Unhealthy food bias in food addiction: behavioral evidence from a probabilistic image choice task

BACKGROUND AND AIMS: Food addiction (FA) shares behavioral and neurobiological similarities with substance use disorders, but objective behavioral measures of FA are lacking. We aimed to adapt the Probabilistic Image Choice (PIC) task to assess behavioral bias toward unhealthy food, and to determine its association with FA and obesity. DESIGN: Cross-sectional study with a 1-year follow-up in a subset of participants. SETTING: Online study with participants from the UK and Ireland. PARTICIPANTS: 415 adults (50% women) recruited via the Prolific platform, with 72 completing a follow-up assessment after 12 months. MEASUREMENTS: Participants completed the Yale Food Addiction Scale (YFAS 2.0), reported height and weight for BMI calculation, and performed the adapted PIC task featuring four image categories: neutral, pleasant, healthy food, and unhealthy food. Primary outcome measures included unhealthy food bias (difference between unhealthy and healthy food image selections) and its relationship with food addiction and obesity status. FINDINGS: Individuals with severe FA or obesity showed greater selection of unhealthy food images compared to those without food addiction or obesity. Gender analyses revealed that unhealthy food bias was larger in women than in men with food addiction. Test-retest analysis showed high stability of the unhealthy food bias over a one-year period. CONCLUSIONS: The FA PIC task provides an operant behavioral measure of unhealthy food bias that is associated with food addiction, particularly in women. The stability of this bias over time suggests it may represent a trait-like characteristic with potential diagnostic and prognostic value for FA and related disorders.