BACKGROUND: Chocolate is the most craved energy-dense food. Yet, most individuals can limit their chocolate consumption. Here, we investigate the cognitive mechanisms underlying chocolate consumption in a chocolate bogus taste test in a cross-sectional experimental design. METHOD: High chocolate cravers abstained from chocolate for a week, followed by a virtual reality chocolate exposure with biometric trajectory recordings of their stopping responses and an ad-libitum bogus taste test of spontaneous chocolate intake. A single-target implicit association task and a computerised stop-signal task served as unstandardised control tasks 1-2 days before chocolate intake. RESULTS: Associations of parameters from all tasks with chocolate intake were small (|r| <
0.23). Elastic net models misestimated food intake by min. 160 kcal (generalisation: 180 kcal) and feature selection was only possible with L1 penalty. At the group level, participants showed a more controlled and delayed movement towards chocolate relative to neutral cues, evidenced by lower peak acceleration and peak velocity and faster stopping latency. DISCUSSION: The findings demonstrate the complex cognitive-behavioural underpinnings of food intake, food craving and abstinence.