Humans regulate arm mechanics in a task-dependent manner through a combination of feedforward and feedback neural mechanisms to complete various functional tasks. However, the relative contributions from these two mechanisms are still not completely understood. In addition, the biomechanical properties of musculoskeletal system may also limit the efficacy of these neural mechanisms, further complicating the problem. The goal of this dissertation was to examine the task-dependent regulation of arm mechanics using a combination of musculoskeletal modeling and simulation as well as experimental approaches.

Arm mechanics can be quantified by endpoint impedance, describing the relationship between externally imposed displacements at the endpoint of the arm (the hand) and the forces generated in response. Endpoint stiffness, the static component of endpoint impedance, is especially important in maintenance of stable arm positions. Firstly, this dissertation examines the muscle properties contributing to the feedforward regulation on the endpoint stiffness. Comparing simulated endpoint stiffness based on two muscle stiffness models with previous experimental data indicates that muscle short-range stiffness is the major contributor to endpoint stiffness and that the resultant musculoskeletal model accurately characterizes endpoint stiffness across various postures and levels of endpoint forces.

Secondly, the developed musculoskeletal model was applied to examine the feedforward regulation on the orientation of the maximum endpoint stiffness. By simulating the controllable range of the stiffness orientation, it was found that the feedforward regulation on the orientation of the maximum endpoint stiffness is highly constrained by the biomechanical properties of the musculoskeletal system and any additional task requirements.

Lastly, this dissertation examined the ability to maintain elbow torque in the face of unpredictable perturbations. Elbow impedance was estimated in a “do not intervene” and a “torque control” task. The results indicate that subjects reduced low-frequency elbow impedance to better regulate elbow torque in the torque control task. Further analyses on electromyographic data suggest this regulation mainly relies on the neural feedback with delays of 100~200ms.