Article Figures & Data
Figures
- Figure 1.
Experiment 1: passively guided movements induce a change along a task-redundant dimension through use-dependent learning. A , Participants moved a line indicating the vertical hand position to a goal box. The horizontal hand position was not relevant to the task. B , Overall movement angle (direction of movement measured from beginning to end). During passive trials (gray box), the robot moved the participants' hand on a trajectory angled 8° rightward (red) or leftward (blue). The manipulation changed the angle of the subsequent free movements. The dashed lines indicate model fit (planned movement angle, w).
- Figure 2.
Experiment 2 shows simultaneous action of use-dependent and error-based learning. A , After a pretest of 15 movements, a tilted trajectory (±8°) was imposed using a force channel (gray box). B , The lateral force exerted against the channel (measured at 200 ms after movement start) increases during the channel trials, indicating force field adaptation (v). C , The initial angle of the subsequent free movements (measured at 200 ms) shows a error-based aftereffect against the direction of the channel, which washes out quickly. D , The overall movement angle shows a persistent aftereffect of use-dependent learning after washout of the force field effect. The dashed black lines indicate model fit.
- Figure 3.
Experiment 3. Shown are force field adaptation in the standard reaching task ( A ) and redundant reaching task ( B ). Initially, the clockwise (blue) and counterclockwise (red) force field caused large errors in the trajectory that is fully corrected in the standard ( C ), but not in the redundant task ( D ). After adaptation, movements in the standard task are nearly straight ( E ), whereas they deviate into the direction of the force field in the redundant task ( F ). G , H , Time evolution of adaptation and fit of the use-dependent learning model. The initial movement angle (y) and fit (ŷ) are shown in blue. In both tasks, there is an equivalent increase in the expected force (v, black dashed line; here scaled by D ). Only in the redundant task there is there a concurrent change in the planned direction (w, red dashed line), leading to a shift of both the initial (blue) and overall (red) movement angle.
- Figure 4.
Estimating amount of force field adaptation and shift in planned direction in experiment 3. A , After adaptation to a clockwise (blue) or counterclockwise (red) force field in the standard task, a channel was applied in the target direction (black dashed line). Participants produced anticipatory forces against the wall of the channel (solid red and blue line), similar to the force field experienced on nonchannel trials (gray line). Model fits (see Materials and Methods) are shown in dashed lines in all panels. B , Before adaptation in the redundant task, a channel was applied in three directions (light, −8.8°; medium, 0°; dark, −8.8°) around the average movement direction. Even after the end of the movement (t > 400 ms), participants pressed against the channel, providing a measure of their planned movement direction. C , After force field adaptation in the redundant task, a channel was applied in three different directions around the respective average direction. Resultant force profiles were modeled as a mixture of force field adaptation and deviations of planned direction (dashed lines). D , E , Parameter estimates indicate equivalent force field learning in standard and redundant tasks ( D ), and a shift of the planned direction in the direction of the force field in the redundant task ( E ). Error bars indicate SEM.
Additional Files
Supplemental Data
Files in this Data Supplement:
- supplemental material - Supplemental Material
