The Seven Sins of Memory: How the Mind Forgets and Remembers

Reasons to forget

Every year I have the same old dilemma when the kids' birthdays, carol concerts and sports days come around. Do I lug about the video camera and risk distancing myself from the action? Or do I just live the moment and not worry about capturing these events for posterity? I split it about fifty-fifty. The children are completely gripped by their own image and I always marvel at how much has already evaporated from my own memory. Sometimes memory seems limitless in potential. In contests, there are people who can remember the order of fifty-four packs of cards shuffled together and who can recite the value of p to 42,000 places (they do it by employing the trick of turning the cards and number sequences into a story). But even ordinary people can do extraordinary things. In one famous experiment, subjects were asked to sit through a slide show of some 10,000 photographs — random images of dandelions, sunsets, street scenes. Even with just a few seconds to inspect each picture, it took five days to plough through the series. The subjects were then tested to see if they could tell which was the already seen image when presented with a choice of two and managed correctly to recognize which felt the more familiar about 90 per cent of the time.

Memory is huge by many measures, but much of our experience is wiped away as quickly as it occurs. Experiments have shown that the curve of forgetting is drastic. When given a list of random words or nonsense syllables to learn, we can score 100 per cent immediately after the training session. But within an hour, over half the list will be forgotten and by the next day, over two-thirds. It is a small comfort that the curve eventually flattens out so that the 25 per cent we remember after a week falls to only 20 after a month. Faulty memory is not just limited to facts and figures, though. Recent advances in DNA forensics have resulted in the release of hundreds of people who turned out to have been wrongfully convicted. In 90 per cent of the cases, faulty eyewitness evidence put the unlucky person behind bars. The more closely psychologists study our power to recollect, the more it appears that we forget, distort, edit, select and generalize. In the 1970s, a group of secondary school pupils were asked if they had been physically punished in their first year at school. A third said they had been. Three decades later, the same group were asked the question again and, astonishingly, nearly 90 per cent now claimed to have been beaten. Surely our brains ought to have evolved to be a bit more retentive, a bit more accurate. In their various ways, Daniel Schacter, Rusiko Bourtchouladze and James McGaugh try to say something about this evolutionary conundrum.

McGaugh is an eminent neurobiologist and in Memory and Emotion he addresses the complex neural processes needed to build a memory trace into the brain and considers how a strong emotional reaction helps to ensure that an experience is vividly fixed. Bourtchouladze, one of a crowd of younger researchers working on the genetic basis of memory, also talks about the forging of memory connections in the brain and is hopeful that gene therapies or pills will eventually be available to boost human memory powers (she now works for a memory drug start-up, Helicon Therapeutics). Schacter is chair of Harvard University's psychology department and easily the most erudite of the three researchers. He considers the seven ways in which human memory can be said to be faulty. As well as ordinary forgetting, Schacter lists the "sins" of absent— mindedness (when we forget to do the things we meant to do); blocking (when we can't quickly retrieve information); persistence (when we cannot shake a traumatic event or just a nagging tune from our minds); and then misattribution; suggestibility; and bias. While each of these last three has its own chapter, to stretch his list of sins to the requisite seven, they are actually all to do with the way that memories are cognitive reconstructions of what we think probably happened, rather than literal records of past experiences.

Yet though all three memory researchers do a reasonable job of recognizing the puzzling fallibility of memory, none of them does what they really should do, which is to challenge the very notion of memory itself. From a strictly biological point of view, the term memory is a misnomer. Brains are evolved not for retrospection and contemplation but for intention and anticipation — for looking forwards rather than backwards, outwards rather than inwards, for being selective rather than merely retentive. Evolution did its job properly; the problem lies in expecting the brain to be something it is not.

Most people consider their memories to be a dusty library of moments and episodes. When one wants to remember something, an inner librarian answers the call. Sometimes a memory record is handed over swiftly; at other times a lot of rummaging is required. Most memory researchers take much the same view. Experiences get etched into nerve networks as wiring changes in the connections between brain cells. These patterns can then lie dormant for decades until a search effort fires them up, causing them to burst into life and dance before the mind's eye once more. The Holy Grail of memory research has thus been to find out exactly how and where memories get captured and stored. As Bourtchouladze recounts in Memories Are Made of This, this belief in dormant memory traces has led science down some strange paths. In the 1960s, many felt that our memories might be written directly onto the DNA or other large molecules inside brain cells. This was because, even with billions of neurons, there did not seem to be enough room in the brain to record all of a person's memories as wiring patterns. So perhaps the information was imprinted onto molecules stored in cells.

One researcher tested this idea by training flatworms to respond to electric shocks, then mashing them up and feeding them to their fellows. Magically, the cannibal worms appeared to absorb the learning through their guts. Soon the brains of rats, goldfish and other lab animals were being mashed and the extract injected directly into the brains of others to see if memory traces could be transferred. It was only after seven labs got together to show that the frequent claims of positive results could not be replicated that the wheels eventually fell off this bandwagon. Yet the memory molecule saga did nothing to diminish a general belief that memories must be captured as fixed traces. It was taken as axiomatic that brains are designed to record the passing moment as completely, faithfully and permanently as possible. Therefore traces must be lodged in some safe repository in the labyrinthine pathways of the brain, kept separate to prevent them from being overwritten by current brain activity, or from interfering with the processing of that new experience. However, this passive warehouse view of memory has been coming under increasing attack from two angles — social psychology and basic neuroscience.

First, social psychology. An obvious consideration is whether animals have memories. Clearly they can retain information and can apply it to recognizing situations and developing appropriate responses. But can your cat or dog recollect? Is there any evidence that even an animal as smart as a monkey spends time reminiscing about its childhood or reliving the fight it nearly had yesterday? As Lev Vygotsky, the great 1920s Russian psychologist, argued, the minds of animals are locked into the present tense. Their responses are intelligent — fully informed by their past experiences — but also completely focused on the needs of the moment. They have no power to roam their memory banks "offline", selecting snippets at will. It is only humans who have developed the extra knack of recollection — or more accurately, imaginative reconstruction.

The Vygotskian story is that all the higher mental abilities of humans are actually socially constructed habits scaffolded by language. We use inner directed speech to shape our thoughts, getting our brains to respond to abstract questions and imaginary scenarios. So we can ask ourselves what we ate for breakfast, whom we sat next to at school, or what has been our saddest memory. Our brains will then react to these words as triggers, filling our heads with candidate images that we can then probe further.

Now are these images really memories? As all three authors agree, research has now made it abundantly clear that our recollections are never literal replays of moments of experience, but rather perceptual reconstructions. Granted, our mental images usually capture the gist of an event; but it is a warped, foreshortened, edited, glossed-over, view of what happened. Schacter tells the tale of the Dutch researchers who asked people about their memories of the El Al cargo plane which crashed into an Amsterdam tower block in 1992 when two engines failed. This is exactly the kind of incident likely to be remembered vividly — a flashbulb experience where we recall precisely how and when we first heard the news. Ten months after the crash, the researchers asked subjects the leading question: "Did you see the television film of the moment the plane hit the apartment building?". More than half said yes, then went on to supply convincing details about the speed and angle of the plane, whether it was on fire, and what happened to its crumpling body as it fell. The catch, however, was that there was no TV footage of the actual moment of impact, only of the terrible aftermath. The simple implication that there might have been was enough for most subjects to invent a memory of what the pictures would have been like.

We should not over-emphasize the reconstructive nature of memory. Psychological tests show that our recollection of events only a few days past can be both vivid and relatively accurate. Our memories of the more distant past do still usually retain the essence of what happened even if the precise details are cloudy. The problem lies in the fact that memory and imagination are supposed to be different — one is about truth, the other about creative fictions. So we get uncomfortable when our recollections don't appear to fit neatly into either category. It all begins to make more sense when psychologists like a trenchant critic of much memory research — Cornell University's Ulric Neisser — talk instead in terms of more naturalistic constructs such as recognition and anticipation. It is easy to see that animals have evolved brains that are excellent at recognizing and anticipating. For them, memory is something that acts online, creating context and allowing an animal to notice what matters, to ignore what doesn't, and to predict what is most likely to follow. Memory serves as a selective filter to focus attention and narrow the field of experience. It is also forward-acting, generating predictions and intentions, rather than being a contemplative and retrospective faculty. So, as Neisser has argued, a recollection ought to be called an anticipatory image. When we ask ourselves what we had for breakfast or whom we sat next to at school, we are tricking our brains into imagining what it would be like if we were just about to see the same situation again. The mental activity is prospective, and is then judged for its apparent accuracy. We think "Cornflakes" or "toast" and the image is recognized as familiar or unfamiliar. This way of looking at the complex process of recollection steers a middle course between our demand that memories be true or false. Anticipations are certainly built from something that is stored — our accumulated experiences. But they are not the inspection of a retrieved trace. Thus, the remaking of a past event is accepted as an inherently creative and generative process. While social psychologists have dissected the act of recall, neurobiologists have found that brains are not designed for the passive storage of anything. For a start, the very molecular structure of the brain is incredibly unstable. In the hot, wet, bustle of a cell's interior, fats and proteins are falling apart almost as soon as they are made. A recent study showed that every molecule making up the region around a synapse — the junction between two brain cells whose subtle alteration is supposed to lock in a memory — is being replaced by the hour. Some molecules need replacing every few minutes. So much for the brain as a repository of memories that might sit untouched for decades. The brain you had even last year will have been rebuilt many times over by now.

Of course, the brain must do a good job of preserving its general pattern of connections in spite of this frenetic molecular turnover. However, other studies have shown that there is still nothing permanent about the forming of memory traces. For the seconds and minutes directly after an experience, it is maintained by temporary changes in the sensitivity of the synaptic junctions used to process the information. The same circuits still "jangle", and so it is reasonably easy to cause the whole pattern to fire again, reinstating the experience. But over hours and days, the memory must be consolidated as some more permanent state of rewiring. Connections must be strengthened and weakened by the sprouting or pruning of synapses so that the memory becomes literally built into the structure of the brain. It is at this stage that there is a heavy editing and generalizing of an experience. Much of what happened will become assimilated as slight changes to circuit patterns that already exist. This is why the details grow blurred in our minds, leaving behind only a gist. We can still prod our brains to yield anticipatory images based on this gist, but it will have to recruit a generalized best guess about how such an event might have looked and felt. The mind's eye picture might be convincingly bright, but we will have no way of telling which aspects of the image are remembered as particular details and which are merely a view based on many such assimilated experiences. Furthermore, neurobiologists have found that the consolidation process never stops. Years and even decades later the brain is still reshuffling its connection patterns. The pace of change may slow, but a memory trace never settles to a fixed shape, or even a fixed location. Again this is a sound design principle if we think of brains as evolved for the efficient processing of current experience rather than for the preservation of a faithful record of the past. The brain must be optimized to extract both what is general, and what is particular, about each passing moment of life if it is to make maximum sense of future moments. For any recent event, there is a reasonable chance that the same situation may crop up again and so the brain preserves more of the particulars. But, as time passes, it is the generalized memory that becomes more useful. It might be useful still to recognize exactly what you had for breakfast yesterday, but for breakfasts of yesteryear, what you really want is an impression of what you typically ate.

All three of these books do discuss the recent neuroscience and point to the growing dynamism of the picture. However, the authors remain wedded to memory — the retrieval of dormant traces — as a scientific construct and so never escape the faulty assumptions built into this everyday term. Schacter at least manages to give a properly rounded account of how it feels to be the owner of a memory, detailing all the frustrations of trying to make it work as specified on the box. He talks about the memory championship contestants who admit they are really very absent-minded in daily life and survive on Post-it notes. He mentions the ageing baby boomers who rank memory problems as their major health worry. As an introduction to the broad range of memory's "failings", his book is highly readable.

Bourtchouladze takes a straight historical approach to the field of memory research. She traces the story from the early psychological studies of memory span and forgetting curves to modern-day work on the genetic mechanisms behind the brain's rewiring and growth changes. The writing is tidy and gets better as the tale becomes more personal. However, the deeper she descends into the neurochemistry of cAMP cascades, non-NMDA receptors and other molecular mysteries, the further she leaves behind any hope of a broad understanding of the brain as an evolved organ with a job to do. McGaugh's is the weakest of the three, and Memory and Emotion reads more like a textbook. He suggests that he is going to talk about something specific — how emotion modulates the forming of memories — yet gets bogged down in the reductionist details. Moreover, emotion is another of the everyday constructs that mind scientists ought to be challenging rather than simply taking at face value. On this count, McGaugh's contribution is doubly unsatisfying.

John McCrone is the author of four books on the evolution of the human mind including Going Inside: A Tour Round a Single Moment of Consciousness, 1999.

• Read more about this book