TO IDENTIFY a suspect, police typically ask eyewitnesses to pick him out of a line-up of similar-looking folk. Alas, this method is often inaccurate. DNA testing has shown that witnesses often fail to detect the culprit or, worse, wrongfully accuse the innocent.
Why should this be? Witnesses may feel under pressure to identify someone as the guilty party, even if they are not sure. Or they could be worried about making a mistake. A new study suggests that witnesses will make more accurate identifications if they do not have to be so precise.
Neil Brewer, a psychologist at Flinders University in Australia, has devised a new type of police line-up, described in Psychological Science. Rather than simply pointing out the perpetrator, witnesses are asked how confident they are when identifying him. And they have to make up their minds quickly. Whereas a typical police officer tells a witness to take time and mull it over, Dr Brewer sets a very short deadline.
Dr Brewer knew from past research that strong memory traces are more rapidly accessed by the brain than weaker ones—and that accurate eyewitness identifications are made significantly faster than inaccurate ones. So he guessed that limiting the time witnesses had to look at suspects would yield better results.
To test this hypothesis, Dr Brewer and his colleagues asked 905 volunteers to watch a series of short films. The volunteers were told to pay close attention, since they would be asked questions about what they had seen. Each film showed either footage of a crime (eg, someone shoplifting or stealing a car), or a seemingly mundane series of events such as people using a cash machine. For the latter, volunteers were told that one of the people in the film was a suspect in a crime that had occurred nearby and that they would be asked to help identify him.
Half the participants—some a week later, others within five minutes—were then shown, on a computer screen, a series of individual pictures from a line-up of 12 people. The witnesses saw each face for three seconds; after two seconds a buzzer went off, indicating that they had to complete their identification in one second. To do that, they had to click on one of 11 on-screen buttons ranging from “100% confident that this is the culprit” to “0%, absolutely certain this is not the culprit”. If they were too slow, the next face appeared.
The other half of the participants were shown the same faces one at a time, but were given as long as they liked to answer. The results were analysed using a statistical test. The witnesses who were given a flexible choice but only a short time to ponder were much more accurate. They picked the correct suspects 67% of the time. The “yes or no” group picked the right man only 49% of the time. A witness delayed may be justice denied.