Impaired reinforcement learning and behavioral activation/inhibition systems in internet addiction

Abstract Background The goal of this study is to examine whether goal-directed control and learning rate are more closely associated with internet addiction (IA) and analyze maladaptive behavior in IA through reinforcement learning processes and personality traits by looking at the personality traits of people with IA, thereby validating its alternative for diagnosing and measuring IA. Methods A total of sixty-one participants with IA and sixty-one healthy participants completed the Internet Addiction Test (IAT), the Behavioral Inhibition System/Behavioral Activation System (BIS/BAS) scale, and the two-step Markov decision task. Through hybrid modeling and multivariate regression, the relationship between reinforcement learning parameters, personality traits, and IA scores was analyzed. Results Significant impairment in goal-directed system was observed in the IA group, although this impairment did not correlate with the degree of IA. In comparison to the healthy control group, the IA group exhibited a significantly higher learning rate, which was positively correlated with the severity of IA and reward sensitivity. Furthermore, the BIS score and learning rate were predictive of IAT scores. Conclusions Maladaptive behavior in IA can be attributed partially to deficits in goal-directed system and an elevated learning rate. Individuals with a higher BIS sensitivity are more likely to experience IA. Incorporating behavioral modeling parameters and personality factors might aid in IA diagnosis.