Dissociation of gravity and limb dynamics in motor control

Abstract Human movements remain accurate and stable despite changes in external forces, yet it is unclear how the nervous system separates and compensates for gravitational loads and intrinsic limb dynamics when generating motor commands. This question is difficult to address under terrestrial conditions, where these forces are inherently coupled. Here we dissociated gravitational acceleration from limb dynamics using parabolic flight. Nine participants performed visually guided pointing movements during alternating microgravity (0g), normogravity (1g), and hypergravity (2g), while upper-limb kinematics and shoulder muscle activity were recorded. Inverse-dynamics analysis revealed that the gravity-independent dynamic component of muscle torques remained largely invariant across gravitational conditions, whereas gravity-related torques scaled proportionally with gravitational acceleration. In contrast, muscle activation exhibited systematic gravity-dependent modulation in both agonist and antagonist muscles, including increased co-contraction consistent with elevated shoulder stiffness. Critically, the dynamic component of EMG-derived torques increased with gravitational load despite minimal changes in the corresponding mechanical torques, particularly during movement deceleration. This dissociation indicates that muscle activity is not solely determined by inverse-dynamics requirements but reflects additional control objectives. A regression-based reconstruction showed that muscle activation patterns can be predicted from a compact set of biomechanical variables, primarily gravity-related torque and movement velocity. Together, these findings demonstrate that motor commands combine predictive compensation for gravitational and inertial forces with gravity-dependent modulation of limb impedance. Scaling joint stiffness with gravitational load provides a robust control strategy for stabilizing movement under variable mechanical conditions. This principle offers a biologically grounded framework for the design of adaptive controllers in prosthetics, wearable robotics, and rehabilitation systems operating in uncertain environments.