Eyewitness Identification: Debiasing the Effects of Composites and Surveillance Images

法医DNA检测的出现帮助确立了这样一个事实，即错误的目击者识别是导致陪审团认定无辜的人有罪的最大因素。因此，目击证人身份识别研究的系统变量方法已开始在法律制度中产生相当大的影响，这种方法旨在减少错误识别，而不损害准确的识别率。尽管近年来取得了相当大的进展，但对于无辜的人成为犯罪嫌疑人的“途径”如何使身份偏向嫌疑人，以及如何在不降低证人认出真正罪犯的可能性的情况下使这些情况偏向(减少错误身份)，人们知之甚少。让无辜的人成为嫌疑犯的两条重要“途径”是使用合成图纸和使用监视图像。这两种途径都保证了目击者在一定程度上倾向于识别无辜的嫌疑人，如果这个人成为嫌疑人的话，部分是基于图像。假设当图像相似路径或合成相似路径运行时，目击者研究人员对选择序列填充物的主要建议不足以偏向序列。这项研究将测试各种去偏技术。这种方法将使用描述恐怖事件的电影(12个版本，涉及12个不同的罪犯)，这些电影将分别向2000多人放映，然后要求他们从一个阵容中辨认出恐怖分子，并说明他们的确定性。无辜的嫌疑人将在一半的阵容中取代恐怖分子。无辜的嫌疑人将根据口头描述、合成图画或监控图像进行选择。去偏向技术将包括选择阵容填充者的方法，使用顺序阵容，以及使用一种名为幻影阵容的新技术。除了传统的分析外，还将使用贝叶斯信息增益分析来评估去偏见程序的有效性。贝叶斯信息增益分析评估了给定阵容程序的定罪和无罪证明价值。这项工作的结果将有助于澄清目击者科学家对错误识别和错误确定过程的基本理解。此外，结果还将告知执法部门，在监控图像和合成图形可能会对结果产生偏差的情况下，可以使用哪些程序来创建公正和信息丰富的阵容。

The advent of forensic DNA testing has helped establish the fact that mistaken eyewitness identification is the largest factor contributing to jury convictions of innocent people. As a result, the system-variable approach to eyewitness identification research, which is designed to reduce mistaken identifications without harming accurate identification rates, has started to have considerable impact in the legal system. In spite of considerable progress in recent years, little is known about how the "paths" through which innocent people become suspects in lineups can bias identifications toward the suspect and how to debias these situations (reduce mistaken identifications) without lowering the chances that the witnesses will identify the actual culprit. Two important "paths" through which innocent people become suspects are the use of composite drawings and the use of surveillance images. Both paths guarantee some propensity for eyewitnesses to identify an innocent suspect if that person became a suspect based in part on the image. It is hypothesized that the dominant recommendation of eyewitness researchers for selecting lineup fillers is not sufficient to debias lineups when either the image-similarity or the composite-similarity paths are operating. This research will test various debiasing techniques. The methodology will use films depicting a terrorist event (12 versions involving 12 different culprits) that will be shown to over 2000 people individually who will then be asked to identify the terrorist from a lineup and state their certainty. Innocent suspects will substitute for the terrorist in half of the lineups. Innocent suspects will be selected based on verbal descriptions, composite drawings, or surveillance images. Debiasing techniques will include methods of selecting lineup fillers, the use of the sequential lineup, and the use of a new technique called the phantom lineup. In addition to traditional analyses, Bayesian information-gain analyses, which assess both the incriminating and exonerating value of a give lineup procedure, will be used to assess the effectiveness of the debiasing procedures. The results of this work will help clarify eyewitness scientists' basic understanding of the processes of false recognition and false certainty. In addition, the results will inform law enforcement about procedures that can be used to create unbiased and informative lineups under conditions in which surveillance images and composite drawings could otherwise bias those results.