Sacha J. van Albada1, Markus Diesmann1,2,3, and Moritz Helias1,3 — 1Inst of Neurosci and Medicine (INM-6) and Inst for Advanced Simulation (IAS-6) and JARA BRAIN Institute I, Jülich Research Centre — 2Department of Psychiatry, Psychotherapy and Psychosomatics, Medical Faculty, RWTH Aachen University — 3Department of Physics, Faculty 1, RWTH Aachen University — 4These authors contributed equally
One of the major challenges of neuroscience is the integration of the available experimental data into a coherent model of the brain. In this endeavor, the exploration of the inevitable uncertainties in anatomical data should be guided by physiological observations. To this end we devise a method based on a mean-field reduction of spiking network dynamics for shaping the phase space of large-scale network models according to fundamental activity constraints, prohibiting quiescence and requiring global stability. In particular, we apply this framework to a multi-area spiking model of macaque visual cortex and obtain plausible layer- and area-specific activity [Schuecker et al. 2015, arXiv:1509.03162] by controlling the location of the separatrix dividing the phase space into realistic low-activity and unrealistic high-activity states. The study systematically identifies modifications to the population-level connectivity within and between areas critical for the stability of the network. Partly supported by Helmholtz association: VH-NG-1028 and SMHB; EU Grant 604102 (HBP).