When we want to make accurate hand movements to for instance a visual target, it is assumed that we combine visual and proprioceptive information to localize our hand precisely. It has been suggested that you do the same for visual targets. In this case, visual localization is based on the target’s position relative to your fovea, and proprioceptive localization is relative to your hand. If you cannot see your hand, this proprioceptive target information will degrade, as updating the location of the target relative to your moving hand adds noise. If this proprioceptive estimate of the target is relevant, movements will become more precise and accurate if we are able to keep a proprioceptive estimate, for instance by placing the other (invisible) hand near the target. This unseen hand does not add new information; it only copies the visual information to proprioception. We tested this prediction by asking subjects to repeatedly move their right index finger between several visual targets on a table. We repeated the experiment after the subjects put their left index finger under one of the target locations (without any feedback). Subjects made considerable errors in positioning their invisible right finger. As predicted, adding a non-informative finger improved the accuracy of the movements (by about 20%). The bias of the invisible right hand after repetitive movements was on average shifted towards the left hand’s location. Our results support the hypothesis that for goal-directed movements, we use proprioceptive estimates of visual targets.

Proprioception; vision; goal-directed movements