Matching dynamically varying forces with multi-motor-unit muscle models: a simulation study

Human muscles exhibit great versatility, not only generating forces for demanding athleticism, but also for fine motor tasks. While standard musculoskeletal models may reproduce this versatility, they often lack multiple motor units (MUs) and rate-coded control. To investigate how these features affect a muscle's ability to generate desired force profiles, we performed simulations with nine alternative MU pool models for two cases: (i) a tibialis anterior muscle generating an isometric trapezoidal force profile, and (ii) a generic shoulder muscle generating force for a reaching movement whilst undergoing predetermined length changes. We implemented two control strategies, pure feedforward and combined feedforward-feedback, each parameterized using elementary tasks. The results suggest that the characteristics of MU pools have relatively little impact on the pools' overall ability to match forces across all tasks, although performances for individual tasks varied. Feedback improved performance for nearly all MU pools and tasks, but the physiologically more relevant MU pool types were more responsive to feedback particularly during reaching. While all MU pool models performed well in the conditions tested, we highlight the need to consider the functional characteristics of the control of rate-coded MU pools given the vast repertoire of dynamic tasks performed by muscles.