Brain Buys (9 page)

That the Pluto/planet association in my brain will probably never be erased is not a bad thing, after all it is relevant for me to know that Pluto used to be considered a planet. If I erased this information entirely, I would be confused by old literary and film references that referred to Pluto as a planet. Other than if I’m ever on
Jeopardy
and presented with the statement “It is the planet most distant from the Sun,” my not being able to delete the Pluto/planet link is of little consequence. As we have seen in the previous chapter, however, there may be consequences to not being able to easily delete other associations; such as Muslims/terrorists, Americans/warmongers, or women/bad at math. Whether or not it would be beneficial to delete specific associations on demand, or erase traumatic memories, is ultimately arguable. What is clear, however, is that our neural hardware is not designed with this feature in mind.

DISK SPACE

When we buy a computer we can choose whether the hard drive stores 500 or 1000 gigabytes. But what is the storage capacity of the human brain? This is a difficult, if not impossible, question to answer for a number of reasons—foremost because it requires defining exactly what we mean by
information
. Digital storage devices can be easily quantified in terms of storage capacity, as defined by how many bytes—that is, how many groups of eight 0s or 1s—can be stored. Although extremely useful because it provides an absolute measure to compare different storage devices, strictly speaking, most of us don’t really care how many bytes fit on a disk. Rather, what we care about is how much information of the flavor that we are interested in can be stored: a professional digital artist may want to know how many Photoshop files can be saved, an electrophysiologist may need to know how many hours of EEG data can be stored, and in the case of an iPod, we’re generally interested in how many songs we can carry around with us. But even for an iPod—a perfectly understood gadget—we cannot precisely answer the question of how many songs it can store, as that number changes based on the length of the songs and the format they are in.

Despite the challenges in estimating the storage capacity of any type of memory, psychologists have attempted to estimate the capacity of human memory by focusing on well-constrained tasks, such as how many pictures people can recall having seen before. Studies in the 1970s suggested that “there is no upper bound to memory capacity.”
¹⁹
But obviously the brain cannot have unlimited memory; as a finite system it can store only a finite amount of information.

The more interesting question is whether the user approaches the capacity limits of his or her memory. Early research indicated that our ability to store images was extremely high. In one such study participants were shown thousands of pictures, each for approximately five seconds. Later, they were shown pairs of pictures, one new and one repeat, and asked to identify which they had already been exposed to. When tested the same day, after having viewed 10,000 pictures, subjects were able to pick the ones they had already seen out of a pair with an accuracy of 83 percent. An impressive feat, which suggested they had recalled 6600 of the pictures.
²⁰
In these experiments, however, each picture was highly distinct from all the others (car, pizza, mountain, peacock), so each one interfered relatively little with the other. Needless to say, if you were shown pictures of 10,000 different leaves, your success rate at identifying which ones you had previously seen would be considerably closer to chance. Furthermore, in these studies subjects always knew they had seen one of the pictures of the pair, so like an eyewitness who believes the criminal is in the lineup, subjects are encouraged to guess. Another study, using 1500 pictures, tested visual memory capacity by showing the photos one by one during the test phase and asking subjects to judge whether each image was “new” or “old.” In this case people classified approximately 65 percent of the pictures correctly, closer to the 50 percent expected by chance alone.
²¹

Our ability to determine whether we have previously seen a specific image is not bad by some measures, but what about our memory capacity for something a bit more practical in the modern world, such as putting a name to a face? This is a task most of us struggle with; study participants who are told the names and professions of 12 pictured people will likely recall only two or three names, but four or five of the professions.
²²
This is, however, after a single exposure and does not address the human brain’s long-term storage capacity for names and faces. Another way to measure memory capacity for face/name pairs would be to determine the total number of people we can name. In theory this could be measured by showing someone pictures of all the people he had ever met or seen and determining how many of them he could name. This would include all possible sources: family, friends, acquaintances, classmates, characters on TV, and celebrities. I am not aware of any estimates of how many people the average human being knows by name,
²³
but speaking for myself I estimate it to be below 1000—and I imagine that the number is well below 10,000 even for those annoying people who seem to remember the names of everybody they have ever met or seen. If someone were reckless enough to try to convert this high estimate of 10,000 into bytes, then he might argue that a reasonable quality picture (and the text for the name) can be easily stored in a 100 KB file, for a total of 1 GB. A respectable but unimpressive number that is approximately the storage capacity of a single sperm cell.
²⁴

MEMORY CHAMPIONS

The study of the memory capacity of humans has been facilitated by the advent of the World Memory Championships, which were first held in London in 1991. Although you’d be forgiven for thinking that the Memory Championships are the clever ploy of some psychologist seeking research subjects, they are actually an earnest competition pitting one mental athlete against another. Memory championships have a number of different events including memorizing the order of all the cards in a deck and sequences of numbers. In the speed number competition, participants are given a sheet of paper with 1000 digits on it. The competitor is given 5 minutes to commit those digits to memory, and 15 minutes later, must reproduce as many digits as possible in the exact order they appeared. In the 2008 U.S. National Memory Championships, the overall champion, Chester Santos, memorized a list of 132 digits. Chester first heard about the World Memory Championship from a TV program in 2000, when he was twenty-three. He competed in his first national championship in 2003, and in just five years managed to become U.S. champion.

One might be inclined to take Chester’s abilities as evidence that human memory is actually quite good, it’s just that the rest of us don’t know how to use it. But in fact the competitors in the World Memory Championships illustrate how poorly suited the brain is for memorizing isolated pieces of information.

Competitors in the World Memory Championships may indeed be naturally blessed with better-than-average powers of memorization, but their feats largely come down to practice and technique. One of the most common methods competitors use for memorizing long sequences of numbers is to learn to associate every possible three-digit number (000, 001, 002,…, 999) with a person, an action, and an object.
²⁵
For example, through months or years of practice you might learn to associate the number 279 with Bob Dylan, playing soccer, and a pickle; the number 714 with Scarlett Johansson, shooting, and a porcupine; and 542 with Einstein, sewing, and clouds. So, if the first nine numbers of a sequence were 2-7-9-7-1-4-5-4-2 then you might visualize Bob Dylan shooting clouds. The next nine digits could lead to Mahatma Gandhi shoveling pizza. Of course, a 90-digit number still requires the difficult task of remembering 10 of these surreal fragments, but the mental imagery of Bob Dylan shooting clouds is much more catchy than a long string of numbers. This person/action/object technique is often supplemented by visualizing these events happening sequentially along the path of a familiar route. In this so-called method of loci one might envision each person/action/object event happening at each one of the stops of the bus taken to work.

The human brain is so ill-equipped for memorizing numbers, then, that the competitors in the World Memory Championship don’t even try to memorize them. They translate them into something much more natural to remember such as people they know, actions, and objects; use these to create stories; and then memorize these stories rather than the numbers themselves. The stories are then translated back to numbers upon recall. From a computational perspective this is, of course, highly inefficient—the neural equivalent of a Rube Goldberg machine. In a computer, numbers are stored as sequences of zeros and ones as opposed to images of the numbers, or as sentence fragments that could have been written by a roomful of monkeys. But if you need to remember the sequence 12-76-25-69, you may be better served by thinking of the associations they evoke: a dozen, American independence, a quarter, and whatever 69 reminds you of.

The person/action/object method relies on first storing a large repertoire of associations in long-term memory the hard way—by rote memorization. This process presumably creates permanent and strong links between specific nodes, for example, “Bob Dylan” and “279.” Once these associations are hardwired into neural circuits, they can be rapidly accessed and stored in short-term memory. The first advantage of this method is that our short-term memory is better suited for memorizing people, actions, and objects than numbers, so it is more natural to visualize people doing something than to visualize strings of numbers. A second, less obvious, benefit of the person/action/object method is that it decreases interference. As we have seen, related concepts can interfere with each other, making it hard to remember the details even though the gist may be recalled. For most of us, a list of numbers, at some point, blends into precisely that, and the individuality of each number is lost. By translating numbers into nonsensical but evocative images, we are performing what neuroscientists call
pattern separation
, referring to how much “overlap” each item on the list has. Simply put, “Bob Dylan” is less similar to “Mahatma Gandhi” than 279 is to 714. By associating each number with totally unrelated concepts, the likelihood that the numbers will interfere with one another is decreased. The masters of the person/action/object method can use it to memorize impressively long lists of digits (the current world record stands at 405 numbers), but perhaps the most telling thing about this feat is how far out of their way mnemonic athletes will go to not have to memorize the numbers themselves.

SELECTIVE MEMORY

We currently inhabit a world in which we are exposed to infinitely more information than we can store. For example, we only remember a fraction of the names and faces we encounter. Evolutionarily speaking, it is no secret that the human brain did not evolve to store the names of a large number of people. The ability to recognize individuals of a social group is an ability shared by many of our mammalian contemporaries; yet we appear to be unique in our ability to use names. Furthermore, early in human evolution the total number of different people any one individual encountered was probably fairly low. Even assuming that 250,000 years ago our ancestors gave each other names, it seems unlikely they were exposed to more than a few hundred different people. Eventually, agriculture and other technological innovations fostered the emergence of villages and cities. Today, further technological advances including photography, TV, the Internet and its social networking have ensured that the number of people we are exposed to is likely orders of magnitude higher than the number of people our distant ancestors would have encountered.

Of the multitude of points in space and time that each of us has experienced, most leave little trace in our neural networks. The fact that I do not remember the faces of every passerby, the name of every person I’ve met, or every sentence I’ve read, may be evolution’s way of avoiding memory bank saturation. It is possible that human memory, whether for names, facts, or autobiographical episodes, currently operates near its capacity limits. Like the progressive decrease in amount of free space on a hard drive, the decrease in ease with which we store information as we age could reflect limited storage capacity of the brain.
²⁶

Early in life, when the cortex is as close to a blank slate as it will ever be, information may be stored in a highly robust and redundant fashion—big bold letters written on multiple pages across thousands and thousands of synapses. Late in life, with relatively few “blank” synapses available, information may be stored in a less redundant and sparser fashion—like small letters written on the margins of a single page—making it more susceptible to the inevitable remodeling, overwriting, and loss of synapses and neurons that happen with the passage of time. This is speculation, but this scenario would explain Ribot’s law, which states that we are more likely to first lose our most recent memories, and the oldest ones are the last to go. It is a phenomenon seen in Alzheimer’s disease in which someone’s life is slowly erased in reverse order—first the ability to recognize or recall the names of their recent friends and grandchildren evaporates, then the memory of their children, and, lastly, knowledge of their spouse and siblings disappears into the void.

What does or does not get stored in memory depends heavily on context, significance, and attention. Most of us remember where we were when we heard about world-shaking events, or about the unexpected death of a loved one. I’ve known people who remember the scores of every baseball game they have attended, yet struggle to remember a new phone number or their spouse’s birth date. Momentous or life-threatening events, as well as those that capture our interests and attention, get privileged access to our memory banks. This is in part the result of the specific cocktail of neuromodulators in circulation in the brain and the amount of attention devoted to these events.
²⁷
For example, the adrenaline that is released in moments of high alert contributes to formation of enduring or “flashbulb” memories. Such mechanisms may ensure that the important events, and those we are most interested in, will be stored, while preventing us from wasting space storing the details of the tedious hours we spend waiting in airports.