Load compensation in human goal-directed arm movements.

We analysed the execution of multijoint pointing movements in humans while weight or spring loads were applied to the pointing hand. Visual feedback on arm and hand position was excluded. Movement paths, final positions, and normalized velocity profiles were found to be load-independent, except for the very first movement after a load change. With increasing size of a weight load movement velocity decreased, and movement duration increased by the same factor, i.e. the velocity profiles were rescaled in magnitude and time. In contrast, under a spring load movement velocity and duration were not different from no-load controls. These findings led us to propose a new hypothesis on load compensation by the motor system. We suggest that an important controlled variable is a fictional force acting externally on the hand, and that the inertia- and gravity-related components of this force are controlled separately; then, loads are compensated by time scaling of the inertia-related, and magnitude scaling of the gravity-related component. The predictions of this hypothesis regarding movement paths and velocities under weight and spring loads are in good quantitative agreement with our experimental data. When specifically asked to do so, our subjects were able to generate velocity profiles under a weight load that were not different from those under no-load conditions, which suggests that alternative control strategies are available when needed.