Fig. 3: Accuracy as a function of set size, delay, drug, and dopamine synthesis capacity.

Accuracy varies by dopamine synthesis capacity in the dorsal caudate nucleus (dCaudate DA) across all three drug sessions: methylphenidate (MPH; n = 45 and 46 for high and low dCaudate DA, respectively—referred to as “High DA” and “Low DA” in the figure), placebo (PBO; n = 46 and 45), and sulpiride (SUL; n = 46 and 46), early (the first two iterations of all stimuli; n = 46 and 46) versus late (the last two iterations; n = 46 and 46) in each block. The effects of A set size and B delay are evident early, but not late in each block. C Means and standard errors reflect that methylphenidate significantly increases, and sulpiride decreases accuracy relative to placebo, both early and late in a block (albeit the increase in early performance on methylphenidate is non-significant). Further analyses reveal that methylphenidate boosts performance in late versus early trials to a greater extent than placebo in separate mixed effect logistic regressions of accuracy, early versus late in a block. *,** indicate p < 0.05 and p < 0.01; main effects of the early model, reveal that the effect of sulpiride versus placebo is significant (in two-sided z-tests at p = 0.026) and the effect of methylphenidate versus placebo is not (p = 0.20). In the late trials, both effects are significant (p = 0.0078 and p = 0.0039, respectively). Error bars indicate ± SEM.