Can eyewitness memory be trusted?

Introduction: Researchers tested whether a question asked after an event can change what people remember about it.

Participants: 140 adults aged 18 to 65 (77 women and 63 men) recruited online in the United Kingdom. Race/ethnicity was not reported.

Method: Participants watched a 2-minute video of a man taking a woman's bag from a café table and walking out. The man wore no hat. The next day, participants answered 12 questions about the video. They were randomly assigned so that, for half of them, one question was, "Did the man adjust his baseball cap before taking the bag?" For the other half, the question was, "Did the man look around before taking the bag?" One week later, all participants described the man in their own words and rated their confidence in each detail they reported (1 = guessing, 5 = certain).

Results and Discussion: In the group asked about the cap, 37% described the man as wearing a cap, compared with 6% of the other group, a statistically significant difference. Participants who reported the cap rated their confidence in that detail at a mean of 3.8, almost as high as their confidence in details that were really in the video (4.0).

Introduction: Researchers examined records of real police lineups to see whether witnesses' confidence when they first chose someone was related to whether their choice was likely to be correct.

Participants: Records of 1,140 witnesses who viewed lineups in one US city over four years. Witnesses were 18 to 81 years old. Gender and race/ethnicity were not reported.

Method: Each lineup contained one suspect and five fillers, people known to be innocent who matched the description. Immediately after choosing someone, each witness stated how confident they were, and an officer who did not know which person was the suspect recorded it as high, medium or low. Researchers counted how often witnesses at each confidence level chose a filler, which is always a mistake. For 210 cases that went to trial, the researchers also recorded how confident the witness said they were when testifying in court, often many months later.

Results and Discussion: Witnesses who stated high confidence immediately chose a filler 6% of the time; those with medium confidence, 19%; and those with low confidence, 34%. At trial, however, 91% of the witnesses who testified said they were highly confident, including many who had stated low confidence at the lineup.

Introduction: Researchers tested whether interviewing witnesses in the place where an event happened helps them remember it.

Participants: Eighty students (46 women and 34 men) aged 18 to 26 at a university in Spain. Race/ethnicity was not reported.

Method: During a lecture, a staged event took place: a visitor entered the classroom, argued briefly with the lecturer and left with a laptop. Two days later, students were randomly assigned to be interviewed either in the same classroom, where they were first asked to picture the room and their own feelings during the event, or in a different building with a standard interview. Researchers compared each statement with a video recording of the event and counted correct and incorrect details.

Results and Discussion: Students interviewed in the classroom reported more correct details (mean of 24) than students interviewed elsewhere (17), a statistically significant difference. The two groups reported about the same number of incorrect details (3.1 and 2.9).