The Memory Illusion

by Dr Julia Shaw

  According to the international organisation the Innocence Project,16 which is dedicated to exonerating convicts who they strongly believe are innocent, faulty memories, particularly of eyewitnesses, are the major contributing factor for wrongful convictions. For example, in 2015, out of 325 cases where modern DNA testing proved innocence beyond reasonable doubt, a whopping 235 cases involved eyewitness misidentification. This suggests that false memories play an absolutely critical role in the imprisonment of the innocent.

  There are no easy answers here, but let’s not deny the reality of false memories. Let’s tell everyone that they exist, that they can look and feel like real memories, and that we can even misremember highly emotional or traumatic events. Let’s increase insight into how our beautiful minds work, accept the malleability of memory as a part of life. Knowledge is power, and ultimately greater knowledge of this subject empowers us all and protects us from interview techniques and assumptions that can spiral out of control.

  Rich false memories exist, whether we want them to or not.

  10. MIND GAMES

  Secret agents, memory palaces,

  and magical realism

  Why we should embrace our faulty memory

  IF I HAVE done my job, your memory should now seem hopelessly fragile, impossibly inaccurate. To bring you to an acceptance that all of us have critically flawed memories is the very reason I wrote this book. You hopefully now appreciate just how plagued memory is by biological flaws, perceptual errors, contamination, attentional biases, overconfidence and confabulation. But where does this leave us? We cannot possibly just write off memory as a lost cause. We still need it. We rely on it, every single day of our lives.

  As I have briefly mentioned once or twice already, metamemory is our knowledge of our own memory and the way in which it functions. This is a type of metacognition, a type of thinking about thinking. Having this ability means that we can muse about why we remember, how we remember, and how good we are at remembering individual pieces of information. One of the first experimental studies on metamemory was conducted in 1965 by Joseph Hart.1 He wanted to understand a particular feature of metamemory, a construct he called the ‘feeling of knowing’.

  When you know you know

  Hart described the feeling of knowing as the sense we have when we think we have something stored in our memory but cannot recall it. He wanted to know whether this feeling is associated with accuracy; whether when we feel like a memory is there but we cannot immediately access it, we are correct in assuming that the memory exists. Over many years of research, he demonstrated that when his participants had this feeling of knowing, they were often correct. However, this information could only be recognised, not recalled. This means that, for example, if we have a feeling of knowing, we are likely to be able to correctly recognise information in a multiple-choice test but will not be able to produce it from scratch in an open-ended question. For open-ended questions we need more than just a feeling of knowing, we need memories that are more accessible.

  In a more recent study of this phenomenon from 2014, psychological scientist Deborah Eakin2 and her colleagues at Mississippi State University wanted to know whether the feeling of knowing was generally accurate regardless of age. They had students, who were on average 19 years old, and seniors, who were on average 72 years old, complete a memory experiment. Participants came into a lab twice to complete a computer task. The first time they came in they were shown a series of pictures of faces. Some of these faces were of famous people, and some were faces of unknown people. This acted as a foundation for the researchers to filter out faces that the participants already knew.

  Participants were then asked to come back a week later. Here they were given the faces of non-famous people again, but this time they were presented with names. The participants were instructed to remember the name–face pairings. After being shown all of the names and faces, the participants were asked how good they would be on a multiple-choice test at recognising the name for each face, from 1 to 100. They were then actually tested, being asked to recognise the name of the face they were shown from a set of three.

  As the researchers had predicted, the university students were better at remembering new faces and names than older participants, which is in line with a general decline in ability to learn new information as we age. However, they found that the feeling of knowing was the same for all participants. Both younger and older participants generally accurately predicted when they would know information.

  This lends support to the idea that we may have an intuitive understanding of the things that we remember and things that we do not. It is the reason why we sometimes say things like ‘I’ll know it when I see it’ – because sometimes we just know we know something, even if we can’t directly remember the information.

  But, wait. Doesn’t this fly in the face of many of the things I have said in this book? I have repeatedly demonstrated that we are overconfident in our memories and really cannot rely on our instincts as a guide regarding memory veracity or accuracy. Am I suddenly backtracking? No, I’m not. While our feelings and thoughts regarding whether we have stored a memory are often linked to that information being there, they also often are not.

  If we take Eakin’s face–name matching task as an example, younger participants rated their feeling of knowing for the multiple choice test as 42 out of 100 for items they later recognised, and 24 out of 100 for those they did not later recognise. In other words, even for information that they could not later remember participants still had a pretty high feeling that they would be able to anyway, an error which is akin to our overconfidence bias from Chapter 6. This shows that this feeling is still far from perfect and can mislead us. A classic feeling of knowing error in everyday life is a situation in which you think ‘I know that guy’ because his face seems familiar, even when in reality you have never seen him before.

  When we start questioning our metamemory insights, like wondering where our feeling of knowing is coming from, we begin to talk about metacognitive monitoring of metamemory. Meta-metamemory. It is at this point that we transcend the regular human pursuit of thinking about how good our memory is, and replace it with a much larger scale question of why we think the way we do about memory. Within this meta-metamemory there is a great deal of frustration, as we come to wonder whether any of our memories are ever trustworthy, but also a great deal of potential for maximising our memory ability. And actually, although I have been at pains to point out the many ways that our memories are faulty and can lead us astray, I must also emphasise how incredible they are – to have a biological system which associatively stores such vast amounts of information is a miracle of evolution and we should all consider ourselves extremely lucky.

  Let us explore some ways in which we can harness our faulty memories to work to our advantage, and to potentially change how we reflect on our own lives.

  Brain games

  Some of us aspire to know everything – or at least to know more. And in the quest to make us cleverer and more knowledgeable we seek out memory training. There is now a huge and growing industry in smartphone and computer games that promise to make you smarter and to improve your memory. The ads say sexy things like ‘scientifically validated’, or that they will give you ‘bespoke brain fitness’, or claim to be designed by neuroscientists. They usually say that their benefit is proven because you get better at the games as you play them. ‘Look! When you first started you could solve a Sudoku in ten minutes, now you can do it in two minutes!’

  Following this argument along we could potentially argue that all games train our brains, since we almost always get better at them with practice. And throwing in a claim that a game can physically change your brain is extra manipulative, as this is of course always true. Technically everything we do changes the physical structure of our brain, if ever so slightly, so of course playing brain training games will also do this.

  These kinds of games also tend to promise ‘transferabl
e’ gains: brain improvement that can be learned in a game but used for tasks in other contexts. Many of them claim that they focus on building our fluid intelligence, which has to do with our capacity to think logically and solve problems in novel situations. Fluid intelligence is dependent on our working memory – how much we can keep in mind at any one time. Naturally the idea that we could play a game to increase our working memory and raise our IQ accordingly is an alluring one.

  One way to measure fluid intelligence is via Raven’s Progressive Matrices test, often referred to simply as the Raven test. This involves completing sets of patterns that get harder as the test progresses. These puzzles are purely visual, and usually involve basic shapes like squares and triangles. The test is supposed to measure two types of fluid intelligence: the ability to make sense of complexity and the ability to store and reproduce information.

  In 2008 two researchers set out to test whether memory games can increase performance on the Raven test. Psychological scientist Susanne Jaeggi and her colleagues at the University of Bern3 wanted to see what would happen if they had participants play a brain game called the n-back, which involves being presented with a series of stimuli – typically pictures, numbers, or letters – each spaced several seconds apart. So you might get L … M … K … M. The participant then has the job of deciding whether the current stimulus (a letter in our example) matches the one that was displayed n trials ago, n being a stand-in for the number of places in the series the participant has been instructed to go back. For example if you are doing a 2-back, you want the participant to say whether the letter in front of them now is the same as the one they were shown two letters back. The more letters back you are asked to go, the harder the game becomes.

  The purpose of the game is to train you to be able to hold an increased number of things in your working memory at once. Jaeggi and her team wanted to know whether doing this difficult task would actually have transferable outcomes on a type of Raven test. They were astonished by their findings. After decades of research ruling out transferable gains from tests intending to improve cognitive capacity, they had done it. They found that before doing the n-back, on average participants could solve 9 or 10 of the 29 Raven questions in 10 minutes, but this was increased by 4.4 questions after 19 days of n-back training. Participants could now keep more items in their working memory, as demonstrated by their enhanced performance on the Raven intelligence test.

  This study opened up researchers, who had for years largely failed to find transferable gains, to the idea that perhaps we could make brain boosting games that actually improve people’s lives. But the idea came crashing down once again when these fluid intelligence boosting effects seemed few and far between and other researchers were not able to find the same results. In 2015 Monica Melby-Lervåg from the University of Oslo, who studies how we can help people with special needs, and her colleague Charles Hulme, a psychological scientist,4 conducted a meta-analysis of whether working memory training is effective. They looked at all the research that had ever been conducted on the topic, and through a statistical analysis of all the results put together, they came to the conclusion that while our trusty iPhone games are likely to make us better gamers, to date ‘there is no convincing evidence that working memory training produces general cognitive benefits’.5

  While memory-enhancing games are an exciting avenue to explore, it seems premature for us to go home and tirelessly play games to make our memories better.

  Secret agent mnemonics

  Occasionally organisations approach me as a consultant, with a view to helping their members improve their memories. Perhaps the most interesting work of this kind that I have been involved with has been with military institutions.

  When you think about how a memory specialist might be able to help the military you may be tempted to think of thriller-like scenarios, perhaps using memory research to improve the interrogation of terrorists, or to change the way the military questions civilian witnesses. You may even jump to my favourite question that people sometimes ask me on this subject, ‘Do you implant false memories in spies?’ I’m sorry to disappoint you but no, I do not. While I’m sure that some of my colleagues have been directly involved in helping the military understand the psychology of interrogation, and even in spy training, I have been approached mostly about one thing: teaching operatives how to identify and retain high-quality information.

  My role has not been to help the military squeeze information out of others, but to help them use evidence-based techniques for their own remembering. The military has strong awareness that they need their operatives to come home with reliable memories. Talking to informants embedded in unfamiliar cultures, military agents get little, if any, time to take notes. They need to know how memory works and how they can best avoid memory illusions, as the stakes are incredibly high and to make good decisions they need to have intelligence they can trust. In order to prevent them from making basic mistakes with their own memories I teach them many of the things discussed in this book: how biases can taint the perception of information; how memories are adaptive to social demands; and how memory naturally changes over time.

  They need to use memory tools to help them remember on the ground – after all, the kind of information they are collecting is hard to come by and they need to avoid it becoming in any way corrupted. So, how do I help them with this? I teach them a few simple mnemonics. Mnemonics are any kind of simple technique to make it easier for us to remember a specific piece of information. They can take many forms – rhymes, acronyms or mental imagery, to name but a few.

  According to a memory mnemonic master, Ed Cooke, ‘What you have to understand is that even average memories are remarkably powerful if used properly.’6 Cooke has become a memory ‘Grandmaster’, for which you need to demonstrate to the World Memory Sport Council (yes, it exists) that you are able to do three things: memorise 1,000 random digits in an hour, memorise the order of ten decks of cards in an hour, and memorise the order of a deck of cards in under two minutes. It is a feat that seems impossible to us mere mortals. Luckily for us, Cooke has dedicated his life to understanding and applying amazing memory techniques. He performs these feats not because of some innate ability but through self-taught mnemonics. Mnemonics that you too can master.

  Mnemonics have been around probably as long as self-reflective memory itself. You probably learned quite a few yourself as a child, perhaps to help you remember the dates of certain historical events, or perhaps the names of planets in the solar system. One of the mnemonics I learned in primary school is ‘Never Eat Soggy Wieners’. It was a mnemonic device for me to remember the four cardinal directions: North East South West. It’s silly, it’s juvenile – but I will never forget it. Now, even though I know the directions, I can’t get rid of it. I have the opposite problem to difficulty with memory retention – I have a memory suppression problem, whereby I can’t think about the compass points without thinking of that phrase. But as a memory aid it worked like gold.

  It worked so well because the sentence is conceptually easy and makes grammatical sense, but more importantly it’s also quite weird and you’d be highly unlikely to hear it in day-to-day life.

  Be weird

  Research clearly shows that, from a memory perspective, weirdness sticks, or to put it differently, unexpected components generally make for the most memorable examples and pieces of information.

  Consider a statement like: ‘Don’t think about pink elephants.’ It is something you can visualise. It is unexpected in a conversation. It is a bit weird. We probably have few previous associations, if any, with pink elephants. This particular statement has the added appeal of automatically making you do exactly the opposite of what it is telling you. For the rest of this chapter, perhaps even long after reading it, you will probably remember that it talked about pink elephants. We don’t care about the pink elephants themselves; we are using them as a mnemonic device to remember the effectiveness of unusual examples for help
ing us remember things. See what I did there? I just used a mnemonic device to help you remember something about mnemonic devices.

  According to research from 2013 published by psychological scientist Lisa Geraci from Texas A&M University and her colleagues,7 the bizarreness effect – our tendency to have a better memory for the unusual – is well documented. As they put it, ‘this bizarreness effect is a robust finding in recall that has been obtained across a variety of encoding tasks and delays.’

  In the typical research paradigm looking into this effect, participants study sentences, some of which are bizarre and some of which are more commonplace. Within these sentences are embedded nouns written in capital letters. The participants are subsequently asked to report back all the nouns they can remember from the sentences in a free recall task. It turns out that nouns embedded within a weird sentence like ‘The DOG rode the BICYCLE down the STREET’ are recalled significantly better than the same nouns given in a typical sentence, such as ‘The DOG chased the BICYCLE down the STREET.’ Similarly, a sentence like ‘The BISCUITS screamed when the OVEN jumped out the WINDOW’ will force you to put more effort into making a connection between the words, and to visualise them more, than ‘The BISCUITS were visible through the OVEN WINDOW’.

  Geraci and her team conducted exactly this kind of experiment. Normally in this kind of study participants are given lists in which half the sentences are bizarre and half are normal. But Geraci had participants work either with a mix of both normal and bizarre sentences or with sets of sentences that were either all normal or all bizarre. What she found was that the memory benefits associated with bizarre sentences only emerged when the participants were presented with a mix of sentences, so the bizarreness of a sentence is no longer useful when everything we need to remember is bizarre. It seems common sense when you think about it – part of what makes the weirdness work is that it stands in contrast to the normality of everything else.