Table 4.
Models testing interactions with age
| Term | A: Estimate | B: Estimate |
|---|---|---|
| Intercept | 0.583 (0.121)*** | 0.508 (0.122)*** |
| NAcc (A) | Value PE (B) | 0.120 (0.063)† | 0.007 (0.054) |
| lPFC (A) | Cognitive control PE (B) | −0.139 (0.057)* | −0.117 (0.061)† |
| Age | 0.107 (0.087) | 0.120 (0.086) |
| Gender | −0.088 (0.171) | −0.056 (0.182) |
| NAcc (A) | Value PE (B) × Age | −0.090 (0.063) | 0.004 (0.056) |
| lPFC (A) | Cognitive control PE (B) × Age | 0.045 (0.057) | 0.019 (0.062) |
| Variance component | A: Estimate | B: Estimate |
| Var(π0ij) | 0.266 | 0.252 |
| Var(π1ij) | 0.033 | 0.006 |
| Var(π2ij) | 0.022 | 0.036 |
| Fit statistic | A: Statistic | B: Statistic |
| AIC | 2890.8 | 2896.5 |
| BIC | 2965.0 | 2970.7 |
Gender was dummy coded (0 = male; 1 = female). NAcc refers to univariate ventral striatum activity; lPFC refers to univariate lPFC activity. Var() refers to a variance component of a given random effect from the model. Results come from a multilevel logistic regression model, with log-odds of a risky choice as the dependent variable. Column A, Results from the classic model (lPFC threshold = 0.25); Column B, Results from the switchboard model (association maps). In order to be concise, differing terms for each model (any term involving a metric of brain activity) are included in the same line of the first column, separated by “|.”
*p < 0.05.
***p < 0.001.
†p < 0.10.