Altered uncertainty processing during adaptive learning in internet gaming disorder

Abstract

Internet gaming disorder (IGD) is characterized by difficulty in stopping online gaming behaviors despite negative consequences. The symptoms may reflect deficits in adaptive learning related to uncertainty representations of changing environments. However, how individuals with IGD process uncertainty information during probabilistic learning has rarely been assessed. A total of 78 participants, including 38 IGD participants and 40 healthy control subjects (HCs), completed a probabilistic learning task, in which reward probabilities and probability reversals were manipulated to create an uncertain environment. Overall and trial-by-trial-based uncertainty indices were estimated using computational modelling. The IGD group showed lower prediction accuracy and a higher proportion of choice switches, as compared to HCs. Computational modeling identified an interaction between task probability difference level and group in trial-based uncertainty. Moreover, the probability-difference-induced uncertainty in the IGD group was associated with poorer task performance. The results suggest that impaired probabilistic learning performance under uncertainty in IGD participants may contribute to insensitivity to environmental changes, which may have a negative impact on their daily choices. The findings highlight a novel, quantifiable approach to advancing understanding of the mechanisms underlying learning deficits in IGD.