The Influence Of Causality On College Students' Bayesian Reasoning

In the case of insufficient information and data collection,Bayesian inference is helpful for people to make analysis and judgment by rational use of the acquired prior knowledge.However,previous studies have shown that people perform poorly on Bayesian inference.To this end,Krynski(2007)proposed the Causal Bayesian framework,suggesting that providing causality can facilitate people's correct assignment of false positives and thus facilitate reasoning.However,the results of subsequent studies were inconsistent.For example,Hayes et al.(2018)found that causality only seemed to reduce people's overestimation of posterior probability,but could not really improve the Bayesian inference performance of subjects.The reason might be related to the effective use of causality.According to the "triple processing mind" model,the reflective mind can initiate suppression of the inferior responses produced by the autonomous mind(e.g.,the transpositional representation bias in Bayesian reasoning),thus helping the subjects to think in a more rational way.Therefore,this study attempts to stimulate the reflective mind of the subjects by setting guidance language,and prompt them to put down their existing concepts and conduct reasoning in an unbiased way,so as to reduce the rational obstacles of individuals in reasoning,achieve better use of causality,and promote Bayesian inference.In addition,most previous studies adopted the method of "deducing the cause from the result and analyzing the final answer",while eye movement technology can observe the Bayesian inference process "directly".In view of this,this paper further explores the role of causality in the process of Bayesian inference from the perspective of reducing the rational barriers of the reasoner,and reveals the mechanism of "reasoning quotient" in Bayesian inference.In this study,338 college students were selected as subjects according to the classic breast cancer text paradigm,and three experiments were designed to investigate the influence of causality and prompt instruction on Bayesian inference.Specifically,experiment 1adopts the experimental design of 2(guidance language: rational/intuitive)×2(causality: with/without)between subjects to investigate the influence of intuitive/rational guidance language and causality on Bayesian inference of college students.It was found in the experiment that(1)The correctness and the accuracy of decision under the condition of causality were significantly higher than that without causality.(2)There was no significant difference in correctness and the accuracy between the subjects under the rational instruction language condition and the intuitive instruction condition.Experiment 2 was further closer to the reasoning problem situation,and the experimental design of 2(with or without prompt instruction)×2(with or without causality)was adopted to investigate the influence of prompt instruction and causality related to breast cancer and hit rate on college students' Bayesian inference.It was found in the experiment that(1)The correctness and the accuracy of decision under the condition of causality were significantly higher than that without causality.(2)There was no significant improvement in the correctness of decision under the condition of prompt instruction,and the accuracy was significantly higher than that under the condition of no prompt instruction.(3)In the absence of causality,the decision-making of the subjects with prompt instruction was significantly higher than that without prompt instruction;Under the condition of no prompt instruction,the decision of the subjects with causality was significantly higher than that without causality.Experiment 3 follows the experimental design in Experiment 2.Eye movement technology is adopted to detect the role of prompt instruction by recording the eye movement information of subjects under the condition of causality and prompt instruction.Results show that(1)the picture overall analysis to the problem,the total fixation time and total fixation times on indicators suggest the main effect of instructions,the main effect of causality,prompt instruction and causality interaction was not significant,while the average fixation time on indicators,suggest no significant main effect of instructions,no significant main effect of causality,prompt instruction and causality interaction was not significant;(2)According to the analysis of eye movement indicators in each ROI,the main effect of prompt instruction was significant in total fixation timeand fixation times,the main effect of causality was significant,and the interaction between prompt instruction and causality was not significant;On the index of return times,the main effect of guide language is significant,while the main effect of causality is not significant,and the interaction between prompt instruction and causality is not significant.(3)After the event comparison of ROI,the total fixation time,fixation times and return times of the ROI of the basic ratio and the problem information are significantly less than the ROI of the hit rate and the false report rate.The results of this study show that:1.Causality has a robust promoting effect on Bayesian inference;2.Simply reminding the subjects to think in a rational way could not improve the individual's reasoning performance.When the prompt instruction related to hit rate was given to the subjects,the reasoning accuracy was not increased,but the reasoning bias was effectively reduced.3.It is obvious that the prompt instruction increases the processing depth of the problem materials,stimulates the process of reflection,and arouses the reflective mind of the reasoner.