People with anorexia nervosa (AN) commonly exhibit elevated anxiety and atypical reward responsiveness. To examine multivariate neural patterns associated with reward and the impact of anxiety on reward, we analyzed fMRI data from a monetary reward task using representational similarity analysis, a multivariate approach that measures trial-by-trial consistency of neural responses. Twenty-five adolescent girls with AN and 22 mildly anxious controls lacking any history of AN were presented personalized anxiety-provoking or neutral words before receiving a reward, and neural response patterns in reward regions were analyzed. Consistent with our preregistered hypothesis, AN participants showed lower representational similarity than controls during neutral-word rewarded trials. Within groups, controls showed significant representational similarity in reward circuit regions including the left nucleus accumbens, left basolateral amygdala, and left medial orbitofrontal cortex, which were not observed in AN. Further, reward-related prefrontal cognitive control areas - left ventrolateral prefrontal cortex and left dorsolateral prefrontal cortex - showed significant representational similarity in both groups, but a larger spatial extent in controls. Contrary to predictions, there were no significant between-group differences for the effects of anxiety-words on reward representational similarity, and representational similarity did not predict longitudinal symptom change over six months. Overall, the results demonstrate relatively inconsistent trial-by-trial responses to reward receipt in the neutral state in AN compared with controls in both reward circuit and cognitive control regions, but no significant differential effects of anxiety states on reward responses. These results add to dynamic understandings of reward processing in AN that have potential implications for planning and guiding reward-focused interventions.
Keywords: Representational similarity analysis; anxiety-provoking word processing; fMRI; monetary reward processing.