Dissociation of gravity and limb dynamics in motor control

Human movements remain accurate across substantial variations in mechanical loading, yet how the nervous system anticipates and compensates for intrinsic and extrinsic forces when generating muscle commands remains unresolved. Because the forces acting on the body are complex and nonlinear, dynamical modeling provides a powerful framework for identifying the computations underlying motor control. Here, we exploited parabolic flight to dissociate gravitational loading from intrinsic limb dynamics and directly test how these mechanical components are represented in motor commands. We analyzed kinematics and shoulder muscle activity from nine participants performing visually guided pointing movements in microgravity (0g), normogravity (1g), and hypergravity (2g). Inverse dynamics revealed that the gravity-independent component of muscle torque remained invariant across gravitational conditions. In striking contrast, muscle moments estimated from muscle activation showed systematic scaling of not only static but also phasic components with gravitational load, particularly during movement deceleration. This divergence arises because activation-derived muscle moments capture cocontraction, and thus impedance-related control, that is absent from net torques obtained by inverse dynamics. Gravity-dependent regression analyses further showed that muscle activation could be reconstructed from a compact set of mechanical variables dominated by gravity-dependent torque and movement velocity, revealing a low-dimensional structure of motor control. Together, these findings support a control architecture in which motor commands integrate predictive compensation for limb dynamics with gravity-dependent regulation of muscle cocontraction. This dual strategy provides a mechanism for maintaining stable and accurate movement across changing mechanical environments and offers a unified framework for understanding motor control under variable external loads.