cognitive psychology
Miller's law
What is Miller's law?
Working memory holds about seven items at once, plus or minus two. Beyond that, things start to slip. It’s not about how smart you are - it’s about the mental space you’ve got.
Also known as: the magical number seven, Millers law
Prefer to watch?
Watch the recap 0:34
The demo
A sequence of digits will flash on screen for 1.5 seconds, then disappear. Type what you saw. Each round adds one more digit. See how far your working memory stretches.
Didn't catch it?
What this demo shows (text version)
A digit-span test: sequences of random digits flash on screen for 1.5 seconds, then disappear. The reader types what they saw. The sequence starts at 4 digits and grows by one each round, up to 10. Most people recall sequences up to 5-9 digits reliably before errors begin, with an average around 7.
The result is Miller's law: working memory holds roughly 7 (+/- 2) items. The test makes the abstract limit personal, letting you find your own span, not a statistic.
People can handle only a few things at once - usually around seven, often less. Break information into small, meaningful chunks (like phone numbers grouped) and you work within your limit, not against it. You can see and try this concept in action on this page.
That's your working memory limit in action. The number wasn't made up: George Miller measured it in 1956 and found the same result across people, senses, and kinds of information.
Miller's original "magical number seven" has been refined since: modern estimates put the chunk limit closer to four, once rehearsal tricks are controlled for. But take care what you do with it - the number is about immediate memory span for unrelated items, not a cap on how many things may appear on screen. There's no "seven navigation items" rule. The real design implication is gentler: working memory is small, so don't make people hold things in their head; let the interface remember for them.
The "+/- 2" caveat matters more than the headline number. Some people manage nine, some five. Designing for the mean fails the people at the edges. Chunking (grouping items into fewer, meaningful units) is the practical answer: it doesn't expand the workspace, but it makes better use of it.