Fig. 1: Task and main effects.
a Reversal-learning paradigm. Two stimuli, a face and a scene, were presented simultaneously. One was associated with a reward outcome, the other with a punishment outcome, with 100% deterministic contingencies. On each trial the computer selected one image, and the participant’s task was to predict whether the highlighted stimulus would be followed by a reward or punishment outcome. Then the actual outcome was presented. The stimulus-outcome associations reversed regularly, which was signaled by either an unexpected reward or unexpected punishment outcome. Accuracy on trials immediately after an unexpected outcome (reversal trials) was the performance measure of interest. Task images were obtained with permission from ref. 48. b fMRI BOLD signal to unexpected versus expected outcomes collapsed across all three sessions (N = 94 participants). In these dual-coded images, color indicates the size of the contrast estimate and opacity codes the height of the t values (plotting procedure: refs. 113, 114). Voxels with t values above the threshold of P < 0.001, uncorrected, are fully opaque. Significant clusters (here, P < 0.05 after whole-brain cluster-level family-wise error correction) are encircled in black for red blobs or in white for blue blobs. The results are overlaid on the group-average T1-weighted anatomical scan in MNI152 coordinate space. c Across all three sessions, unexpected outcomes increased face/scene stimulus-specific BOLD signal in visual association cortex (N = 94 participants; main effect of expectancy in ANOVA: F(1,93) = 182.97, P = 2.2e-16). The stimulus-specificity index represents the outcome-related BOLD signal in the contrast (FFA: faces – scenes) – (PPA: faces – scenes). Boxplots show the median and 25th and 75th percentiles, with the whiskers extending max. 1.5 * interquartile range. Round dots next to the data density kernel represent the mean value. d, e Bayesian mixed-effects model coefficients for the main effects of methylphenidate and sulpiride (relative to placebo) on accuracy and response times (N = 88 participants). Boxplots defined as in panel c. Source data are provided with this paper. MPH methylphenidate, SUL sulpiride, arb units arbitrary units.
