Spatio-temporal correlations and visual signaling in a complete neuronal population

J W Pillow, J Shlens, L Paninski, A Sher, A M Litke, E J Chichilnisky, and E P Simoncelli

Published in Nature, vol.454(7206), pp. 995--999, Aug 2008. Published online, 23 July 2008. In journal, 21 Aug 2008.
DOI: 10.1038/nature07140

Download:
  • Reprint (pdf)
  • Supplementary Materials (pdf)
  • Editor's summary
    • Statistical dependencies between the responses of sensory neurons govern both the stimulus information conveyed by population responses and the means by which downstream neurons can extract it. Although a variety of measurements indicate the existence of such dependencies, their functional origin and impact on the neural code are still not understood. Here we address these issues by analyzing the functional interactions in a complete population of primate parasol retinal ganglion cells, using a model of multineuron spike responses. The model, with parameters fit directly to physiological data, simultaneously captures both the stimulus dependence and detailed spatio-temporal correlations in population responses, and provides two insights into the structure of the neural code. First, neural encoding at the population level is less noisy than expected from the variability of individual cells. Individual spike times are more precise when correlations are taken into account, explaining a significant fraction of the trial-to-trial response variability that would otherwise be attributed to single-cell noise. Second, correlations carry significant sensory information: a model-based decoder that exploits the response correlation structure can extract 20% more information about the visual scene than a decoder that assumes independent responses, and 40% more information than an optimal linear decoder. This model-based approach reveals the importance of statistical dependencies for retinal coding of visual stimuli, and provides a general framework for understanding the influence of correlated spiking activity on population coding in other neural circuits.
    Related:
  • Superseded papers: Pillow08aThe effects of correlated neural activity on single-neuron spiking variability in the primate retina
    by J W Pillow, J Shlens, L Paninski, A Sher, A M Litke, E J Chichilnisky, and E P Simoncelli    , Pillow07bCorrelations and coding with multi-neuronal spike trains in primate retina
    by J W Pillow, J Shlens, L Paninski, A Sher, A M Litke, E J Chichilnisky, and E P Simoncelli    , Pillow07aDeciphering Correlations: Bayesian Decoding of Multi-neuronal Spike Trains in Primate Retina
    by J Pillow, J Shlens, L Paninski, A Sher, A Litke, EJ Chichilnisky, and E P Simoncelli    , Pillow05aModeling multineuronal responses in primate retinal ganglion cells
    by J W Pillow, J Shlens, L Paninski, E J Chichilnisky, and E P Simoncelli    , Pillow05bModeling the correlated spike responses of a cluster of primate retinal ganglion cells
    by J W Pillow, J Shlens, L Paninski, E J Chichilnisky, and E P Simoncelli
  • Related model using stochastic integrate-and-fire: JN-05Prediction and decoding of retinal responses with a probabilistic spiking model
    by J W Pillow, L Paninski, V J Uzell, E P Simoncelli, and E J Chichilnisky
  • Online Publications