Understanding the neural control of motor unit ensembles is a challenging task. Historically, this challenge has been addressed by studying 2-3 motor units concurrently in muscles from diverse compartments and during different voluntary tasks. However, due to the low number of decoded motor units, it was not possible to extract a more structured view of the synaptic inputs shared between motor units during voluntary movement. With the advent of novel intramuscular and surface EMG electrodes and decomposition techniques, it has been shown that motor units exhibit a high level of correlated activity during isometric contractions. In this talk, I will discuss the behaviour of motor units controlling the hand (first dorsal interosseous and thenar muscles) during concurrent index abduction and thumb flexion, and leg muscles (vastus lateralis and medialis) during knee extension isometric contractions.

By applying a factorization method directly to the motor unit discharge timings, we were able to study the neural dimensions that control a relatively large number of motor units in the hand and vastii muscles. We first validated the factorization method with synthetic motor unit discharge times generated using an integrate-and-fire model with known distributions of shared and independent synaptic inputs. Subsequently, in the experiments, we found that the factorization method extracted a single dominant common input that was muscle-specific. On average 75% of the motor units for the thenar muscles and first dorsal interosseus were dominated by a single common input that was specific for the muscle they resided. On the other hand, in the vastii muscles, although we also found two independent muscle-specific shared inputs, the motor units exhibited a continuous distribution of correlation with these two dominant muscle-specific common inputs. The proportion of the muscle-specific motor unit common input was 60% for vastus medialis and 45% for vastus lateralis. The other motor units were either correlated with both muscles (shared inputs) or belonged to the common input for the other muscle (15% for vastus lateralis). These results indicate that correlated discharge rates of motor units arise from at least two independent sources of common input among the motor neurons innervating synergistic muscles, and that the hand and leg muscles significantly differ in their distribution, likely due to anatomical constraints.

In the second part of the talk, I will discuss how we used the same methods applied to subjects with C3-C6 complete motor and sensory spinal cord injury and healthy individuals during dynamic hand movements. By recording the spared motor units from the forearm muscles, we found high correlations between these spared motor units' discharge timings and the kinematics of a virtual hand. We then studied the dimensions of these motor unit firing times in comparison to healthy subjects. The results showed that even after 15 years of chronic hand paralysis, there are still motor units encoding flexion and extensions of the paralyzed fingers. We then connected these motor units in real-time through a novel brain-computer interface software and demonstrated that these patients can reliably control a virtual hand and exoskeleton controlling the paralyzed hand.