Berlin 2018 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
BP: Fachverband Biologische Physik
BP 34: Neuroscience
BP 34.4: Vortrag
Donnerstag, 15. März 2018, 16:00–16:15, H 1058
Optimal detection of a localized perturbation in random networks of integrate-and-fire neurons — •Davide Bernardi1,2 and Benjamin Lindner1,2 — 1Bernstein Center for Computational Neuroscience, Berlin, Germany — 2Humboldt University, Berlin, Germany
Cortical networks operate in a chaotic regime, as both theoretical and experimental studies have shown. Therefore, neural coding is believed to rely on the mean activity of many cells. However, there is evidence that the brief stimulation of a single neuron can elicit a behavioral response, a theoretically unexplained result. We study how large recurrent networks of integrate-and-fire neurons react to the perturbation of one single cell and propose a simple readout mechanism to detect the perturbation. Biasing the readout towards specific neurons leads to detection rates similar to experimentally observed values, as our numerical simulations and analytical estimates show. We observe near-optimal detection for intermediate values of the mean coupling between neurons. Current research aims to capture how detection rates depend on the temporal structure of the stimulation.
Ref: Bernardi and Lindner, Phys. Rev. Lett. 118 (2017)