Proprioceptive control of interjoint coordination.

This paper reviews a series of experiments comparing intact controls with functionally deafferented patients to determine the role of proprioception in controlling dynamic interactions between limb segments during movement. We examine the control of hand path in a planar movement-reversal task and in a familiar three-dimensional gesture with similar biomechanical characteristics. In the planar task subjects had to move their hand out and back along a series of straight-line segments in the horizontal plane without visual feedback. The lengths and directions of the target line segments were chosen to require different amounts of shoulder motion while requiring the same elbow excursion. In controls, hand paths were, as required, straight with sharp bends at the outermost point. In patients, however, distinctive errors appeared at movement reversals, consisting of widened hand paths resulting from desynchronization in the reversals of elbow and shoulder motions. These errors reflected an inability to program elbow muscle contractions in accord with interaction torques produced at the elbow by variations in acceleration of the shoulder. The reversal errors were substantially reduced after patients had practiced for a few trials while visually monitoring movements of their arm. The improvement was not limited to the direction where they had practiced with vision, but also extended to other directions in which the elbow torques were different. This suggests that practice with vision of the arm served to improve the general rules that subjects used to plan movement, rather than simply improving the performance of a specific response. Similar to their performance on the planar task, the patients made errors in interjoint coordination during unconstrained three-dimensional gestures with movement reversals. We conclude (i) that both the planning and the learning of movement required an internal model of the dynamic properties of the limb that takes account of interaction torques acting at different joints; (ii) that this internal model is normally established and updated using proprioceptive information; but (iii) that when proprioception is lacking, vision of the limb in motion partially substitutes for proprioception.