e., the average time series from the LFPC seed), as well PD0325901 as six motion parameters as regressors of no interest. To further investigate the results of the PPI analysis, we conducted a conjunction analysis by finding the intersection of voxels that were significant
in the willpower contrast at p < 0.05 whole-brain cluster-level corrected and that also showed significant precommitment-related functional connectivity with LFPC at p < 0.001 uncorrected with an extent threshold of 10 voxels. We tested for statistical significance using small-volume correction (p < 0.05, family-wise error corrected at the cluster level) in a priori regions of interest (ROIs) identified from the literature in DLPFC, IFG, PPC, and LFPC (Table S8). ROI masks were constructed as bilateral 10 mm spheres centered on peak coordinates from previous studies of value-based decision making (Supplemental
Experimental Procedures). We also note results outside our regions of interest that survive whole-brain cluster-level corrections. Images are displayed at a threshold of p < 0.005, k > 10 to show the extent of activation in the significant clusters. Results are reported using the MNI coordinate system. For the ROI analyses, we extracted contrast-specific parameter estimates for each ROI (identified from the literature, as above). To test for the effects of condition on responses in each ROI, we conducted repeated-measures ANOVA on the parameter estimates in SPSS v21. One subject was excluded from this analysis for having parameter estimates more than two SDs higher AZD2281 supplier than the group mean. For the cross-region comparison ANOVA, we were not interested in differences in average parameter estimates across regions but rather in the within-region differences across tasks. We therefore first z transformed the parameter estimates for each region separately by subtracting each region × task parameter estimate from the mean parameter estimate for that region (collapsed across tasks) and dividing by the SD of the parameter estimates
for that region across tasks. For the mediation analysis, Axenfeld syndrome we used hierarchical linear regression as outlined in Baron and Kenny (1986). Indirect effects in the mediation model were estimated using the SPSS procedure described in Preacher and Hayes (2004). All parameter estimates used in the mediation analyses were extracted from coordinates derived from previous studies (Table S8) to avoid nonindependence issues. vmPFC parameter estimates were extracted from the Precommit > Opt-Out LL contrast. DLPFC parameter estimates were extracted from the PPI contrast (the interaction between the neural activity in the LFPC seed and a vector coding for the main effect of decision type [1 for Precommitment, −1 for Opt-Out LL]). M.J.C. is supported by the Sir Henry Wellcome Postdoctoral Fellowship. T.K.