THE RELATIVE CONTRIBUTIONS OF VISION AND PROPRIOCEPTION TO JOINT ANGLE ESTIMATION AND MOTOR PLANNING 

Samuel J. Sober and Philip N. Sabes

Dept. of Physiology, University of California, San Francisco, San Francisco, CA,  94143, U.S.A.

Visually guided reaching tasks require that an experimental subject combine visual information about a target's location in space with visual as well as proprioceptive information concerning the location and state of the reaching limb. We have investigated the relative contributions of vision and proprioception to subjects' estimates of the intrinsic state (joint angles) of their reaching arm. Our task consisted of a sequence of center-out reaches performed with neither online visual feedback nor knowledge of the reach's accuracy. Three blocks of center-out reaches were performed when the subject received normal visual feedback prior to the movement as well as feedback that had been visually displaced (prism shifted) in two opposite directions by means of a virtual reality environment. These data were analyzed in the context of a model that uses the subjects' initial reach velocity, endpoint, and prism-adaptation state to assess the relative contributions of vision and proprioception to estimating the intrinsic state of the arm and planning the upcoming reach. Our data suggest that when planning a movement vector in extrinsic space, subjects rely on visual rather than proprioceptive data when assessing joint angles and calculating the change in joint angles necessary to execute a visually specified movement.