The dual-feature approach-avoidance task: validity, training efficacy, and the role of contingency awareness in changing food preference

The approach-avoidance task (AAT) probes tendencies contributing to unwanted behaviours, like excessive snacking, by measuring RT differences between approach and avoidance responses to different stimuli. It retrains such tendencies using repeated avoidance of appetitive stimuli and approach of healthy alternatives. The most common paradigm, the irrelevant-feature AAT, conceals these stimulus-response contingencies by requiring approach or avoidance based on features irrelevant to the tendencies (e.g. frame colour). Unfortunately, it is an unreliable measure and not always successful as a training, likely because the stimuli can be ignored. In the novel dual-feature AAT, the <i>combination</i> of stimulus and irrelevant feature determines responses: participants approach foods and avoid objects surrounded by frame A, and vice versa given frame B. We trained 219 online participants to approach fruit and avoid chocolate using active (7:1 stimulus-to-frame contingency) and sham (1:1 stimulus-to-frame contingency) versions of these two trainings. Compared to sham, active irrelevant-feature training was associated with more selection and desire to eat fruit, and active dual-feature training increased approach bias for fruit. Participants' <i>perceived</i> rate of approaching fruit versus chocolate correlated with many outcome measures, suggesting contingency awareness plays a major role in AAT training effectiveness, challenging implicit accounts. While the dual-feature paradigm shows potential, its high error rates, RTs, and difficulty mandate improvement.