TY - JOUR T1 - Analysis of Behavior of Drug Addicts in Lowa Gambling Task Using Reinforcement Learning Model TT - تحلیل رفتار معتادان در آزمون قمار آیوا با استفاده از مدل یادگیری تقویتی JF - icss JO - icss VL - 12 IS - 2 UR - http://icssjournal.ir/article-1-90-en.html Y1 - 2010 SP - 15 EP - 28 KW - Reinforcement Learning KW - Gambling Task KW - Drug Dependence KW - Computational Modeling KW - Decision Making Process N2 - Objective: The aim of this study was to assess decision making disorders in opioid dependents in riskful situations, using cognitive modeling. Method: First, the decision making process of participants (treatment-seeking dependents and the normal group) was evaluated using the Farsi version of Iowa gambling task; and then the data were analyzed using cognitive modeling. Different models of decision making in riskful situations in the gambling task were designed based on reinforcement learning and gambling task circumstances. In the next step, the models were simulated in gambling task environment and were separately fitted on the data of both groups. Thereafter, the model with higher correspondence with the subject behavior in each group was determined. Results: Both the group of treatment-seeking dependents (n=127) and the control group (n=130) showed a weak performance in the task (pure scores lower than 10). In addition, the performance of the dependent group was weaker than that of the normal group (p<0.07). Modeling results indicated that the optimum model in both groups was the model that was focused only on the frequency of reward and punishment (not on their size), in the assessment of different choices. Sensitivity revealing method revealed that the optimum model for the dependent group, in addition to merely focusing on the frequency of reward and punishment, is affected by harm-avoidance more than the control group. Conclusion: The above-mentioned results indicate that the reason for the low scores obtained in both groups, is ignoring the size of reward and punishment and focusing on their frequency. Moreover, the reason for the lower output in the dependent group in comparison with the control group is a lack of balance between harm avoidance and reward dependence and a bias toward harm avoidance. The mentioned bias toward harm avoidance could be associated with the specific circumstances of treatment seeking drug dependents. M3 ER -