Abstract
The saccade generator accounts for previous saccades in order to ‘update’ internal representations for subsequent saccades. Here we tested the role of limb proprioception on updating the locations of hand-held objects for the saccade generator. We measured radial saccades (8 directions, approximately 7 and 13 degrees eccentricity) from a central target in 6 human subjects under the following conditions: 1. Visual controls (saccade to the peripheral flashing LED, with or without memory delay); 2. Proprioception alone (after hand-held LED target was extinguished, hand was moved passively or actively from the center to the periphery, followed by saccade to the perceived target location); 3. Combination of vision and proprioception (like #2, but target flashed again after the hand movement). In all trials, the eye held the central fixation point until an auditory command told them to saccade to the peripherally shifted target (in complete darkness). We recorded eye and arm movements with search coil and optotrak systems respectively. Subjects were able to use the spatial updating from proprioception to generate approximately accurate saccades, but tended to overshoot the targets. Proprioception alone trials were least accurate, followed by combination of vision and proprioception, and then vision alone (which was best). In addition, proprioception-related errors were different for different movement directions (being greater for horizontal and centripetal movements). Surprisingly, Compared to visual controls, the combination of proprioception and vision gave reduced saccade accuracy, perhaps due to the proportionate weighting of each of these two different inputs during the integration.