Article Figures & Data
Figures
- Figure 1.
Experimental design. A, General setup and convention for aiming landmarks and target. Participants were provided with a ring of numbered landmarks—spaced 5.625° apart—to report their aim prior to each reach. The target was always presented directly ahead of the starting location at 0°. The calculation of the implicit component of adaptation was performed by subtracting the participant’s self-reported explicit aim from the reach angle for each trial. B, Experimental protocol. In the baseline block (yellow), participants completed 48 trials with veridical end point feedback. In baseline plus report (green), participants practiced reporting their aim while receiving veridical end point feedback for eight trials. In the rotation plus report, participants were introduced to a 45° CW visuomotor rotation while still verbally reporting their aiming location for 320 trials. In the aftereffect block (purple), participants were instructed to reach directly for the target in the absence of both aiming landmarks and end point feedback for 40 trials. The rotation was removed from these trials to measure the amount of sensorimotor recalibration present. Last, the washout block (gray) restored veridical end point feedback for 40 trials. No-feedback, no-rotation single catch trials (purple dashed lines) were periodically collected every 40 trials during the rotation block. These trials had the same format as the aftereffect block (purple).
- Figure 2.
Target and aim groups. The catch trials (green) and aftereffect (black) were measured at various locations in the workspace relative to the 0° target location (key to the right illustrates catch and aftereffect locations relative to the target). Implicit aim (magenta) was calculated from subtracting explicit aim (red) from reach angle (blue). The light blue shaded areas represent the target area. The purple shaded region during the aftereffect (trials 377-417) denotes no-feedback trials. The black error bar (trial 377) displays the magnitude of the first aftereffect trial. Error bars and shaded error regions denote the SEM. A, Target group. The catch trials (green) and aftereffect (black) were measured at the 0° trained target location. B, Aim group. The catch trials (green) and aftereffect (black) were measured 30° CCW of the target location, corresponding to the most frequently reported aim location. Moving the catch trial/aftereffect location to the 30° CCW aiming location removes the offset between the calculated implicit (magenta) and catch trial measurements (green) when collected at the target location observed in A.
- Figure 3.
Extreme CCW and extreme CW groups. The trained target remained at the 0° location, while the catch trials and aftereffect were probed in untrained regions of the workspace. A, Extreme CCW group. The catch trials (green) and aftereffect (black) were measured 60° CW of the target location. B, Extreme CCW group. The catch trials (green) and aftereffect (black) were measured 90° CW of the target location. Error bars and shaded error regions denote the SEM.
- Figure 4.
Cursor and hand groups. The trained target remained at the 0° location, while the catch trials and aftereffect were probed in untrained regions of the workspace. A, Cursor group. The catch trials (green) and aftereffect (black) were measured 30° CW of the target location. B, Hand group. The catch trials (green) and aftereffect (black) were measured 45° CCW of the target location, corresponding to the location where the participants reached their hands most frequently during the rotation block. Error bars and shaded error regions denote the SEM.
- Figure 5.
Implicit adaptation generalizes around the aiming location (30° CCW). φ denotes generalization, calculated as the proportion of the catch trial magnitudes (Figs. 2-4, green error bars) to the calculated implicit (Figures 2-4, magenta trace). When plotted as a function of angle relative to the trained target location, the φ values reveal a generalization curve that is centered around the aiming location (blue). Error bars denote the SEM. This is overlaid on fitted Gaussian distributions for the cursor (green; R 2 = 0.996), reported aim (red; R 2 = 0.969), and hand (gold; R 2 = 0.996) angles for all 70 participants during the rotation block. The distributions are normalized to their respective means.
- Figure 6.
Mixed aim/target group to explore the within-subject generalization effect. The catch trials (green) were measured 30° CCW of the target location, which is similar to that of the aim group in Figure 2B, corresponding to the location where the participants reached their hands most frequently during the rotation block. The aftereffect trials (black) were measured at the target location. Note that there is a drop from the catch to the aftereffect trials; this is because the catch trials were performed at the aiming location, and aftereffect trials at the target location. Error bars and shaded error regions denote the SEM.
- Figure 7.
Within-subject generalization. For each participant, the mean implicit learning magnitude was calculated at each reported aiming location. Aiming locations that had data available for >10 of 70 subjects were included. A, Implicit learning magnitude for all 70 subjects. Data for individual subjects were centered on their most frequently reported aiming locations. Thus, 0° here denotes the aiming location that the subject reported most frequently during the rotation block. The green curve denotes a best-fit cosine function. B, Correlational analysis revealed a negative correlation (correlation coefficient = −0.89, p = 0.0029) between the absolute angle away from the most frequent aiming location and the generalization of implicit learning. This collapsed all negative and positive aim angles shown in A. As participants reached farther from their most frequently reported aim, the generalization of implicit learning decays. The black trace is a best-fit linear regression (R 2 = 0.796). Error bars denote the SEM.
In this issue
