Eyewitness memory under threat: what experiments and police procedures establish
A gunpoint robbery does not permanently fix a robber's face in eyewitness memory. The National Research Council's 2014 assessment describes high stress as a condition that can impair facial identification. A frightened witness can be sincere and still make an inaccurate identification. The intensity of the experience and the strength of later confidence supply no automatic guarantee of accuracy.
The council separates estimator variables from system variables. Lighting, distance, exposure time, stress and weapon presence concern the original encounter. Lineup instructions concern the investigative procedure. A carefully administered identification test cannot improve the original view. It can avoid adding an investigator's expectations to the recollection being tested. Conditions during the crime and later investigative decisions require separate evaluation.
The council also recorded substantial limits in the research. Its discussion of stress noted variation across studies, small effects and wide confidence intervals in a major synthesis, and the difficulty of ethically reproducing extreme threats in a laboratory. Military survival-training research suggested that even memories of intensely stressful encounters could be altered by misinformation introduced afterward. Those findings support caution about eyewitness memory without implying that every stressed witness must fail or that every experiment measures the same experience.
A 2020 experiment by Thomas Nyman and colleagues illustrates the variation. Thirty-seven participants watched eight threatening or non-threatening scenes, presented either on a conventional screen or in immersive virtual reality. They then described the person and viewed six-person photographic lineups, some containing that person and some excluding them. When the target was present, identification accuracy was higher after conventional video than after virtual reality: 91 percent against 66 percent. Greater immersion did not automatically produce a better identification.
Threat did not uniformly reduce accuracy in that experiment. Across the relevant conditions, target-present accuracy was 86 percent for threatening scenes and 70 percent for non-threatening scenes. The corresponding benefit did not appear for target-absent lineups. The authors treated the work as a small demonstration and did not collect physiological stress measurements. It therefore cannot establish how a real robbery affects an individual victim's eyewitness memory. It does establish why a simulated threat, a measured stress response and an actual violent encounter should not be treated as interchangeable.
Weapon focus requires similar precision. In 2024, Hannes Körner, Franz Faul and Antje Nuthmann recorded eye movements while participants watched a staged robbery involving a knife, a water bottle or a plunger. The design compared a threatening object with an ordinary object and an unusual but non-threatening one. In the eye-tracking experiment, total viewing time did not differ significantly across the three objects. A weapon shifted attention from the perpetrator's face toward her body, but memory for her appearance did not differ by object condition.
An additional online experiment with the same videos did find poorer memory associated with the weapon; that difference disappeared when sound was removed. The result complicates a simple picture in which every weapon always captures the gaze in the same way. The researchers also distinguished the memory outcome from the eye movements presumed to explain it. Remembering fewer details and looking longer at an object are separate measurements. Their findings make weapon presence a relevant condition to examine, while resisting an automatic prediction about every witness.
Police procedures address a different source of error: influence during the identification itself. In January 2017, the Justice Department announced department-wide photo-array procedures for the FBI and other federal investigative agencies. Except in exceptional circumstances, an administrator should either be blind to the suspect's identity or use a blinded arrangement that prevents seeing which photograph the witness is viewing. The purpose was to reduce even unintentional suggestions about which face investigators expected the witness to choose. Prosecutors were also directed to review identification procedures before deciding whether to charge a suspect identified through a photo array. That instruction covered identifications obtained by state and local officers as well as federal personnel.
The same policy directed investigators to preserve the witness's confidence at the initial identification and document the procedure through recording or a close transcription. The National Institute of Justice's 1999 evidence guide had already organized eyewitness work into distinct stages, including the first response, follow-up interviews, show-ups and photographic or live lineups. Its initial-response section began with the emergency call and obtaining information at the scene. Its follow-up procedures included assessing individual elements of a statement and maintaining contact with the witness. It explicitly treated recording recollections and identification results as part of evidence preservation. Eyewitness memory therefore reaches court through a sequence of observations and investigative decisions. A reliable evaluation needs the original viewing conditions, the first identification procedure and the witness's words at that time; fear alone cannot substitute for any of them.
- National Research Council, Identifying the Culprit (2014), Chapter 5: Applied Eyewitness Identification Research — www.nationalacademies.org
- Nyman et al. (2020), Eyewitness identifications after witnessing threatening and non-threatening scenes — original experiment, PLOS ONE — journals.plos.org
- Körner, Faul and Nuthmann (2024), Is a knife the same as a plunger? — original eye-tracking and memory experiments — pmc.ncbi.nlm.nih.gov
- U.S. Department of Justice, Department-Wide Procedures for Eyewitness Identification, January 6, 2017 — www.justice.gov
- National Institute of Justice, Eyewitness Evidence: A Guide for Law Enforcement (1999), official publication description — nij.ojp.gov
- Innocence Project, How Eyewitness Misidentification Can Send Innocent People to Prison, April 15, 2020 — interview supporting the quiz answer — innocenceproject.org