Fig. 3: Group differences in exploration and learning on the horizon task.

Plots depict participants’ computational parameter estimates for the Horizon Task model, where parameter values in task runs with and without the anxiety induction (breathing resistance) for each participant are connected using thin lines. Thick lines and surrounding confidence ribbons represent the mean and standard error for parameter values in each group. Results indicate that the MUD group showed lower levels of directed exploration, random exploration, and learning rates compared to HCs, both with and without the breathing resistance manipulation.