Supplementary Figure 1: Emotion intensity ratings for scenario stimuli.

A separate sample of participants was recruited to provide ratings of affective responses to GL and PL scenarios (GL group: n = 21; PL group: n = 20). The design of this study was identical to our imaging study, except that subjects performed the study in a laboratory at Vanderbilt rather than in an MRI scanner, and instead of punishment decisions they were asked to report on how strong their emotional experience was for six discrete emotions using a 0-9 scale, where 0 indicated “not at all” and 9 indicated “extreme”. The emotions rated were anger, contempt, disgust, fear, and sadness. Across all emotion categories we found a trend-level main effect of graphic language, such that individuals in the GL group gave higher ratings than those in the PL group (F_(1,39) = 2.98, p < 0.092), as well as a significant Group X Emotion interaction (F_(1,36) = 3.13, p = 0.027). This interaction was primarily driven by low ratings for fear across both groups; when fear was dropped from the ANOVA, a significant main effect of graphic language was present across the remaining 4 emotions, such that individuals in the GL group gave higher ratings than those in the PL group (F_(1,39) = 4.54, p = 0.040). Reported p-values indicate results of post-hoc between-groups t-tests (*p > 0.05, **p > 0.01, one-tailed). Error-bars represent standard error of the mean.