A phone number you already know how to remember
Read this sequence once, look away, and try to say it back: 1 9 4 5 2 0 0 1 1 4 9 2.
Most people stumble somewhere in the middle. Twelve loose digits is more than working memory likes to hold. But watch what happens when the same digits are regrouped: 1945, 2001, 1492. Suddenly it is three dates—the end of a war, a year that needs no explanation, a voyage. The raw material did not change. What changed is that you stopped carrying twelve things and started carrying three.
That move has a name. Psychologists call it chunking, and it is one of the most reliable ways human memory beats its own limits. Understanding how it works changes how you study, how you write notes, and how you decide what is worth memorizing at all.
Why working memory is so small
In 1956 the psychologist George Miller published a paper with a famous title: "The Magical Number Seven, Plus or Minus Two." He noticed that across wildly different tasks—recalling digits, tones, words—people seemed to hit a wall at around seven items. Later researchers, particularly Nelson Cowan, argued the true limit is smaller still, closer to four independent items when you strip away rehearsal tricks.
Either way, the headline is humbling. The mental workspace where you hold information you are actively thinking about is tiny. It is the bottleneck every learner fights.
Here is the crucial part Miller emphasized: the limit is not measured in bits of information. It is measured in chunks—meaningful units. A chunk can be a single digit or an entire date, a single letter or a whole word. Working memory counts units, not raw data. So if you can pack more meaning into each unit, you can smuggle far more information past the bottleneck without the bottleneck ever noticing.
That is the whole game. Chunking does not expand working memory. It makes each slot worth more.
What actually makes something a chunk
A chunk is not just any grouping. It is a group your long-term memory already recognizes as a single, familiar pattern. "1492" is one chunk for someone who knows the date and four separate digits for someone who does not. The grouping has to connect to knowledge you already hold.
This is why chunking and expertise are inseparable. The classic demonstration came from chess. In the 1970s William Chase and Herbert Simon showed chess masters and novices a board mid-game for a few seconds, then asked them to reconstruct it. The masters were dramatically better—but only when the pieces formed a real, plausible game position. When the pieces were scattered randomly, the masters' advantage almost vanished. They were not seeing twenty-five pieces. They were seeing a handful of familiar configurations—a castled king, a known pawn structure—each one a single chunk built from years of play. Take away the meaning and the pieces became twenty-five loose items again, and the expert's memory was suddenly ordinary.
The lesson is quietly profound: a large part of what we call expertise is a vast private library of chunks. The expert is not holding more in working memory than you are. They have simply learned to see bigger units.
You can build chunks deliberately
The most striking proof that chunking is trainable came from a single undergraduate, known in the literature as S.F., studied by Ericsson, Chase, and Faloon. He began with an ordinary digit span—about seven numbers. Over roughly two years of practice, he worked his way up to recalling around eighty digits in a row.
He did not enlarge his working memory; tests confirmed it stayed normal-sized. He was a competitive runner, so he started hearing strings of digits as running times. "3492" became "3 minutes 49.2 seconds, near world-record mile pace." Each meaningful time was one chunk. As he ran out of running associations he layered on dates and ages, then grouped the chunks themselves into super-chunks. He had built, from scratch, the same kind of recoding machinery a chess master uses.
The encouraging takeaway for the rest of us: chunking is a skill, not a gift. You assemble it by relentlessly connecting new material to patterns you already own.
How to chunk on purpose
The principle translates into a few concrete habits.
Group by meaning, not by appearance. When you face a long list, resist memorizing it in raw order. Ask what natural categories live inside it. A grocery list of fifteen items is brutal as a sequence and easy as "four vegetables, three dairy, the taco things, breakfast." You are trading fifteen units for four.
Find the structure that is already there. Phone numbers, license plates, and account numbers come pre-chunked with dashes and spaces for exactly this reason. When material does not come grouped, group it yourself. Musicians do this constantly—reading a phrase of music as one shape rather than eight separate notes.
Build hierarchies. The most powerful chunking nests small chunks inside bigger ones. This is why an outline beats a flat list, and why a well-organized chapter is easier to recall than the same facts scattered. You hold the top level in mind and let each heading unpack the details beneath it on demand.
Earn your chunks before you lean on them. A chunk only saves space once the underlying pattern is genuinely familiar. Trying to treat unfamiliar material as a single unit just hides complexity you have not actually absorbed. The grouping has to be real to you.
Where chunking meets its limit
Chunking is powerful, but it is a strategy for encoding, not a substitute for durability. Grouping the digits into dates gets them through the working-memory door. It does not, by itself, keep them there next week. Memories fade unless they are revisited, and the spacing of those revisits matters enormously—the well-documented forgetting curve does not care how cleverly you packaged the material on day one.
So the complete picture has two moves. First, chunk—recode complex material into meaningful units so it is learnable in the first place. Then space your review of those chunks over expanding intervals, so the patterns harden into the long-term knowledge that makes future chunking possible. The two reinforce each other: every chunk you make permanent becomes raw material for a bigger chunk later. That is, quite literally, how expertise compounds.
Where Recall fits in
This is the rhythm Recall is built around. A good flashcard is already an exercise in chunking—you decide what counts as one meaningful unit, one question worth holding, instead of trying to swallow a chapter whole. And because Recall uses modern FSRS spaced repetition, the chunks you build don't quietly decay; each card returns just as you're about to forget it, turning a fragile new grouping into durable, reusable knowledge. You can import the decks you already have from Anki or Quizlet, and it all works fully offline, so your review happens wherever your attention actually is. If you've ever felt the ceiling of how much you can hold at once, Recall is a way to keep raising it: recall.lumenlabs.works.