The Memory Illusion

by Dr Julia Shaw

  All this ties in quite naturally and cyclically with memory. According to an article published in 2014 by David Ross and colleagues at Vanderbilt University,20 the reason we are able to recognise faces at all is because we have these face-learning strategies. Ross suggests that it is because we have a strong set of memories of what faces look like that we are able to identify new faces. More specifically, he suggests that faces are represented in our brains by their similarity to exemplars of previously experienced faces. In other words, we remember new faces in relation to our database of faces that we already have: How similar is this new face to old faces?

  This so-called exemplar-based model of remembering faces means that our existing memory bank matters. It allows us to optimise the way we analyse faces, minimising the time and effort we must invest in learning a new face. However, these strategies that we apply then taint our memories of new faces, and can backfire when a face has too many new features. It’s almost as if our face database can’t handle too much newness at once. Luckily for us, while we may initially be bad at recognising faces from other ethnicities, each new face we encode results in a small update to our database of what faces look like. This relates to an idea known as the contact hypothesis, generally attributed to American psychologist Gordon Allport in 1954. This is the idea that people generally become more understanding of other groups, and more appreciative of their points of view, the more contact they have with them. This may also apply to facial recognition. Indeed, according to a review by Steven Young at Tufts University and his colleagues in 2012,21 there is some, albeit mixed, evidence to support the idea that the more exposure we have to other-race faces, the better we get at identifying them.

  In addition to the own-race bias, we have also been shown to have an own-age bias and an own-gender bias. Siggi Sporer22 from the University of Giessen suggests that this is because otherness, in general, is bad news for memory. In a review conducted in 2001, Sporer argues that not only are we bad at identifying people from other groups of any kind, but we are generally overconfident in our ability to perform well at this. As with many other tasks discussed in this book, we think we are going to be good at identifying others, even if we think others may not be good at identifying us.

  So, while eyewitness identification is at the core of most legal trials, research shows that there are fundamental memory characteristics that make any such identification a whirlwind of possible errors. In such cases, independent pieces of corroborating evidence are needed if we are to feel at all confident that an identification is correct. The Dutch have a great saying that applies here: one witness is no witness.

  Making monsters

  When I am not lecturing or conducting memory research, I sometimes work on criminal trials. These typically involve issues related to the memory and identification topics we have discussed up to now. Since lawyers and the police generally only bring in a false memory expert when something very bad has happened, I tend to work on the kinds of highly unpleasant cases that make people worry about the fundamental nature of humankind. Murder. Abuse. Sexual assault.

  While most cases I work on are shocking, nothing has ever worried me as much as my first. Note that I am going to obscure the details for the sake of the integrity of the case. In this case it was alleged that a number of teachers and members of the clergy had abused – both sexually and non-sexually – a large number of the pupils at a religious-run school. It was a historical case, so these crimes were said to have taken place over 40 years previously. The case had already been investigated twice before, both times being abandoned due to insufficient evidence for criminal proceedings, leaving a shelf filled with binders of documentation to be waded through. Some of the alleged perpetrators and witnesses had already passed away. The police staff had changed multiple times since the case was first opened, and none of the detectives who had originally worked on the case were still on it. It seemed as though everyone had to start from scratch. Well, everyone except for the people leading the charge on behalf of the victims, as they had told their side many times to many people over the years, as police came and went.

  I had been contacted by a detective at the police force that was dealing with the case, who had asked me to stop by to talk about ways I might potentially be able to become involved with their work. I was both pleased and impressed because that sort of thing does not happen very often – we academics can seem inaccessible to the very professionals we would like to help. However, here was the detective asking about my research.

  ‘We need someone to help us catch these monsters,’ he said.

  ‘These suspects’, I corrected him.

  In a world where catching the bad guys is the motivation to get up every day and deal with tremendously difficult people and situations, using a term like ‘monsters’ makes sense. Of course they want to catch monsters. But in some situations we do not know who the bad guys are, or whether they exist at all.

  This was one of those situations. Of those involved in the case, the overwhelming majority claimed that nothing untoward had ever happened to them, at least nothing outside the bounds of acceptable forms of punishment that were available to teachers at the time – those were the days when it was legal to use basic forms of physical punishment. Only a very small group of former pupils claimed to remember bits and pieces of unusual, problematic situations. What made things even trickier was that these bits and pieces seemed to have drastically changed and grown over time.

  Still, I was armed with my notepad and a pen, determined to establish who said what to whom when. Just because a case is tricky, and memories and accounts look contradictory at first glance, this does not mean that that the case should be dismissed – it just means that people need to be aware that allegations are being dealt with, not facts. Only if there are too many red flags, and insufficient independent corroborating evidence, might we begin to think there is reasonable doubt and that a case is based on false allegations or false memories.

  This particular case is still ongoing, and it will likely take years to figure out whether the claims are reliable. So far, there are a number of red flags, including victims’ accounts drastically changing over time, and denial by most of those who were at the school that any of this happened at all. A large number of denials, combined with not much corroborating evidence, will make it a difficult case. Either way, it is important for police in such situations to at least acknowledge that all of us can succumb to tunnel vision and not to get too caught up in the assumption that there must be monsters to catch.

  Making Monsters is an excellent book by sociologist Richard Ofshe and Ethan Watters23 which highlights the way confidence and assumptions about memory can lead to the personal acceptance of false memories of victimisation. Processes involving cascades of such assumptions can take hold in the justice system, radically affecting the course of a case; one overconfident witness or victim can start a domino effect which, in the worst-case scenario, could end up putting innocent people in prison.

  It is crucial that the justice system becomes aware of overconfidence factors, memory illusions and the problems we have with identification, since these can lead to atrocious situations where we sometimes build cases based on nothing but air. There is a tendency when a crime has been committed, especially a particularly horrific one, to assume guilt of an accused party. While of course it would be terrible for someone guilty of a terrible crime to go unpunished, it is surely equally terrible to end up punishing the innocent as a result of poor practice or inadequate understanding.

  It is tempting to think that perhaps the police, under pressure to secure convictions, might be more vulnerable to making such errors than the rest of us. That’s simply not the case – we are all defective detectives when it comes to identifying real memories or true accounts of events, be they of crimes or everyday occurrences. We are all vulnerable to the same kinds of memory and confidence illusions. And we need to realise that confidence is not key in these kinds of situations. To me, h
igh confidence is often instead a warning sign. WARNING, this person may not fully appreciate their biases. WARNING, this person may not be aware of memory illusions and shortcomings. WARNING this memory is too good to be true. I approach high levels of confidence with high levels of caution because if it is overconfidence, it can be incredibly destructive.

  7. WHERE WERE YOU WHEN 9/11 HAPPENED?

  Flashbulbs, memory hacking,

  and traumatic events

  Why our memory for emotional events is flawed

  IN 2015, BRIAN Williams, the most-watched television news anchor in America, was suspended from his job at NBC Nightly News amid confusion and recriminations. In 2003, Williams had gone to the front lines of the war in Iraq to give a news report and while he was under way his helicopter platoon came under fire. Ten years later he described the event during a TV interview with David Letterman:

  Two of our four helicopters were hit by ground fire, including the one I was in. RPG and AK47 … We were only at 100 feet doing 100-forward knots. We landed very quickly and hard and … we were stuck, four birds in the middle of the desert and we were north out ahead of the other Americans. … They started distributing weapons and we heard a noise. It was Bradley fighting vehicles and Abrams tanks coming. They happened to spot us. This was the invasion. The US invasion. They surrounded us for three days during the sandstorm that was so big that it suspended the war effort. It was called ‘Orange Crush’. And they got us out of there alive.

  Two years later, in 2015, he reflected upon the situation once again whilst on air:

  The helicopter we were travelling in was forced down after being hit by an RPG. Our travelling NBC News team was rescued, surrounded, and kept alive by an armour mechanised platoon from the US Army 3rd Infantry.

  This is a very intense story with many specifics. Williams repeated it on numerous occasions and clearly had enough confidence in it to relate it on television for everyone to see. And everyone did see it, including the men who were present in the helicopter that went down. ‘Sorry dude, I don’t remember you being on my aircraft,’ wrote one of them on the NBC News Facebook page,1 in response to a video of Brian Williams describing the event. Another responded to the Facebook post: ‘He was actually on my aircraft and we came in behind you about 30–40 minutes later.’

  It transpired that Brian Williams had told the story of what had happened to an aircraft in front of him – he was never in the helicopter that was attacked, and since it was an event with so many witnesses his account was easy to debunk. A media storm ensued. Everyone immediately assumed that Williams had purposely embellished his experience in Iraq in order to bolster his own reputation. He apologised, but the damage was done and his credibility was shattered.

  However, given the nature of the work that I do, I cannot help but try to contextualise the immediate jump to the conclusion that he was making it up. To me it seems premature to pull out the pitchforks when we aren’t sure why someone is giving an inaccurate account, mostly because we can unfortunately never actually tell unintentional and intentional fabrication apart, unless the person later tells us that they were lying. What it definitely does bring to light, however, is a core assumption that many people make about memory. To me it very much seems like what happened in the Chopper Whopper scandal was that Williams was accused of being a liar, at least in part, because our general assumption is that no one could possibly misremember such an emotional event. But is that really the case?

  Highly emotional

  If you are like most people, you probably believe that trauma memories are special. But your understanding of these trauma memories is probably self-contradictory. You probably believe that on the one hand we often forget or repress incredibly emotional events, but on the other that we can have nightmares and flashbacks to them. If this is true, then you probably believe that our memories of traumatic events are simultaneously worse and better than non-emotional memories. But, which is it?

  In a 2001 paper aptly entitled ‘Is traumatic memory special?’, Stephen Porter at Dalhousie University and Angela Birt at the University of British Columbia suggest2 that there are a number of different viewpoints when it comes to memories of highly emotional events.

  The first of these, they say, is the traumatic memory argument, which is the view that we remember traumatic events differently, and often worse, than other kinds of events. The underlying assumption behind this argument is that the high emotional impact of traumatic situations overrides our other processing abilities. Proponents of this viewpoint would argue, for example, that a soldier in a war zone may be so severely shell-shocked that they have trouble encoding or recalling a coherent memory of a battle. We can actually trace this idea all the way back to Aristotle: ‘Memory does not occur in those who are in a rapid state of transition, whether owing to some perturbing experience or their period of life; just as no impression would be left if a seal were stamped on running water.’

  This view assumes that our memories of traumatic events are thus stored as fragmented images, emotions and sensations without a coherent structure. A soldier might remember the smell of the battlefield, the sound of gunshots and the taste of blood, but not remember any particular events. Proponents of the traumatic memory argument claim this is why PTSD sufferers sometimes experience powerful flashbacks – they are remembering little fragments of traumatic memories rather than whole events.

  On top of the idea that we may store such memories as individual fragments without a structure, proponents of the traumatic memory argument also often contend that people can ‘dissociate’ during highly emotional events. The word dissociation is used in many ways, but it most often describes symptoms including derealisation, feeling as if the world is not real, and depersonalisation, feeling as if you are not real.3 Such dissociation can supposedly occur during a highly emotional event, when a person feels like they are not actually there, or after such an event, when a person has persistent feelings of not being real.

  Proponents of the traumatic memory argument, such as professor of applied psychology Judith Alpert, who in 1998 formed a working group with colleagues to investigate the nature of trauma memories,4 suggest that dissociation is ‘a psychological defense against the impact of trauma and the mental mechanism that most likely accounts for the amnesia and hypermnesia commonly experienced by traumatised individuals’. In other words, things can be so traumatic that our mind has trouble dealing with them at all, essentially making such memories inaccessible. The statement is also inherently contradictory because it simultaneously says that we have both amnesia, which is severe forgetting, and hypermnesia, which is enhanced recall of an event.

  From my own experience, the belief that this can happen is widespread amongst both practitioners and the general public. But, is the assumption of dissociation as a response to trauma reasonable? This is an important question, as according to medical doctor Angelica Staniloiu and her colleague Hans Markowitsch in a review of the science of dissociation in 2014,5 ‘dissociative amnesia is one of the most enigmatic and controversial psychiatric disorders’.

  Researchers such as Porter and Birt say that claims in support of the traumatic memory argument are poorly supported. Most memory researchers since the early 2000s argue that while dissociation may be possible, people usually do not dissociate during emotional events, and that there is no evidence to support a special fracturing process of memory in trauma situations. It is also unlikely that there is such a thing as repression, hiding emotional memories from direct access, but we will get back to that in a later chapter.

  Related to, but different from, the traumatic memory argument is the medical argument of trauma encoding. This also suggests a diminished memory for such events but is different because it focuses exclusively on principles of brain biology. It includes the basic premise that accidents or violent assaults may cause physical trauma to the brain, which can result in amnesia. In other words, actually physically damaging parts of the brain can lead to memory los
s.

  The medical argument also includes the idea of mnestic block. This posits the existence of a type of amnesia due to problems in the way the brain works rather than physical damage. According to Hans Markowitsch from the University of Bielefeld and his colleagues6, mnestic block syndrome is ‘related to an altered brain metabolism which may include changes in various transmitter and hormonal systems (GABA-agonists, glucocorticoids, acetylcholine)’.

  These sorts of medical impairments, unlike the vague traumatic memory argument, are not thought to cause fragmented memories, and rather than leaving a memory gap for the single inciting event, are generally thought to cause weeks or even years of amnesia. They are also scientifically sound and medically indisputable; breaking the parts of your brain responsible for memory processing, whether it be in structure or function, will disrupt your memory. However, such specific disruptions to the brain’s memory system are very rare, barring pathological memory-impairing conditions that befall many of us in later life, such as Alzheimer’s or dementia.

  In spite of assertions about the negative impact of trauma on memory, the majority of current research actually supports the idea that, barring a broken brain, there is a ‘trauma superiority effect’ on memory. In 2007 Stephen Porter and Kristine Peace at Dalhousie University7 published a study that looked into this issue. They recruited participants who had recently experienced trauma, and interviewed them immediately, then 3 months later, and then about 3.5 years later. They asked them about their traumatic memories and also about highly positive emotional memories at the same time, focusing on features such as vividness and clarity, the overall quality of memory relative to other memories, and the existence of sensory components to the memory such as sights, sounds and smells.