PSYCH OpenIR  > 社会与工程心理学研究室
跨期决策是否基于“折扣计算”:来自过程检验的证据
其他题名Is Intertemporal Decision-Making Based on a Temporal Discounting Process? Evidence from Process Testing
张阳阳
学位类型博士
导师李纾 ; 梁竹苑
2016-05
学位授予单位中国科学院研究生院
学位授予地点北京
学位专业心理学
关键词跨期决策 延迟折扣计算 眼动追踪 脑功能连接 眼动轨迹分析
摘要跨期决策指对发生在不同时间点上的结果之间进行的权衡。跨期决策的作用小至日常生活,影响个人的健康财富,大至环境污染和可持续发展,影响国家的繁荣与昌盛。但是对人们如何做出跨期决策的基本认识,目前的研究仍有分歧。前人分别从无限理性和有限理性的角度提出了关于跨期决策的折扣模型和非折扣模型,这两类模型的分歧点在于:跨期决策是否基于折扣计算的规则。前人研究通过结果预测或者模型拟合的基于结果的检验方法,虽然能够较好地预测选择结果,但是不能检验潜在的决策过程。过程检验的方法和技术(眼动追踪技术、磁共振成像技术等)可以直接探查决策背后的认知过程,因而可以有效地解决跨期决策是否基于折扣计算的分歧。所以,本研究使用基于过程验证方法探讨跨期决策是否基于“折扣计算”过程。
为实现上述目的,本研究设计了一套新的实验范式,其中包含两个实验任务:符合折扣计算假设的基线折扣任务和被试自主进行跨期决策的自主跨期任务。本研究利用双分离逻辑,分别选择了仅影响基线折扣任务或自主跨期任务的操作变量与调节变量。其中,操作变量关注实验材料的参数属性对任务的影响,在两任务的操作变量分别为选项的计算难度和选项中结果的大小;调节变量关注与任务相关的个体差异变量(认知能力和人格倾向),在两任务中的调节变量分别为计算能力、最大化倾向和认知反思风格、冲动性人格。
    研究一使用认知行为实验,旨在检验新实验范式的有效性。研究一发现:1)基线折扣任务的任务表现上符合折扣计算假设的预测:在反应时上,低难度条件下,基线折扣任务的反应时要短于高难度条件;在正确率上,基线折扣任务的正确率受到计算难度和计算能力的影响,计算难度越低,计算能力越高,正确率越高。2) 自主跨期决策任务的的任务表现与基线折扣任务不同:自主跨期任务的反应时短于基线折扣任务;且其选择偏好(选择SS  选项的比例)受选项结果大小的影响:结果越大,选择SS 选项的比例越少。此外,小结果条件下被试越倾向于基于直觉(直觉的认知反思风格分数越高),选择SS  选项的比例越大。研究一的结果表明,研究所设计的新实验范式在检验折扣计算假设中有效,且从任务表现层面看,自主跨期任务并不符合折扣计算假设的预测。
研究二采用眼动追踪技术, 旨在为检验跨期决策折扣计算假设提供基于眼动特征的过程检验证据,并尝试探索自主跨期决策的眼动特征是否遵循非折扣计算模型——齐当别模型的预测。研究二发现:1)其认知行为结果与研究一一致。2)在加工深度上,基线折扣任务符合折扣计算过程的假设,且与自主跨期任务不同。即基线折扣任务中的平均注视点时长大于自主跨期决策任务,并且随着计算难度越高,平均注视点时长越大。3)在眼动轨迹相似性上,自主跨期任务和基线折扣任务具有不同的整体认知过程:二者任务内的相似性程度均高于两任务间。4)自主跨期任务的注意分配更符合齐当别模型的预测:选项结果越大,自主跨期任
务中分配在 LL 选项上的注视点个数更多,且分配在 LL 选项中的结果或时间上的注视点个数,均比SS 选项结果更多。但在基线折扣任务中没有此效应。研究二的结果表明,本研究设计的实验任务范式有效并可靠,从眼动特征层面上看,自主跨期任务并不符合折扣计算假设的预测,而是符合齐当别模型的预测。
研究三采用了磁共振成像技术, 旨在为检验跨期决策的折扣计算过程假设提供基于脑功能活动的证据,并从该角度验证自主跨期决策的脑功能活动更符合齐当别模型的预测。研究三发现:1)其认知行为结果与研究一和二一致。2)在脑激活和脑功能连接上,基线折扣任务符合折扣计算过程的假设,且与自主跨期任务不同。具体来说,在基线折扣任务中,计算难度调节表征时间的额下回与表征价值的顶下沟之间的功能连接强度:计算难度越高,功能连接越强;而在自主跨期任务中没有发现此效应;且在基线折扣任务中,随着计算难度的增大,表征数字计算的壳核、左侧额下回和双侧楔前叶的激活程度更高,但选项的结果大小不影
响基线折扣任务的脑活动。3)自主跨期任务中的脑激活更符合齐当别模型的预测:随着选项结果的增大, 表征认知控制与决策冲突的左侧背外侧前额叶和双侧楔前叶的激活程度更高。但选项的计算难度不影响自主跨期任务的脑活动。研究三的结果表明,从脑功能活动的层面上看,自主跨期任务并不符合折扣计算假设的预测,而是更符合齐当别模型的预测。
总之,本研究设计并使用了新的实验范式,基于双分离逻辑,结合行为、眼动追踪和磁共振成像的汇聚性证据,逐层深入地证明了跨期决策并没有遵循折扣家族模型所预期的折扣计算过程,并证明跨期决策的过程更加符合齐当别抉择模型的预测。这一结果也说明,基于过程的检验方法更适宜于对跨期决策的理论模型进行更深入全面的检验。本研究为更好地回答“人们是如何做出跨期决策”的问题进行了方法学上的有益探索,并提供了多层面的实证证据。
其他摘要Intertemporal choice refers to the tradeoff among outcomes that occur at different time points. It covers a wide range of topics at individual and national levels, from daily trivial, including personal health and wealth, to developmental issues that concern the future prosperity of a country. Current studies have different findings about the basic theoretical models of intertemporal choice. The family of discounting and non-discounting models is proposed based on unbounded rationality and bounded rationality, respectively. These models disagree on whether intertemporal choice is based on discounting calculation. Previous studies could well predict decision results through testing methods that are based on  outcomes, such as result prediction and model fitting. However, the decision process remains unknown. The methods and techniques (e.g., eye tracking and fMRI) of process testing can be applied to explore directly the cognitive process of decision making, and thus they can also be used to solve this contradiction. Thus, this study aims to determine whether intertemporal choice is based on discounting calculation by using process testing methods.
A new set of experimental paradigm is designed. The design includes two experimental tasks, namely, the baseline discounting task and the free intertemporal choice task. The former task is in line with the hypothesis of discounting calculation, and the subjects in the latter task make intertemporal choice on their own. Based on the basic logic of double dissociation, this study chooses manipulated variables and moderator variables that only affect baseline discounting task or free intertemporal choice task. Manipulated variables cover the effect of attribute of stimuli on these two tasks, including  computational  difficulty and  outcome magnitude  of each option, respectively. Moderator variables cover cognitive ability and personality traits which were correlated with these two tasks, including  computational ability  and maximization preference, and cognitive style and impulsive personality, respectively.  
Study 1 aims to test the validity of the new experimental paradigm through cognitive behavioral experiment. The results of  Study  1 are as follows. First, the behavioral performance of discounting baseline task met the prediction of discounting calculation hypothesis. In terms of reaction time, the baseline discounting task exhibited shorter reaction time under low difficulty condition than high difficulty condition. In terms of accuracy, the task was influenced by  computational difficulty and  computational ability. High  computational ability  and low  computational difficulty resulted in high accuracy. Second, the performance of the free intertemporal choice task was different from that of the baseline discounting task, i.e., the reaction time of intertemporal choice task was shorter than that of baseline discounting task. Its choice of preference (the percentage of choosing the SS option) was influenced by outcome magnitude. Specifically,  large  magnitude  of outcome  results in less percentage in choosing the SS option. Moreover, subjects were more inclined to choose based on intuition (high score on intuition in the cognitive style test) under small outcome magnitude condition, which results in greater percentage in choosing the SS option. Results of  Study  1 indicate that the new experimental paradigm is effective in testing the hypothesis of discounting calculation. Based on task performance, the free intertemporal choice task did not meet the prediction of
discounting calculation hypothesis.
Study 2 aims to provide process testing evidence to the assumption of discounting calculation by using eye tracking technique. This study also explores whether the eye tracking characteristics of intertemporal choice meet the prediction of equate-to-differentiate model, which is a non-discounting model. Study 2 obtained the following results. First, the behavioral results accorded with those of Study 1. Second, in terms of processing depth, the baseline discounting task met the hypothesis of discounting calculation, in contrast to the free intertemporal choice. The average  fixation duration of the baseline discounting task was longer than the free intertemporal choice task, and was positively correlated with computational difficulty. Third, in terms of scanpath similarity, the two tasks had different cognitive processes; the within-task  similarity score was higher than the between-task similarity score. Fourth, the attention allocation of free intertemporal choice task met the prediction of equate-to-differentiate model; the larger the  outcome magnitude  was, the more fixation number fell  in LL option in the free intertemporal choice task, instead of in SS option. The number of fixation in the outcome or time of LL option was higher than that of SS option. The baseline discounting task did not show this pattern. Results from  Study  2 indicate that the experiment paradigm is valid and reliable. Based on the eye movement perspective, the free intertemporal choice task met the prediction of  equate-to-differentiate model, but not the discounting calculation hypothesis.
Study 3 aims to provide neuroimaging evidence in testing intertemporal choice discounting process and to verify whether the brain activity pattern of free intertemporal choice meets the prediction of the equate-to-differentiate model. Study 3 obtained the following findings. First,  the behavioral results agree with those in Studies 1 and 2. Second, in terms of brain activity and functional connectivity, the baseline discounting task accorded with the discounting calculation hypothesis, whereas the free intertemporal choice task did  not. Specifically, for in the function connectivity between the inferior frontal gyrus (representing time) and the inferior parietal sulcus (representing value) in the baseline discounting task,  computational difficult resulted in strong function connectivities. This effect was not observed in the free intertemporal choice task. In the baseline discounting task, the activities of brain areas that are associated with calculation, including putamen, left inferior frontal gyrus, and bilateral precuneus, were stronger with increasing computational difficulty. However, the  outcome magnitude  did not affect the brain activities of baseline discounting task. Third, the brain activity of free intertemporal choice met the prediction of equate-to-differentiate model better. The activities of the left dorsolateral prefrontal cortex and the bilateral precuneus (representing cognitive control and decision conflicts) were stronger with increasing option outcomes. However, computational difficulty did not affect the brain activity of free intertemporal choice task. Results from Study 3 indicate that, based on the brain activity perspective, the free intertemporal choice task better met the prediction of  equate-to-differentiate model, but not the discounting calculation hypothesis.
In summary, this study designed and tested a new experimental paradigm, and integrated behavioral, eye tracking, and fMRI evidences based on the logic of double dissociation to test the discounting calculation hypothesis. The obtained results verified that the intertemporal choice does not follow the discounting calculation process of discounting family model through the new experimental paradigm. Instead, the process of intertemporal choice better fit the equate-to-differentiate model. Results of this study indicate that process testing methods are suitable for testing theoretical models of intertemporal choice. This study sheds light on the methodologies to better answer the question of “how people make intertemporal choice” while providing practical evidence at multiple levels.
学科领域应用心理学
语种中文
文献类型学位论文
条目标识符http://ir.psych.ac.cn/handle/311026/19870
专题社会与工程心理学研究室
作者单位中国科学院心理研究所
推荐引用方式
GB/T 7714
张阳阳. 跨期决策是否基于“折扣计算”:来自过程检验的证据[D]. 北京. 中国科学院研究生院,2016.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
张阳阳-博士论文整合版.pdf(2330KB)学位论文 限制开放CC BY-NC-SA浏览 请求全文
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[张阳阳]的文章
百度学术
百度学术中相似的文章
[张阳阳]的文章
必应学术
必应学术中相似的文章
[张阳阳]的文章
相关权益政策
暂无数据
收藏/分享
文件名: 张阳阳-博士论文整合版.pdf
格式: Adobe PDF
所有评论 (0)
暂无评论
 

除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。