Mistaken Identity: How Feedback "After the Fact" Influences Eyewitnesses
Test Your Eyewitness Aptitude
Witness a simulated crime, then try to identify the suspect from a lineup. For best results, take this test before you read the article. Select "View Video" at right.
New research reveals that giving feedback to eyewitnesses of crimes after they have identified the suspect from a lineup or photospread distorts a witness's memories of both the original event and the identification decision. Surprisingly, this "post-identification feedback effect" is not tempered by the passage of time, even when the feedback is delayed by as much as 48 hours.
In a series of NSF-funded experiments led by Dr. Gary L. Wells of Iowa State University, eyewitnesses viewed a staged crime and were then shown lineups that, unbeknownst to them, did not contain the culprit. Regardless, all the eyewitnesses made mistaken identifications. They were then randomly assigned one of several post-identification conditions: they were either given feedback which confirmed their false identification, told that they had identified someone who was not the culprit, or told nothing.
The figure on the right shows the results of one of the experiments. Confirming feedback produced strong, statistically significant differences in eyewitnesses' memories of key aspects of the witnessing experience, including:
• their confidence level at the time they identified the suspect
• how good a view they had of the event
• their ability to make out facial details
• the ease of identification
• whether they had a good basis (enough information) to make their decision
More often than not, lineups are conducted by police officers who know which person in the lineup is the suspect. As a result, eyewitnesses often learn, either through direct feedback or confirmatory body language, whether they chose the suspect or not. For this reason, Dr. Wells is a long-time proponent of "double-blind" lineups, wherein the individual administering the lineup doesn't know the identity of the suspect, thereby eliminating any chance of informing the witness. But double-blind lineups are only starting to be introduced in a few of the over 16,000 independent law enforcement agencies across the United States, primarily because police cannot help but think that to require double-blind lineups is to question their integrity and honesty.
"The need for double-blind testing is not because we do not trust police," explains Dr. Wells. "It is necessary because police are human and humans unintentionally (and commonly without awareness) influence the person they are testing. Psychologists understand this very well, but outside of psychology, this is a hard sell."
Whether or not eyewitnesses exude a high level of confidence about their identifications of criminal suspects is of immense importance in the courtroom. Judges and juries both rely heavily on witnesses' confidence when making their own decision about the witnesses' accuracy. The findings demonstrated by Dr. Wells and his colleagues raise serious concerns about the fallibility of this system. If feedback can influence the confidence a witness demonstrates in court, then feedback is capable of distorting one of the most important cues that judges and juries use in making decisions about witness accuracy.
Effects of Delayed Feedback
Although the post-identification feedback effect has been well documented in previous studies by Dr. Wells and others, the current research adds a new twist by examining delayed feedback. In previous experiments, feedback had never been given more than 3 minutes after the eyewitnesses made their identifications. In Dr. Wells' new study, he not only tests the effect of delaying the feedback itself, but he also tests delaying the measures of witnesses' memories of the original event, and of the identification experience.
The unexpected results showed that even delaying feedback by as much as 48 hours did not moderate the post-identification feedback effect; it remained strong and present. This has practical implications for real-world cases, since there are often cases in which witnesses do not receive immediate feedback from detectives or other authorities involved with the case but are later told they picked out the perpetrator. Yet even such delayed feedback apparently compromises the reliability of the witness.
In his efforts to advance double-blind testing and other measures to ensure eyewitness accuracy, Dr. Wells has presented his findings at hundreds of universities, law-enforcement associations, legal conferences and prestigious professional organizations throughout the United States and Canada over the last 25 years.
"I don't expect police, prosecutors, judges and others in the legal system to read the scientific journals," Dr. Wells says. "Therefore, the only way to give psychology away is to do the hard work of getting out there and delivering the information to the people who can use it. Otherwise, the research doesn't get applied."
As to why eyewitnesses tend to become so much more confident when their identifications are confirmed, the answer has yet to be discovered, but it probably lies somewhere in the realm between rational thought and emotion.
"Cold cognitive mechanisms include the normal tendency for people to recall things in a manner consistent with present knowledge," Dr. Wells explains. "For example, if I tell you that the stock market rose dramatically today, you might recall things consistent with that outcome. In other words, you "knew it all along" and were confident that you knew it. We have not been quite as effective in researching the "hot" cognitive processes, like the good feelings that come from confirming feedback, or the ego stroke that makes us boastful."
The results of Well's research, along with the related work of others, allows a deeper understanding of how eyewitness confidence develops, why it changes and how confidence becomes dissociated from accuracy. It adds to a body of prior research on eyewitness identification that has had, in some police agencies, a substantial influence on lineup and related identification procedures, making it more likely that the guilty will be caught and the innocent not ensnared.
In related NSF-funded research, Dr. Wells is studying the issues surrounding mistaken identifications resulting from computer-generated composites.