How does the memory test work?
This memory test measures your digit span, the longest number you can see once and type back correctly. It starts with a 3-digit number. It stays on screen for one second plus 0.6 seconds for every digit, then disappears, and you type it back. Get it right and the next number is one digit longer. The first mistake ends the game, and your score is the longest number you recalled correctly. Psychologists call that figure your digit span.
| Length | Time shown |
|---|---|
| 3 digits | 2.8 s |
| 5 digits | 4.0 s |
| 7 digits | 5.2 s |
| 9 digits | 6.4 s |
| 12 digits | 8.2 s |
Source: this test: 1 second plus 0.6 seconds per digit.
The extra time per digit matters. With a fixed time, long numbers would fail because you could not finish reading them, which would measure reading speed rather than memory. (Speed has its own tests: the reaction time test and the typing test.) When the game ends you see the number and your answer side by side, with the wrong digits marked, so you can tell a dropped digit from two swapped ones.
What is the magical number seven?
Memory span tests are among the oldest in psychology. In 1887 the journal Mind published "Experiments on prehension" by J. Jacobs, who read out lists of growing length until listeners could no longer repeat them in order. The idea became famous in 1956, when the Harvard psychologist George A. Miller opened a paper with the line "My problem is that I have been persecuted by an integer."
That integer was seven. Miller noticed that across very different tasks, from judging tones to repeating lists, people topped out at around seven items, give or take two. His paper, "The magical number seven, plus or minus two," is one of the most cited in psychology. Its subtler point is often lost: Miller argued that the limit applies to chunks, meaningful units, not raw digits. A string such as 1-9-8-4 is four digits but one chunk for anyone who has read Orwell.
Is the real memory limit closer to four?
The University of Missouri psychologist Nelson Cowan reviewed decades of results in 2001 and concluded that the core capacity of short-term memory is about four chunks, not seven. In a 2010 summary he put it at three to five meaningful items in young adults.
Spans of seven appear, Cowan argues, because people quietly help themselves. They rehearse the list under their breath, which can hold about two seconds of speech, and they group digits into chunks. Block those strategies and the number drops. In running span tasks, where a list stops without warning so there is no chance to rehearse, people recall only the last three to five digits.
| Source | Estimate | What is counted |
|---|---|---|
| Miller (1956) | 7 ± 2 | Items or chunks in span-type tasks |
| Cowan (2001) | about 4 | Chunks, with rehearsal and grouping controlled |
| Cowan (2010) | 3 to 5 | Meaningful items, young adults |
| Running span | 3 to 5 | Final digits of a list that stops without warning |
Source: Miller, Psychological Review (1956); Cowan, Behavioral and Brain Sciences (2001) and Current Directions in Psychological Science (2010).
Both numbers can be right. Seven is a fair description of what adults do on a digit span test with their usual tricks. Four is an estimate of the storage underneath the tricks.
What makes a memory span longer or shorter?
| Factor | Effect | Evidence |
|---|---|---|
| Grouping into chunks | Longer | Miller (1956); Cowan (2010) |
| Rehearsal blocked | Shorter, toward 3 to 5 | Cowan (2010) |
| Digits that take longer to say | Shorter | Welsh and English digits, Ellis and Hennelly (1980) |
| Long practice with a mnemonic | From 7 to 79 digits | Ericsson, Chase and Faloon (1980) |
Source: studies listed under Sources.
The language effect is a neat demonstration of rehearsal. In 1980 N. C. Ellis and R. A. Hennelly tested bilingual Welsh and English speakers. Welsh number words take longer to say, and the same people had shorter spans in Welsh than in English. The authors argued this, not any difference in ability, explained why Welsh children's scores on a digit span intelligence subtest ran lower.
The practice result is the most striking. In a study published in Science in 1980, a student with an ordinary span of seven practiced for more than 230 hours in the lab and reached 79 digits. A keen runner, he coded digit groups as running times, turning meaningless strings into a few familiar chunks. His underlying capacity had not grown; his chunks had.
Read the number in groups of three, as you would a phone number, and say it silently while it is on screen. Most people gain a digit or two. That gain is chunking and rehearsal, the very strategies Cowan describes.
How is this different from a lab digit span?
Psychologists usually read the digits aloud, one at a time, and the person repeats them. This version shows the whole number at once on screen, which lets you see and group it. Treat your score as a game result, comparable with your own earlier games, rather than with published norms.
| Typical lab test | This page | |
|---|---|---|
| Presentation | Spoken, one digit at a time | Shown all at once |
| Direction | Often forward and backward | Forward only |
| Answer | Said aloud to a tester | Typed |
| Purpose | Research or assessment | A game and a demonstration |
Source: The Daily Summit.
How does the sequence game work?
The second mode of the memory test measures spatial memory instead of digits. Squares on a 3x3 grid light up one after another; you tap them back in the same order. Each round repeats the whole sequence with one more square added. It resembles the Corsi block-tapping task, a spatial counterpart to digit span that psychologists have used since the 1970s. Most people find it harder to rehearse positions than digits, because there is no word to say under your breath.
Common questions
What is a good digit span?
Around seven is the classic figure for adults hearing digits read aloud. This test lets you see and group the whole number, so your score here can differ from a spoken test. Compare it mainly with your own earlier games.
Can you train your memory with this test?
You can raise your score, mainly by getting better at chunking. The 1980 runner study shows how far that goes, and also that the gain came from strategy, not from a bigger memory store.
Is this memory test an IQ test?
No. Our IQ test uses reasoning puzzles instead. Digit span is one small part of some intelligence test batteries, but on its own it measures one narrow skill, and this game version is not a clinical tool.
Where is my best score stored?
In your browser's local storage on this device. The page does not send it to a server, though the site does count page views. In a private window it may not be kept at all.
Sources
- Miller: The magical number seven, plus or minus two, Psychological Review (1956), full text at Classics in the History of Psychology
- Cowan: The magical number 4 in short-term memory, Behavioral and Brain Sciences (2001)
- Cowan: The magical mystery four, Current Directions in Psychological Science (2010)
- Ericsson, Chase and Faloon: Acquisition of a memory skill, Science (1980)
- Ellis and Hennelly: A bilingual word-length effect, British Journal of Psychology (1980)
Last reviewed November 9, 2026. For information and fun only; not a medical or psychological assessment.