Figure 3.

Results Computational Learning Model

(A) Group-averaged parameter differences in the computational learning model between L-DOPA and placebo. Parameter w represents a measure of model-based over model-free control. Error bars represent SEM. (B) Single-subject data for parameter values in (A). Each data point represents the parameter value of a single subject. Subjects above the diagonal (circles) had higher parameter values in the L-DOPA compared to placebo condition, while subjects below (crosses) had smaller parameter values. The relative degree of model-based control was higher in the L-DOPA condition in 14 out of 18 subjects. See Figure S1 and Table S1 for validation of the winning model and Table S2 for model comparison details.