The untapped potential of lineups: using eyewitness memory to rule out innocent suspects

ABSTRACT

Over the past 45 years, psychological scientists have invested a tremendous amount of effort into increasing the accuracy of suspect identifications from eyewitness lineup procedures. Those efforts have paid dividends, leading to the development of several practices that increase the accuracy of suspect identifications. Meanwhile, lineup rejections have largely been written off as lacking diagnostic value and little effort has been placed into understanding (1) why lineup rejections are less diagnostic than suspect identifications, and (2) how to increase the diagnostic value of lineup rejections. We show that preventing innocent-suspect identifications is not equivalent to demonstrating innocence and highlight several consequences of not being able to clear innocent suspects from police suspicion. We then review recent evidence, grounded in Signal Detection Theory, suggesting that lineup rejections are less diagnostic than suspect identifications because it is harder to reject a lineup of several faces than it is to identify a single face. When lineups are modified so that witnesses provide rejection ratings for each non-identified lineup member, rejection accuracy drastically increases. We highlight several questions for future research examining the potential for eyewitness memory to demonstrate innocence and question what impact suspect rejections might play in multiple-witness cases.

KEYWORDS:

• Eyewitness memory
• eyewitness lineups
• eyewitness identification
• memory
• Signal Detection Theory

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 For suspect identifications, d' is computed as follows: ${\mathrm{\Phi }}^{-1}\left(\mathrm{G}\mathrm{S}\right)-{\mathrm{\Phi }}^{-1}\left(\mathrm{I}\mathrm{S}\right)$. ${\mathrm{\Phi }}^{-1}$ represents the standard normal inverse function and GS and IS refer to the guilty-suspect identification rate and innocent-suspect identification rate, respectively.

2 For rejections, d' is computed as follows: ${\mathrm{\Phi }}^{-1}\left(\mathrm{C}\mathrm{R}\right)-{\mathrm{\Phi }}^{-1}\left(\mathrm{F}\mathrm{R}\right)$. ${\mathrm{\Phi }}^{-1}$ represents the standard normal inverse function and CR and FR refer to the correct-rejection rate and false-rejection rate, respectively.

3 Readers who are familiar with the eyewitness literature might note the similarity between our conceptualization of a lineup’s purpose and the lineup-as-experiment analogy (Wells & Luus, 1980). The lineup-as-experiment analogy proposes that the purpose of a lineup is to test the police investigator’s hypothesis that the suspect is guilty. The lineup-as-psychometric-instrument analogy proposes that the purpose of a lineup is to measure the match between the suspect and the witness’ memory for the culprit. These two analogies are complementary, but where the lineup-as-experiment analogy focuses on the outcome, the lineup-as-psychometric-instrument analogy focuses on the process. It was only by thinking through the process that we came to discover that the standard lineup was a flawed measurement device (Smith & Ayala, 2021).