欺骗识别的准确率:影响因素与提高途径

The Accuracy of Deception Detection:Influence Factors and Improving Approaches

欺骗是一种常见的社会现象,通过观察他人的行为表现识别欺骗则是人们的一项重要能力。研究表明,人们的欺骗识别能力仅仅略微高于随机水平。本文关注基于行为线索的欺骗识别研究。首先,介绍欺骗识别的准确率;然后,结合Brunswik的透镜模型从欺骗线索的有效性和欺骗线索的利用两方面分析识别准确率的影响因素;并在此基础上探讨了提高识别准确率的途径。最后,对未来可能的研究方向进行展望。

Deception and deception detection is a common phenomenon in our daily lives. In social interaction, when a person makes a deliberate attempt to mislead others, no matter successful or not, the behavior is defined as deception. Deception detection is the behavior to make a judgment about whether the person is telling the truth or not. In psychological studies, human deception detection ability has long been an important research concern. The present study focuses on human beings’ performance in deception detection, and further analyzes the reasons and points out ways to improve. First, the study introduced general findings on the accuracy of deception detection. In deception detection, the typical paradigm is the lie detection test. During the test, the participants were asked to watch some videotapes carefully and make judgments about whether the target in the video was lying or not. The participants were not given any additional information about the targets in the test, therefore, the only information that could be used was the targets’ verbal and nonverbal behavior. Meta-analysis studies showed that the accuracy of deception detection was about 54%, which was only slightly better than chance. Moreover, accuracy is not related to the judge’s confidence level, age, experience, education, and sex. Thus, human beings are poor at detecting deception. Then, the study investigated factors contributing to the accuracy of deception detection in social interactive perspective, and did further analysis in the framework of Brunswik’s lens model. From a social interaction perspective, two aspects are the most relevant in explanation of the failures in deception detection. The first is whether the liars leak valid cues to deception. According to cue theories, the difference of mental states between liars and truth-tellers would be reflected on the observable behaviors. However, meta-analysis studies showed that cues to deception were generally faint and unreliable. The fact that there is no Pinocchio’s nose in deception makes deception detection a difficult task. The second is whether the observers rely on valid cues to detect deceit. Survey studies by the self-report method have shown that wrong beliefs about cues to deception are universal, which might result in using wrong cues(e.g., gaze aversion) in deception detection. However, studies using correlational measurement revealed that people indeed used valid cues in deception detection. The controversial results indicate that there is a dual-processing system in deception detection. Although we could use valid cues in the unconscious level, the judgment might be interfered by invalid cues. Besides, the Brunswik’s lens model help to determine which is the primary reason for the low accuracy, and the meta-analytic results suggested that no objective cues to deception was the initial reason. Third, based on the result of Brunswik’s lens model analysis, the study suggested two approaches to improve deception detection accuracy. The first approach is strategic questioning, which includes interviewing with an information-gathering style, asking unanticipated questions, and strategic use of evidence, etc. The second is imposing cognitive load, such as telling the story in reverse order and keep eye contact with the interview, etc. Both ways benefit the deception detection accuracy. Last but not least, we proposed two directions that are worth researching. One is about extending application of the Brunswik’s lens model in deception detection, such as in different populations and situations. The other is to explore the influence of top-down factors, e.g., stereotypes and prejudice in person perception, on the accuracy of deception detection.