Figure 1.

Both groups performed the task well, with no group differences. Participants in the (A) HDGC (n = 35) and (B) control (n = 35) groups learned to choose the most rewarding symbol in the gain (G) trials and avoid the most punishing symbol in the loss (L) trials. Mean participant choice data are shown by the dashed lines, with shaded regions showing ±SEM, green for gains and red for losses. The connected dotted green and red lines show mean computational model performance for gains and losses, respectively, showing computational model–predicted behavior well in both groups. (C) Both groups showed marginally better performance in learning from gains than in learning from losses with small positive reward bias (mean percentage correct in gains minus loss). (D) Percentage correct in gains and losses by group. Bars show mean correct, with error bar showing ±SEM. No significant group difference was observed in these metrics.