The Memory Illusion

by Dr Julia Shaw

  For infants, especially young infants, this is likely to be an impossible task. Babies do not yet have the ability to decipher which part of the magical glowing screen they have been placed in front of is most significant. They may not even find it the most attention-worthy thing in the room. This makes the chances of their managing to filter out and retain the take-home messages which their parents are so keen for them to absorb seem remote. Never mind that in the very early months a baby may not even be able to physically see the complex educational video they have been placed in front of, since their eyes are still unable to focus more than a few inches past their own noses. So, no, unfortunately your baby probably isn’t creating useful memories from the TV while you do the dishes.

  But don’t just take my word for it. In a research report from 2010, ‘Do babies learn from baby media?’ Judy DeLoache and her team from the University of Virginia3 studied how well 12- to 18-month-old children learned language from a popular brand of baby media. They found that children who viewed the educational videos for four weeks did not learn any more or any fewer words than if parents were given no instruction to teach their children language at all. However, they did find that the tots learned significantly more words if they were not exposed to any video but instead were taught words during everyday activities. It seems babies prefer the live show. Other studies have produced similar results. Live presentations of language, and of tasks, have been shown to be far more effective for developing babies’ memories than any kind of media simulation.

  Usually the effect of media on baby development has been found to be non-existent, but one large-scale study on the industry actually points to negative results – in 2007 Frederick Zimmerman4 and his team from the University of Washington found baby television exposure to have highly detrimental effects on language development. They called 1,008 parents of young children and asked them about their children’s media viewing habits. They also asked them to complete the short form of the MacArthur-Bates Communicative Development Inventory which measures language development in children. In this correlational exercise, for every hour of baby media watched per day by infants between 8 and 16 months, they were found to know six to eight fewer words.

  The ineffectiveness of baby media is considered so well evidenced within the academic and professional communities that major paediatric bodies have provided clear-cut guidelines on the issue. For example, in 2011 the American Academy of Pediatrics (AAP)5 clearly said that children under two should have no screen time at all, meaning no iPads, iPhones, laptops or TVs. Instead, parents should use play and live interaction if they want to give their babies the best possible developmental help. This may be bad news for the 90 per cent of parents who let their babies regularly interact with screens, and in practical terms may not be possible, but an effort can be made to limit screen exposure as much as possible. Next time you are looking after a young child and need to get on with some chores, perhaps choose crib time, grandma time, or playpen time, instead of screen time.

  All this goes to show that attention works with a complex array of other physiological and psychological processes to enable memory formation. Simply looking at something and reacting to parts or all of it is insufficient. We may even ask ourselves whether that kind of interaction is a true representation of attention at all. But if that isn’t attention, what is?

  Happy to be blind

  You are blind. And you should be happy about it. What’s more, you are not alone. We are all blind. You see, scientists may disagree on some nuanced points about attention, but they all generally agree that it is the selection of some information for further processing and the simultaneous inhibition of other information. Paying attention to something requires you to be blind to the overwhelming majority of the information you are receiving from both your external and internal environment. It is your trusty filter that allows you to sift through the constant chatter of your senses and thoughts, the chatter that tries to tell you that you are hungry, that you are a bit cold, that the person sitting next to you is wearing a neon shirt, that you need to call your parents later, that your knee hurts, that there’s an interesting conversation happening in front of you, that you like this song, and, oh yeah, how about getting some work done. Amazing human being that you are, you can usually handle it all without even noticing that you are doing it.

  There are a number of experimental studies that have demonstrated just how ‘blind’ we can be when paying attention to something. One of the most famous was conducted by Daniel Simons and Christopher Chabris at Harvard University and published in 1999.6 They asked participants to watch a short video of a group of people passing a ball, and to count the number of times the ball was passed. After the video ended, participants were immediately asked to write down the number of passes, before being asked a series of unusual follow-up questions, which included ‘Did you see a gorilla walk across the screen?’ Of course, one of the most frequent responses was ‘Why would I have seen a gorilla?’ There was every reason for them to have seen a gorilla – during the video a woman in a full-body gorilla costume had walked right through the group of people passing the ball, at normal walking pace. In a phenomenal demonstration of our selective blindness when paying attention, 46 per cent of participants had failed to notice the gorilla due to their preoccupation with the task of counting the passes. This effect is called change blindness.

  Change blindness can happen not only when we observe photos or videos, but also in real life. Psychological scientists Daniel Simons and Daniel Levin published a study in 19987 which showed us that when a stranger asks us for directions, if we are briefly distracted and the stranger swaps out with another person in the middle of the conversation, we are unlikely to notice that we are suddenly talking to someone else. Another psychological scientist, Ira Hyman, and his team from Western Washington University showed in 20108 that we can even fail to notice a unicyling clown if we pass them on the street while we are on our cell phones. Change blindness is the reason that in our everyday lives we miss our partner’s haircuts, or say things like ‘he came out of nowhere’ when we are driving. The phenomenon is common or potentially even universal among animals, having recently been observed in both pigeons and chimps. It appears that even when we look sometimes we don’t see.

  Change blindness is a function of two bottlenecked processes, processes that need to filter a great deal of information and can only do so much at once. The first is our limited ability to perceive the world through our senses. The second is our limited short-term memory capacity. As mentioned in Chapter 1, our short-term memory really is super-short-term, lasting only about 30 seconds, and has a very limited capacity. That means that when we experience a complex scene we cannot possibly remember all of the details in it.

  There may also be a third reason. Ira Hyman argues that we experience change blindness because we have conceptual representations of our experiences in our memories. These representations can be rather abstract – they are the gist memory traces we discussed in Chapter 3. Hyman explains: ‘I have, for example, a rather vague picture of what my friends look like. Surprisingly, my interaction with my friends and the world is better because of this. I need to recognize my friends in different clothes, lighting, locations, and after they get a haircut.’9 These memory functions that seem like failures may thus exist because they offer us larger adaptive advantages.

  It appears that many of us are not just change blind, we are change double-blind. In 2000 Daniel Levin and his team at Kent State University10 demonstrated that most of us engage in a metacognitive error called change blindness blindness. They asked participants to rate how likely it was that they would notice change in four different situations. Three of these situations had been previously tested and had produced change blindness rates in 100 per cent of participants; the fourth was the experiment just mentioned where participants were approached by a lost pedestrian asking for directions and the person switched during the conversation after being briefly hi
dden from view. However, across the four conditions, between 70 per cent and 97.6 per cent of participants thought they would detect the changes described, and they did so with high confidence ratings – we apparently grossly overestimate our live scene processing abilities and underestimate our own change blindness.

  So it seems that our facility for paying attention overwhelmingly works to make us notice only a small amount of information so that we have a chance of actually processing it, and, in certain situations, remembering it for the future. Memory feeds into attention to tell it what ‘important’ information is, based on past experience, and attention feeds back into memory to update our internal representations of the world. But while researchers may not all agree exactly how this process does and does not work itself into lasting memories, they all agree that sleep is an inherently inattentive process. And yet, it sometimes feels like we can learn or remember things that happen while we sleep – as the experiment where sleeping participants were exposed to different scents while different musical tones were played seems to suggest. So what actually happens to our memories when we sleep?

  Replay

  I generally consider sleep an annoying necessity. If I could, I would skip my nightly time-out altogether. What makes sleep all the more frustrating is that scientists are not even entirely sure why we need it. We know how much we need, seven to nine hours. We know what it is like, split between being almost completely unconscious and hallucinating vividly. We also know under which circumstances it is most likely to occur: total quiet and cool darkness. But why do we need it? Simply for rest and replenishment seems like a decidedly insufficient answer, as we could presumably also gain those benefits by lounging on a couch without actually losing consciousness.

  Biological psychologists Gordon Feld and Susanne Diekelmann at the University of Tübingen in Germany argue in a review paper from 201511 that dreaming is a state of ‘active offline information processing essential to the appropriate functioning of learning and memory’. They also suggest that our memory engrams, and the connections between them, are played back to us during deep sleep, like a replay of the day. In particular they claim that something called ‘active system consolidation theory’ can help us understand the association between memory and sleep. This theory suggests that during a type of sleep called slow wave sleep, the memories that we just formed while we were awake are strengthened. This, they argue, is how sleep helps to consolidate memories: by repeating the connections between neurons, and replaying our experiences, which makes some of our memories last.

  According to neuroscientist Gordon Wang and colleagues at Stanford University in 2011,12 sleep particularly appears to be important for bringing down brain activity from the levels it reaches during the day, and perhaps diminishing some of the less important connections to increase brain efficiency – the previously mentioned process of synaptic pruning. Wang’s team argues that this process allows us to keep our most important memory traces and get rid of the less important ‘noise of daily experience’.

  The other thing that seems to necessitate this down-regulation of the brain provided by sleep is the reliance of our brains on glutamate. Most of us know the word ‘glutamate’ from the food additive monosodium glutamate, MSG, and MSG is chemically related to the glutamate found in our brains. Glutamate is the most common neurotransmitter in the brain, and works to open up some of the main channels of communication between cells. These channels allow calcium to flow into cells, which activates them and allows the chemical encoding of engrams, enabling us to generate and access the networks of information required for complex memories. Thus, our brains release glutamate as part of the chemical process that underlies memory formation. This glutamate mainly remains in the brain until it is processed and drained when we sleep. But while we need glutamate to make memories, too much of it is bad for us. An excess of it can cause excitotoxicity, in which brain cells are damaged and killed due to an overactivation of glutamate receptors causing an excess build-up of calcium. It seems that sleep allows us to drive down our overall glutamate production, essentially preventing brain cell self-destruction.

  However, not everything is necessarily down-regulated during sleep. In 2014 researchers including Guang Yang from the New York University School of Medicine found that catching some z’s after learning increases the formation of synaptic spines,13 which are considered one of the foundations of memory storage. Synaptic spines are tiny doorknob-shaped bumps on our dendrites (the connections between our brain cells), and they are where most of the synapses in the brain are located. Generally, by increasing synaptic spines, we improve our memory. Yang and team found this when they had mice learn a new motor skill, running on a fast rotating rod. They then looked into the brains of the mice, having injected them with a protein that made the relevant motor cells fluorescent yellow so their growth over time could be monitored. They found that mice that had been sleep-deprived after the learning task formed significantly fewer spines than those that were allowed to sleep, and thus had worse memories.

  These memory processes that occur during sleep may even help us understand why we dream – we often dream about events, people, situations or emotions similar to those that transpired during the day. We know that memories are being variously pruned or reinforced as we sleep; in the process related memory engrams may be activated simply because of their associations, and it may be that they then manifest as dreams. Of course quite often dreams can be bizarre combinations of engrams that could never occur in reality.

  So, sleep seems to be a way for us to strengthen, reorganise and transform memories. When it comes to consolidating new or complex memories, the old adage is right; it is indeed ‘best to sleep on it’. But, could we tap into this dream state and learn new complex information?

  Psycho-phone

  The 1920s saw the invention of a device called the psycho-phone. It was formally patented in 1928 by Alois Saliger, a businessman from New York who, according to a 1933 interview in the New Yorker, was ‘a tall, spare, thin-lipped man with piercing eyes and a wide forehead’.14

  The device he had created was a record player triggered by a clock, so the device could activate itself once the owner was asleep. Once this ‘time-controlled suggestion machine’15 triggered it would begin to play a recording of Saliger speaking. He would talk in the first person, in a soothing voice, starting with a note about how the buyer was asleep and telling them that their subconscious would now follow his spoken guidance. Then he would begin the sleep therapy, repeating phrases such as ‘Money wants me and comes to me. Business wants me and comes to me … I am rich. I am a success …’16. According to Saliger the device worked because ‘it has been proven that natural sleep is identical with hypnotic sleep and that during natural sleep the unconscious mind is most receptive to suggestions.’17

  A quick search online reveals that similar audio files aimed at helping the buyer achieve their dreams are still widely available, sold as ‘sleep learning’ or ‘subliminal learning’. They make a wide range of promises, from ‘develop extreme motivation’, to ‘overcome social anxiety’, ‘think yourself thin’ or ‘increase your memory by as much as 75%’. There are even anti-ageing programs on offer. According to a scientific review by health psychologist Madalina Sucala at Icahn School of Medicine in New York, along with her international team,18 in 2013 there were at least 1,455 hypnosis smartphone apps offering, essentially, a high-tech version of the psycho-phone.

  These sorts of products make tall claims with major potential applications. It is easy to see how the military, professional organisations and educational outlets would be immediately interested in the possibilities should they prove effective, as indeed they were during the last century. In order to test these claims of the possibility of subliminal learning more scientifically, in 1956 a series of studies was conducted by weapon researchers Charles Simon and William Emmons at RAND Corporation, a company that conducts research for the US Armed Forces.19 Presumably they wanted to know
whether this was something they could use in military training, or even something they could weaponise. They tested responses to material presented at various levels of wakefulness, and used electroencephalography (EEG) to confirm that their participants were actually asleep. In fact, while it seems like a basic prerequisite for such research, they were among the first scientists to ensure that their participants were actually sleeping during a sleep study.

  Their conclusion: ‘The results support the hypothesis that learning during sleep is unlikely.’ They found that exposing participants to learning materials during sleep had no discernible effect. This led to such claims falling generally out of scientific favour. Most researchers considered the issue a closed case, and thought that there was no need for further studies. But a few always remained hopeful, and research did continue, albeit at a very slow pace.

  In a 1995 series of neuroimaging and behavioural experiments examining fear responses in rats, Elizabeth Hennevin and her team at the University of Paris20 claimed that animals could form new associations during sleep, and that information to which an animal was exposed during sleep could have behavioural implications for when they awoke. In particular, they claimed these effects were achievable during a type of sleep known as paradoxical sleep, which exists alongside our slow wave deep sleep. Paradoxical sleep is characterised by rapid eye movements (REM) and by brainwave patterns similar to wakefulness – hence the paradox, the brain is acting as though awake, but the person is not. Hennevin and her colleagues went on to generalise that the same sleep-learning processes should take place in humans, since our brains are similar to those of rats in many important ways. If the brain is in an awake-like state then might the person be able to attend to stimuli, at least in a basic sense? If so, paradoxical sleep could indeed be the key to sleep learning.