Fig. 4.

Psychiatric factors (dimensions) and decision‐making processes. (a) Main effects of past rewards and past choices on the current choice in the reward‐seeking task (mean $\pm$ SEM, n = 461 679). The mean and SEM of the effects were estimated with a generalized linear mixed‐effect model (GLMM). **P < 0.01, FDR‐corrected by the number of tests (i.e. 2) in two‐tailed t‐test (Past reward: corrected and uncorrected Ps < 0.001; Past choice: corrected and uncorrected Ps < 0.001). (b) Main effects of past no‐losses and past choices on the current choice in the loss‐avoidance task (mean $\pm$ SEM, n = 472 271). Same format as in (a) (Past no‐loss: corrected and uncorrected Ps < 0.001; Past choice: corrected and uncorrected Ps < 0.001). (c) Interaction effects on the current choice in the reward‐seeking task (mean $\pm$ SEM, n = 461 679). The mean and SEM of the interaction effects were estimated with the same GLMM as in (a). Left, interaction effect between past reward and CIT; and right, interaction effect between past choice and CIT. **P < 0.01, FDR‐corrected by the number of tests (i.e. 2) in two‐tailed t‐tests (Past reward x CIT: corrected and uncorrected Ps < 0.001; Past choice x CIT: corrected and uncorrected Ps < 0.001). CIT, compulsive behavior and intrusive thought. (d) Interaction effects on the current choice in the loss‐avoidance task (mean $\pm$ SEM, n = 472 271). Same format as in (c) (Past no‐loss x CIT: corrected and uncorrected Ps < 0.001; Past choice x CIT: corrected and uncorrected Ps = 0.934).