alcohol on impaired posterror adjustment was mediated independently by both reduction in the ERN and by the relationship between NA change and the ERN, considering that the multiply mediated path contributed significantly to explaining the variance in posterror adjustment. However, it is important to acknowledge that, although the causal link from beverage consumption to both NA change and the ERN is clear, and although NA change was measured before the ERN, it is not clear that NA change caused changes in the ERN. Thus, although consistent with the idea of a temporally prior mediating variable, the causal direction from NA change to the ERN cannot be inferred from our analysis. The temporal ordering of the ERN relative to posterror adjustment is also unclear, and therefore a potential causal effect in that part of the model also must be interpreted with caution.
