by Joshua Foer
Photographs, videos, and digital recordings are, like books, prosthetics for our memories—chapters in the long journey that began when the Egyptian god Theuth came to King Thamus and offered him the gift of writing as a “recipe for both memory and wisdom.” Lifelogging is the logical next step. Maybe even the logical final step, a kind of reductio ad absurdum of a cultural transformation that has been slowly unfolding for millennia.
I wanted to meet Bell and see his external memory at work. His project would seem to offer the ultimate counterargument to all the effort I was investing in training my internal memory. If we’re bound to have computers that never forget, why bother having brains that remember?
When I visited his immaculate Microsoft Research office overlooking the San Francisco Bay, Bell wanted to show me how he uses his external memory to help find things that have gone missing in his internal memory. Because memories are associative, finding the odd misplaced fact is often an act of triangulation. “The other day, I was trying to find a house I had looked at online,” Bell told me, leaning back in his chair. “All I could remember about it is that I was talking to the realtor on the phone at the time.” He pulled up a time line of his life on his computer, found the phone conversation on it, and then immediately pulled up all the Web sites he was looking at while he was on the phone. “I call them information barbs,” says Bell. “All you need is to remember a hook.” The more barbs there are stored in one’s digital memory, the easier it is to find what you’re looking for.
Bell has a wealth of external memories at his fingertips. By far the biggest problem Bell faces is how to avoid the fate of Funes and S and keep from drowning in a sea of meaningless trivia. So much of remembering happens at the moment of encoding, because we only tend to remember what we pay attention to. But Bell’s lifelog pays attention to everything. “Don’t ever filter, and never throw anything away” is his motto.
“Do you ever feel burdened by the sheer volume of memories you’re collecting?” I asked him.
He scoffed at the notion. “No way. I feel this is tremendously freeing.”
The SenseCam is not a beautiful machine. It’s a black box, about the size of a pack of cigarettes, that dangles around Bell’s neck. Inconspicuous it’s not. But then again, the first computers took up entire rooms and the earliest cell phones were the size of cinder blocks. It doesn’t take much imagination to see how future versions of the SenseCam could be embedded in a pair of eyeglasses, or inconspicuously sewn into clothing, or even somehow tucked under the surface of the skin or embedded in a retina.
For now, Bell’s internal and external memories don’t mesh seamlessly. In order for him to access one of his stored external memories, he still has to find it on his computer and “re-input” it into his brain through his eyes and ears. His lifelog may be an extension of him, but it’s not yet a part of him. But is it so far-fetched to believe that at some point in the not-too-distant future the chasm between what Bell’s computer knows and what his mind knows may disappear entirely? Eventually, our brains may be connected directly and seamlessly to our lifelogs, so that our external memories will function and feel as if they are entirely internal. And of course, they will also be connected to the greatest external memory repository of all, the Internet. A surrogate memory that recalls everything and can be accessed as naturally as the memories stored in our neurons: It would be the decisive weapon in the war against forgetting.
This may sound like science fiction, but already cochlear implants can convert sound waves directly into electrical impulses and channel them into the brain stem, allowing previously deaf people to hear. In fact, they’ve already been installed in more than 200,000 human heads. And primitive cognitive implants that create a direct interface between the brain and computers have already allowed paraplegics and patients with ALS (Lou Gehrig’s disease) to control a computer cursor, a prosthetic limb, even a digital voice simply through the force of thought. These neuroprosthetics, which are still highly experimental and have been implanted in only a handful of patients, essentially wiretap the brain, and allow direct communication between man and machine. The next step is a brain-computer interface that lets the mind exchange data directly with a digital memory bank, a project that a few cutting-edge researchers are already working on, and which is bound to become a major area of research in the decades ahead.
You don’t have to be a reactionary, a fundamentalist, or a Luddite to wonder whether plugging brains into computers and seamlessly merging internal and external memory would ultimately be such a terrific idea. Today bioethicists work up sweats over such hot-potato topics as genetic engineering and neurotropic “cognitive steroids,” but these kinds of enhancements are just tweaking the dials compared with what it would mean to fully marry our internal and external memories. A smarter, taller, stronger, disease-resistant person who lives to 150 is still, in the end, just a person. But if we could give someone a perfect memory and a mind that taps directly into the entire collective knowledge of humanity, well, that’s when we might need to consider expanding our vocabulary.
But perhaps instead of thinking of these memories as externalized or off-loaded—as categorically different from memories that reside in the brain—we should view them as extensions of our internal memories. After all, even internal memories are accessible only by degrees. There are events and facts I know I know, but I don’t know how to find. Even if I can’t immediately recall where I celebrated my seventh birthday or the name of my second cousin’s wife, those facts are nevertheless lurking somewhere in my brain, waiting for the right cue to pop back into consciousness, in just the same way that all the facts in Wikipedia are lurking just a mouse click away.
We Westerners tend to think of the “self,” the elusive essence of who we are, as if it were some starkly delimited entity. Even if modern cognitive neuroscience has rejected the old Cartesian idea of a homuncular soul that resides in the pineal gland and controls the human body, most of us still believe there is a distinct “me” somewhere up there driving the bus. In fact, what we think of as “me” is almost certainly something far more diffuse and hazier than it’s comfortable to contemplate. At the least, most people assume that their self could not possibly extend beyond the boundaries of their epidermis into books, computers, a lifelog. But why should that be the case? Our memories, the essence of our selfhood, are actually bound up in a whole lot more than the neurons in our brain. At least as far back as Socrates’s diatribe against writing, our memories have always extended beyond our brains and into other storage containers. Bell’s lifelogging project simply brings that truth into focus.
EIGHT
THE OK PLATEAU
If you visited my office in the fall of 2005, you would have seen a Post-it note—one of my external memories—stuck to the wall above my computer monitor. Whenever my eyes strayed from the screen, I saw the words “Don’t Forget to Remember,” a gentle reminder that for the next several months until the U.S. Memory Championship, I needed to strive to replace my regular procrastination patterns with more productive mnemonic exercises. Instead of browsing the Web or walking around the block to cool my eyes, I’d pick up a list of random words and try to memorize it. Rather than take a magazine or book along with me on the subway, I’d whip out a page of random numbers. Did I understand, at the time, how weird I was becoming?
I started trying to use my memory in everyday life, even when I wasn’t practicing for the handful of arcane events that would be featured in the championship. Strolls around the neighborhood became an excuse to memorize license plates. I began to pay a creepy amount of attention to name tags. I memorized my shopping lists. I kept a calendar on paper, and also one in my mind. Whenever someone gave me a phone number, I installed it in a special memory palace.
Remembering numbers proved to be one of the real world applications of the memory palace that I relied on almost every day. I used a technique known as the “Major System,” invented around 1648 by Johann Winkelmann, which is n
othing more than a simple code to convert numbers into phonetic sounds. Those sounds can then be turned into words, which can in turn become images for a memory palace. The code works like this:
The number 32, for example, would translate into MN, 33 would be MM, and 34 would be MR. To make those consonants meaningful, you’re allowed to freely intersperse vowels. So the number 32 might turn into an image of a man, 33 could be your mom, and 34 might be the Russian space station Mir. Similarly, the number 86 might be a fish, 40 a rose, and 92 a pen. You might visualize 3,219 as a man (32) playing a tuba (19), or maybe a person from Manitoba (3,219). Likewise, 7,879 would translate to KFKP, which might turn into a single image of a coffee cup, or two images of a calf and a cub. The advantage of the Major System is that it’s straightforward, and you can begin using it right out of the box. (When I first learned it, I immediately memorized my credit card and bank account numbers.) But nobody wins any international memory competitions with the Major System.
When it comes to memorizing long strings of numbers, like a hundred thousand digits of pi or the career batting averages of every New York Yankee Hall of Famer, most mental athletes use a more complex technique that is known on the Worldwide Brain Club (the online forum for memory junkies, Rubik’s cubers, and mathletes) as “person-action-object,” or, simply, PAO. It traces its lineage directly back to the loopy combinatorial mnemonics of Giordano Bruno and Ramon Llull.
In the PAO system, every two-digit number from 00 to 99 is represented by a single image of a person performing an action on an object. The number 34 might be Frank Sinatra (a person) crooning (an action) into a microphone (an object). Likewise, 13 might be David Beckham kicking a soccer ball. The number 79 could be Superman flying with a cape. Any six-digit number, like say 34-13-79, can then be turned into a single image by combining the person from the first number with the action from the second and the object from the third—in this case, it would be Frank Sinatra kicking a cape. If the number were instead 79-34-13, the mental athlete might imagine the equally bizarre image of Superman crooning at a soccer ball. There’s nothing inherently Sinatraish about the number 34 or Beckhamesque about 13. Unlike the Major System, those associations are entirely arbitrary, and have to be learned in advance, which is to say it takes a lot of remembering just to be able to remember. There’s a big fixed cost in terms of time and effort to compete on the memory circuit. But what makes this system so potent is that it effectively generates a unique image for every number from 0 to 999,999. And because the algorithm necessarily generates unlikely scenes, PAO images tend, by their nature, to be memorable.
The sport of competitive memory is driven by an arms race of sorts. Each year someone—usually it’s a competitor who is temporarily underemployed or a student with an unstructured summer vacation—comes up with an ever more elaborate technique for remembering more stuff more quickly, forcing the rest of the field to play catch-up.
Ed had just spent the previous six months developing what he described as “the most elaborate mnemonic behemoth ever brought to bear at a memory championship.” His new system, which he referred to as “Millennium PAO,” represented an upgrade from the two-digit system used by most European competitors to a three-digit system consisting of a thousand different person-action-object images. It would allow him to convert every number from zero to 999,999,999 into a unique image that would hopefully be impossible to confuse with any other. “While before I had a little two-digit laser boat that could dart through numbers like a tuna on amphetamines, now I have a three-digit sixty-four-gun Man of War,” he boasted. “It is enormously powerful, yet potentially difficult to control.” If the system worked, he figured, it would advance the sport of competitive memory by a quantum leap.
Mental athletes memorize decks of playing cards in much the same way, using a PAO system in which each of the fifty-two cards is associated with its own person/action/object image. This allows any triplet of cards to be combined into a single image, and for a full deck to be condensed into just eighteen unique images (52 divided by 3 is 17, with one card left over).
With Ed’s help, I laboriously created my own PAO system, which involved dreaming up fifty-two separate person/action/object images. To be maximally memorable, one’s images have to appeal to one’s own sense of what is colorful and interesting. Which means that a mental athlete’s stock of PAO images is a pretty good guide to the gremlins that live in someone’s subconscious: in my case, 1980s and early 1990s TV icons; in Ben Pridmore’s case, cartoon characters; in Ed’s case, lingerie models and Depression-era English cricketers. The king of hearts, for me, was Michael Jackson moonwalking with a white glove. The king of clubs was John Goodman eating a hamburger, and the king of diamonds was Bill Clinton smoking a cigar. If I were to memorize the king of hearts, king of clubs, and king of diamonds in order, I would create an image of Michael Jackson eating a cigar. Before I could memorize any decks of cards, I first had to memorize those fifty-two images. No minor job.
But my PAO system pales in comparison to the system that Ben Pridmore uses for cards. In the fall of 2002, he quit the job he’d held for six and a half years as an assistant accountant at a meat factory in Lincolnshire, spent a week in Vegas counting cards, and then came back to England and spent the next six months watching cartoons, getting qualified to teach English as a second language, and developing an entirely new mnemonic nuclear arsenal. Instead of creating a single person-action-object image for each card in the deck, Ben spent dozens of hours dreaming up a unique image for every two-card combination. When he sees the queen of hearts followed by the ace of diamonds, that’s a unique image. When he sees the ace of diamonds followed by the queen of hearts, that’s a different unique image. That’s 52 times 52, or 2,704, possible two-card combinations for which Ben has an image pre-memorized. And like Ed, he places three images at each of his loci. That means he’s able to condense an entire pack of cards into just nine loci (52 divided by 6), and twenty-seven packs of cards—the most he’s ever been able to memorize in a single hour—into just 234 places.
It’s hard to say which is the more admirable component of this feat: Ben’s mental or manual dexterity. He has developed an ability to quickly thumb two cards at a time off the top of the deck, in the process spreading them just enough to reveal the suit and number in the corner of both. When he’s going at top speed, he looks at each pair of cards for less than a second.
Ben developed a similarly Byzantine system for memorizing binary digits, which enables him to convert any ten-digit-long string of ones and zeros into a unique image. That’s 210, or 1,024, images set aside for binaries. When he sees 1101001001, he immediately sees it as a single chunk, an image of a card game. When he sees 0111011010, he instantaneously conjures up an image of a cinema. In international memory competitions, mental athletes are given sheets of 1,200 binary digits, thirty to a row, forty rows to a page. Ben turns each row of thirty digits into a single image. The number 110110100000111011010001011010, for example, is a muscleman putting a fish in a tin. At the time, Ben held the world record for having learned 3,705 random ones and zeroes in half an hour.
Every mental athlete has a weakness, an Achilles heel. Ben’s is names and faces. His scores in the event are always near the bottom of the pack. “I don’t tend to look at people’s faces when I talk to them,” he told me. “In fact, I have no idea what lots of people I know really look like.” To get around this problem, he has been developing a new mnemonic system for the event that would assign numerical codes to eye color, skin color, hair color, hair length, nose, and mouth shape. He figures that if people’s faces could only be turned into a string of digits, they’d be a cinch to remember.
When I first set out to train my memory, the prospect of learning these elaborate techniques seemed preposterously daunting. But Anders Ericsson and I struck a deal. I would give him the meticulous records of all my training, which would be useful data for his research on expertise. In return Tres and Katy, his graduate students, woul
d analyze that data in search of ways I could perform better. After the memory championship, I had promised to return to Tallahassee for a couple days of follow-up testing so they could get a journal article out of the whole enterprise.
Ericsson has studied the process of skill acquisition from dozens of different angles in almost as many different fields, and if there were any general secrets to becoming an expert, he was the person most likely to reveal them. What I already knew from talking with him extensively, and from reading almost every book and paper he’d written, was that in domain after domain, he’d found a common set of techniques that the most accomplished individuals tend to employ in the process of becoming an expert—general principles of expertise acquisition. Those principles would be my secret weapon.
Over the next several months, while I toiled away with PAO in my parents’ basement, Ericsson kept close tabs on my development. I kept him apprised of my evolving thoughts about the impending competition, which I noticed had gradually begun to shift from innocent curiosity to zealous competitiveness. When I’d get stuck, I’d call Ericsson up for advice, and he’d inevitably send me scurrying for some journal article that he promised would help me understand my shortcomings. At one point, a few months into my training, my memory stopped improving. No matter how much I practiced, I couldn’t memorize a deck of playing cards any faster. I was stuck in a rut, and I couldn’t figure out why. “My card times have hit a plateau,” I lamented to him.
“I would recommend you check out the literature on speed typing,” he replied.
When people first learn to use a keyboard, they improve very quickly from sloppy single-finger pecking to careful two-handed typing, until eventually the fingers move so effortlessly across the keys that the whole process becomes unconscious and the fingers seem to take on a mind of their own. At this point, most people’s typing skills stop progressing. They reach a plateau. If you think about it, it’s a strange phenomenon. After all, we’ve always been told that practice makes perfect, and many people sit behind a keyboard for at least several hours a day in essence practicing their typing. Why don’t they just keep getting better and better?
