To test our hypothesis that urgency traits would be related to ROI activation, we entered negative urgency as a covariate in the previously conducted Mood (3) × Odor (2) linear mixed-effects model. Adding the covariate significantly improved model fit in the left vmPFC (X2 difference = 7.87, p = 0.025) and the right vmPFC (X2 difference = 9.20, p = 0.01), as indicated by log likelihood difference tests, with negative urgency being a significant predictor of BOLD in left (p = 0.03) and right vmPFC (p = 0.03). There was no improvement of fit in the left or right mPFC. In addition, when negative urgency was added to the models, the main effects of odor were no longer significant (p = 0.51 in the left vmPFC and p = 0.91 in the right vmPFC). Thus, after accounting for urgency, there were no longer any differences in the magnitude of responses between AcO and AppCo. Positive urgency was not a significant variance covariate in any of the ROIs.
