认知与发展心理学研究室知识库

心理学研究的重要目的之一发现心理干预的途径和方法。但截至目前,有效干预人类被试因果推理过程的手段尚不丰富,干预手段的效果并不稳定。本研究采用完全随机设计开展两个实验,分别探讨频率树是否影响大学生被试在反事实提问和能力提问因果推理问题上的作答表现。结果显示:(1)在两个实验中都发现了明显的图形促进效应,大部分被试在借助提供嵌套集合关系频率树(而非隐藏嵌套集合关系频率树)辅助推理时使用PPC值估计因果强度;(2)频率树类型和提问方式共同影响被试的因果强度估计模式,提供嵌套集合关系频率树+反事实提问的组合促使最多被试使用PPC估计因果强度。结果说明:明确数据之间的嵌套集合关系能极大地提高被试使用PPC估计因果强度的概率,关注焦点集信息有助于被试明确数据间的嵌套集合关系。

There are lots of evidences showing that participant’s performance on Bayesian inference, syllogistic reasoning and probability reasoning could be promoted by cumulative frequency tree. However, very few study focuses on the promotion effect of frequency tree on causal reasoning. This study carried out two experiments to investigate the effect of frequency tree on causal strength inference. The research hypotheses include: (a) Frequency tree featuring a explicit nest-sets structure (ENS) can improve the rationality of participant’s reasoning, while the frequency tree featuring a concealed nest-sets structure (CNS) can’t improve rationality of reasoning; (b) Participants estimate the causal strength of different contingencies by different modes in experimental treatment which used frequency tree featuring a CNS; and (c) There are more participants estimate the causal strength by Power-PC model in preventive contingency rather than in productive contingency. 2 (Frequency tree, level 1: featuring a ENS, level 2: featuring a CNS) × 2 (causal direction, level 1: productive, level 2: preventive) × 3 (contingency, level 1: P = 0.33 and Power - PC = 0.5; level 2: P = 0.33 and Power - PC = 0.83; level 3: P = 0.67 and Power - PC = 0.83) completely random design were used in two experiments. 469 undergraduate students participated in Experiment 1 which adopted counter-factual question, and 463 undergraduate students participated in Experiment 2 which adopted ability question. Contingency was offered by a booklet which contains 30 pages, and each page presents one sample related to the causality. Participant completed a frequency tree based on contingency, and estimated the causal strength of contingency individual. The frequency tree featuring a ENS consists of three types of information: the number of total samples, the number of samples in focus set, and the number of samples that represent effect emerge or not, while frequency tree featuring a CNS consists of the number of total samples and samples that represent effect emerge or not. The study found that (a) There are three common models of causal reasoning: p, Power-PC and P (E/C) for productive contingency (or P(-E/C) for preventive contingency), the most popular model changes with different experiment treatments; (b) 70.06 % of participants estimate causal strength by Power-PC model when they used frequency tree featuring a ENS, and only a few participants (about 21.28 %) estimate causal strength by Power-PC model when they used frequency tree featuring a CNS; (c) The type of frequency tree and the format of question have combining influence on causal strength evaluation, and the type of frequency tree have more influences on strength evaluation than the format of question; (d) Both contingency effect and causal direction effect are present from the experimental treatment which used frequency tree featuring a CNS. Experiment results significantly support research hypotheses (a), (b) and (c). These results indicate that frequency facilitating effect depends on supply nest-sets structure or not, whether in counter-factual question treatment or in ability question treatment. According to above two experiments, it is suggested that participant tends to make rational inference when they use frequency tree featuring a ENS or they were questioned by counter-factual format.