Fig. 6: Interference.

`Anterograde interference' refers to the phenomenon where humans sometimes adapt slower to a condition A when they were previously exposed to the opposite condition B, that is, with the belt speed differences reversed between the two belts. a We performed simulations of two split-belt adaptation protocols: first, T-A-T-A-T-A, alternating between tied-belt conditions T and the split-belt condition A, and second, T-A-T-B-T-A, where one of the A phases is replaced with the opposite condition B. We compare the adaptation between the two protocols during the final A phase (denoted as yellow shaded region). We find that the two protocols are not significantly different in their initial response to the perturbation or the early change in the step length asymmetry for the final adaptation period (yellow shaded region), as shown by Malone et al⁵². b To test if this non-interference remains in the absence of washout, we performed prospective experiments in the absence of such a tied-belt washout phase: we compared protocols T-A with T-B-A with both simulations and human participant experiments. In experiments, we found that the initial step length asymmetry (first step of A) was significantly higher when B was present and the time constant of adaptation during A was not significantly different under the two conditions. This confirmed our model simulations, which predicted that the initial transients for A will be higher after B. The model also predicted no statistically significant difference in the adaptation rate constant in the presence of inter-participant variability of magnitude similar to that in the experiment. All box-plots show the median (red bar), 25–75% percentile (box), non-outlier range (whiskers), and individual data points (pink circles). The time-series shows median as thick colored line and light gray lines are individual participant data overlaid. See Supplementary Table 1 for statistical details of comparisons.