Previous studies have used tendon vibration to investigate kinesthetic illusions in the isometric limb and end point control in the moving limb. These previous studies have shown that vibration distorts the perceptions of static joint angle and movement and causes systematic errors in the end point of movement. In this paper we describe the effects of tendon vibration during movement while human subjects performed a proprioceptively coordinated motor task. In an earlier study we showed that the CNS coordinates this motor task-a movement sequence-with proprioceptive information related to the dynamic position and velocity of the limb. 2. When performing this movement sequence, each subject sat at a table and opened the right hand as the right elbow was passively rotated in the extension direction through a prescribed target angle. Vision of the arm was prevented, and the movement velocity was changed randomly from trial to trial, leaving proprioception as the only useful source of kinematic information with which to perform the task. 3. In randomly occurring trials, vibration was applied to the tendon of the biceps brachii, a muscle that lengthens during elbow extension. In some experiments the timing of tendon vibration was varied with respect to the onset of elbow rotation, and in other experiments the frequency of vibration was varied. In each experiment we compared the accuracy of the subject's response (i.e., the elbow angle at which the subject opened the hand) in trials with tendon vibration with the accuracy in trials without tendon vibration. 4. The effect of tendon vibration depended on the frequency of vibration. When the biceps tendon was vibrated at 20 Hz, subjects opened the hand after the elbow passed through the target angle ("overshooting"). Overshooting is consistent with an underestimate of the actual displacement or velocity of the elbow. Vibration at 30 Hz had little or no effect on the elbow angle at hand opening. Vibration at 40 Hz caused subjects to open the hand before the elbow reached the target angle ("undershooting"). Undershooting is consistent with an overestimate of the actual displacement or velocity of the elbow. The size of the error depended on the velocity of the passively imposed elbow rotation. 5. The effect of tendon vibration also depended on the timing of vibration. If 40-Hz vibration began at the onset of movement, the subject undershot the target. If 40-Hz vibration started 5 s before movement onset and continued throughout the movement, the undershoot error increased in magnitude. However, if 40-Hz vibration started 5 s before movement onset and then stopped at movement onset, the subject overshot the target. When vibration was shut off during movement, a transition occurred from an over-shooting error to an undershooting error at a time that depended on the velocity of elbow rotation. 6. In a separate experiment, subjects were instructed to match either the perceived dynamic position or the perceived velocity of rotation imposed on the right elbow by actively rotating the left elbow. In both matching tasks, tendon vibration produced oppositely directed errors depending on the frequency of vibration. Vibration at 20 Hz produced a perception of decreased elbow velocity and a bias in dynamic position in the flexion direction, and vibration at 40 Hz produced the opposite perceptions. 7. We conclude that muscle spindle afferents, which are activated by tendon vibration, are an important source of the dynamic position and velocity information that the CNS uses to coordinate this movement sequence task. The observed effects of vibration timing and frequency suggest that perceptual changes evoked by vibration cannot be explained by the simple summation of sensory input evoked by movement and by vibration. Rather, the bias in perception produced by vibration appears to be related to the difference between vibration- and movement-evoked activity in muscle spindle afferents.
