(a) Participants in the MEG study were assigned to one of three pharmacological groups. At the start of each experimental session, they orally took 40 mg atomoxetine (Strattera), 5 mg donepezil (Aricept), or placebo. Since the time of peak plasma concentration is 3 hr for donepezil (Rogers and Friedhoff, 1998) and 1–2 hr for atomoxetine (Sauer et al., 2005), we used a placebo-controlled, double-blind, double-dummy design, entailing an identical number of pills at the same times before every session for all participants. Participants in the donepezil group took 5 mg of donepezil 3 hr, and placebo 1.5 hr before starting the experimental session. Participants in the atomoxetine group took placebo 3 hr, and 40 mg of atomoxetine 1.5 hr before the experimental session. Those in the placebo group took identical-looking sugar capsules both 3 and 1.5 hr before starting the session. This ensured that either drug reached its peak plasma concentration at the start of the experimental training. The drug doses were based on previous studies with healthy participants (Chamberlain et al., 2009; Rokem and Silver, 2010). Blood pressure and heart rate were measured and registered before subjects took their first and second pill. In the 3 hr before any MEG or training session, participants waited in a quiet room. In total, 19 people in the placebo, 22 in the atomoxetine, and 20 in the donepezil group completed the full study. (b, c) Choice history biases separately for each pharmacological group. Since we did not observe differences in choice history bias between these groups, we pooled all observers for the main analyses.