Evaluating Neural Interaction in Motor Cortex During Reach-to-Grasp Task from the Spike Train Data

In this work, we recorded the spike train data from a rhesus monkey's motor cortex while it was trained to perform reach-to-grasp task at different directions and orientations. The neurons were divided into several classes according to their neuronal activities, and each class of neuron encoded a kind of functional parameter, such as direction-related, orientation-related and so on. Then we applied the generalized partial directed coherence (PDC) to evaluate the neural interaction. This method is a frequency domain measure established on the multivariate vector autoregressive model (MVAR). Since MVAR works primarily for continuous data, the spike trains were converted into the time series of the instantaneous firing rate prior to the MVAR estimation. The results indicated that the neural interactions commonly existed among neurons, and there were different network topology and coupling strength of the neural interaction corresponding to different targets. And also, our study still provided the more evidence for the hypothesis that there probably existed a common pathway in motor cortex that controlled both direction and orientation during the reach-to-grasp task.