The unreliability of recalling events

The brain connects fragmented information with made-up bits, which is why it’s so hard for crime investigators to get reliable facts from witnesses.  The current interrogational method of recalling events backwards has just been shown to be less effective than free recall.  From the September 3, 2011 The Economist:

Forensic psychology:Backwards and forwards

A modern approach to interviewing witnesses of crimes may make things worse

Sep 3rd 2011 | from the print edition

Can you work back to the truth?PEOPLE love to tell tales. Indeed, even when someone’s memory is patchy, he will still do his best to spin the information he has into a credible yarn. This is not a matter of deceit. Rather, it is an established psychological phenomenon in which the brain tries to make sense of fragmentary information. Although such behaviour is natural and normal, it is a nuisance for the forces of law and order when they are trying to find out what happened during an incident by taking statements from witnesses. For this reason, psychologists working with the police often advocate asking witnesses of crimes to say what they saw in reverse order, to stop them making things up to help the story run smoothly. It sounds like sensible advice, and police forces in Australia, Britain, New Zealand, Norway, Spain and Sweden have all adopted it. But a new study suggests that far from improving recall, it makes things worse.

Coral Dando of Lancaster University, in Britain, showed 54 volunteers a short film of a staged mobile-phone robbery. The participants were then split into three groups and, two days later, interviewed about what they remembered from the film. All were asked to describe what they had seen twice, with no wait between the two descriptions. However, the way they were asked to make these descriptions differed from group to group. In one, participants were first told to recall the robbery freely, and then to recall it in reverse order. In another, they were told to recall the robbery in reverse order first and then to recall it freely. The third group was a control. Participants were told to recall the robbery freely on both occasions. All the interviews were recorded and passed to coders who were unaware of the purpose of the study, but who knew all the details of the film. These coders scored every apparent recollection in each interview by noting which items were correct, which were inaccurate (saying a dog was brown when it was really black, for example), and which were complete confabulations—things or events that bore no resemblance at all to anything in the film. A participant’s final score for each type of recollection was the number of such items recalled or invented in at least one of his two debriefings.

Dr Dando and her colleagues report in Cognition that reverse-order recall had a significant effect on the average number of correct items participants remembered—and not a good one. The control group, with no reverse recall, averaged 48.7 correct observations about the incident. The group that started with reverse recall and finished with free recall managed an average of only 42.2. The group that started with free recall but finished with reverse recall did worst, averaging 38.7 correct observations. And though the number of inaccurate recollections did not differ significantly between the groups, the tendency to make things up completely did.

Among the control group, an average of 0.2 such confabulations were created by each participant. Among those who freely recalled the robbery first and then recalled events in reverse order, this value climbed to 0.7. Among those asked to recall the robbery in reverse order first and then recall it freely, it was higher still, averaging 1.4 pure inventions per participant. Moreover, when the researchers analysed when these confabulations were mentioned, the majority (0.6 of the 0.7, and 1.2 of the 1.4) were told during the part of interview that involved reverse-order recall.

Why this is so is a mystery, for it is clearly not what psychology predicts. It does, however, point out the dangers of taking even logically plausible ideas on trust, rather than testing them. Psychologists are often accused by laymen of doing experiments to prove the obvious. In this case, a little more such testing of the obvious might have been sensible.

Advertisement