Fig. 5

Neuroimaging results. a–d Whole brain: Results of RFX GLM, n = 33, p < 0.0005, cluster-size correction, x = 0 (MNI coordinates). Model regression: \(\begin{array}{l}{\mathrm{BOLD}} = \beta _0 + \beta _1{\mathrm{RT}} + \beta _2{\mathrm{MMI}}_{{\mathrm{trial}}\_{\mathrm{specific}}} + \beta _3{\mathrm{SV}}+ \beta _4{\mathrm{priceratio}} + \beta _5{\mathrm{endowment}}\end{array}\). Six additional motion-correction regressors were included as regressors of no interest. a Neural correlates of the trial-specific MMI. b Neural correlates of the SV: We present results for the frontal lobe. Other activations are detailed in Supplementary Table 4. c Conjunction analysis. d Overlay. e ROI: The ROI analysis revealed that choice inconsistency was correlated with activation in the dACC, vmPFC (p(Bonferroni) < 0.05) and PCC (p < 0.0005, cluster-size correction), but neither with vStr nor with V1. RFX GLM, n = 33, regression model as in a–d. For illustration purposes, we set the threshold to p < 0.001 (more stringent than an FDR correction). f Subject level: Subject-level analysis representing overlap of the SV and trial-specific MMI in the vmPFC, dACC, and PCC ROIs. For each subject, we conducted a conjunction analysis on the brain areas that significantly tracked trial-specific MMI and SV. Most subjects had an overlap region in the vmPFC (21 of 33), dACC (18 of 33), and PCC (21 of 33). FFX GLM, regression model as in a–d. We set a liberal threshold of p < 0.15 due to lack of statistical power. For three subjects, the threshold was set to p < 0.2, MNI coordinates (see Supplementary Table 5 as well). g Comparison of subject-level mean effects (% signal change, β values) of trial-specific MMI and SV in the vmPFC, dACC and PCC, using two-sided Wilcoxon sign-rank test (n.s. not significant, **p < 0.01). In all panels a–f, results are shown on the Colin 152-MNI brain