Estelle Raffin

Background Despite advances in the etiology of anorexia nervosa (AN), a large subgroup of individuals does not profit optimally from treatment. Perfectionism has been found to be a risk factor predicting the onset, severity, and duration of AN episodes. To date, perfectionism has been studied predominantly by the use of self‐report questionnaires, a useful approach that may, however, be impacted by demand characteristics, or other distortions of introspective or metacognitive access. Methods Here we circumvent these problems via a behavioral paradigm in which participants perform a modified Go/NoGo task, whilst self‐evaluating their performance. We compared a group of 33 adolescent females during their first episode of AN (age = 16.0) with 29 female controls (age = 16.2), and 23 adolescent girls recovered from AN (age = 18.3) with 23 female controls (age = 18.5). The controls were closely matched by intelligence quotient and age to the two clinical groups. Results First‐episode AN and control participants performed equally well on the task (reaction time and errors of commission), whereas the recovered group displayed significantly faster reaction times but incurred the same error rate. Despite performing at least as good as and predominantly better than control groups, both clinical groups evaluated their performances more negatively than controls. Conclusion We offer a novel behavioral method for measuring perfectionism independent of self‐report, and we provide tentative evidence that this behavioral manifestation of perfectionism is evident during first‐episode AN and persists even after recovery.