We investigated whether a representation of a visual target can be stored in memory and used to support the online control of reaching movements. To distinguish between the use of a stored target representation for movement planning versus online control, we employed a novel movement environment in which participants could not fully plan their action in advance of movement initiation; that is, the spatial mapping between the movement of a computer mouse and the on-screen movement of a cursor was randomly varied from trial to trial. As such, participants were required to use online control to reach the target position. Reaches were examined in full-vision and three memory-dependent conditions (0, 2, and 5 s of delay). Absolute constant error did not accumulate between full-vision and brief delay trials (i.e., the 0-s delay), suggesting a stored representation of the visual target can be used for online control of reaching given a sufficiently brief delay interval. Longer delay trials (2 and 5 s) were less accurate and more variable than brief delay trials; however, the residual accuracy of these memory-dependent actions suggests that the motor system may have access to a stored representation of the visual target for online control processes for upwards of 5 s following target occlusion.
Matthew Heath and David A. Westwood
Matthew Heath, David A. Westwood and Gordon Binsted
The goal of the present investigation was to explore the putative contributions of feedforward- and feedback-based processes in the control of memory-guided reaching movements. Participants (N = 4) completed an extensive number of reaching movements (2700) to 3 midline targets (20, 30, 40 cm) in 6 visual conditions: full-vision, open-loop, and four memory-guided conditions (0, 200, 400, and 600 ms of delay). To infer limb control, we used a regression technique to examine the within-trial correspondence between the spatial position of the limb at peak acceleration, peak velocity, peak deceleration, and the ultimate movement endpoint. A high degree of within-trial correspondence would suggest that the final position of the limb was largely specified prior to movement onset and not adjusted during the action (i.e., feedforward control); conversely, a low degree of within-trial correspondence would suggest that movements were modified during the reaching trajectory (i.e., feedback control). Full-vision reaches were found to be more accurate and less variable than open-loop and memory-guided reaches. Moreover, full-vision reaches demonstrated only modest within-trial correspondence between the spatial position of the limb at each kinematic marker and the ultimate movement endpoint, suggesting that reaching accuracy was achieved by adjusting the limb trajectory throughout the course of the action. Open-loop and memory-guided movements exhibited strong within-trial correspondence between final limb position and the position of the limb at peak velocity and peak deceleration. This strong correspondence indicates that the final position of the limb was largely determined by processes that occurred before the reach was initiated; errors in the planning process were not corrected during the course of the action. Thus, and contrary to our previous findings in a video-based aiming task, it appears that stored target information is not extensively (if at all) used to modify the trajectory of reaching movements to remembered targets in peripersonal space.