Input synchrony strengthens correlation transmission via noise suppression

Abstract

Event Abstract Back to Event Input synchrony strengthens correlation transmission via noise suppression Matthias Schultze-Kraft1, 2, 3*, Markus Diesmann4, 5, 6, Sonja Grün4, 7 and Moritz Helias4 1 Berlin Institute of Technology, Machine Learning Group, Germany 2 Bernstein Center for Computational Neuroscience, Germany 3 Bernstein Focus: Neurotechnology, Germany 4 Research Center Jülich, Computational and Systems Neuroscience, Germany 5 RIKEN Brain Science Institute, Japan 6 RIKEN Computational Science Research Program, Japan 7 RIKEN Brain Science Institute, Japan Whether precise spike timing [1] conveys information in cortical networks or whether the firing rate alone is sufficient is still matter of controversial debates. While it is often argued that synchrony is an epiphenomenon caused by shared afferents among densely interconnected neurons [2], several studies have reported task related modulation of spike synchrony, lately in primary visual cortex [3] and motor cortex [4]. Recently, cortical networks were found to exhibit very weak correlations [5], thus suggesting either population-rate codes or potentially providing a suitable substrate on top of which spike synchrony can represent information. In this work we theoretically investigate the efficacy of common synaptic afferents on the one hand and synchronized inputs on the other hand to contribute to closely time-locked spiking activity of pairs of neurons [6]. We employ direct simulation and extend earlier analytical methods based on the diffusion approximation [7] to pulse-coupling, thus allowing us to describe correlations on the level of synchronized presynaptic sources. We investigate the transmission of correlated synaptic input currents by pairs of integrate-and-fire model neurons. We therefore generate input currents exhibiting zero to full correlation and realize each correlation value by different proportions of common afferents and spiking synchrony. The comparison of the count correlation of the output spike trains of neurons driven by these currents allows us to address the question how much synchrony is caused by presynaptic synchronous activity and how much is due to the high convergence and divergence the cortex' connectivity. We find that, at a fixed working point, the transmission of a particular input correlation is boosted when it contains synchronous events (Fig. 1, black curve) compaired to when induced by common input alone (Fig. 1, gray curve), giving rise to a higher output correlation and sharpening the correlation function (Fig. 1, inset). In the regime of high input correlation this boosting is even more striking, resulting in an output correlation higher than the total correlation in the input. Making use of recent theoretical insights [8] into the non-linear response properties of neurons, we provide a quantitative understanding of how the threshold acts as a noise suppression mechanism, explaining the correlation transmission gain > 1. Figure 1 Acknowledgements Partially supported by BMBF Grant 01IB001A (brain@work), the Helmholtz Alliance on Systems Biology, the Next-Generation Supercomputer Project of MEXT, EU Grant 15879 (FACETS), EU Grant 269921 (BrainScaleS), DIP F1.2, and BMBF Grant 01GQ0420 to BCCN Freiburg. All network simulations carried out with NEST (http://www.nest-initiative.org).