Table 5.
Two-way repeated-measures ANOVA of value representation in ventral and dorsal subregions of vmPFC
| Factor | Mean square | F | Significance |
|---|---|---|---|
| Region | 0.002 | F(1,17) = 0.035 | 0.854 |
| Outcome | 0.242 | F(2,34) = 10.597 | 0.000442 |
| Region × outcome | 0.072 | F(2,34) = 4.764 | 0.015 |
In this analysis, the dependent variables were the mean percentage change in BOLD activity in the ventral and dorsal subregions of vmPFC (Fig. 5) during the second to fourth TRs following outcome. These TRs were selected because they covered the entire duration of the rise in BOLD to peak. The three TRs immediately before the beginning of a trial were used as the baseline. There were no outliers in the data as assessed by inspection of a boxplot. The dependent variables were normally distributed, as assessed Shapiro–Wilk’s test of normality, except for a minor violation for responses to no reward in the ventral subregion (all p > 0.05, except for no reward in the ventral subregion p = 0.023). Given the known robustness of repeated-measures ANOVA to minor normality violations, we proceeded with the raw, untransformed data. In this analysis, the assumption of sphericity for region and its interaction with outcome was automatically met, because it only had two levels. The assumption of sphericity for outcome was met, as assessed by Mauchly’s test for sphericity (p = 0.528). As shown above, there was a statistically significant two-way interaction between region and outcome, F(2,34) = 4.764, p = 0.015. In the dorsal subregion, there was a statistically significant simple main effect of outcome (F(2,34) = 13.68, p = 0.0001), which was not observed in the ventral subregion (F(2,34) = 1.11, p = 0.34). A follow-up pairwise comparison of the dorsal subregion’s response to different outcomes revealed significant differences between no reward and $10 reward (p = 0.00024, Bonferroni corrected) and between $5 and $10 (p = 0.022, Bonferroni corrected), but not between no reward and $5 (p = 0.19). Boldface values indicate effects that are statistically significant at 0.05 level.