Fig 1.

Manipulating expected efficacy and assessing learning in Study 1. A) Trial Schematic. On each trial, participants saw a cue (gray circle), predicting the onset of a Stroop stimulus (target), and were then sequentially presented with reward and efficacy feedback. On half of the trials, efficacy feedback was presented first, and on the other half, reward feedback was presented first. Every 2–4 trials, participants were subsequently asked to estimate their current efficacy level (“How much do you think your rewards currently depend on your performance?”) or reward rate (“How often do you think you are currently being rewarded?”). B) Efficacy manipulation. We let the probability of performance-based versus random feedback continuously drift over the course of the experiment (inversed for one half of the sample). Arrows mark time points with low and high efficacy, respectively. When efficacy was low, rewards were more likely to be random, whereas when efficacy was high, rewards were more likely to be performance-based.