Number Memory
How many digits can you hold in your head?
🎮 Our new game is coming
Town Guardians launches in 2026 — drop your email and be first to play.
More from Pong Studio →About the Number Memory Test
A number flashes on screen. It disappears. You type it back from memory. Get it right and the next number is one digit longer; get it wrong once and the run is over — there are no lives here, no second chances on a level you've already reached. That single rule makes this a different kind of test from most of the memory games on this site, and it's worth knowing before you play.
How it works
- The first number is 3 digits long. Every correct answer adds one more digit, up to a maximum of 20.
- The number stays on screen for 900 milliseconds plus 650 milliseconds per digit — so a 3-digit number shows for about 2.85 seconds, and a 10-digit number shows for about 7.4 seconds. Longer numbers get more time, not less.
- Type the number back and submit. One wrong answer — even a single misplaced digit — ends the run immediately.
Your score is the longest number you typed back correctly. There's no retry built into a single run: get it wrong once and you're done.
What one wrong digit costs you
This test has no lives and no partial credit. One digit out of place — even the last one — ends the game at whatever length you'd already cleared. That's a sharper rule than our Chimp Test, where a mistake costs a life and repeats the round instead of ending it. The two games measure related things, but a Number Memory score is closer to "where you first broke" and a Chimp Test score is closer to "your best run across a few tries." Neither is more valid than the other — they're just different questions, and it matters which one you think you're answering.
It also means a single distracted moment costs you exactly as much as a genuine limit. If you glance away for a second on digit 9 and mistype it, your score is 8 — identical to what it would be if 8 digits were a hard ceiling for you. A span test like this one is noisier than it looks; treat one result as a data point, not a verdict.
Why "7 digits" isn't the rule you think it is
The number everyone quotes is George Miller's 1956 paper, usually shortened to "seven, plus or minus two." It's a real, frequently cited paper — but Miller himself described the recurrence of the number seven across different memory experiments as something close to a coincidence, not a hard limit on human capacity. He was making a narrower point about how the mind counts in "chunks," not handing down a law that everyone's short-term memory holds exactly seven of anything.
A more careful estimate comes from Nelson Cowan's 2001 review, which found that when researchers block the tricks people normally use to hold numbers in mind — repeating them under your breath, grouping them into pairs — the actual capacity looks closer to about four chunks, not seven. That's not a contradiction of Miller's number so much as a different question: Miller's seven describes what people can hold onto with their normal strategies running; Cowan's four describes what's left when those strategies are taken away.
Which brings up the honest part: chunking digits into pairs or triples the way you'd read a phone number is not cheating on this test, and it will genuinely raise your score. That's fine — it just means the number you end up with measures your strategy about as much as it measures raw memory, and that's exactly the distinction Cowan's research is about. We don't have a sourced average for this browser format to compare yourself against; the live spread of real scores, and how many people it's drawn from, is on our stats page.
Frequently asked questions
What is a good number memory score?
We don't have a solid research figure for this exact browser test to hand you as a benchmark. What we can say is that the popular "7 digits" reference point comes from Miller's 1956 paper, and Miller himself treated the number seven as closer to a coincidence than a rule. Our Elite and Pro labels are tiers we set for this game, not scientific thresholds — check our stats page for the real spread of scores.
Is the "7 plus or minus 2" rule real?
The paper is real and frequently cited, but the "rule" gets overstated. George Miller himself said the repeated appearance of the number seven across different memory tasks looked like a coincidence, not evidence of a fixed capacity. A 2001 review by Nelson Cowan found that when common memory tricks like chunking are blocked, capacity looks closer to about four items — which isn't a contradiction, just a stricter version of the question.
Why does one wrong answer end the whole game?
That's a deliberate design choice, not a bug: this test has no lives, so your score is exactly the longest number you got completely right before the first mistake. It's a stricter format than our Chimp Test, which gives you three lives and repeats a failed round instead of ending your run.
Is it cheating to chunk the digits into pairs?
No — grouping digits the way you'd read a phone number is a normal memory strategy, and it will raise your score. That's worth knowing, though: it means this test measures your strategy about as much as your raw memory capacity, which is the same distinction researchers draw between memory "with tricks allowed" and memory with them blocked.
Why does the number stay on screen longer as it gets longer?
The display time scales with length — roughly 900 milliseconds plus 650 milliseconds for every digit — so a longer number gives you more total time, not the same fixed flash. It's still a tight window relative to how much you have to hold in your head, which is the point of the test.
Does this measure the same thing as the Chimp Test?
Related but not identical. This test ends on your first mistake, so your score reflects the first point where memory broke. The Chimp Test gives you three lives and lets you retry a failed round, so its score reflects your best run under a few attempts. Both are memory span tests; they answer slightly different questions about you.