Digit span: the 60-second memory task, explained

How the tasks work · updated 2026-09-05

If you have ever been shown a short string of numbers and asked to repeat them, you have taken a digit span task. It is more than a century old, it takes under a minute, and it still appears in almost every standard cognitive battery. Its longevity is a good sign: simple tasks that keep earning their place usually measure something real.

Forwards and backwards

The task comes in two forms, and the difference between them is the whole point.

Forwards you see or hear a sequence, then reproduce it in the same order. This is close to a pure test of how much you can hold in short-term storage. Most adults manage somewhere around seven items, which is where the familiar "seven plus or minus two" figure comes from, though modern estimates for genuinely unrehearsed material are lower.

Backwards you reproduce the sequence in reverse. This is harder for almost everyone, and not because it needs more storage. It needs you to hold the sequence and operate on it at the same time. That combination, holding plus manipulating, is roughly what psychologists mean by working memory as distinct from short-term memory.

Forwards asks how much you can carry. Backwards asks whether you can rearrange it while carrying it. Most people are noticeably worse at the second, and the size of the gap varies a lot between people.

Why chunking makes the score slippery

Digit span is easy to game, which is worth knowing before you take your result too seriously. The standard trick is chunking: instead of holding 4 7 1 9 as four items, you hold "47" and "19" as two, or recognise 1947 as a year.

This is why competitive memory performers can reproduce absurdly long strings without having unusual raw capacity. They have a practised system for converting digits into larger, more memorable units. It is a real skill, but it means their span score is measuring the system, not the storage.

For the same reason, your own score will improve if you take the test repeatedly, and that improvement is mostly strategy rather than capacity.

What moves your span on the day

• Rehearsal. If you can say the digits to yourself, you will do better. Tasks that block rehearsal, by making you speak or count at the same time, produce much lower spans.
• Presentation speed. Digits presented faster than about one per second reduce most people's span, because there is no time to rehearse.
• Modality. Heard and seen sequences do not always produce the same score in the same person.
• Interference. Anything you have to hold at the same time, including a conversation, cuts the score sharply.

Reading your own result

Two things are worth paying attention to, and neither is the headline number.

The first is the gap between forwards and backwards. A large gap suggests you store well but find manipulation costly; a small gap suggests the opposite profile. Neither is better in general, but they suit different kinds of work.

The second is where you break down. Most people do not degrade gradually; they are near-perfect up to a length and then fall off a cliff. Noticing your own cliff edge is more informative than the score.

Can you improve it?

You can certainly improve your score, quickly, by learning to chunk. Whether training a span task improves anything else is a long-running and genuinely unsettled argument in the research literature, and the honest summary is that transfer to unrelated tasks has been hard to demonstrate reliably.

Which is a reasonable note to end on: digit span is a good, cheap, durable measure of something specific. It is not a measure of intelligence, it is not a diagnostic, and improving it mostly makes you better at digit span.