Representational geometries reveal differential effects of response correlations on population codes in neurophysiology and functional magnetic resonance imaging

Abstract

Two sensory neurons usually display trial-by-trial response correlations given the repeated representations of an identical stimulus. The effects of such response correlations on population-level sensory coding have been the focal contention in computational neuroscience over the past few years. In the meantime, multivariate pattern analysis (MVPA) has been the leading analysis approach in functional magnetic resonance imaging (fMRI), but the effects of response correlations in voxel populations remain underexplored. Here, instead of conventional MVPA analysis, we calculate linear Fisher information of population responses in human visual cortex and hypothetically remove response correlations between voxels. We found that voxelwise response correlations generally enhance stimulus information, a result standing in stark contrast to the detrimental effects of response correlations reported in neurophysiological literature. By voxel-encoding modeling, we further show that these two seemingly opposite effects actually can coexist. Furthermore, we use principal component analysis to decompose stimulus information in population responses onto different principal dimensions in a high representational space. Interestingly, response correlations simultaneously reduce and enhance information on high- and low-variance principal dimensions, respectively. The relative strength of the two antagonistic effects within the same computational framework produces the apparent discrepancy in the effect of response correlations in neuronal and voxel populations. Our results suggest that multivariate fMRI data contain rich statistical structures that are directly related to sensory information representation, and the general computational framework to analyze neuronal and voxel population responses can be applied in many types of neural measurements.