The Problem of Eyewitness Memory

Reading about the recent exoneration of Robert Dewey , after 16 years behind bars in the United States for a rape and murder that he did not commit, I was reminded of the great work being done by the Innocence Project. A not-for-profit organisation based in the US, the Innocence Project, like its sister organisations across the world, provides legal help to those wrongfully convicted of crimes, usually through the use of DNA testing. Since it was established in 1989, the project has played a part in the exoneration of 290 individuals in the US alone.

An equally important role of the Innocence Project is to attempt to prevent wrongful convictions from occurring in the future. The Project estimates that 70% of those wrongfully convicted are the victims of false eyewitness identification, and puts a lot of effort into teaching people how to take steps to reduce the incidence of false identification. To give just one example, when eyewitnesses come in to perform an identification, they are often provided with simultaneous line-ups, where they can see all members of the line-up at once. This can result in the eyewitness choosing the member of the line-up who is the “best fit” – that is, the member who looks most like the perpetrator, even if the actual perpetrator is not even in the line-up. Providing images of members of a line-up sequentially can help prevent these relative judgments from being made. More information is provided in this video produced by the Innocence Project.

There are a number of other ways to improve the techniques used in line-ups, and much research has been conducted in this area (see this paper for a great review). However, understanding why false eyewitness identification occurs also relies on an awareness of the fallibility of memory. We generally think of ourselves as having a good memory for events that occur in our life; indeed, our whole sense of self relies on these memories. Yet time and again studies have shown that our memories are highly malleable. Thus it might not only be that eyewitnesses are making erroneous heuristic judgments, but they may actually have a distorted memory of the events and people involved in the crime.

A couple of now classic studies conducted in the 1970s demonstrated that memory for events can be distorted by providing information after having experienced the event itself, a phenomenon called the “misinformation effect”. In 1974, Loftus and Palmer had participants watch a video of two cars in an accident. Participants were either asked “How fast were the cars going when they smashed into each other?”, or the same question with less emotive verbs such as “contacted” or “bumped”. The experimenters found that participants who were asked the “smashed” question generally gave a higher estimation of the cars’ speed. Moreover, at a follow-up test this group were more likely to erroneously remember seeing broken glass in the film.

In a second study in 1978, Loftus and colleagues got participants to watch slides of a car driving along a road and hitting a pedestrian. Half of the participants were shown the car stopping at a stop sign, and the other half saw it stop at a yield sign (that’s “give way” to us Brits!). Within each group, half of the participants were asked whether another car had passed by when the car was stopped “at the stop sign”, and the other half whether another car had passed when it was stopped “at the yield sign”. Finally, participants were shown one slide with the car at the ‘Stop’ sign and one with the car at the ‘Yield’ sign, and had to choose which they had originally seen. Participants who had been given consistent information (e.g. originally saw the yield sign and were asked about the yield sign) were much more accurate at choosing the correct slide than those who had been given inconsistent information (e.g. saw the yield sign but had been asked about the stop sign). It seems that members of the inconsistent group often had a false memory of which sign they saw, because of the inconsistent verbal information they received.

It is not hard to see how the results of these studies might transfer to real life situations, and it is clear that the misinformation effect could be responsible for some cases of false eyewitness testimony. Just as memories of the car accident were influenced by the phrasing of the question in Loftus and Palmer’s study, the phrasing of questions made by the police or in court could have a profound influence on eyewitness memory, potentially affecting memory for the course of events or even of the appearance of the perpetrator. We should be wary of the use of leading or emotive questions in these situations. Similarly, just as participants’ memories were influenced by the presence of additional, conflicting information in the second study, the presence of additional information after witnessing a crime could affect memory for the incident. Indeed, in the video above, it is suggested that Jennifer Thompson’s memory for her attacker could have been influenced by the long time she spent viewing the picture of the man she was to wrongly identify and comparing him to the other options. The result was that she really did “remember” this man attacking her, even though she had never seen him until after the attack.

These are just two of the hundreds, if not thousands, of studies that have demonstrated how easily memories of events can be altered. There is now a wealth of research on the factors involved in the misinformation effect, as well as other distortions of memory. But what seems most incredible about all this research is that it often seems simple, yet has such profound social implications. Even just a basic knowledge of the science of memory amongst the right people could be the difference between an innocent man’s freedom and his incarceration.