The influence of pre-training evaluative responses on approach-avoidance training outcomes.

Approach-avoidance training (AAT) has been shown to be effective in both clinical and laboratory research. However, some studies have failed to show the effects of AAT. Therefore, finding moderators of the AAT effect is a priority for further research. We investigate the moderating effect of pre-training evaluative responses towards familiar AAT targets. In particular, we test predictions: (a) that congruent responses (i.e. approach to positive targets and avoidance of negative targets) increase liking, whereas incongruent responses decrease liking; (b) that training is more effective when it can strengthen existing positivity or negativity; and (c) that ambivalence increases AAT effects. Two experiments (total N = 132) implemented an AAT with local soft-drink brands after measuring initial positive/negative explicit evaluative components and implicit liking towards the brands. Results show no reliable evidence for training effects on consumption or rating of drinks, but participants showed more implicit liking of approached drinks than avoided drinks. Furthermore, the magnitude of implicit liking measured pre-training was positively related to the size of the training effect. Ambivalence had no direct effect on the training outcomes. These results partially support the congruency prediction and underline the importance of implicit liking prior to AAT as a moderator for AAT effects.