Figure 7.

Dependence of both behavior and β ERS on error in last trial and trial history. A, In random perturbation sessions, attenuation of β ERS after trials with large perturbation recovered with repetitive presentation, despite the absence of any difference in the error in the response between successive bins. B, Correlation between β ERS and error weighted by the squared bias and variance of previous trials in constant perturbation sessions. C, VAF of a model describing performance on the next trial through error on the last trial weighted by the squared bias and variance of previous error history in constant perturbation sessions. Both correlations and VAF were more enhanced when the previous 20 trials were taken into account than when only unweighted error was considered. Circles and bars represent mean ± SEM. *Significant difference with Bonferroni correction for multiple comparisons.