Eyewitness confidence: what changes between an identification and a trial
Strong eyewitness confidence can accompany a mistaken identification, especially after feedback has changed a witness's recollection of the original decision. Timing is crucial. Certainty expressed immediately during a fair first identification has a different evidentiary history from certainty expressed months later in court. Research distinguishes those situations rather than treating every confident witness alike. The distinction concerns how a person's certainty developed after an identification, including influences that may no longer be obvious to the witness.
The courtroom's attention to certainty has a long history. In Manson v. Brathwaite, decided in 1977, the Supreme Court considered an undercover officer's identification following a heroin purchase in Hartford, Connecticut. Officer Jimmy Glover made the purchase on May 5, 1970. Two days later, he viewed a single photograph supplied by another officer and identified Nowell Brathwaite. At the January 1971 trial, Glover expressed complete certainty. Brathwaite denied being at the apartment during the transaction. The jury convicted him of possession and sale of heroin.
The Supreme Court allowed the identification evidence despite the unnecessarily suggestive single-photograph procedure. A federal appeals court had ordered relief, but the Supreme Court reversed that judgment. It rejected a rule that would automatically exclude identification evidence solely because the procedure was unnecessarily suggestive. Its majority weighed certainty alongside the opportunity to observe, attention, the earlier description, and the interval before identification. Justice Thurgood Marshall's dissent challenged the usefulness of the officer's certainty as an indicator of accuracy. These were competing judicial assessments within a ruling about admissibility. The conviction and the majority's legal conclusion did not make confidence an independent scientific measurement of whether a particular recollection was accurate.
Gary Wells and Amy Bradfield isolated a later influence on eyewitness confidence in a 1999 experiment. Their sample comprised 156 participants who had made false identifications. Some received feedback confirming their selections. The researchers reported that this feedback inflated more than remembered certainty: it also changed retrospective judgments of the quality of the view, the attention paid, and the speed of identification. A later report of excellent observation could therefore reflect information received after the choice, even though the original identification remained mistaken.
The experiment also varied when participants privately considered their confidence and viewing experience. Thinking about those matters before the feedback reduced its effects; the same exercise after feedback did not. Even without feedback, private reflection on certainty had some tendency to increase confidence. The result concerned retrospective judgments under the study's conditions, not a percentage of mistaken identifications in criminal trials. It demonstrated that later reflection does not necessarily recover an untouched description of how certain a person initially felt. Reconsidering a decision and preserving its original confidence are different operations.
Later research refined the relationship between eyewitness confidence and accuracy. In their 2017 synthesis, John Wixted and Gary Wells reviewed mock-crime experiments and police field studies. They found a strong relationship under carefully controlled identification conditions: an initial, uncontaminated test, a fair lineup, no influence from the administrator, and an immediate confidence statement. Among adults tested in those circumstances, high-confidence suspect identifications were highly accurate. That finding gives initial certainty potential value; it does not make every later declaration equally informative.
The same synthesis warned that unfair lineups can seriously compromise even a highly confident identification. A lineup ordinarily combines one suspect with fillers known to be innocent. If only the suspect resembles the description, a confident choice may reflect the lineup's construction rather than accurate recognition. The researchers distinguished suspect identifications from filler identifications: selecting a filler is an error, however firmly expressed. Eyewitness confidence at trial cannot replace the uncertainty recorded when the witness first selected a photograph. Fairness and the timing of the confidence statement both affect its evidentiary value.
The Justice Department put the timing distinction into federal policy in January 2017. Its procedures directed agents, except in exceptional circumstances, to use blind or blinded photo arrays. A blind administrator does not know which person is the suspect; a blinded procedure prevents the administrator from seeing the photographs' order or arrangement. The announced policy applied to the FBI, DEA, ATF, and U.S. Marshals Service. It also directed prosecutors to review identification procedures before charging cases involving photo-array evidence, including evidence collected by state or local officers.
The policy called for recording eyewitness confidence at the initial identification, through audio or video or a transcription as close as possible to the witness's own words. This preserves the statement before subsequent investigative contact can reshape its expression. The department explicitly recognized that initial confidence is often more informative than confidence at trial. A preserved first statement and the conditions surrounding it provide evidence that a later emphatic identification cannot recreate.
- Innocence Project, Judge Kozinski on Myths of Fairness in the Criminal Justice System and Consequences for the Innocent — innocenceproject.org
- U.S. Supreme Court, Manson v. Brathwaite, 432 U.S. 98, June 16, 1977, opinion and dissent via Cornell LII — www.law.cornell.edu
- Gary L. Wells and Amy L. Bradfield, Distortions in Eyewitnesses’ Recollections: Can the Postidentification-Feedback Effect Be Moderated?, Psychological Science, 1999, original-study abstract — journals.sagepub.com
- John T. Wixted and Gary L. Wells, The Relationship Between Eyewitness Confidence and Identification Accuracy: A New Synthesis, 2017 — journals.sagepub.com
- U.S. Department of Justice, Department-Wide Procedures for Eyewitness Identification, January 6, 2017 — www.justice.gov