Reading: Remember Forgetting
THE SCIENCE OF LEARNING TO LEARN
READING: REMEMBER FORGETTING
We all forget. It might take a few days—or just a few minutes. But learning often leaves as soon as it arrives. When it comes to memory, the brain is a lot like a sieve. Many recollections will vanish after just a few moments. Still worse, even the details that do stay often disappear with time.
This happens in schools all the time, even after dedicated bouts of studying. Medical students, for instance, sometimes fail to remember more than 50 percent of what they’ve learned within a few months, according to one study. So while an aspiring doctor might ace an anatomy test during her first year at Med U., she’s likely to bomb the same exam less than a year later.
Granted, we want to believe that we will remember everything that we see or do or experience. It’s painful to think that we won’t remember key memories—a big graduation, a close friend, our first kiss. But when cognitive scientists study forgetting in the lab, they find that our memories seem to come with a type of timer. If the timer rings—and we have not re-engaged that memory again—the memory is forgotten. Researchers call it the “forgetting curve.”
While the research on memory and forgetting has been around for decades, it’s generally lived in dusty research journals, obscure books written by 19th century academics. At least until Roger Craig came along. Ever since Craig was a kid, he has loved playing games. Chess, Scrabble, poker, baseball, he’s been devoted to them all. “I’m very competitive,” Craig told me. “I like to win.”
So in grad school, Craig decided to compete at Jeopardy. As a kid, Craig had watched the game a lot with his grandparents. In graduate school, some of his buddies had tried to get on the show, too. Craig also thought that he might have an edge after coming across a Wired magazine article that detailed the value of spacing learning out over time.
From nearly the moment [a researcher] described the spacing effect, psychologists have been urging educators to use it to accelerate human progress.
“All students have been warned not to cram,” a Wired article argued. “But the efficiencies created by precise spacing are so large, and the improvement in performance so predictable, that from nearly the moment [a researcher] described the spacing effect, psychologists have been urging educators to use it to accelerate human progress.”
Craig had a sense that the Wired article might provide an important advantage. When Craig would study for tests in college at Virginia Tech, he would often revisit ideas so that he wouldn’t forget them. He even wrote a short computer program to help him space out his learning so that he would revisit ideas at regular intervals.
But the Wired magazine article promised a much deeper approach to spacing out learning, and Craig soon downloaded a bit of software called Anki. Using a highly developed algorithm, the software relied on the principles of spaced learning, and it would quiz people at the very edges of their memory. Or as the software’s developer argues: “Only practice the material that you're about to forget.”
Armed with a database of past Jeopardy questions, Craig started to hone his game-show skills, and he would review facts—details about presidents, names of old movies—in alignment with his rate of forgetting. So if Craig got a fact wrong, Anki would ask him about the fact again in a few minutes. If Craig landed the fact correct, then the question might not reappear again for a few days. If Craig landed the fact correct a second time, the fact would not appear again for a few months.
A more visual way to think about this approach is to consider this chart of the “forgetting curve.” It shows how long you remember something that you’ve just learned, and it suggests that after a few days—or even a few minutes—of learning something, you’ve probably forgotten it.
But the key thing are the green lines. That’s the sign of learning. That’s the sign of remembrance. When we come across something repeatedly in spaced fashion, we do actually remember it. And that’s what Roger Craig did with as many Jeopardy answers as he possibly could.
Craig landed on the show for the first time in September 2010, and in a small room with Alex Trebek and his two competitors, Craig basically didn’t miss a question. He destroyed his two competitors. He banged though Jeopardy item after item, eventually setting a record for the most amount of money won in a single game, blowing away the record that Ken Jennings had set some years earlier. (You can see a YouTube video of the event here.)
When Craig got back to his Los Angeles hotel that night, he was almost more surprised than excited. He knew that he would do well at the game; the effect of spacing out practice was rooted in some very robust science. But Craig just did not think that he would do that well, and Craig gave such a dominant performance that he thought: “Oh wow, maybe it’s worked too well.”
Craig had a hard time sleeping on that night. Would Jeopardy invite him back? Would Trebek think that he had somehow cheated? Craig hadn’t done anything wrong—he had simply relied on basic memory research to hone his skills. In the end, Jeopardy did ask Craig to return, and he won a half-dozen more games during his first appearance and also eventually aced the Tournament of Champions, which pits the game’s all-stars against each other.
Today, Craig lives in New York City. He works as a data scientist. He often uses Anki for work--and different games. As for spacing out learning, Craig argues that it’s simply a better way to grapple with knowledge. It ensures that we don’t forget. “Everybody that wants to succeed at a game is going to practice at the game," Craig once told an NPR reporter. "You can practice haphazardly, or you can practice efficiently. And that's what I did."
To be sure, Roger Craig’s call to end “haphazard learning” has not gone unanswered. At least a half-dozen software programs now promise to help people spread their learning out over days, weeks, months, or even years. SuperMemo is the grandfather of the field, perhaps the oldest program. More recently, VocApp allows users to include pictures to their spaced learning. DuoLingo’s approach focuses exclusively on foreign languages, allowing people to spread the learning of their Spanish language words out over time.
The spacing approach has spread to other fields too. Some corporate training programs have tried to take a more distributed approach to learning. Verizon, for instance, will now send follow-up training material to an employee’s computer to help keep their memories fresh. Some of Thompson Reuter’s learning tools use the approach to keep strategies “top-of-mind.”
Yet for the most part, learning remains concentrated. A spaced approach has not caught on, and in ways big and small, people cram. Instead of spreading their practice out over time, they will try and do it all in one afternoon. Or, we won’t revisit important ideas or details. Most people, for instance, can't recall the name of the most important battle of the Revolutionary War? (Hint: It wasn’t Lexington). Why? It’s because they simply have not revisited the material.
Cumulative tests, which encourage more spaced learning, are often rare or limited to the end of the year.
For their part, schools often encourage cramming. Besides perhaps for a “review” session at the start of the year, there’s little revisiting of material. Cumulative tests, which encourage more spaced learning, are often rare or limited to the end of the year. Many textbooks also fail to include dedicated review sessions, besides for a few questions at the close of a chapter.
Yet studies suggest that adding even a small amount of spacing can improve outcomes. Spreading out learning by just a tiny amount can boost scores. Without question, my favorite example of this idea comes from Nate Kornell. While Kornell was landing his post-doc at the University of California, Los Angeles, he noticed something curious about how the students approached their school work.
As Kornell wandered campus, he found that some of the undergrads would quiz themselves with small piles of flashcards, using a half-dozen cards or so to test their understanding of a specific idea or detail. With just a few cards in each pile, the students would flip through the small batches and then tuck all of the flashcards away, feeling like the material was well learned.
Other students on the campus who used flashcards took a very different approach. They would create one large pile of flashcards. In some instances, these stacks of cards were big—an inch or two thick—and so the students learned the material in a more distributed, spaced-out sort of way. More exactly, there'd be longer intervals of time between the revisiting of a flashcard, with students more likely to come across a card that they might have forgotten.
Kornell knew the students themselves didn’t dwell much on the difference. But he thought that the size of the stacks might shift what’s being learned. So he drew up a quick experiment. In the lab, one group of subjects would practice learning vocabulary words using a large pile of flashcards. Then Kornell had a second group, who learned the exact same material using four, smaller piles of cards. All of the students needed to know the meaning of the words, and they were high-level words that they had not heard before, words like “effulgent” and “abrogate.”
Another way to think about Kornell’s experiment is to imagine that you’re preparing for a talk. Maybe you’re giving a presentation to a client—or perhaps you have a speech to give at some family event. The question is: Are you better off practicing the full speech once a day for five minutes over a four-day period (like the students with one large stack of flashcards)? Or should you split the speech into four smaller sections and focus on one section each day for five minutes (like the students with four smaller stacks of flashcards)?
In the experiment, just about all of the students who practiced using one large stack of flashcards scored higher...
For their part, the subjects in the study voted for the smaller piles. They wanted to batch their studying, to concentrate their learning, and before the experiment began, just about all of the students indicated that they would learn more if they used four small heaps of cards. In other words, most people think that the best way to prepare for a speech is by splitting the talk into four short sections.
But Kornell found the exact opposite, and the results of spacing the learning—of distributing the practice —were dramatic. In the experiment, just about all of the students who practiced using one large stack of flashcards scored higher, even if the amount of time spent learning was the same. What’s more, many of the students who studied with one large pile of cards learned about a third more.
For anyone who wants to learn anything, the take-home is clear: Anything we do to distribute our learning over time pays off, and people should space out the development of a skill. If you’re practicing the violin don't just rehearse a melody for a few hours. Also return to the melody periodically so that it stays burned in your memory.
Want to ace a high-stakes exam? Start early so the learning can be spread out over time and re-quiz yourself every few weeks to ensure that you know the material. In my house, we’ve started doing less homework on weeknights and more on the weekends for the simple reason that it spaces out learning.
Today, Kornell is a professor at Williams College. He still often sees students quizzing themselves with a small stacks of cards—and shakes his head every time that he sees it. “Spacing doesn't require any more time. No extra resources. You don’t have to buy an iPad,” he told me. “It's kind of like a gift—you learn a lot more—and it’s free.”