Berlin 2015 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

BP: Fachverband Biologische Physik

BP 2: Neurophysics I

BP 2.5: Vortrag

Montag, 16. März 2015, 10:30–10:45, H 1058

Information filtering by partial synchronous spikes in a neural population — •Alexandra Kruscha^1,2 and Benjamin Lindner¹ — ¹Bernstein Center for Computational Neuroscience, Berlin, 10115, Germany — ²Institute for Physics, Humboldt-Universität zu Berlin, Berlin, 12489, Germany

Synchronous firing of neurons is a prominent feature in many brain areas. Here, we are interested in the information transmission by the synchronous spiking output of a noisy neuronal population, which receives a common time-dependent sensory stimulus. Experimental and theoretical work revealed that synchronous spikes encode preferentially fast components of the stimulus, i.e. synchrony acts as an information filter. In these studies a rather strict measure of synchrony was used: all neurons in the population have to fire within a short time window. Here, we generalize the definition of the synchronous output, for which only a certain fraction γ of the population has to fire in synchrony. We present an analytical approach to characterize the information transfer in dependence of this fraction and the population size, by deriving the cross-correlation and the coherence function between the stimulus and the partial synchronous output. We show that there is a critical synchrony fraction, namely the probability at which a single neuron spikes within the predefined time window, which maximizes the information transmission of the synchronous output. At this value, the partial synchronous output acts as a low-pass filter, whereas deviations from this critical fraction lead to a more and more pronounced band-pass filtering effect. We confirm our analytical findings by numerical simulations for the leaky integrate-and-fire neuron.