The Memory Illusion

by Dr Julia Shaw

  Given that identifying the interrogator could be essential intelligence that the navy would want to know should you get released, and since you have specifically been trained to bring home this kind of information, you paid a lot of attention to his features. You certainly had enough time to look at him. So, if I put two photographs in front of you, do you think you would be able to identify which showed your interrogator?

  In 2013, PTSD researcher Charles Morgan from Yale University and a number of colleagues published a study17 that examined whether individuals in this exact situation could be prone to the same kinds of misinformation effects that have been observed in lab studies.

  How did they do it? Simply by showing the mock prisoners a single mugshot photograph for a few minutes while they were in solitary confinement. The photo was not the interrogator, but was introduced by the person who brought it in as if it were. The original interrogator had medium-length curly brown hair and round features, while the person the participants were shown had no hair and very narrow features. They looked totally different. In spite of this, when the participants were subsequently asked to identify their interrogator, the overwhelming majority – 84 to 91 per cent – misidentified them and picked the false photo. The researchers had intentionally introduced misinformation, and it had replaced the participants’ memory of the real face.

  In the same study Morgan and his colleagues also demonstrated that they could change whether participants reported the presence of neutral items such as glasses or a telephone in the room, or even more critical pieces of information such as the presence of a uniform or weapon. Depending simply on how the participants were asked about their experiences when they returned to the barracks, the presence or absence of these features in their memories underwent marked changes. When leading questions were used, such as ‘Was the uniform worn by your interrogator green with red boards or blue with orange boards?’ or ‘Did your interrogator allow you to make a phone call? Describe the telephone in the interrogation room’, 85 per cent and 98 per cent of participants respectively described that they saw the uniform, and that there was a phone in the room. It’s true that even when no misinformation was presented some participants included inaccurate details, but this was comparatively rare. In general it seems that simply being shown photographs or asked particular questions can plant false details in our memories, even for incredibly emotional events.

  Better left unsaid

  It’s not only outside sources that can dramatically alter our recollections of emotional events; we are also prone to distortion from internal influences. One way this can happen is through sharing our memories with others, something that most of us are likely do after important life events – whether it’s calling our family to impart some exciting news, reporting back to our boss about a big problem at work, or even giving a statement to police. In these kinds of situations we are transferring information that was originally encoded visually (or indeed through other senses) into verbal information. We are turning sensory inputs into words. But this process is not flawless; every time we take images, sounds or smells and verbalise them we potentially alter or lose information. There is a limit to the amount of detail we are able to communicate through language, so we have to cut corners. We simplify. This is a process known as verbal overshadowing, a term coined by psychological scientist Jonathan Schooler.

  Schooler, a researcher at the University of Pittsburgh, published the first set of studies on verbal overshadowing in 1990 with his colleague Tonya Engstler-Schooler.18 Their main study involved participants watching a video of a bank robbery for 30 seconds. After then doing an unrelated task for 20 minutes, half of the participants spent 5 minutes writing down a description of the bank robber’s face, while the other half undertook a task naming countries and their capitals. After this, all the participants were presented with a line-up of eight faces that were, as the researchers put it, ‘verbally similar’, meaning that the faces matched the same kind of description – such as ‘blonde hair, green eyes, medium nose, small ears, narrow lips’. This is different from matching photos purely on visual similarity, which may focus on things that are harder to put into words, such as mathematical distances between facial features.

  We would expect that the more often we verbally rehearse and reinforce the appearance of a face, the better we should retain the image of it in our memory. However, it seems that the opposite is true. The researchers found that those who wrote down the description of the perpetrator’s face actually performed significantly worse at identifying the correct person out of the line-up than those who did not. In one experiment, for example, of those participants who had written down a description of the criminal, only 27 per cent picked the correct person out of the line-up, while 61 per cent of those who had not written a description managed to do so. That’s a huge difference. By rehearsing only details that could be readily put into words, the participants had de-emphasised the nuances of their original visual memory, making it harder to access.

  This effect is incredibly robust, as indicated by quite possibly the biggest replication effort of all time in psychology.19 This was a massive collaborative effort by almost 100 scholars and 33 labs, including Jonathan Schooler and Daniel Simons, and published in 2014. All researchers followed the same protocol, and they found that even when the experiment was conducted by different researchers, in different countries, and with different participants, the verbal overshadowing effect was constant. Putting pictures into words always makes our memories for those pictures worse.

  Further research by Schooler and others has suggested that this effect may also transfer to other situations and senses. It seems that whenever something is difficult to put into words, verbalisation of it generally diminishes performance. Try to describe a colour, taste or music, and you make your memory of it worse. Try describing a map, a decision or an emotional judgement, and it becomes harder to remember all the details of the original situation. This is also true when others verbalise things for us. If we hear someone else’s description of a person’s face, a colour or a map, our memory of it is also impaired. Our friends may be trying to help us when they give their verbal account of something that happened, but they may instead be overshadowing our own original recollections.

  According to Schooler,20 besides losing nuances, verbalising non-verbal things makes us generate competing memories. We put ourselves into a situation where we have both a memory of the time we described the event and a memory of the time we actually experienced the event. This memory of the verbalisation seems to take precedence over our original memory fragment, and we may subsequently remember the verbalisation as the best account of what happened. When faced with an identification task where we need all the original nuances back, such as a photo line-up, it then becomes difficult to think past our verbal description. In short, it appears our memories can be adversely manipulated by our own misguided attempts to improve them.

  This does not mean that verbalising is always a bad idea. Schooler’s research also shows21 that verbalising our memories does not impair performance – and may even enhance it – for information that was originally in word form: word lists, spoken statements, or facts, for example.

  Another way we try to hang on to the past is by taking pictures. We tell ourselves we are making memories, that these photos will help us remember our lives. But if we can overshadow our memories by verbalising them, can we also overshadow them with pictures? Of course, photos inherently retain some of those nuances that we lose when we verbalise things, but the issue of potentially creating competing memories remains. In 2011, Linda Henkel at Fairfield University conducted research22 to examine the impact viewing photographs can have on memory. She had participants complete a series of tasks such as breaking a pencil, crushing a paper cup, or opening an envelope. She then had them return a week later and simply asked them to match photos with descriptions of tasks. Some of the photos were of the tasks the participant had completed, while others were not. Tw
o weeks later the participants returned again, and this time they were asked to identify which of 80 tasks described on a list they had completed during the first part of the experiment.

  Simply seeing photos of completed tasks made it more likely for participants to think that they actually had completed them, even without the researcher making that implication – they thought they had done things simply because they saw photos of them. Seeing a photo made participants about four times more likely to say they had done something they had not done.

  This effect extends to more complex autobiographical experiences as well. Research from 2008 by Alan Brown at Southern Methodist University and Elizabeth Marsh at Duke University23has demonstrated that simply showing people photos of particular locations makes them more likely to erroneously report having visited those places when asked a week or two later. Participants were more likely to misremember visiting places that were mundane than unique places. Because their study was investigating memory for visiting locations on a college campus, mundane locations included things that exist on all campuses, including classrooms, libraries and streets. Unique locations included photos of statues, artwork and particularly ornamental buildings. When questioned, 87 per cent of their participants claimed to have visited at least one mundane location and 62 per cent claimed to have visited at least one unique location. None of the photos were from the campus the students actually visited, they were from an entirely different campus, so it was impossible for the students to have seen the depicted locations during the campus tour. A possible explanation is that it is easier for us to picture and accept visiting places that are mundane, since we can draw on real previous memories of similar places that we can mistake as a memory of the fake visit.

  When researchers manipulate images or introduce misinformation to suggest that people did things that they never did, the problem unsurprisingly becomes even worse. In 2002, research by Kimberly Wade and Maryanne Garry at the Victoria University of Wellington, along with their colleagues Don Read and Stephen Lindsay from the University of Victoria, showed24 that half of the participants in a study could come to recall details of a hot-air balloon ride they had never taken simply through being asked to remember the supposed event while being shown a photoshopped image of themselves in the balloon basket.

  Another study, published in 2004 by Stephen Lindsay from the University of Victoria and his colleagues,25 showed that the photos didn’t necessarily even have to be altered. The team had half of their participants imagine experiencing three events from childhood, while the other half were asked to do the same while looking at a real photo of their former school classmates. Participants were then asked to recall their memories of the events in question. Two of these had actually happened (information about these true events had been provided ahead of time by the participants’ parents) but the third was a fictional event which had been invented by the team.

  Of those who were asked to picture the event happening, 45 per cent formed false memories of it, while an astonishing 78.2 per cent of those who pictured the event and were exposed to true pictures of old classmates formed false memories. So giving pictures to the participants who were trying to remember events made them more likely to create memories of things that never actually happened. These real pictures served as a foundation that the participants could meld into their false accounts, making them feel more real.

  It seems that photos can quite severely mislead our memories, especially when coupled with deliberate misinformation. One of the main reasons for this is presumably similar to the cause of verbal overshadowing; when we see a photo we create a new memory of that occasion which can interfere with our memory of actually experiencing (or not experiencing) an event. When we think about the event we may then have trouble distinguishing between our memory of the photo and our actual experience – possibly even entirely replacing a real visual memory with another. Emotional or not, verbal or visual, our memories can be readily manipulated.

  Critical incident stress debriefing

  Given all this, what should we do when someone experiences a highly emotional event? Think about it. If someone has just been involved in a train crash, or witnessed an attack, what should we do? We may feel unclear about how to handle such a situation, wanting to offer support but perhaps worrying about distressing the person further by forcing them to revisit painful memories.

  Those who respond to such catastrophes may use something called critical incident stress debriefing to try to help people through this difficult time. It is a process first introduced by emergency health services researcher Jeffrey Mitchell from the University of Maryland in 1983,26 and it is often referred to as psychological first aid. It is a structured process, administered by trained crisis interventionists. The technique is quite simple, and is founded on the notion that people who have been through an extremely emotional event usually have a need to share their experience with others. In what is known as the recoil phase, people work through what happened to them, and often seek out others who have had similar experiences.

  In critical incident stress debriefing people are brought together in small groups, 24 to 72 hours after an event has taken place. Every person attending is encouraged to tell their story of what happened. The intention is to allow people to go from only sharing cold facts, describing the incident with as little detail as possible, to later on exploring their thought processes during the event in detail. It is a gradual and guided disclosure of the event and its repercussions. Participants in the process are also encouraged to focus on their reactions and symptoms, asking questions such as ‘What is the very worst thing about this event for you personally?’ Finally, participants are educated as to what normal recovery from an event like the one they experienced looks like. It sounds as if it is a well-rounded intervention and clearly it is done with the best of intentions. However, I am afraid to say that I fundamentally disagree with pretty much every part of the process.

  It was clearly not formulated by a memory scientist. For one thing, group recollection situations like this are the poster child for people’s memories melding into one another’s – for better or worse. Due to verbal overshadowing effects, both our own descriptions and the descriptions of others may now become a permanent part of our memory records. Every new account we hear has the potential to taint and re-taint our memories.

  I’m not alone in my concern. According to a 2003 review of the academic literature by Grant Devilly and Peter Cotton at the University of Melbourne,27 the techniques of critical incident stress debriefing can have toxic effects and may even allow for vicarious traumatisation. Vicarious traumatisation occurs when someone tells another person about an event and they experience adverse trauma-like symptoms as a result. Let’s imagine that person A and person B were both at an event, but that person A saw gory details that person B was not privy to. In a group session person A talks about these details and explains the terrible effect of seeing them. Later, when person B thinks of the event they think of both their own account and the horrible details told to them by person A. Person B would have been better off had that adverse memory fragment not been imparted to them.

  This approach also catastrophises an already delicate situation. Not all people will react to so-called potentially traumatic experiences (PTEs) in the same way. A PTE is an event which is generally considered extremely negative, such as surviving an attack or a natural disaster where your life is in danger. But there is no such thing as an inherently traumatic event – the event only becomes traumatic when the person who experienced it suffers from severe psychological consequences.

  The number of people exposed to PTEs varies widely from country to country. According to researcher Dean Kilpatrick from the Medical University of South Carolina and his team,28 almost 90 per cent of people in the US will experience a PTE at some point in their lives. They found that 8.3 per cent of those exposed to a PTE suffered enough symptoms to be clinically diagnosed with PTSD at some point in their lives. Indeed
, when we look at results from various countries around the world, results consistently suggest that only about 1 in 10 of those who experience a PTE suffer severe long-term clinical consequences from it.

  So, while some people who experience a PTE will be traumatised and go on to develop post-traumatic stress disorder, others may have almost no emotional response, and yet others may even feel somehow gifted and enhanced by surviving an experience. However, by setting up the expectation that everyone who experienced a particular event probably has – or should have – a severe adverse response, critical incident stress debriefing has the potential to adversely homogenise people’s reactions, pushing their memories and responses to be more negative than they naturally would have been.

  If a person goes into the debriefing with the view that they have been through a bad situation, but for them personally it wasn’t actually too awful, are they really going to say that in the company of people who are tearful and obviously suffering? Instead they may re-evaluate their stance and think to themselves that it should be a big deal. They may therefore reframe their original experience and remember the whole situation as worse for them than it actually was, and ultimately have a worse prognosis regarding their ability to deal with it. When we ask people about just how terrible an event must have been, or about how a situation has fundamentally changed them, we are setting up social expectations that they are likely to have a hard time avoiding. We are trying to help but are actually making things worse.