Fast, online control of movement is an essential component of human motor skills, as it allows automatic correction of inaccurate planning. The present study explores the role of two types of concurrent signals in error correction: predicted visual reafferences coming from an internal representation of the hand, and actual visual feedback from the hand. While the role of sensory feedback in these corrections is well-established, much less is known about sensory prediction. The relative contributions of these two types of signals remain a subject of debate, as they are naturally interconnected. We address the issue in a study that compares online correction of an artificially induced, undetected planning error. Two conditions are tested, which only differ with respect to the accuracy of predicted visual reafferences. In the first, “Prism” experiment, a planning error is introduced by prisms that laterally displace the seen hand prior to hand movement onset. The prism-induced conflict between visual and proprioceptive inputs of the hand also generates an erroneous prediction of visual reafferences of the moving hand. In the second, “Jump” experiment, a planning error is introduced by a jump in the target position, during the orienting saccade, prior to hand movement onset. In the latter condition, predicted reafferences of the hand remained intact. In both experiments, after hand movement onset, the hand was either visible or hidden, which enabled us to manipulate the presence (or absence) of visual feedback during movement execution. The Prism experiment highlighted late and reduced correction of the planning error, even when natural visual feedback of the moving hand was available. In the Jump experiment, early and automatic corrections of the planning error were observed, even in the absence of visual feedback from the moving hand. Therefore, when predicted reafferences were accurate (the Jump experiment), visual feedback was processed rapidly and automatically. When they were erroneous (the Prism experiment), the same visual feedback was less efficient, and required voluntary, and late, control. Our study clearly demonstrates that in natural environments, reliable prediction is critical in the preprocessing of visual feedback, for fast and accurate movement.
Read full abstract