Purchase this article with an account.
Melchi M. Michel, Robert A. Jacobs; The costs of ignoring high-order correlations in populations of model neurons. Journal of Vision 2005;5(8):672. doi: 10.1167/5.8.672.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Background: Investigators debate the extent to which neural populations use high-order statistical dependencies among neural responses to represent information about a visual stimulus. A number of recent studies, using either performance or information measures, suggested that correlations between neurons play only a minimal role in encoding stimulus information. An important limitation of these studies is that, in approximating the joint distribution of responses, they considered only pairwise or 2nd-order correlations, thereby ignoring the information contained in higher-order correlations. Methods: To study this issue, we used three statistical decoders to extract the information in the responses of model neurons about the binocular disparities present in simulated pairs of left-eye and right-eye images: (i) the Full Joint Probability Decoder considered all possible statistical relations among neural responses as potentially important; (ii) the Dependence Tree Decoder also considered all possible relations as potentially important, but it approximated high-order statistical correlations using a computationally tractable procedure; and (iii) the Independent Response Decoder which assumed that neural responses are statistically independent, meaning that all correlations should be zero and, thus, can be ignored. We used both information-theoretic and performance measures to calculate the information extracted by the different decoders.
Results: Simulation results indicate that high-order correlations among model neuron responses contain significant information about binocular disparities, and that the amount of this high-order information increases rapidly as a function of neural population size. The results also highlight the potential importance of the Dependence Tree Decoder to neuroscientists as a powerful, but still practical, way of approximating high-order correlations among neural responses.
This PDF is available to Subscribers Only