Control signal dimensionality depends on limb dynamics.

Neural control of movement has to overcome the problem of redundancy in the multidimensional musculoskeletal system. The problem can be solved by reducing the dimensionality of the control space of motor commands, i.e., through muscle synergies or motor primitives. Evidence for this solution exists; multiple studies have obtained muscle synergies using decomposition methods. These synergies vary across different workspaces and are present in both dominant and non-dominant limbs. We explore the effect of biomechanical constraints on the dimensionality of control space. We also test the generalizability of prior conclusions that muscle activity profiles can be explained by applied moments about the limb joints that compensate for dynamic and gravitational forces during reaching. These muscle moments derived from motion capture represent the combined actions of muscle contractions that are under the control of the nervous system. Here, we test the hypothesis that the control space dimensionality is shaped by the complexity of dynamic and gravitational forces. To achieve this, we examined muscle activity patterns across reaching movements in different directions, starting from different postures performed bilaterally by healthy individuals. We used principal component analysis to evaluate the contribution of individual muscles to producing muscle moments across different reaching directions and in both dominant and non-dominant limbs. Extending our earlier work, we find that muscle activity profiles are described well by muscle moment profiles during reaching by both dominant and non-dominant arms. Our results further show that the dimensionality of control signals depends on the complexity of muscle moments, supporting the primary hypothesis. Our results suggest that the neural control strategy for limb dynamics compensation involves the modulation of the co-contraction of proximal and distal antagonistic muscles that change limb stiffness.