Fig. 2: Experiment 1A, generalization of recalibration.

a Experimental protocol (as in Fig. 1a). b Speed match task setup. Participants controlled the right belt speed and tried to match it the left “reference” speed. c Predictions for perceptual results made by three competing hypotheses: generalization of forward model recalibration is (1) none (top row), (2) full (middle row), or (3) partial (bottom row). “baseline speed match”: schematic time course of the left (dashed gray) and right (solid black) belt speed during a baseline speed match task. Participants accurately match the right speed to the left speed. “adaptation speed match”: schematic for the task performed at the end of adaptation. Participants overshoot the right speed. We compute bias as right minus left speed at the end of the task. Baseline and adaptation predictions are based on previous work and do not differ across hypotheses. “overground right leg feels”: prediction for the overground post-adaptation questionnaire. Perceptual bias is predicted by the “full” and “partial” generalization hypotheses, but not the “none” hypothesis. “treadmill decay in bias”: prediction for the first speed match task in treadmill post-adaptation. The (1) “none”, (2) “full” and (3) “partial” generalization hypotheses respectively predict (1) same perceptual bias as adaptation speed match (decay is none), (2) no bias (full decay), or (3) smaller bias (partial decay). d Experimental results for the predictions in (c). Solid black line with solid gray shade depicts right speed (group mean ± SE); dashed gray line depicts left speed. “baseline speed match” (average of three iterations) and “adaptation speed match” are consistent with the predictions in (c). “overground right leg feels”: all ten participants reported a perceptual bias at the start of overground post-adaptation (pink bar), consistent with the “full” or “partial” generalization hypotheses. “treadmill decay in bias”: speeds in the first post-adaptation speed match task. The bias is less than the adaptation speed match task, consistent with the “partial” generalization hypothesis. e Perceptual treadmill decay measure (bar ± error bar = group mean ± CI, dots = individual participants, asterisk indicates significant difference from 0 and 100%).