Strong cravings for unhealthy foods and implicit tendencies to approach them threaten the physical and mental health of vulnerable populations. Yet, implicit measures of food approach tendencies have methodological limitations, as existing approach-avoidance tasks (AAT) are often unreliable and require specialized hardware. We propose a novel method to measure approach biases: on a touchscreen, participants slide their hand either toward a food item (and away from control images) or away from it (and toward control images) in separate blocks. Adequate attention to the stimuli is ensured by the coupling of stimulus category to the required response. We found that this touchscreen-variant of the AAT yielded reliable bias scores when approach and avoidance were defined as movements relative to the stimulus rather than to the body. Compared to control images, we found an approach bias for low-calorie foods but not for high-calorie foods. This bias additionally varied on a food-by-food basis depending on the participant's desire to eat individual food items. Correlations with state and trait cravings were inconclusive. Future research needs to address the order effects that were found, in which participants avoiding foods in the first block showed larger biases than participants approaching food in the first block, likely due to insufficient opportunity to practice the task. Our findings highlight the need for approach bias retraining paradigms to use personalized stimulus sets. The task can enrich the methodological repertoire of research on eating disorders, obesity and cognitive bias modification.
