Two modes of error processing in reaching.

Several processes are devoted to error reduction for the production of purposeful actions. When motor responses deviate from their goal, online corrections can be performed either under voluntary control with additional sub-movements or under fast automatic control with smooth velocity profiles. When errors cannot be corrected online and are repeated over trials, subsequent responses can be improved iteratively through adaptation, a progressive adjustment of motor commands that acts to reduce the magnitude of error. It has been argued that reaching adaptation results essentially from a conflict between actual sensory feedback and expected sensory feedback. Here, we specifically compare two innovative hand-reaching paradigms that provide the subject with undistorted hand sensory feedback. Both paradigms induce motor planning errors unknown to the subjects. Experiment 1 yields a continuous retinal and visuomotor feedback which allows fast and complete automatic online corrections. In experiment 2, all visual feedback is eliminated during movement execution. This prevents online correction and provides information on hand-to-target visual error at movement end only. Despite a reiterated motor planning error and an automatic online correction of the whole error, experiment 1 shows a complete lack of adaptation. In contrast, experiment 2 which yields the same motor planning error exhibits a robust and generalized adaptation, although devoid of limb inter-sensory mismatch. These results demonstrate independence between the induced motor adaptation and automatic online correction, both characterized by the lack of any cognitive interference. Despite these quite different processes acting upon either motor planning or motor control, the general structure of the movement kinematics remains unaltered. A putative visuomotor cerebro-cerebellar network accounting for our results is proposed.