This article is part of our Dishonesty Uncovered series. Dishonesty is a core component of fraud, which globally is estimated to cost a staggering £3.89 trillion each year – and is on the rise. In the UK alone, the cost of fraud to business is estimated at £130 billion. But what do we know about dishonesty and is it possible to learn how to spot it?
Widely held beliefs about what dishonesty looks and sounds like are routinely debunked by science. Our biases – whether they be conscious or otherwise – can cloud or impede our ability objectively to evaluate evidence. This presents a particular problem when it comes to witness testimony.
In conjunction with Professor Mark Howe and Dr Lauren Knott of the Centre for Memory and Law at City, University of London, Philippa Rees, Gareth Minty, Sofia Berggren and James Watson explore over a series of articles what science tells us about relying on memory, and how this intersects with the perception of dishonesty.
Changing Recollections Over Time
In our first article, we looked at how we encode memories and what that means for how we assess inconsistencies between witnesses to the same event. In our second article, we will look at inconsistencies within a single witness' account and how memories can change over time. In particular, we explore how memories are retrieved, and consider the risk of cross-contamination and how post-event information may be incorporated into the 'original' recollection.
Changing recollections over time may simply be a function of how memories are retrieved
Understanding the way that memories are retrieved is fundamental to a proper understanding of memory and can explain why a person's recollection may change over time. If a witness later recalls something that they had not mentioned previously, this can often be seen as an inconsistency and evidence that the witness is lying or at the very least, mistaken. However, it is equally possible that a changing recollection over time is merely a function of how our memories are retrieved.
This is because, when it comes to retrieving memories, we do so in a (re)constructive manner. We do not recall the same event in the same way each time it is recalled. Instead, what we remember about an event at any point depends on what cues are available at the time of retrieval, including how we are questioned, and how we retrieve the memory of the event in question. As the cues themselves can vary across time, different aspects of the same experience may be retrieved.
It is particularly important to understand this in the context of legal proceedings, where multiple interviews of the same witness may take place over the course of several weeks or even months. Using different retrieval cues (or questions) across interviews can be a valuable way to elicit further details from a witness, leading to recollection of different details that had not previously been recalled (known as reminiscence). As long as leading questions are not used, the more varied the retrieval cues in each interview, the more reminiscence will be expected to occur, though importantly with no increase in contradictory items. That is, new information may be brought to light in later interviews with this technique but there is little research evidence to suggest this information will be inaccurate1. We therefore see that a changing recollection can occur as part of a normal functioning of memory.
Memories are not static but are vulnerable to the incorporation of post-event information or to cross-contamination
Caution does need to be exercised, however. Memories are not static and, once encoded and stored, information in memory does not remain intact in that form until it is subsequently retrieved. New memories can interact with already established memories and, as we saw in our first article, this is in and of itself an important function that keeps our memories current and up-to-date. Indeed, this is one way in which memory helps us to understand the present and anticipate the future. By “overwriting” or modifying previous memories with ones that are more current and perhaps more relevant to understanding our environment, we can better understand the meaning of current experiences and better anticipate and prepare for the future. Thus, memories old and new do not remain static and unalterable but are dynamic and can change as a function of their interaction.
Whilst this is a helpful process in the context of the everyday functioning of memory, it can be more problematic in the context of legal proceedings, for example if post-event information, such as that obtained from media reporting or third party accounts, is incorporated into the original memory. We can see an example of this in the case of the Oklahoma City bombing in April 1995, when 168 people were tragically killed and 600 injured after a bomb exploded outside the Alfred P Murrah Federal Building in Oklahoma City.
Timothy McVeigh was arrested and later tried and convicted of the bombing. The main evidence for the conviction of McVeigh came from several witnesses who worked at Elliot's Body Shop where McVeigh had rented the truck used in the bombing. One of the three eyewitnesses, when first questioned, stated that McVeigh had been with a second man, who he described as wearing a blue baseball cap with a zigzag pattern. The presence of this 'second man' was used by defence lawyers to suggest that someone other than the defendant was responsible for the bombing. However, a year and a half after the bombing, it transpired that the man described by the witness was in fact another customer who had rented a van the day after McVeigh had been in the body shop. Post-event information had become incorporated into the original memory.2
This case also highlights another, related, issue of memory cross-contamination. When this happens, memories of details from various sources can be amalgamated with the original memories of an event3. In the Oklahoma City bombing case, at the first interview, the other two witnesses gave very sketchy descriptions of McVeigh. However, after the initial interviews, their memories transformed and became more detailed and confidently held, with one coming to give details about the baseball cap the second man was wearing, with blue 'lightning stripes'. In all likelihood, the witnesses were incorporating information picked up from other sources, including media coverage and the first witness' recollection.
A study on this issue found a “memory conformity” effect whereby post-event information provided by a co-witness was later included in the eyewitness report of the second witness4. Using a novel procedure, they showed each member of a pair different videos of the same event. Each video contained unique event items seen only by one witness. The groups of pairs were then split into a control group and an experimental group, with the pairs in the experimental group being encouraged to discuss the event. The control group did not discuss the event. Finally, each witness performed an individual recall test.
Performance was compared between the two groups. Seventy-one percent of witnesses who had discussed the event went on to mistakenly recall items that had in fact been acquired during the discussion with the co-witness, compared with 0% from the control group. These errors are believed to be due to both cognitive factors, such as the source confusion we have been discussing5, but also to social factors, such as the individual’s need for social approval6.
Such 'contamination' can also occur as a result of the way that the witness was questioned. Leading questions that include new information can lead to "uptake" of these details in an interviewee's later recollection. For example, a study by Loftus & Palmer (1974) found that subjects who were shown a film of a traffic accident and then asked "about how fast were the cars going when they smashed into each other?" provided a much higher estimate of speed than subjects who were asked "about how fast were the cars going when they hit each other?"
Witnesses are unable to disentangle new information from the original recollection
The most challenging part of this from a legal perspective is that the witness will not be aware that this is happening and will be unable to disentangle the post-event information or contamination from the original memory. In research that examines source attributions, participants have been asked, after exposure to misinformation, to recall the event, but state if specific pieces of information came from the event or from the post-event information they were provided with. Misleading details, that could only have come from the later post-event information, are often believed to have been seen in the original event7. Even when it is explained to the witness that this may have happened, such details, once processed, cannot be easily undone.
As Gary Wells, who acted as an expert witness in the Oklahoma City bombing case, explained in direct questioning during the trial:
"Q: Now, suppose the eyewitness was told to shut the television image out of their mind and only rely on their memory of the previous event. Wouldn't that take care of the problem?
A (Gary Wells): No. You're asking a person to do something they can't do. Their memory by this point is blended. In other words, the first time in which they saw the person which they are being asked about has been blended with – with this second time – in which they may have seen the person on television, and so there's no real way for them to – for people to sort that out. It's just a single memory."8
This leads to an interesting conclusion – a witness can be honestly convinced that what they are saying is a true account of what they witnessed, when in fact that may not be the case. This in turn requires us to reassess our natural association of honesty with truth. Although we consider them to be one and the same thing, we can see in this scenario that a witness may simultaneously be being honest and at the same time recounting details that they did not personally witness but have instead picked up from another source or even from another event.
So what does this mean for how we assess evidence?
This presents a fundamental problem for those who assess evidence. If the witness' own memory is fallible, how can third parties expect to uncover the truth? There is no straightforward answer to this question because there is no litmus test for the veracity of what has been remembered, unless there is separate corroborating evidence.
What is key is that assessors of memory evidence – i.e. judges and jurors – must have a proper understanding of these issues and, when assessing evidence, consider several intervening factors. How long has it been since the event took place? Who has the person spoken to since the event took place? What was discussed, how was it discussed, and what has the witness been shown? Were there multiple witnesses and have they ever spoken to each other about the event?
As we have seen, all these factors, which will, by their nature, be common issues facing the legal system, impact the accuracy of memory. Each case has to be considered individually, assessing a multitude of the above factors and having knowledge of the impact of each.
The critical issue to consider therefore is how we can improve the understanding of the main participants in the legal process – judges, jurors, lawyers, and investigators. As we will examine in our next (and final) article, important strides are already being made in this respect in both the civil and criminal spheres, but more remains to be done.