THREE
I Forgot to Remember to Forget You
EVERYONE HAS A UNIQUE LIFE STORY, AND EVERYONE’S story is wrong. Our memories of events are notoriously unreliable. Storing and retrieving autobiographical memory is not like writing in a book that is later opened to reveal perfect sentences. It’s not even like taking a photograph and leaving it out in the sun to fade, with the fine details gradually obscured by time. Rather, our memories of events fail in particular, reproducible ways, even in those of us who pride ourselves on having good memories.
Put simply, memories are not objective recordings of events; they are the unreliable traces of our individual experiences of events. Two people, standing side by side, will have different experiences of the same event based on their prior life history. If I had a traumatic experience with fire in the past, then my experience—and hence my later memory—of seeing a house fire will be different than yours, even as we watch the fire engine roll up to the scene together. And the memory for an event can continue to change long after it occurs. After memories are stored in the brain, they can be altered, both by subsequent experience and the mere act of recollection. While memories formed by a particular life course are central to human individuality, their malleability makes it clear that our most strongly held beliefs about ourselves are continually and messily constructed and reconstructed.
ON THE MORNING OF April 19, 1995, Timothy McVeigh parked a truck containing a huge bomb at a drop-off zone in front of the Alfred P. Murrah Federal Building in Oklahoma City. He lit a slow-burning fuse and walked to his escape car, parked a few blocks away. The fuse ignited a horrifically effective bomb, which had been constructed by McVeigh and his friend Terry Nichols. The bomb was made from stolen commercial blasting explosives, ammonium nitrate fertilizer, racing fuel, diesel fuel, and acetylene cylinders, all packed into barrels. The enormous blast collapsed the front of the building, and blew out window glass and damaged adjacent structures within a sixteen-block radius. One hundred and sixty-eight people were killed, including nineteen children, most at an on-site day-care center for federal employees that was situated directly above the bomb.
The FBI sprang into action and, within a few hours, one of the truck’s axles, bearing a legible vehicle identification number, was found among the wreckage. This quickly led the feds to Elliott’s Body Shop, a Ryder truck rental location in nearby Junction City, Kansas. They called the body shop to say that they were sending an agent over to interview the workers who had seen the rental transaction: owner Eldon Elliott, mechanic Tom Kessinger, and bookkeeper Vicki Beemer. All of them recalled that the man who had rented the truck two days earlier had given the name Robert Kling, but only Kessinger remembered another man with him. An FBI artist raced to the scene and, working with Kessinger, produced sketches of the two men (figure 6). Robert Kling was called John Doe #1 and his accomplice was John Doe #2.
FBI agents went door-to-door in Junction City, showing the sketches and searching for leads. They got a hit at the Dreamland Motel, where the owner identified John Doe #1 as a man who had checked in on April 15, stayed through April 18, and kept a Ryder truck parked outside his room. The owner recalled that, after briefly sputtering (presumably having forgotten his alias of Robert Kling), the man had given his name as Timothy McVeigh. When the agents typed McVeigh’s name into police computers, they couldn’t believe their luck. McVeigh was, at that very moment, held in jail in a small town about a two-hour drive north of Oklahoma City. His getaway car had no rear license plate, and when he was pulled over for this infraction, the local patrolman spotted a concealed pistol and arrested him. The address on McVeigh’s forged driver’s license was a farm in Michigan owned by Terry Nichols. A few hours later, the farm was raided and Nichols was arrested. Bomb-making supplies and a hand-drawn map of Oklahoma City were found, with the location of the Murrah Federal Building and McVeigh’s getaway car marked in red ink.
This was fine investigative work, but there was one remaining problem. While John Doe #1 was clearly Timothy McVeigh, John Doe #2 didn’t look anything like Terry Nichols. When Attorney General Janet Reno announced the arrests of McVeigh and Nichols on TV, she emphasized that “John Doe #2 remains at large and he should be considered armed and dangerous.”1
FIGURE 6. The eyewitness sketches of the presumed Oklahoma City bombers that were distributed by the FBI. John Doe #1 (Timothy McVeigh) is on the left and John Doe #2 is on the right. Used with permission of the FBI.
The eyewitness sketch of the Oklahoma City bombing suspects may be the most famous one in the history of American criminal investigation. It was in every newspaper and magazine and in constant rotation on TV news. John Doe #2 was said to have a tattoo of a snake on his left bicep and wear a baseball cap with blue and white markings. A $2 million price was put on his head. Leads poured in to the hotline established by the FBI, and over ten thousand agents and others were assigned to follow them up. By various accounts, John Doe #2 had been seen with Timothy McVeigh in an Oklahoma City strip club or running out of the Ryder truck just before it exploded or buying fertilizer, presumably to construct the bomb. None of these stories could be verified. Fourteen men who resembled the sketch of John Doe #2 were taken into custody, but all were released with solid alibis. After many weeks, it became clear that the most intensive manhunt in the history of the FBI had failed.
It is almost certain that the search failed because there was never anyone with Timothy McVeigh when he rented the Ryder truck. Later, it was discovered that the day after McVeigh’s visit, two men came in to Elliott’s Body Shop to rent a truck. They were US Army Sergeant Michael Hertig, who, like McVeigh, was blond and fair skinned, and his friend, Private Todd Bunting, who was dark haired, muscular, and a dead ringer for John Doe #2. Tom Kessinger, the mechanic, had, despite the best of intentions, committed an error of memory. He correctly described the features of the innocent Todd Bunting but attributed his presence to the McVeigh episode that had occurred on the previous day. This merging of separate incidents is one of the typical ways in which memory fails us in daily life.
IT HAPPENS EVERY HOUR in police stations around the world. A crime has been committed, and a suspect is paraded in front of an eyewitness in a lineup with several others, often of the same general appearance. In this situation, if the real perpetrator is not among those in the lineup, some eyewitnesses, mostly acting without malice, will pick the person who most closely matches their memory of the perpetrator. The accuracy of eyewitness identification is not improved when a “six-pack” of mug shots is used in place of a live lineup. These lineups have been the cause of many wrongful convictions over the years.
It’s hard to know the true prevalence of convictions based on mistaken eyewitness testimony, because most mistakes remain unrevealed.2 We can make an estimate, however, based on simulated lineups in the laboratory. Experimental subjects were shown a video of a crime, in which the face of the “criminal” is clearly visible, and then presented with a lineup of six suspects, none of whom appeared in the video. In this situation, about 40 percent of the subjects nonetheless picked someone out of the lineup. Usually, but not always, this was the person who was the closest physical match to the criminal. If the subjects were told that others have already identified a particular suspect in the lineup and they just need to confirm or deny it, the rate of false recollection rises to 70 percent. Furthermore, when the subjects who picked a suspect were asked how confident they were in their identification, most said that they were absolutely certain.3
OUR AUTOBIOGRAPHICAL MEMORIES ARE subject to all kinds of distortions—what psychologist Daniel Schacter cheekily calls “sins of commission.”4 In addition to misattributions of time, as in the case of John Doe #2, and the suggestibility of eyewitnesses, there’s bias, which is the warping of one’s recollection to mold to present beliefs, knowledge, and feelings. For example, after a bad breakup, people’s recollections of the early stages of the relationship, previously recalled with pleasure, often turn darker. Or people will say, “I always knew that candidate X would win the election,” even when they had voiced doubts about that outcome beforehand.
Some of the ways autobiographical memory fails are well known. Generally, our memories of recent events are more accurate and detailed than our memories of the distant past. But there are other, less obvious changes. If I ask you to recall a recent event, you are most likely to imagine it from your own point of view, with the camera, as it were, in your own eyes. This is called field memory. But if I ask you to recall a memory from your childhood, there is a much greater probability that your point of view will shift to that of an observer; you will see yourself in the scene rather than seeing the event though your own eyes. Furthermore, if asked to recall the emotional tone of a past event, you are more likely to evoke a field memory, while if asked to recall facts of an event, you are more likely to call forth an observer memory. The key point here is that the way we recall the memory is not set in stone. It can be strongly influenced by the task at hand.5
Another time-related phenomenon is that repetition of experience renders memories generic. If you’ve only been to the beach once, then you are likely to remember many details of that experience. But if you’ve been over fifty times, you’re unlikely to remember details of visit number thirty-seven, unless something emotionally affecting occurred. Perhaps visit number thirty-seven was the day a dead whale washed up on the beach, or the day you met your future spouse. Then the details of that day would likely be written into your memory deeply and retained with greater detail and fidelity. Emotions, both positive and negative, are the currency of autobiographical memory. Emotions cause the brain to store memory in a stronger and more permanent fashion, set down in bold type and italics. This reinforcement of emotional memories is mostly good and sometimes bad. It’s good because emotionally salient events are often the ones you most need to remember later in life. However, in some cases, memory can become pathologically persistent, as when the memory of a traumatic experience—like an assault or a soldier’s time in combat—is recollected incessantly.
IF OUR MEMORIES FOR events are often so inaccurate and changeable, then why do we even have them? What is memory for? The main answer is that memory allows us to learn: to adjust our behavior based on individual experience and therefore efficiently find food, avoid predators, find and attract mates, and so on. In other words, memory does for the individual what evolution of the genome does for the species over many generations: it allows us to respond to the environment in a way that increases the chance of surviving and passing genes on to the next generation. That is endlessly useful. For example, a newborn mouse has an inborn fear of foxes, even if it is the descendant of many generations of lab mice with no exposure to foxes at all. This is a useful adaptation for mice in the wild, but it is not a good general strategy for dealing with a changing world. It is not possible to encode all useful behavioral responses into the genome in order to have a newborn equipped to deal with every eventuality. It’s both more efficient and flexible to have animals remember and learn, even if they don’t do so perfectly. And there is another benefit. The act of recollection allows us to mentally time travel to a past event, and this allows us to imagine a future as well as a past. Memory releases our mental life from the tyranny of the present moment. And imagining a future allows us to make predictions, which is a requirement for decision-making.
Another answer to the question of what memory is for is that the particular failures of autobiographical memory are actually features rather than bugs. For memory to be useful, it must be updated and integrated with subsequent experience, even if it alters the memory of the original event. In that way, it’s helpful for recollection to render the memory of an event malleable, so that it may be integrated with the present. In most situations, a generic memory compiled from many trips to the beach is more useful in guiding future decisions and behavior than fifty stand-alone, detailed, and accurate beach trip memories. The repetition-driven loss of detail allows for the efficient use of the brain’s limited memory resources.
In other words, it’s not surprising that our memories for events are often inaccurate, because the particular way in which memories are compromised is often useful. What’s surprising is that we mostly fail to recognize this in our daily lives. We humans all have an inborn tendency to create a plausible story out of memory fragments. Because of this ongoing narrative construction, we are often confident about the veracity of blurred memories and allow them to form the basis of our core beliefs about ourselves.
IN ADDITION TO MEMORIES of events, we also store memories of facts or concepts that are not linked to any particular event. For example, I can tell you that the capital of Mongolia is Ulaanbaatar, even though I cannot recall where or when I learned that fact. Or, I might have the fact right but get the source and time wrong—recalling that I learned it in high school, forty years ago, when I really read it on Wikipedia just last year. Likewise, I might be able to explain the concept of transitivity in mathematics without remembering when or how I learned it.
Psychologists call this decoupling of facts or concepts from events “source amnesia,” and everyone has it to some degree, although it becomes more prevalent with normal aging.6 On average, our memories for both facts and concepts, while not perfect, are usually more accurate than our memories for events. This may reflect that they are less rich in detail and context than events, which involve all of the senses. Facts and concepts are, in a way, already distillations of raw experience.
Bringing these concepts down to the level of individuals, we may say that “Fred has a bad memory” or “Sally has a good memory.” But, of course, memory is not a single phenomenon. Not only are some people good at remembering facts and concepts but poor at events, but there is considerable variation in people’s ability to remember particular types of facts and concepts. We all know people who have an impressive recall for comedy routines but struggle to commit music to memory. Others might be poor at putting names to faces but great at remembering what they’ve read. And among those people with a good memory for written material, there are different strategies involved. Some may recall a visual image of the page with the text laid out, while others recall the sounds and meanings of the words without a corresponding image of the page.
Memories for events, facts, and concepts all fall into a category called explicit memory: specific information that can be brought to mind with conscious mental effort. Explicit memory is what we usually mean when we talk about memory in everyday conversation. However, there’s another type of memory that is equally important but less discussed. That is implicit memory: memory that is acquired and used subconsciously, without mental effort. Implicit memories are mostly acquired through repeated practice rather than from a single event.7 Generally speaking, implicit memories are more stable than explicit ones and are stored in different brain circuits. That is why you may have to hunt around the house to find where you left your wallet (an explicit memory), but you’re unlikely to forget how to ride a bike (an implicit memory). Our individuality is formed as much by our subconscious implicit memory as it is by the explicit memory for facts, events, and concepts that gets so much attention.
Figure 7 summarizes many years of work attempting to categorize types of long-term memory. The evidence for making these distinctions often comes from the analysis of patients who have sustained brain damage. For example, people with injuries to the medial temporal lobe can have a profound anterograde amnesia—an inability to form new memories, starting when the injury was sustained. They also display some degree of retrograde amnesia, which is the erasure of memory for a period of months or years before the injury, leaving older memories intact.8 Originally, it was thought that medial temporal lobe amnesiacs had no ability to form new memories at all. But over the years, it has been revealed that the ability to form new implicit memories remains intact.
FIGURE 7. A taxonomy of human long-term memory. Explicit memory includes memories of events (autobiographical memory) or facts and concepts (semantic memory). Implicit memory is acquired and used without conscious attention, yet can guide decisions and behaviors. It includes memory for skills and habits (procedural memory), as well as simple associative learning (like eyelid conditioning) or nonassociative learning (like habituation of the orienting response).
Reading text reflected in a mirror is a task that is difficult at first but can be gradually improved with practice. If someone with medial temporal lobe damage practices mirror reading for thirty minutes a day for three days and is then tested on mirror reading on the fourth day, their speed will have improved. But if you ask them if they have ever tried mirror reading before, they will say no. They will have no memory of the task, the room, or the person who helped them.9 Recalling the event of training for mirror reading is a form of explicit memory that requires intact medial temporal lobe circuits. But improvement in mirror reading is a skill, and hence a form of implicit memory that can be stored and recalled even when these circuits are damaged. While mirror reading is a cognitive skill, motor skills, like improving one’s tennis swing with practice, are also preserved in patients with temporal lobe amnesia.10
Another type of implicit memory is eyelid conditioning, a type of associative learning. If I play a soft tone, you will not blink your eyelid in response to it. But if I direct a brief puff of air to your cornea, you will blink reflexively. The blinking is not a conscious decision; it will happen whether you intend to blink or not. Then, if I pair the tone with the air puff—such that the start of the tone precedes the start of the air puff, and they end together—and I repeat this pairing many times, you will gradually learn that the tone will predict the air puff. The result is that you will begin to blink your eye earlier, so that your eyelid is at least partially closed and your cornea is protected when you expect the air puff to arrive. Again, eyelid conditioning is subconscious. You cannot override it or acquire it faster with force of will. It will happen no matter what you do.
FIGURE 8. Mirror reading is a skill that can be improved with regular practice. Even people with brain damage who have profound anterograde amnesia for facts, events, and concepts can still improve in a mirror reading task. Here, the two-axis reversed text is a famous line from Kurt Vonnegut’s 1969 novel Slaughterhouse-Five.
An even simpler form of subconscious learning is habituation, a type of nonassociative learning. If I were to stand just outside your field of view and drop a book on the floor, you would rapidly swivel your head to investigate. This behavior is called the orienting reflex, and it was first described by the Russian physiologist Ivan Sechenov in 1863.11 The orienting reflex is a response to novelty, and so if I were to continue to drop the book, say, once per minute, you would soon learn to ignore it—to habituate—and the orienting reflex would be suppressed. Habituation is specific to a particular stimulus. A bright light will also elicit an orienting response. But repeated book drops will not diminish your response to the first bright light or vice versa.12
Interestingly, once you have habituated to some stimulus that’s regularly repeated, then its absence becomes novel and can elicit an orienting reflex. To illustrate this phenomenon, the neurophysiologist Karl Pribram told the story of the Third Avenue elevated railway line, which once ran along the Bowery in New York City and was famously loud.13 The train ran on a regular schedule through the night, and the residents of the apartments along the Bowery had become habituated to the intermittent noise. When train service on the line was stopped, in 1954, the police started receiving calls from local people who had been awakened out of a sound sleep by some occurrence they could not entirely define but which they assumed must be a prowler. An astute detective, who found no unusual evidence of creeping miscreants, soon realized that these calls were clustered at the times when the elevated train used to pass by. The strange occurrences that woke people were the orienting responses evoked by roaring silence when noise was expected. Once people habituated to the absence of elevated train noise, their panicked calls to the police subsided.
WE LIKE TO IMAGINE that we are fundamentally creatures of free will. We reliably call certain facts, events, and concepts to mind. We make conscious decisions and act volitionally. Our individuality is inextricably bound up with a deep sense of agency and autonomy. To a large degree, this is a trick our brain plays on us. Most of our behavior is subconscious and automatic. In the words of neuroscientist Adrian Haith: “Almost everything you do is a habit.”14 A habit is not just a behavioral routine that is formed and then performed at a subconscious level, it must become divorced from an ultimate goal. Your goal might be to stop off at the Thai restaurant for takeout after work, but instead you habitually drive straight home. The same habit can be either beneficial or detrimental, depending on the context. You may type rapidly and automatically on a standard QWERTY keyboard, but this habit will fail you if only a Dvorak keyboard, with a different layout of letters, is available.
Generally speaking, when you are learning a new task, your behavior is flexible and goal directed at the outset, but becomes automatic and habitual with repeated practice. For example, when you first learn to drive a car, you must think carefully about every action: steering, braking, signaling, scanning the road ahead. But with time, these actions become mostly automatic. Driving has become a habit and no longer requires your full mental attention.
While habits have the limitation of being inflexible, they have the advantage of being easy. The sad truth is that much of life is predictable and boring, so the inflexibility of habits is rarely a problem. Crucially, when behaviors become habitual, the conscious mind is free to ponder, predict, and plan. All of us have a collection of learned behaviors acquired over a lifetime of practice. We master one task over time, render it habitual, and then move on to the next one. In this way, we each assemble a vast library of habits and skills that can be called up automatically. As Haith writes, “Atop this massive conglomeration of habits sits a thin sliver of cognitive deliberation that steers only the highest-level decisions that we need to make.” Without habits, our brain would be instantly overwhelmed with a multitude of tiny decisions better left to rapid, automatic processes.
These examples of subconscious learning that we’ve explored—including the learned skill of mirror reading, habituation of the orienting reflex, and associative eyelid conditioning—all rely on implicit memory and so are spared in patients with medial temporal lobe amnesia.15 Indeed, implicit memory mostly depends on circuits involving other brain regions.16
ALL MEMORIES, EITHER EXPLICIT or implicit, must be stored in the brain. The very short-term memory that you need to keep a phone number in mind as you dial is encoded by reverberating electrical activity that passes back and forth between three brain regions: the thalamus, the frontal cortex, and the cerebellum.17 This working memory is also what you need to keep the beginning of a long sentence in mind as you read to the end. It is easily disrupted by competing mental activity (like someone speaking to you while you dial or read) and is discarded almost immediately after it is used.
Longer-term memories require more enduring alterations. Patterns of electrical activity associated with particular experiences must give rise to changes in the interconnected networks of neurons that make up the brain. Signaling in the brain has a mixed electrical and chemical character. Neurons convey information though rapid, all-or-none electrical signals called spikes. A spike travels down the long, thin, information-sending fiber of a neuron called the axon. When the spike invades specialized active zones in the axon, it triggers the release of chemical neurotransmitter molecules. These diffuse across a tiny saltwater-filled gap and activate receptors on the information-receiving part of the next neuron in the signaling chain, called the dendrite, sometimes producing an electrical response in the next neuron in the network. These locations where neurotransmitters are released by one neuron and then received by another are called synapses.18
Let’s play God for a moment. If you are the Great Engineer and you want to build memory storage in the brain, there are two main options. First, you could have experience-driven patterns of electrical activity persistently change the strength of chemical transmission across synapses. This could take the form of making synapses stronger (or growing new ones) or making synapses weaker (or eliminating existing ones). Together, these changes are called synaptic plasticity. Or, you could have experience alter the electrical signaling properties of whole neurons. For example, you could alter neurons to make making them more or less likely to fire spikes, or to fire spikes in different temporal patterns. These experience-driven processes are called intrinsic plasticity. It turns out that both intrinsic and synaptic plasticity are involved in storing long-term memories, although much more attention has been paid to the latter. Because each neuron in the brain receives an average of five thousand synapses, the information storage capacity of synaptic plasticity in much greater than that of intrinsic plasticity. Intrinsic and synaptic plasticity interact in complex and useful ways to store memories.19
Just as important is what’s not changed to store memory. Experience does not modify the sequence of DNA in the cells of the brain, so this process cannot be the substrate of memory. Rather, memory is yet another example, albeit a specialized one, of how experience changes gene expression to produce lasting changes.20 It’s not unlike the example we discussed in chapter 2 about how the ambient temperature in the first year of life determines the degree of sweat-gland innervation. Only in this case, the tissue being altered by experience is not the peripheral nerves or the skin but the brain, and the changes in gene expression give rise to synaptic and intrinsic plasticity, the stuff of memory.
Much of the biology underlying the recollection of memory for facts and events remains poorly understood, but there are some general features we understand. Recollection of memory typically involves electrical activity in at least some of the neurons and synapses that were active during the original experience. However, the story is more complicated than that, as the neural circuits and brain regions involved in storing memories can sometimes shift over time. As mentioned earlier, people who sustain damage to a brain region called the medial temporal lobe will typically lose their memories for facts and events from a period of months or years prior to their injury, also known as retrograde amnesia. Older memories for facts and events remain intact, however, suggesting that they have been transferred from the medial temporal lobe to other brain regions.
In emotional situations, certain neurotransmitters (like dopamine and norepinephrine) and hormones (like epinephrine and corticosterone) are released. Some are released and act within the brain, while others are released in the body and make their way to the brain. These emotion-associated chemical signals can increase the extent of synaptic and intrinsic plasticity driven by experience, thereby strengthening memory. Importantly, this strengthening doesn’t just happen at the initial time the memory is laid down. If the act of recalling a memory evokes emotional responses, then this chemical process can further strengthen (and warp) the memory every time it comes to mind.
IS MEMORY STORAGE IN the brain an unlimited resource, or can we run out of space? Can training in one particular skill or task crowd out our ability to excel at another, or is there room for limitless self-improvement? Disappointingly, there are some reasons to believe that memory resources are limited.
To become a licensed taxi driver in London, one must accomplish a prodigious feat of learning. The exam requires comprehensive memorization of the city’s twenty-five thousand streets, as well as hotels, restaurants, landmarks, and the optimal routes between them, a body of information called “the Knowledge.” Even after several years of study, many drivers do not pass the exam and must try again or give up. There was great excitement in the field when a careful study by Eleanor Maguire and her coworkers revealed that, on average, compared to age- and education-matched controls, licensed London taxi drivers have an enlarged brain region called the posterior hippocampus.21 This region is thought to have a special role in the processing of spatial information. This result could mean that the intensive training performed in order to pass the exam caused the volume of the posterior hippocampus to increase as it came to contain a detailed mental map of the city. Alternatively, it could mean that those blessed with a large posterior hippocampus prior to training were more skilled at spatial cognition and hence more likely to successfully acquire the Knowledge and pass the exam.
A more recent study followed prospective London taxi drivers with repeated brain scans before and after their extensive bout of memorization. It showed that drivers who studied and passed the test significantly enlarged their posterior hippocampus, while those who failed or dropped out did not, and neither did the control subjects of similar age.22 Thus, acquisition of the Knowledge appears to cause enlargement of the posterior hippocampus.
As the posterior hippocampus grows with spatial learning, this expansion comes at the expense of an adjacent brain region, the anterior hippocampus, a structure that is not involved in spatial cognition but rather in formation of new visual, but nonspatial, memories. This is likely to explain why, on average, London taxi drivers perform somewhat worse on visual memory tests than matched control subjects or drivers who failed to pass the exam. This finding indicates that at least some mnemonic and cognitive resources are limited in the brain and can be dynamically assigned to the tasks at hand by extensive training. Interestingly, retired London taxi drivers slowly return to the control state, with smaller posterior hippocampi, larger anterior hippocampi, improved visual memory scores, and a fading recollection of London’s twisting streets.23
There are two reasons why London taxi drivers are considered an unusually good population for examining training-evoked changes in the brain. First, acquiring the Knowledge is a difficult task, but one that doesn’t require unusually high intelligence. The average intelligence for London taxi drivers is about the same as that of the general population in the United Kingdom. Second, unlike musical or sports training, which often begins in childhood and hence conflates brain development with learning, learning the Knowledge only begins in adulthood, after the taxi drivers’ brains are mature.
The London taxi driver result prompts the question: Is the enlargement of a brain region (and the related shrinkage of a neighboring brain region) a general property of intensive training during adulthood, or is there something special about the taxi exam? After two years of study, medical students in Germany must take a grueling comprehensive exam, the Physikum, which tests knowledge in chemistry, physics, anatomy, and biology. Students are allotted three months of daily study sessions to prepare. Arne May and his coworkers scanned the brains of medical students and matched control subjects before the three-month-long period of studying, a day after the exam, and then again three months after the exam. Over the learning period, the volume of three brain regions increased in the medical students compared to the control subjects: the posterior parietal cortex, the lateral parietal cortex, and our old friend the posterior hippocampus. The increases in these three brain regions were sustained when measured three months later. As with the London taxi drivers, there were also reductions in an adjacent structure, the occipital parietal lobe,24 suggesting that competition for brain space following intensive training in adulthood is a general principle. Unfortunately, we don’t know whether there were particular cognitive impairments that correlated with these shrunken brain regions in the medical students.
It’s likely that the growth and shrinkage of brain regions produced by studying for exams is not as long-lasting as that produced by acquiring and then using the Knowledge. This was suggested by a study by Eleanor Maguire and her team that focused not on medical students but on working doctors, a group that must acquire and deploy a great deal of knowledge during years of extensive training. Unlike taxi drivers, doctors score better than average on intelligence tests. As a result, they were compared with a population that had the same intelligence test scores, but didn’t have university attendance or other intensive training (like trade school). In this comparison, neither the posterior hippocampus nor any other brain region was found to be enlarged in doctors compared to the control group.25 This result suggests that acquiring a lot of information over years of training is not sufficient to produce lasting changes in the gross structure of brain regions.
One clever way to search for rapid, training-induced brain changes in adults is to perform scans of their brains before and after teaching them to juggle. Researchers divided an age- and sex-matched pool of volunteers into two groups, jugglers and non-jugglers, and gave them all brain scans before training. The group of jugglers were given three months to learn a classic three-ball cascade juggling routine and sustain it for one minute. When the juggling group was tested soon after having mastered that skill, they showed expansion of two areas in the brain—the mid-temporal cortex (on both sides) and the posterior intraparietal sulcus (on the left side only)—when compared to the control non-juggling group. The expansion of these particular regions makes sense, because the former is involved in tracking the speed and direction of moving objects while the latter is involved in attention and sensory-motor coordination. After a break from practicing for several months, most people in the juggler group could no longer juggle on their first attempt and the enlargement of the brain areas was partially reversed.26 This reversal is probably similar to what’s happening with the doctors. They study hard briefly to pass their exams, but most would be hard-pressed to pass again in mid-career without brushing up. The expansion and shrinkage of brain regions with memorization—whether an implicit sensory-motor memory like juggling, or the explicit memory of the London street map or the Physikum exam—seems to last only as long as that knowledge is actively used.
IN ORDER FOR A brain region to grow larger, there must be a significant addition of cellular material. Given what we know about synaptic plasticity, a large part of this memory-associated expansion of brain regions must involve enlargement of existing synapses, as well as growth of new synapses and the dendrites and axons on which they are formed. It is also possible that entirely new cells are added to brain regions. Glial cells are continually produced by cell division in the brain and so could contribute to regional growth.27 In some very limited parts of the brain, like the hippocampal dentate gyrus, which is part of a brain circuit for memory of facts and events, new neurons are formed after birth. But while it is clear that new neurons are created in the brains of adult birds and rodents, there is a robust debate about whether the adult human brain can form new neurons, or if that is limited to early life.28
It’s important to highlight that seeing a brain region expand with memory storage is an extreme case produced only by sustained training. In most situations, memories are stored without overt changes in the size of brain regions. You can imagine that if experience adds or strengthens some synapses while removing or weakening others in a brain region, there will be no overall change in the volume of that region, even as memories are stored by changes in the function of its neural circuitry. This type of change can lead to functional plasticity, even in the absence of detectable changes in the size of brain regions.
A good example of functional plasticity comes from the study of serious musicians. When players of string instruments, like the cello or guitar, are placed in a brain scanner and compared to controls, it is revealed that more of their brain is devoted to touch sensation from the left hand but the brain space dedicated to the right hand is unaltered. Importantly, the structure of bulges and grooves on the brain is identical between musicians and controls. There’s no overall growth or shrinkage of brain regions. Rather, one particular area, called the primary somatosensory cortex, has more of its volume dedicated to the left hand at the expense of processing touch sensations from other parts of the body.29 This occurs because the left hand performs the highly dexterous fingering motions, which require finer touch sensation and motor control than the strumming or bowing performed by the right hand. Presumably there are some deficits in touch sensation on the left side of the body of string players, produced by expansion of the hand area and the consequent shrinkage of brain space dedicated to the rest of the body surface, but these have yet to be investigated.
ALMOST ALL OF OUR instincts about memory are wrong. We feel like creatures of free will, with detailed and unlimited recall of those events that have helped to form us as individuals. In reality, most of our behavior is composed of learned, subconscious habits and skills with only a thin veneer of decision-making at the surface. Our recollection of specific events is unreliable and subject to further distortion every time we recall them. Our memories of facts and concepts are only marginally better. When asked about how confident we are about the veracity of a particular memory, our estimation bears no relationship to the truth. We feel like we can learn more and more with no limit, yet intensive training in one type of memory seems to degrade our ability to store some other forms of memory.
Our memories are suboptimal, and yet we hold them close. They feel true and important. They are central to our sense of individuality and agency. The mismatch between how we revere our memories and how often they fail is striking. Why we should feel more agency than we really have is an interesting and open question. I tend to see it as more of a feature than a bug. When we feel that we are in charge of our behavior, that we are making decisions based on accurate recollection, this allows for more rapid decision-making in those cases of the “thin sliver of cognitive deliberation,” where it is truly required. In other words, when we don’t have to stop and second-guess whether we are in charge, we can be decisive when it really matters.