The Memory Illusion

by Dr Julia Shaw

  In another study on this topic, psychological scientists Mika Koivisto and Eveliina Rientamo from the University of Turku in Finland found that the same kind of effect can occur for a task identifying animals. They found that flashing a picture of an animal so quickly that the participant could not consciously perceive it made the participant faster at classifying another picture of that animal. For example, participants would be faster at deciding ‘animal or not’ for a picture of a horse if they previously saw a picture of a horse non-consciously. However, the researchers found that this effect only worked for this very basic level of ‘animal or not’, and did not improve participant reaction times for anything else, like different kinds of animals. These kinds of results indicate that this non-conscious process is likely to have incredibly limited ability to influence our behaviour.

  While these kinds of subliminal effects are still poorly understood, and cannot always be replicated, they are likely due to an effect called priming. Priming is another memory phenomenon. It is a function of implicit memory, the process that allows our previous experiences to inform our present or future experiences despite us not consciously realising we are being influenced by those particular memories at the time. It is a form of memory that cannot be accessed in the way we normally think about memory – as a sort of visual, acoustic or tactile representation; this kind of memory is generally argued to be more primitive, closer to being an impression or feeling.

  I like this idea. I trust this brand. I feel like I should slow down. This seems dangerous. Perhaps a smaller amount of attention, so small that it is not perceptible, is all we need to encode these deep-rooted feelings as memories. These are the kinds of friend-or-foe feelings that have allowed us as a species to make snap decisions that have enabled our survival over millennia. They are still memories, and they can have strong influences on us, but we cannot remember their origins.

  When priming effects were first discovered, some people thought that this meant that any kind of subliminal stimulus would elicit effects, including things such as backward messages woven into songs (a process referred to as backmasking). The public feared that these kinds of effects could be used like brainwashing, for evil. Of course researchers also wanted to know whether messages played backwards could influence people. In a series of studies described in a paper from 1985 entitled ‘Subliminal messages: Between the devil and the media’, Jon Vokey and Don Read at the University of Lethbridge32 set the issue to rest. They set themselves the question ‘Is there any evidence to warrant assertions that such messages affect our behavior?’ Their answer? ‘Across a wide variety of tasks, we were unable to find any evidence to support such a claim.’ According to them, and scientists who followed in their footsteps, we can neither process nor remember such backward messages, so we can rest assured that they can have no impact on our beliefs or behaviours.

  So I hope this chapter has demonstrated that we need some form of attention to be able to create memories, and that sleep is crucial for the consolidation and strengthening of those memories. I also hope that it has shown that intelligence-enhancing baby videos, sleep learning, and hypnosis or subliminal messaging as ways to influence ourselves or others, are all best considered creative fictions.

  6. DEFECTIVE DETECTIVE

  Superiority, identity crisis,

  and making monsters

  Why we are overconfident in our memory

  AS A CRIMINAL psychologist, when I prepare expert testimony I sometimes find myself siting at my desk reading case files where one thing after another seems to have gone horribly wrong. From problematic eyewitness testimony, to unreliable victim statements, to detectives who seem to misremember how evidence was collected, I find myself with a host of issues that warrant concern.

  Conspiratorialists may argue that the police cannot be trusted, perhaps even suggesting that they may intentionally distort the facts of a case. I, however, choose to believe that it is not the integrity of the police that is problematic. I am confident that they almost always want to do their jobs as effectively as possible; they want to catch criminals and to protect the public. The problem is that they have been assigned an impossible task, where they need to piece together the past in a way that is entirely reliable. But, as we know, memory almost never is entirely reliable.

  Unfortunately the police are often under-equipped to deal with the complexity of the memory problems that can plague investigations. In a study I published in 2015 with Chloe Chaplin1 at London South Bank University, I investigated whether or not British police officers knew more about memory and other psychological processes than members of the general public. We distributed a 50-item questionnaire and found that, overall, the police held as many misconceptions about issues in psychology and law as the general public, but that they were more confident in their responses. Of our confidently wrong police, 14 per cent endorsed the myth that ‘Memory is like a video camera’ and 18 per cent believed that ‘People cannot have memories of things that never actually happened’. This research points to a lack of police education and potential problems with overconfidence, the latter being something we will explore throughout this chapter.

  As much as we may wish our justice system to be infallible, and as much as we hope that the police always catch the right culprit, we know that in truth this is not always the case. There are plenty of instances of people being wrongfully convicted and imprisoned for horrific crimes. The Innocence Project2, an organisation dedicated to getting innocent people exonerated through DNA testing, has helped to release at least 337 people who were wrongfully convicted. On average, these people served 14 years in prison for a crime they did not commit. Faulty memory played a role in at least 75 per cent of those cases. Those figures are just for the US, and just for cases in which DNA was available, so around the world there are significantly more people who have been wrongfully imprisoned.

  When such cases are subsequently examined it often becomes clear that the police officers involved did everything in their power to get a suspect convicted. It might be easy to assume that the police have been terribly negligent or, worse, deliberately tried to frame someone they knew to be innocent. Perhaps sometimes that is the explanation, but it is also perfectly plausible that they simply got caught up in a string of psychological biases. The police can develop ‘tunnel vision’, where they overvalue evidence that supports their argument and discredit or ignore information that contradicts it.

  And it’s not just the police – this kind of process can happen to anyone, because incorrect information can seep into any of the coherent stories we construct to understand reality. To use a term stolen from one of the world’s leading legal psychologists, Peter van Koppen, we can all be ‘defective detectives’, struggling to be unbiased evidence collectors.

  As we will see in this chapter, when we need to make sense of an event, but do not have enough information to do so, we tend to import other plausible content to fill in the gaps. Events in our minds need to have a linear progression, connections, reasons. Once we have this kind of plausible narrative, we can become incredibly confident in its accuracy. But what exactly is the relationship between confidence and accuracy, and how does it all tie in with memory?

  Above average

  Let’s switch gears for a second. Do you think you are a good driver? How about compared to your peers? As Ola Svenson from the University of Stockholm asked her participants in a 1981 study:3 ‘We would like to know about what you think about how safely you drive an automobile. All drivers are not equally safe drivers. … We want you to indicate your own estimated position in this experimental group.’

  Rather than being interested in the self-assessed driving ability of participants, this was actually one of the first studies to look at overconfidence. It was found that the overwhelming majority of Americans and Swedes thought they were both safer and more skilful than the average driver. Svenson had even asked them to compare themselves specifically to the average in the particular study sampl
e, which means that they thought they were better than peers of similar age and intellectual capacity.

  People responding this way intuitively makes sense to me, since whenever I find myself on the road I cannot help but think that most of the other drivers are idiots. Similarly, other studies have shown that most people think they are more intelligent, more attractive and more competent than average. We may not necessarily believe we are absolutely brilliant at everything – far from it – but we do generally think that we are better than average at pretty much everything. Which is, of course, statistically impossible – if everyone thinks they are above average, clearly a lot of people are wrong. Yet, studies have found this overconfidence effect in all kinds of areas. Police are overconfident in their ability to detect liars. Students are overconfident about their course grades. CEOs are overconfident in their business decisions. Teachers are overconfident in their teaching ability. The problem is so persistent that in a 2011 article published in Nature, social scientists Dominic Johnson at the University of Edinburgh and James Fowler at the University of California4 argued that ‘Humans exhibit many psychological biases, but one of the most consistent, powerful, and widespread is overconfidence.’

  One reason for this may be the superiority illusion, which suggests that we have a tendency to overestimate our positive qualities and to underestimate our negative traits. This is a characteristic that is inherently linked to memory, because in order to think about our positive traits we need to be able to remember the good things we have done in our lives that provide evidence of those traits. For example, you may think about all the times you have done chores around the house, and think to yourself that you are a really good spouse. You took out the trash, bought groceries, cooked, and did the dishes. Go you. However, you may be forgetting or diminishing the times when you did not do any of those things and actually made more work for your spouse, leaving them frustrated and with extra work to do.

  In 2010 the web company Cozi conducted a survey of 700 men and women with children who were either married or in a committed relationship.5 It aimed to find out how much each partner thought they contributed to household chores, and how much they thought their spouse contributed. This type of research is sometimes called a ‘chore wars’ survey. Perhaps unsurprisingly, women were thought to do more chores by both partners. But much more interesting than this finding was people’s opinion on the comparative distribution of chores. If the percentages each partner thought they contributed to each chore were added together, they frequently came to over 100 per cent. Let’s take, for example, the item ‘scheduling of events and appointments’. On average dads claimed to do 50 per cent of the scheduling, and moms claimed to do 90 per cent. Of course, it’s not possible to do more than 100 per cent of any given chore, so what is happening here? Perhaps the participants simply misunderstood how to assign percentages to these tasks, but I propose an alternative explanation: our memory is selfish.

  We are less likely to remember something done by someone else than something we did ourselves. This is partly because watching a partner do a chore, or having them report to us that they did it, provides us with a far less rich and complex memory trace than if we had performed the task ourselves because there is simply less sensory input. The memory trace being weaker means that we are probably more likely to forget that it happened in the long run. On the other hand, we will always have stronger and more meaningful memories of those occasions when we have done chores. This means that unfortunately the game is always rigged against our partners – our recollections of our own contributions are always likely to be stronger and more significant.

  Besides the superiority illusion, we also suffer from survivorship bias. This is an error whereby we tend to focus on successes and overlook failures, literally focusing on people or things that survived a process. This is the kind of mistake people make quite blatantly when they say things like ‘Steve Jobs was a university dropout, so I’m going to drop out in order to make a success of myself.’ By focusing on one success, they are failing to think about all the people they have never heard of who were in a similar situation and did not achieve fame and fortune.

  In 2003, a study was published on survivorship bias amongst investment bankers. In it, hedge fund manager Gaurav Amin and Harry Kat from City University, London, looked at how hedge funds were invested between 1994 and 2001.6 They found that many investments were ended early, and were omitted from a database that was used to calculate the risk of new investments. They argued that because generally it was bad investments that dropped out of this database it overly focused on investments that worked, leading to a survivorship bias. The authors argue that this bias means that ‘hedge fund returns may be overestimated and risk may be underestimated’. Many economists have identified overly optimistic investing as a contributory factor in financial crashes, and such investing obviously arises from a basic bias as a result of only thinking about successes.

  Similarly, a police officer may discount the times they elicited false confessions from suspects, at least partly because they may not even know that such failures happened – they have therefore omitted this information from their internal database of failures and successes, in a similar way to our investment bankers. If a suspect is in jail a case is generally considered closed – even if the person ended up there due to poor police practices. In such cases the police officer is likely to miss any failures on the part of themselves or their colleagues and instead count them as successes, falsely bolstering their self-appraisal. Lack of visibility of our own failures along with an excessive-focus on achievements leads to overconfidence in our abilities and assessment of opportunities, hence the survivorship bias.

  There is one more illusion that may play into our tendency to be overconfident. It is related to the greater strength and accessibility of our memories of our own actions and insights compared to those of others – the illusion of asymmetric insight. In 2001 Emily Pronin at Stanford University and her colleagues published a paper7 on this bizarre bias, appropriately entitled ‘You don’t know me, but I know you’.

  Over six studies the team demonstrated that we think we know our close friends and roommates better than they know us. For example, in the first study participants were asked to think of a close friend and answer a number of questions about how well they knew them, including the extent to which they thought they understood their friend’s feelings, thoughts, motivations and personality. Finally, they were asked whether they could see the ‘essential nature’ of their friend. This was done by telling the participants that we are all like icebergs, with part of our true selves observable by others and part hidden from view. The participants were then asked to pick a picture of an iceberg that best represented their friend from a selection showing icebergs at various levels of submersion. Then, participants did the reverse task, thinking about how their friend would answer these same questions about them. Five other similar studies were also conducted to examine this bias for different types of relationships, including for roommates and strangers.

  Pronin and her team found that participants believed that their own quintessential qualities, including their intimate thoughts and feelings, were mostly kept internal but that those of others were more likely to be observable. They were more submerged icebergs, while other people were more visible icebergs. This makes sense from a memory perspective because we have direct access to our own thoughts and feelings and so appreciate that they can be complicated and nuanced – which makes them difficult for other people to understand. On the other hand, it can be difficult or even impossible to appreciate the complexity of the thoughts and feelings of others in anything other than a basic ‘surface’ way – we tend towards assuming that is all that there is to understand. Our general outlook is ‘I’m a riddle, but my friend is an open book.’

  This bias turns out to be really important for our decision-making and arguing skills. In their final study, Pronin and her team asked 80 participants to complete a backgro
und questionnaire on a number of politically relevant topics, including such items as whether they identified as liberal or conservative, or whether they were pro-life or pro-choice. Then, several weeks later, they asked questions about how well the participants thought their in-group knew their out-group, and vice versa. So, for example, they asked self-identified conservatives how much conservatives as a whole know about liberals, and they asked them how much they thought liberals knew about conservatives. They found that liberals and conservatives both claimed to know the other side better than the other side knew them, as did those on either side of the abortion debate.

  Asymmetric insight helps explain why in arguments and debates we may believe that the other side will never understand our point of view. We may also think we perfectly understand their point of view, perhaps also bolstered by the superiority illusion that we are smarter and more informed than our opponents. As Pronin suggests at the end of her paper, we can begin to think ‘I know everything about the other party, and I know they are wrong. They don’t even try to understand my arguments. If only they knew more about it, they would be on my side.’ It is an easy trap to fall into, and one that is a staple of political shouting matches.

  So, overconfidence has far-reaching implications, from bias in our everyday internal dialogue when evaluating relationship fairness, to our inability to give our failures equal weight and acknowledgement to our successes, and our problematic assumptions about the knowledge other people have of us and we of them. It touches every aspect of our lives. Even if we wish to be humble and take pains to avoid overconfidence illusions, we may not be able to – they are largely the by-product of selective memory processes we cannot control.