11, p =

0.059, whole-brain FWE corrected). Next, we examined the LFPC’s involvement in the three tasks involving explicit decisions (Precommitment, Choice, and Opt-Out). We extracted E7080 manufacturer parameter estimates from our ROI in LFPC based on a previous study (−34, 56, −8; Boorman et al., 2009) for LL decisions in the three decision tasks and conducted a repeated-measures ANOVA to compare LFPC activation across tasks (Figure 4C). This analysis demonstrated a significant main effect of task on LFPC activity (F(3,17) = 5.573, p = 0.008). Pairwise post hoc comparisons revealed that LFPC activation was significantly greater during precommitment choices than during LL choices in the Opt-Out task (t(19) = 3.83, p = 0.003, Bonferroni corrected). The LFPC mean parameter estimate for precommitment choices was also greater than that for LL choices in the Choice task, but the difference did not survive correction for multiple comparisons, mirroring our behavioral self-control findings (compare Figure 4C with Figure 2C). We note that the Choice task, like the Precommitment task, also involves the opportunity to make a binding choice for LL; our results therefore support the notion that the LFPC is sensitive to the opportunity to make binding choices for large, but delayed, rewards. For comparison, we also investigated whether regions involved in willpower (DLPFC, IFG, and PPC) were sensitive to opportunities

to precommit. We extracted parameter estimates from these regions (using RO4929097 ROI coordinates from previous studies; Table S8) during LL choices in the three decision tasks and subjected them to a repeated-measures ANOVA. None of these regions were sensitive to opportunities to precommit (Figure S1); the effect of task was not significant for DLPFC (F(3,17) = 1.676, p = 0.215), IFG (F(3,17) = 1.209, p = 0.322), or PPC (F(3,17) = 0.924, p = 0.415). Thus, DLPFC, IFG, and PPC showed activation

patterns consistent with their role in self-control more generally but were not sensitive to opportunities to precommit. Finally, we subjected the parameter estimates from LFPC, DLPFC, IFG, and PPC for the three decision tasks to a repeated-measures ANOVA with region and task as within-subjects of factors. Parameter estimates were z transformed to control for differences in mean parameter estimates across regions. This analysis revealed a significant interaction between region and task (F(6,114) = 3.989, p = 0.001), confirming our above observations that the LFPC was differentially activated across decision tasks, but the regions engaged during willpower (DLPFC, IFG, and PPC) were not. We next investigated the possibility that LFPC implements decisions to precommit by controlling activity in the DLPFC, in line with theories positing that the LFPC sits at the top of a cognitive control hierarchy from which it orchestrates different courses of actions represented in DLPFC (Tsujimoto et al.