Recurrence of correlation structure in hippocampal neuronal ensembles during spatial behavior

TitleRecurrence of correlation structure in hippocampal neuronal ensembles during spatial behavior
Publication TypeConference Paper
Year of Publication2010
AuthorsNeymotin, S. A., Fenton A. A., & Lytton WW.
Conference NameSociety for Neuroscience 2010 (SFN '10)
KeywordsSFN, Society for Neuroscience
Abstract

The correlation structure of neurons is a candidate mechanism for information representation. We developed a method, Population Coordination (PCo) to visually assess the correlation structure of ensembles of single neurons. The first step is to form a population correlation (PCorr) vector, consisting of pair-wise correlations between ensemble elements at a given time instant. Next, we form the PCo matrix, consisting of correlations between PCorr vectors occurring at different times. Visual inspection of the matrix allows assessing recurrence of activity patterns over time. We compared PCorr against a common representation of ensemble activity, population activity vectors (AV), where the nth dimension of the vector is the number of spikes of neuron n per time. To assess the utility of PCorr we compared the error in decoding location from place cell ensemble discharges using each type of representation. Location decoding begins by assigning each population vector (PCorr or AV) the average location of the animal during the vector's time interval. An average vector is then formed and assigned to each spatial location. Decoding involves selecting the best match between the transient population vector at each time step and the average templates. We found that PCorr provided localization errors that were less than or equal to those provided by AV at the given spatial and temporal resolutions, which ranged from 0.55-9.375 cm/pixel and 5-120s respectively. This suggests that the correlation structure of neuronal ensembles could provide a potent signal for representing spatial information through co-activation of different cell subsets. We next assessed whether the correlation structure of the ensembles changed significantly with place cell remapping when the rat was in different environments. We were able to detect a change in correlation structure in these different environments in several ways. The first, a coarse approximation, was done by noting a difference in average PCorr in each environment. Next we used PCo to demonstrate that the PCorr activity was more self-similar within than across environments. These results indicate that place field remapping is a form of altered correlation structure of the ensemble. We compared these results with those obtained from simulated place cells, who's firing fields matched the real place cells, but with random Poisson firing times. We found higher recurrence (0.3+/-0.004 vs 0.15+/-0.004, p<0.01) of correlation structure in the real place cell ensembles. Our results suggest that the correlation structure of hippocampal place cells represents spatial information at different spatio-temporal scales.