by Joshua Foer
Everyone has a great memory for something. We’ve already seen the mnemonic gifts of London cabbies, and the scientific literature is filled with papers about the “superior memories” of waiters, the vast capacities of actors to remember lines, and the memory skills possessed by experts in a wide variety of other fields. Researchers have studied the exceptional memories of doctors, baseball fans, violinists, soccer players, snooker players, ballet dancers, abacus wranglers, crossword puzzlers, and volleyball defenders. Pick any human endeavor in which people excel, and I’ll give you even odds that some psychologist somewhere has written a paper about the exceptional memories possessed by experts in that field.
Why is it that veteran waiters don’t have to write down orders? Why are the best violinists in the world so good at memorizing new musical scores? How come, as one study proved, elite soccer players can glance at a soccer match on TV and reconstruct almost exactly what was happening in the game? One possible explanation might be that people with good memories for dinner orders get channeled into the food-service industry, or that the soccer players with the best memory for arrangements of players have a knack for clawing their way up to the premier league, or that people with a great eye for chicken ass naturally gravitate to the Zen-Nippon Chick Sexing School. But that seems unlikely. It makes much more sense to believe the causality works in the opposite direction. There is something about mastering a specific field that breeds a better memory for the details of that field. But what is that something? And can that something somehow be generalized, so that anyone can acquire it?
The Human Performance Lab, which Ericsson runs with a group of other FSU researchers, is where experts come to have their memories—and much else—tested. Ericsson is probably the world’s leading expert on experts. Indeed, he has achieved a degree of popular fame in recent years thanks to his research showing that experts tend to require at least ten thousand hours of training to achieve their world-class status. When I called him up and told him that I had been thinking about trying to train my own memory, he wanted to know whether I had started yet. I said I hadn’t really begun. He was thrilled; he told me he almost never gets the chance to study a novice in the process of becoming an expert. If I was serious, he said, he wanted to make me his research subject. He invited me down to Florida for a couple days to take a few tests. He wanted to get some baseline measurements of my memory before I started trying to improve it.
The Human Performance Lab occupies a plush office complex on the outskirts of Tallahassee. The bookshelves that line the walls overflow with an eclectic catalog of titles that have been relevant to Ericsson’s research: The Musical Temperament, Surgery of the Foot, How to Be a Star at Work, Secrets of Modern Chess Strategy, Lore of Running, The Specialist Chick Sexer.
David Rodrick, a young research associate in the lab, gleefully described the place as “our toy palace.” When I arrived a couple weeks after my initial phone call with Ericsson, there was a floor-to-ceiling nine-by-fourteen-foot screen set up in the middle of one of the rooms displaying life-size video footage of a traffic stop. It was shot from the perspective of a police officer walking up to a stopped car.
For the previous few weeks, Ericsson and his colleagues had been bringing members of the Tallahassee SWAT team and recent graduates of the police academy into his lab and placing them in front of the big screen with a Beretta handgun loaded with blanks holstered to their belt. They bombarded the officers with one hair-raising scenario after another and watched how they responded. In one scenario, the officer saw a man walk toward the front door of a school with a suspicious bulge that looked like a bomb strapped to his chest. The researchers wanted to know how officers with different levels of experience would react.
The results were striking. Experienced SWAT officers immediately pulled their guns and yelled repeatedly for the suspect to stop. When he didn’t, they almost always shot him before he made it into the school. But recent graduates of the academy were more likely to let the man with the bomb stroll right up the steps and into the building. They simply lacked the experience to diagnose the situation and react properly. At least that would be the superficial explanation. But what exactly does experience mean? What exactly did the more senior officers see that the younger recruits didn’t? What were they doing with their eyes, what was going through their minds, how were they processing the situation differently? What were they pulling from their memories? Like the professional chicken sexers, the senior SWAT officers had a skill that was difficult to put into words. Ericsson’s research program can be summarized as an attempt to isolate the thing we call expertise, so that he can dissect it and identify its cognitive basis.
In order to do that, Ericsson and his colleagues asked the officers to talk aloud about what was going through their minds as the scenario unfolded. What Ericsson expected to learn from these accounts was the same thing he’s found in every other field of expertise that he’s studied: Experts see the world differently. They notice things that nonexperts don’t see. They home in on the information that matters most, and have an almost automatic sense of what to do with it. And most important, experts process the enormous amounts of information flowing through their senses in more sophisticated ways. They can overcome one of the brain’s most fundamental constraints: the magical number seven.
In 1956, a Harvard psychologist named George Miller published what would become a classic paper in the history of memory research. It began with a memorable introduction:
My problem is that I have been persecuted by an integer. For seven years this number has followed me around, has intruded in my most private data, and has assaulted me from the pages of our most public journals. This number assumes a variety of disguises, being sometimes a little larger and sometimes a little smaller than usual, but never changing so much as to be unrecognizable. The persistence with which this number plagues me is far more than a random accident. There is, to quote a famous senator, a design behind it, some pattern governing its appearances. Either there really is something unusual about the number or else I am suffering from delusions of persecution.
In fact, we are all persecuted by the integer Miller was referring to. His paper was titled “The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information.” Miller had discovered that our ability to process information and make decisions in the world is limited by a fundamental constraint: We can only think about roughly seven things at a time.
When a new thought or perception enters our head, it doesn’t immediately get stashed away in long-term memory. Rather, it exists in a temporary limbo, in what’s known as working memory, a collection of brain systems that hold on to whatever is rattling around in our consciousness at the present moment.
Without looking back and rereading it, try to repeat the first three words of this sentence to yourself.
Without looking back
Easy enough.
Now, without looking back, try to repeat the first three words of the sentence before that. If you find that quite a bit harder, it’s because that sentence has already been dropped by your working memory.
Our working memories serve a critical role as a filter between our perception of the world and our long-term memory of it. If every sensation or thought was immediately filed away in the enormous database that is our long-term memory, we’d be drowning, like S and Funes, in irrelevant information. Most of the things that pass through our brain don’t need to be remembered any longer than the moment or two we spend perceiving them and, if necessary, reacting to them. In fact, dividing memory between short-term and long-term stores is such a savvy way of managing information that most computers are built around the same model. They have long-term memories in the form of hard drives as well as a working memory cache in the CPU that stores whatever the processor is computing at the moment.
Like a computer, our ability to operate in the world, is limited by the amount of information we can juggle at one time. Unless we repeat th
ings over and over, they tend to slip from our grasp. Everyone knows our working memory stinks. Miller’s paper explained that it stinks within very specific parameters. Some people can hold as few as five things in their head at any given time, a few people can hold as many as nine, but the “magical number seven” seems to be the universal carrying capacity of our short-term working memory. To make matters worse, those seven things only stick around for a few seconds, and often not at all if we’re distracted. This fundamental limitation, which we all share, is what makes us find the feats of memory gurus so amazing.
My own memory test did not occur in front of the Human Performance Lab’s floor-to-ceiling projection screen. There were no guns holstered to my belt, no eye-tracking devices attached to my head. My humble contribution to human knowledge was extracted in Room 218 of the FSU psychology department, a small windowless office with a stained carpet and old IQ tests strewn across the floor. Ungenerously, it might be described as a storage closet.
The man administering my tests was a third-year PhD student in Ericsson’s lab named Tres Roring. Though his flip-flops and blond surfer mop might not suggest it, Tres grew up in a small town in southern Oklahoma, where his father is an oil man. At age sixteen, he was the Oklahoma State Junior Chess Champion. His full name is Roy Roring III—hence “Tres.”
Tres and I spent three full days in Room 218 taking memory test after memory test—me wearing a clunky microphone headset attached to an old tape recorder, Tres sitting behind me, legs crossed, with a stopwatch in his lap, taking notes.
There were tests of my memory for numbers (forward and backward), tests of my memory for words, tests of my memory for people’s faces, and tests of all sorts of things that seemed unlikely to have anything to do with my memory—like whether I could visualize rotating cubes in my mind’s eye, and whether I knew the definitions of “jocose,” “lissome,” and “querulous.” Another multiple-choice exam called the Multidimensional Aptitude Battery Information Test gauged my Trivial Pursuit skills with questions like:
When did Confucius live?
a. 1650 A.D.
b. 1200 A.D.
c. 500 A.D.
d. 500 B.C.
e. 40 B.C.
and:
In a gasoline engine, the main purpose of the carburetor is to
a. mix gasoline and air
b. keep the battery charged
c. ignite the fuel
d. contain the pistons
e. pump the fuel into the engine
Many of the tests Tres administered were lifted directly from U.S. Memory Championship events, like the fifteen-minute poem, names and faces, random words, speed numbers, and speed cards. He wanted to see how I’d do on them before I’d ever tried to improve my memory. He also wanted to test me on a few of the events that are only used in international memory competitions, like binary digits, historical dates, and spoken numbers. By the end of my three days in Tallahassee, Tres had collected seven hours of audiotaped data for Ericsson and his grad students to analyze later. Lucky them.
And then there were the extensive interviews conducted by another graduate student, Katy Nandagopal. Do you think you have a good natural memory? (Pretty good, but nothing special.) Did you ever play memory games growing up? (Not that I can think of.) Board games? (Only with my grandmother.) Do you enjoy riddles? (Who doesn’t?) Can you solve a Rubik’s cube? (No.) Do you sing? (Only in the shower.) Dance? (Ditto.) Do you work out? (Sore subject.) Do you use workout tapes? (You need to know that?) Do you have electrical wiring expertise? (Really?)
For someone who wants to know what’s being done to him so that he might someday tell other people about it, being the subject of a scientific study can be exceedingly trying.
“Why exactly are we doing this?” I’d ask Tres.
“I’d rather not tell you everything right now.” (If there was something I was going to be tested on later—and as it turned out, there was—he didn’t want me to know.)
“How did I do on that last test?”
“We’ll let you know when this is all done.”
“Can you at least tell me about your hypothesis?”
“Not now.”
“What’s my IQ?”
“I don’t know.”
“High, though?”
The mind-numbing memory exam that SF, the Carnegie Mellon undergraduate, took over and over again for 250 hours for two years is known as the digit span test. It is a standard measure of a person’s working-memory capacity for numbers. Most people who are given the test are like SF when he started: They’re only able to remember seven plus-or-minus two digits. Most people remember those seven plus-or-minus two numbers by repeating them over and over again to themselves in the “phonological loop,” which is just a fancy name for the little voice that we can hear inside our head when we talk to ourselves. The phonological loop acts as an echo, producing a short-term memory buffer that can store sounds just a couple seconds, if we’re not rehearsing them. When he began participating in Chase and Ericsson’s experiment, SF also used his phonological loop to store information. And for a long time his scores on the test didn’t improve. But then something happened. After hours of testing, SF’s scores started inching up. One day he remembered ten digits. The next day it was eleven. The number of digits he could recall kept rising steadily. He had made a discovery: Even if his short-term memory was limited, he’d figured out a way to store information directly in long-term memory. It involved a technique called chunking.
Chunking is a way to decrease the number of items you have to remember by increasing the size of each item. Chunking is the reason that phone numbers are broken into two parts plus an area code and that credit card numbers are split into groups of four. And chunking is extremely relevant to the question of why experts so often have such exceptional memories.
The classic explanation of chunking involves language. If you were asked to memorize the twenty-two letters HEADSHOULDERS-KNEESTOES, and you didn’t notice what they spelled, you’d almost certainly have a tough time with it. But break up those twenty-two letters into four chunks—HEAD, SHOULDERS, KNEES, and TOES—and the task becomes a whole lot easier. And if you happen to know the full nursery rhyme, the line “Head, shoulders, knees, and toes” can effectively be treated like one single chunk. The same can be done with numbers. The twelve-digit numerical string 120741091101 is pretty hard to remember. Break it into four chunks—120, 741, 091, 101—and it becomes a little easier. Turn it into two chunks, 12/07/41 and 09/11/01, and they’re almost impossible to forget. You could even turn those dates into a single chunk of information by remembering it as “the two big surprise attacks on American soil.”
Notice that the process of chunking takes seemingly meaningless information and reinterprets it in light of information that is already stored away somewhere in our long-term memory. If you didn’t know the dates of Pearl Harbor or September 11, you’d never be able to chunk that twelve-digit numerical string. If you spoke Swahili and not English, the nursery rhyme would remain a jumble of letters. In other words, when it comes to chunking—and to our memory more broadly—what we already know determines what we’re able to learn.
Though he’d never been properly taught the technique of chunking, SF figured it out on his own. An avid runner, he began thinking of the strings of random numbers as running times. For example 3,492 was turned into “3 minutes and 49 point 2 seconds, near world-record mile time.” And 4,131 became “4 minutes, 13 point 1 seconds, a mile time.” SF didn’t know anything about the random numbers he had to memorize, but he did know about running. He discovered that he could take meaningless bits of information, run them through a filter that applied meaning to them, and make that information much stickier. He had taken his past experiences and used them to shape how he perceived the present. He was using associations in his long-term memory to see the numbers differently.
This, of course, is what all experts do: They use their memories to see the world differently. Ove
r many years, they build up a bank of experience that shapes how they perceive new information. The experienced SWAT officer doesn’t just see a man walking up the front steps of the school; he sees a nervous twitch in the man’s arm that calls up associations with dozens of other nervous twitches he’s seen in his years of policing. He sees the suspect in the context of every other suspicious person he’s ever come across. He perceives the current encounter in light of past encounters like it.
When a graduate of the Zen-Nippon Chick Sexing School looks at a chick’s bottom, finely honed perceptual skills allow the sexer to quickly and automatically gather up a stock of information embedded in the chick’s anatomy, and before a conscious thought can even enter his or her head, the sexer knows whether the chick is a boy or a girl. But as with the senior SWAT officer, that seemingly automatic knowledge is hard earned. It is said that a student of sexing must work through at least 250,000 chicks before attaining any degree of proficiency. Even if the sexer calls it “intuition,” it’s been shaped by years of experience. It is the vast memory bank of chick bottoms that allows him or her to recognize patterns in the vents glanced at so quickly. In most cases, the skill is not the result of conscious reasoning, but pattern recognition. It is a feat of perception and memory, not analysis.