Fig 5. Pattern of compensation of a typical participant in the: A) reinforcement group, B) reinforcement + error group, and C) error group.

All experimental trials are represented in the line graph (bins 1–50), where each point represents the average of 10 trials. The circles at bin 0 represent the average of the first three trials and show each participant’s behaviour immediately after perturbation onset, which was similar across groups (see Fig 6). For each group, the upper dashed line represents the optimal compensation (location to aim the hand) that maximizes the probability of hitting the target [$x_{aim}^{max\left(hits\right)}$]. The lower dashed line represents the optimal compensation (location to aim the hand) that minimizes squared error [$x_{aim}^{min\left(erro{r}^2\right)}$]. It can be seen that the Reinforcement participant had a pattern of compensation that on average maximized target hits. Conversely, both the Error participant and the Reinforcement + Error participant learned a compensation that on average minimized approximately squared error. This behavior was consistent across participants. Error bars represent ±1 standard deviation.