Empirical models of spiking in neural populations

摘要: Neurons in the neocortex code and compute as part of a locally interconnected population. Large-scale multi-electrode recording makes it possible to access these population processes empirically by fitting statistical models unaveraged data. What structure best describes concurrent spiking cells within local network? We argue that cortex, where firing exhibits extensive correlations both time space typical sample neurons still reflects only very small fraction population, most appropriate model captures shared variability low-dimensional latent process evolving with smooth dynamics, rather than putative direct coupling. test this claim comparing dynamical realistic observations coupled generalised linear spike-response (GLMs) using cortical recordings. find approach outperforms GLM terms goodness-of-fit, reproduces temporal data more accurately. also compare whose are either derived from Gaussian or point-process models, finding non-Gaussian provides slightly better goodness-of-fit spike counts.