Neural noise and spatio-temporal patterns of movement-related activity in the supplementary motor area

Bruno Averbeck

University of Rochester

We analyzed the variability and coding capacity of simultaneously recorded pairs of neurons in the primate supplementary motor area (SMA). The coding performance was analyzed to determine whether the temporal precision of spike arrival times and the interactions within and between neurons improve the prediction of the upcoming movement direction. There were three main findings in the analyses. First, analysis of neuronal variability showed that the variance of spike counts was smaller than its mean in many SMA neurons, suggesting that a Poisson process is not a good model for the neural responses we recorded. Second, we found that the correlation in spike count variability between pairs of neurons was concentrated at low frequencies (10 Hz). Third, we tested multiple decoding models that differed in the temporal resolution and the correlations in the spike trains which were utilized for decoding. The results showed that in about 68% of neuron pairs, the arrival times of spikes at a resolution between 66 and 40 ms carried more information than spike counts in a 200 ms bin. In addition, in 24% of neuron pairs, inclusion of within- or between-neuron correlations in spike trains significantly improved decoding accuracy. These results suggest that in some SMA neurons there is more information in the spatio-temporal pattern of activity than there is in the rate code of independent neurons.