Control of muscle synergies by cortical ensembles

Since its introduction in the early 1980s, the concept of a "preferred direction" for neuronal discharge has proven to be a powerful means of studying diverse properties of individual neurons in the motor areas of the brain. More recently, the activity recorded from ensembles of neurons, each with an identified preferred direction, has been used to predict hand movement, both off-line, and in real-time. Our recent experiments have addressed similar issues, but have focused on the relation between primary motor cortical discharge and muscle activity, rather than limb kinematics. We recently introduced the concept of a "muscle-space" preferred direction (PD(M)), that is analogous to the familiar hand-space preferred direction (PD(H)). In this manuscript, we show that there is considerable variety in the direction of these PD(M) vectors across neurons, but that for a given task and neuron, two successive measurements of PD(M) are very similar. We found that these vectors tend to form clusters in particular regions of the muscle space that may reflect neurons that control synergistically important groups of muscles. We have also shown that the discharge measured from neural ensembles can be used to predict the activity of individual muscles, in much the way that kinematic signals have been predicted by other groups. In fact, the accuracy of these predictions is similar to that of kinematic signals, despite the stochastic nature and greater bandwidth of the EMG signals. PD(M)s represent a divergence from one neuron to numerous muscles, while the prediction of muscle activity represents convergence from many neurons to individual muscles. We are continuing to investigate the nature of this complex matrix of functional interconnections.