Nicholas G. Hatsopoulos, Qingqing Xu, and Yali Amit, Encoding of Movement Fragments in the Motor Cortex, The Journal of Neuroscience, May 9, 2007 - 27(19)
This paper explores the spatio-temporal response fields of M1 neurons. The authors were able to reconstruct spiking activity based on movement trajectory information within -100 to +300 ms. They use the average area under the receiver operating characteristic (ROC) curve for predicted spikes to quantify the predictive power of their model.
The reconstruction was based on smoothed and normalized velocity over time, mean speed, and mean position. This reconstruction had a mean ROC curve area of 0.60. The authors also substituted torque for the velocity vector and achieved comparable performance. The preferred trajectories ("pathlets") of individual neurons changed over time and space, but not randomly: trajectories were greatly and significantly correlated across space and time. Nearby neurons also had trajectories more correlated than average.
The authors investigate whether neurons encode kinematics (movements and positions only) or kinetics (forces that cause the movements), and find kinematic variables are better encoded. They note that rostral M1 may have more kinematic neurons and caudal M1 may have more kinetic neurons.