Automatic control during hand reaching at undetected two-dimensional target displacements

The aim of this study was to demonstrate that goal-directed pointing movements, executed at normal speed to a small visual target, but without vision of the movement, do not rely on preprogrammed commands (open-loop process); by contrast these responses are under the control of a feedback loop, which compares the ongoing response and the goal (or its internal representation). When the location of this goal is changed at the onset of the movement, an automatic correction of the path occurs. Modification of the goal was obtained by presenting a target in the peripheral visual field that the subject had to look at and point at as quickly and accurately as possible. When the orienting ocular saccade reached its peak velocity, statistically corresponding to the hand movement onset, the target was suddenly shifted 10 degrees in a random direction. This perturbation was undetected by the subject because of the absence of perception during the saccade. For the compensation to occur, the initial orientation of the movement and also its extent had to be modified. The results revealed 1) a nearly complete compensation of the movement path and a 66- to 80-ms duration lengthening; 2) relatively short reaction times to the perturbations (from 145 to 174 ms, with effective reaction times even 40 ms shorter); 3) nearly identical spatiotemporal movement characteristics to the perturbations, regardless of whether vision of the hand was allowed, suggesting that corrections were subserved by the same mechanisms. 2. The spatiotemporal characteristics of these unconscious corrections were similar to those observed in the classical double-step experiments investigating the intentional modifications of ongoing movements and suggest that they might share some common low-level mechanisms. That is, they could rely on visuokinesthetic feedback loops, which compare the updated information provided by the eye at the end of the saccade and the proprioceptive information of the end point effector (the fingertip here); they could also rely on feed-forward processes detecting the discrepancy between an efference copy of the movement and the new goal; or they could rely on a combination of those two main processes.