Figure 3.

(A) Protected exceedance probabilities for the 6 candidate models. (B) Model-fit context-independent learning rates. (C) Model-fit context-dependent learning rates. (D) Prediction reset factors. Each circle denotes one participant. There were no significant differences in (E) context-independent learning rates, or (F) context-dependent learning rates according to the best-fit reset parameter. (G) There was a trending positive relationship between context-independent and context-dependent learning rates. (H) Mean response time on the attentional flexibility learning paradigm as a function of the quintile of model-derived unsigned PE. Error bars denote 1 between-subjects SEM.