Cornell University Law School Social Science and Law

Witnesses: Eyewitness Identification

Few forms of evidence at a trial hold greater sway with a jury than the testimony of an eyewitness claiming at that she saw the accused commit the crime.  Traditionally, the credibility of this testimony has been bolstered by the witnesses’ ability to pick the correct suspect out of a police lineup.  However, since the 1970’s a growing body of social science evidence has cast doubt on the reliability of eyewitness testimony—more troubling yet, faulty eyewitness testimony is the most common uniting factor in cases shown to have resulted in wrongful conviction.  This in turn has led courts to confront difficult questions concerning the role, if any, that expert testimony concerning the pitfalls of eyewitness identification should play at trial.

Part I: Legal Approaches to Eyewitness Testimony
A.  Lineups and Suggestive Eyewitness Identification

The Supreme Court’s 1977 decision in Manson v. Braithwaite remains the last judicial word on eyewitness identification.  Manson addresses the validity of eyewitness identifications provided under faulty methods.  For example if the witness is simply presented with a picture of the suspect and asked if the individual in the picture is the same one the witness saw at the crime scene, the identification is flawed because it is suggests to the witness that the police have other reasons to suspect that the individual in the photograph is guilty.  By contrast, a properly conducted photographic lineup in which the witness selects the suspect’s photograph from among several possibilities lacks this suggestive element.  Manson provides for a two-prong inquiry, in which both prongs must be satisfied for the identification to be excluded at trial.  Under the first prong, the identification must be shown to be flawed.  Having decided that the technique was flawed, however, the trial court weights five criteria in deciding whether the identification was so unreliable that it should be excluded at trial:

1.  The eyewitness’s opportunity to view the perpetrator at the crime scene.
2.  The degree of attention the eyewitness focused on the perpetrator.
3.  The accuracy of the witness’s description of the perpetrator.
4.  The time elapsed between the witness’s identification of the suspect and witnessing the crime.
5.  The certainty of the witness’s identification of the suspect.

If these factors weigh against the reliability of the identification, then the identification is to be excluded at trial.

B.  The Admissibility of Expert Testimony on Eyewitness Identification

Another line of cases deals with the admissibility of testimony by social scientists explaining to the jury the counterintuitive empirical findings regarding the reliability of eyewitness testimony.  Typically, defendants seek this testimony to counteract damning eyewitness testimony. 

Objections to such testimony have been raised on the grounds that under Daubert v. Merrell Dow Pharms., Inc.,509 U.S. 579 (1993), the expert testimony must fit the facts of the case to satisfy the Federal Rules of Evidence.  Since expert testimony on eyewitness identifications only addresses the unreliability of such identifications in general, the reasoning goes; the testimony is inadmissible because it does not fit the facts of the specific case.  Thus, in the 1970’s, 1980’s and 1990’s, many courts disallowed expert testimony by social scientists:

We find no abuse of discretion in the trial court's ruling [excluding testimony by the defense’s expert witness, Buckhout]. Buckhout did not comment specifically on the identification made by the two government witnesses, but instead testified generally as to problems with eyewitness identification and that pilots as a group were not better equipped than ordinary witnesses to make identifications. To admit such testimony in effect would permit the proponent's witness to comment on the weight and credibility of opponents' witnesses and open the door to a barrage of marginally relevant psychological evidence. Moreover, we conclude, as did the trial judge, that the problems of perception and memory can be adequately addressed in cross-examination and that the jury can adequately weigh these problems through common-sense evaluation. United States v. Thevis, 665 F.2d 616 (5th Cir. 1982).

However, recent decisions have often taken a different approach to the issue leading to a different result:

The court was convinced at trial that the psychological research supporting [the identified flaws in eyewitness testimony] is both reliable and helpful and that the constantly increasing knowledge social scientists are obtaining about the inner workings of the human animal are likely not commonly understood or obviously apparent to jurors (or, for that matter, judges). Therefore, educating the jury about this research does not (and, in this case, did not) run afoul of Rule 702, and, indeed, it is an important step along the road to using improved scientific knowledge to create more accurate and fair legal proceedings. It would be anachronistic to categorically bar courts from employing the latest reliable scientific evidence in their effort to make sure that the trials that they administer resemble as closely as possible a search for truth; such a search requires diligently pursuing better understandings of human decisionmaking, including the flaws, weaknesses, and biases that characterize human life. Particularly for cases like this one, in which the reliability of eyewitness testimony is so important and so linked to well-established flaws in human perception and memory, such testimony may be crucial to fair, thorough, informed, and rigorous decisionmaking. It can only help to make factfinders more informed. Applying this research to the facts of this case, however, is within the sole province of the jury. United States v. Smith, 621 F. Supp. 2d 1207, 1219 (M.D. Ala. 2009).

Part II: The Social Science of Eyewitness Identification

Since the 1970’s a wide range of social science experiments have dramatically increased the scientific understanding of the dynamics of eyewitness identification.  The trend in these studies has been to cast significant doubt on the reliability of eyewitness testimony and on the ability of a jury to properly weigh eyewitness testimony.  Consider that in the more than 200 cases in which DNA has exonerated an individual who had been convicted of a crime, more than three-quarters of the cases featured an incorrect eyewitness identification at trial:

Block quote from Innocence Project
http://www.cbs.com/thunder/player/thunder.php?pid=aDRlRhkUMnd0qTN3VktIC0Ne7qpQ29Yl Caption: A story from 60 Minutes detailing the story of exoneree Ronald Cotton and the roll faulty eyewitness identification played in his trial.

A.  The Social Science Critique of Manson v. Braithwaite

A.  Although Manson v. Braithwaite remains law, it has come under fire from social scientists.  In a nuanced criticism, several social scientists simultaneously allow that the approach in Manson makes conceptual sense, but criticize Manson because the particulars of the Manson analysis do not mesh with empirical findings:

Perhaps the most serious discrepancy between eyewitness science and Manson is evident in the second inquiry, which comes into play when an identification procedure is found to be suggestive. The Manson criteria (view, attention, certainty, time, description) were meant to clarify the idea that the ultimate issue is the reliability of the identification, not suggestiveness per se. Reasonable people can disagree, but we believe that the Court’s general conceptualization was appropriate and defensible… Although we do not take issue with the broad assumption of the Manson Court that some set of reliability factors might justifiably trump concerns about suggestiveness in a given case, the science on the five factors in Manson points to very serious problems. First, none of the five criteria are unequivocally related to the accuracy of identifications. But, the most serious problem is that three of the five criteria(certainty, view, and attention) are self-reports by the eyewitnesses and these self-reports are themselves products of suggestive procedures. Gary L. Wells & Deah S. Quinlivan, Suggestive Eyewitness Identification Proceduresand the Supreme Court’s Reliability Test in Light of Eyewitness Science: 30 Years Later, 33 Law Hum Behav1 (2009).

http://www.psychology.iastate.edu/~glwells/theeyewitnesstest.html caption: Try your own hand at eyewitness identification from a police lineup

B.  The Social Science of Eyewitness Identification

While the social science on eyewitness identification is voluminous, a few especially interesting and counterintuitive findings are highlighted here:

1. Witness certainty in the accuracy of an identification does not correlate with the accuracy of that identification:
Based on a large number of eyewitness identification experiments conducted in dozens of different labs, a meta-analysis of the certainty-accuracy correlation showed that if the analysis is restricted to only those witnesses who made an identification the average correlation could be as high as 0.41… What does a 0.41 correlation mean? One way to think about a 0.41 correlation is to compare it to something with which people have some experience. For instance, the correlation between height and gender in humans is considerably greater than 0.41. That means that we could better predict whether someone was male or female based on their height than we could predict whether a witness was accurate or inaccurate based on their certainty. Gary L. Wells & Deah S. Quinlivan, Suggestive Eyewitness Identification Procedures and the Supreme Court’s Reliability Test in Light of Eyewitness Science: 30 Years Later, 33 Law Hum Behav 1 (2009).

However, recent research provides new evidence that suggests that an eyewitness's confidence during the first identification of a perpetrator (and under ideal conditions when there is little opportunity for suggestion) may in fact be related to accuracy. John T. Wixted et al., Initial Confidence Reliably Predicts Eyewitness Identification Accuracy, 70 American Psychol. 515 (2015). See also John T. Wixted & Gary L. Wells, 18 Psychol. Sci. Pub. Interest 10 (2017).

2. Cross-race identifications are less reliable than inter-race identifications:
The chance of a mistaken identification is 1.56 times greater in other-race than in same-race conditions and that the witnesses were 1.4 times more likely to correctly identify a previously viewed own-race face as they were to identify an other-race face. [Witnesses] were more than 2.2 times as likely to accurately categorize own-race faces as new versus previously viewed as they were to accurately categorize other-race faces. Gary L. Wells, Amina Memon, and Steven D. Penrod, Eyewitness Evidence: Improving Its Probative Value, 7 Psychological Science in the Public Interest 45 (2006).

3. The method by which a police lineup is conducted has a great impact on itsvalidity:
Here, social scientists presented four rules for insuring the validity of police lineups.

            Rule 1: Who Conducts the Lineup
The person who conducts the lineup or photospread should not be aware of which member of the lineup or photospread is the suspect.
            
             Rule 2: Instructions on Viewing
Eyewitnesses should be told explicitly that the person in question might not be in the lineup or photospread and therefore should not feel that they must make identification.  They should also be told that the person administering the lineup does not know which person is the suspect in the case.

             Rule 3: Structure of Lineup or Photospread
The suspect should not stand out in the lineup or photospread as being different from the distractors based on the eyewitness’s previous description of the culprit or based on other factors that would draw extra attention to the suspect.

             Rule 4: Obtaining Confidence Statements
A clear statement should be taken from the eyewitness at the time of the identification and prior to any feedback as to his or her confidence that the identified person is the actual culprit.

Gary L. Wells et al, Eyewitness Identification Procedures: Recommendations for Lineups and Photospreads, 22 Law & Human Behavior 603 (1998).

Further Study:

1. United States v. Thevis, 665 F.2d 616 (5th Cir. 1982). http://openjurist.org/665/f2d/616/united-states-v-g-thevis
2. United States v. Smith, 621 F. Supp. 2d 1207, 1219 (M.D. Ala. 2009).
3. Gary L. Wells & Deah S. Quinlivan, Suggestive Eyewitness Identification Procedures and the Supreme Court’s Reliability Test in Light of Eyewitness Science: 30 Years Later, 33 Law Hum Behav 1 (2009).  Available from http://www.psychology.iastate.edu/~glwells/
4. Eyewitness Evidence: Improving Its Probative Value, 7 Psychological Science in the Public Interest 45 (2006). Available from http://www.psychology.iastate.edu/~glwells/
5. Gary L. Wells et al, Eyewitness Identification Procedures: Recommendations for Lineups and Photospreads, 22 Law & Human Behavior 603 (1998). Available from http://www.psychology.iastate.edu/~glwells/
6. http://www.eyeid.org/  A site for practicing defense attorneys on litigation eyewitness identification issues (accessible only to members of the defense community).
7. Special Master Appointed by N.J. Supreme Court Calls for Major Overhaul of Legal Standards for Eyewitness Testimony See available link here
8. Marlee Kind Dillon, et al., Henderson Instructions: Do They Enhance Evidence Evaluation?, 17 J. Forensic Psychol. Research & Practice 1 (2017).

Additional Materials:

• A video from the Innoncence Project which discusses the Ronald Cotton case in more detail, as well as potential issues that can arise during eyewitness identifications.


• An episode of "Brian Games" which covers memory failures that con contribute to misidentifications.


• Gary Wells, accepting the James McKeen Cattell Fellow Award at the Association for Psychological Science's May 2017 meeting, discusses his research on eyewitness identification.