Fig. 1.

(a) Schematic of the PRL task. Subjects chose one of three stimuli. The timeline of a trial is depicted: stimuli appear, a choice is made, the outcome is shown, a fixation cross is presented during the intertrial interval, stimuli appear for the next trial (etc.) (RT, reaction time). One stimulus delivered positive feedback (green smiling face) with a 75% probability, one with 50%, and one with 25%. The probabilistic alternative was negative feedback (red sad face). Midway through the task, the contingencies for the best and worst stimuli swapped. s, seconds. (b) Better initial learning was predictive of more perseveration on LSD and not on placebo. Shading indicates ± 1 standard error of the mean (s.e.). (c) Trial-by-trial average probability of choosing each stimulus, averaged over subjects during the placebo session. A sliding 5-trial window was used for smoothing. The vertical dotted line indicates the reversal of contingencies. R-P indicates mostly rewarded stimulus, later mostly punished. N-N indicates neutral stimulus during both acquisition and reversal. P-R indicates mostly punished stimulus, later mostly rewarded stimulus. Shading indicates ± 1 s.e. (d) Trial-by-trial average probability of choosing each stimulus, averaged over subjects during the LSD session. A sliding 5-trial window was used for smoothing. The vertical dotted line indicates the reversal of contingencies. R-P indicates mostly rewarded stimulus, later mostly punished. N-N indicates neutral stimulus during both acquisition and reversal. P-R indicates mostly punished stimulus, later mostly rewarded stimulus. Shading indicates ± 1 s.e. (e) Distributions depicting the average per-subject probability (scattered dots) of choosing each stimulus while under placebo (shown in dark blue) and LSD (light blue). The mean value for each distribution is illustrated with a single dot at the base of each distribution, and the mean values for the probability of choosing different stimuli in each condition are connected by a line. Black error bars around the mean value show ± 1 s.e. Horizontal dotted line indicates chance-level ‘stay’ behaviour (33%). The global probability of choosing each stimulus did not differ between the placebo and LSD conditions. (f) Raw data measures of feedback sensitivity were unaffected by LSD. Distributions depicting the average per-subject probability (scattered dots) of repeating a choice (staying) after receiving positive or negative feedback under placebo (dark blue) and LSD (light blue). The horizontal dotted line indicates chance-level ‘stay’ behaviour (33%).