Step changes in torque were applied to the elbow or ankle joint of normal human subjects who exerted constant levels of effort. They were instructed to not react to the torque but to allow their limbs to move to a new equilibrium position. In this experimental paradigm, the joint may be characterized by a nonlinear compliant element. The aim of this study was to characterize the elastic properties of the compliant element. 2. Joint elasticity is described by an S-shaped relation between torque and angle (a "compliant characteristic curve"). The stiffness of a joint is greatest for small perturbations and decreases as the size of the perturbation is increased whether the limb is loaded or unloaded from its initial equilibrium. 3. The S shape of the compliant characteristic curve is relatively constant when measured at different initial joint angles from the same initial joint torque. 4. Higher levels of initial muscle torque increase the steepness of the compliant characteristic curve. 5. All changes in initial joint torque and angle preserve the S shape. The inflection point of the characteristic curve is always at the initial equilibrium angle and torque. This shifting of the inflection point of the torque-angle relation implies a fundamental plasticity in joint compliance. The elastic component is not invariant but changes with the joint's initial equilibrium state. 6. Changes in muscle tension and length that result from a perturbation are accompanied by changes in muscle activation. The relationship between perturbation torque and mean equilibrium EMG is similar to that found for voluntary isometric contraction. It is not possible to conclude what proportion of the late EMG response to perturbation is mediated by segmental reflex mechanisms. 7. At the levels of torque used here, changes in joint stiffness are highly correlated with changes in tonic contraction of the muscle opposing the load. This change in stiffness is not the result of antagonist coactivation, which was minimal. 8. The compliant characteristic curves of elbow and ankle are qualitatively similar. The principal difference is due to the greater passive stiffness of the ankle. 9. Our findings are inconsistent with aspects of the theory of invariant characteristics or with models of movement and load compensation that postulate a control scheme based only on the setting of muscle and reflex equilibrium points. The data are also incompatible with models that only control the elastic stiffness of the muscle.
