Fig. 2: Participant- and model-generated behavior during reward learning in the high eating-disorder (‘HED’, red) and low eating-disorder (‘LED’, gray) symptom groups.

A The average proportion of correct choice for the more rewarding category in the control and food tasks (across blocks), and B within the control task between the first block (before the emotional arousal manipulation, ‘pre-arousal’) and second block (after the emotional arousal manipulation, ‘post-arousal’); participants are represented by dots. C The average trial-by-trial proportions of correct choice within the food task and D control task (dotted line at trial 20 indicates a ‘change-point’ event where there was a switch in the rewards associated with each category). The LED and HED groups did not differ in overall learning performance within or across (food/control) tasks and (pre-/post-arousal) blocks. Diamond points indicate average model predictions from the best-fitting model across participants. Error bars represent the standard error of the mean.