Fig. 4: Cognitive restraint as mediator between peripheral and central insulin action.

Graphic illustrates mediation model adjusted for sex, BMI and age. Cognitive restraint (based on three factor eating questionnaire) positively mediated the relationship between peripheral insulin sensitivity and the insular cortex BOLD activity (insulin minus placebo) in response to high-caloric food cues. Path coefficients and corresponding p-values are shown next to the arrows; path a indicates the relationship between peripheral insulin sensitivity and cognitive restraint, path b indicates the relationship between the cognitive restraint and the insula BOLD activity in response to high-caloric food cues; path ab indicates the indirect effect (not standardized) of peripheral insulin sensitivity on the insular cortex activity via the cognitive restraint score; path c’ indicate the direct effect of peripheral insulin sensitivity on the insular cortex activity.