【作者】 沈悦；

【导师】 戴俊毅；

【作者基本信息】 浙江大学 ， 心理学， 2022， 硕士

【摘要】 爱荷华赌博任务(IGT)是一项基于经验的风险决策任务,常应用于临床领域来检测决策缺陷。为了揭示IGT背后的心理机制,研究者们提出了一系列认知模型,包括以第二类前景效价学习模型(PVL2 model)为代表的强化学习模型和以赢留输走模型(Win-stay-lose-shift model,WSLS model)为代表的启发式模型,但很少有模型考虑到被试可能在IGT中发生了策略转换。本文在强化学习模型和启发式模型的基础上提出了一种策略转换模型(Switching-strategy-once model,SSO model),来考察是否至少有部分被试在IGT中发生了策略转换。研究一用SSO模型拟合了过往研究中的617个健康被试的IGT选择数据。结果表明,和WSLS模型、PVL2模型以及价值附加坚持模型(Valence-plus-perseverance model,VPP model)相比,SSO 模型在拟合一半以上的被试的个体数据时表现最为优异。为了验证被试的确会在IGT中发生策略转换,研究二通过心理学研究和数学分析相结合的方式进行了进一步的探究。研究二在IGT当中加入了反应时测量,并在其前后分别加入了按键反应时测量。160名大学生参与了研究二。模型拟合结果表明,不论从整体角度还是个体角度,SSO模型的表现都非常优异,但反应时分析的结果并不支持出现了策略转换。为了进一步提供有关策略转换的证据,研究三通过将IGT的试次数从研究二中的100次提升到200次,考察了 IGT试次数和发生策略转换的被试比例之间的关系。161名大学生参与了研究三。模型拟合的结果支持有部分被试在IGT中进行了策略转换。此外,模型结果的对比分析表明,在试次数为200的IGT中进行策略转换的被试比例显著高于在试次数为100的IGT中进行策略转换的被试比例。反应时分析的结果部分支持出现了策略转换。综合三项研究的结果,至少有部分个体在IGT中出现了策略转换,且策略转换的可能性随着试次数的增加而有所上升。因此,对于个体在IGT中的行为表现的认知模型分析,需要考虑策略转换的可能性,以便对背后的认知机制和影响因素,得出更加合理可信的结论。更多还原

【Abstract】 The Iowa Gambling Task(IGT)is an experience-based risky decision-making task to detect decision deficits in clinical populations.To reveal the underlying psychological mechanisms,researchers have proposed a series of cognitive models,including reinforcement learning models exemplified by the PVL2 model,and heuristic models exemplified by the Win-stay-lose-shift model(WSLS model).However,few existing models have considered the possibility that subjects may switch their strategy during the IGT.Therefore,three studies combining psychological experiments with mathematical modeling were conducted in the current research to investigate whether at least some participants switch their strategy in IGT.Study 1 proposed a class of strategy switching models(SSO models)based on the PVL2 model and the WSLS model and used the SSO models to fit IGT choice data from 617 healthy participants from previous studies.The results showed that individual data from more than half of the participants were better fitted by the SSO models than the WSLS model,the PVL2 model,and the Valence-plus-perseverance model(VPP model).To verify that some participants do switch strategies in the IGT,Study 2 recruited 160 undergraduate students to perform the IGT and further measured response time of the IGT and baseline key-pressing time in a simplified choice task for converging evidence.The model-fitting results were generally consistent with those of Study 1,that is,the SSO models performed very well at both aggregate and individual levels.However,response time data did not provide supportive evidence for the existence of strategy switching.To provide further evidence for strategy switching,Study 3 examined the relationship between the possibility of strategy switching and number of trials in the IGT by increasing the number of trials from 100(as in Study 2)to 200.161 college students participated in Study 3.The model-fitting results were consistent with those of the first two studies.The results of the response time analysis partially supported the existence of strategy switching.A comparative analysis of the results of modeling analysis between Study 2 and Study 3 showed that the proportion of participants who made a strategy switch in the IGT with 200 trials was significantly higher than the proportion of subjects who made a strategy switch in the IGT with 100 trials.Overall,the results of the three studies supported the existence of strategy switching among at least a portion of participants and the likelihood of such switch increased with the number of trials in the task.Therefore,cognitive modeling analysis of individual behavior in the IGT shall take this possibility into consideration for more reasonable and credible understanding of the underlying cognitive mechanisms and influential factors.更多还原