Brain Science
Chunking groups information into meaningful units so working memory can hold more. How it works, the research from Miller to chess masters, and how to use it every day.
Try to remember this string of letters after reading it once: F B I C I A U S A N A S A. Thirteen letters is far more than most people can hold. Now look again, grouped: FBI, CIA, USA, NASA. Suddenly it is four items, and almost anyone can recall it.
Nothing about the letters changed. What changed is how you packaged them. That packaging is called chunking, and it is the closest thing memory has to a cheat code.
Your working memory is the mental space where you hold information while you use it: a phone number while you dial it, a total while you add the next item, the start of a sentence while you read to the end. Its capacity is small.
In 1956 the psychologist George Miller published one of the most cited papers in psychology, titled "The Magical Number Seven, Plus or Minus Two". He noticed that people could hold roughly seven items in immediate memory, whether those items were digits, letters or words. Later research, notably by Nelson Cowan, suggested the true limit is closer to four when people cannot rehearse or group items.
Miller's key insight was not the number itself. It was that the limit applies to chunks, not to raw pieces of information. Seven digits and seven words take up roughly the same space, even though seven words contain far more letters. If you can make each chunk bigger, the same small working memory holds much more.
A chunk is any group of items you already know as a unit. "NASA" is one chunk for you because you have seen it thousands of times. For someone who has never heard of it, it is four letters.
That is why chunking depends on knowledge stored in long-term memory. When you read, you do not process letters one at a time. Familiar words arrive as single units, and familiar phrases can arrive as one unit too. A child learning to read is chunking letters into words for the first time, which is why reading is so tiring at first.
Some of the most striking research on chunking comes from chess. In the 1970s, William Chase and Herbert Simon showed chess positions to players of different levels for five seconds and asked them to rebuild the board from memory.
Masters were dramatically better than beginners at recalling positions from real games. But when the pieces were placed randomly, the masters' advantage largely disappeared. Their memory was not better in general. They recognised groups of pieces as familiar patterns, an opening structure or a castled king, and stored each pattern as a single chunk. Random boards had no patterns, so they had to remember piece by piece like everyone else.
A later study made the point even more dramatically. Psychologists Anders Ericsson and William Chase trained a university student, known in the research as SF, to remember long strings of digits. His starting span was about seven digits. After many months of practice, he could recall around eighty. He did it by chunking digits into running times, because he was a keen runner: 3492 became "3 minutes 49.2 seconds, near world record mile time". His working memory had not grown. His chunks had.
You already chunk constantly, often without noticing:
When you need to hold a number, break it into groups of two to four digits and attach meaning if you can. A PIN of 1945 is a year. A sequence like 2 4 6 8 is a pattern. A number with a repeated pair is easier than one without. The more meaning, the more stable the chunk.
Before you try to learn a list, spend a moment sorting it. A shopping list of fifteen items is hard. The same list sorted into "dairy, vegetables, cleaning" becomes three categories of about five items each. Each category then acts as a cue for its members.
When learning something complex, master small units until they feel automatic, then combine them. A guitarist learns chord shapes, then chord changes, then whole progressions. Each layer becomes a single chunk for the next.
Chess shows that bigger chunks come from knowing more patterns. Whatever you want to get better at remembering, whether historical dates, medical terms or code, study the common structures. Prefixes and roots in vocabulary, common shapes in data, recurring dates in a timeline: each one you learn becomes a ready-made chunk.
Memory games are a good place to practise chunking because you get instant feedback. In a sequence task, you will quickly find that remembering "up, up, left, left" as "two ups, two lefts" is easier than four separate moves. In a grid of numbers, noticing that several cells form a pattern lets you store them as one unit.
MyRin's Number Sequence and Memory Match games are built around this kind of grouping. As levels get harder, the players who improve fastest are usually not those with the biggest raw span. They are those who start looking for structure.
Chunking does not enlarge working memory itself. SF's digit span for letters, rather than numbers, stayed at about six. The skill was specific to the material he had built chunks for. That is consistent with what we know from other research, including the debate around dual n-back training: improvements tend to be specific to what you practise.
That is not a disappointment. It means that the way to remember more in any area that matters to you is to learn its patterns well. The limit stays at about four. What you can fit into each of those four slots keeps growing.