More by Stick Than by Carrot: A Reinforcement Learning Style Rooted in the Medial Frontal Cortex in Anorexia Nervosa

Anorexia nervosa (AN) is characterized by a relentless pursuit of thinness, despite serious implications for health and social relations. In a previous study wielding the power of computational psychiatry, we found alterations in learning from negative feedback and in neural activity in the posterior medial frontal cortex (pMFC) in young acutely underweight AN patients (acAN). Here we ask whether these abnormalities are merely a state-related consequence of the illness or whether they might constitute a trait marker predisposing individuals to AN. To address this question, we employed the same reinforcement learning paradigm during fMRI with 31 female former AN patients after complete weight-recovery (recAN) and 31 age-matched healthy volunteers (15-28 years). Participants performed a decision task that required adaptation to changing reward contingencies. Data were analyzed within a hierarchical Gaussian filter model, which captures interindividual variability in feedback learning and decision-making under uncertainty. Similar to acute patients, individuals recovered from AN appear to emphasize negative over positive feedback when updating expectations regarding changing reward-punishment contingencies (difference in learning rate between punished and rewarded trials was increased in recAN: p = .006, d = .70. This behavioral pattern was mirrored in hyperactivation of the pMFC following negative feedback (FWE p < .001). Because the previously observed alterations in acANs are also evident after recovery and do not correlate with state variables like weight, altered feedback learning might be a trait marker of AN. The neural underpinnings of these alterations may lie in the pMFC.