What this data is, and isn't
Working-memory span — most commonly measured via the digit span test — is one of the most-studied measures in cognitive psychology and clinical neuropsychology. The norms on this page are drawn from the clinical-neuropsych literature, where digit span is a standard subscale on tests like the WAIS (Wechsler Adult Intelligence Scale).
They are clinical norms used for context, not for self-diagnosis. Digit span is genuinely used in clinical settings to assess working memory, and a trained clinician interprets the results in context of the full assessment. Self-administering the test from a webpage is not a clinical assessment.
They are not Senwitt-collected data. Senwitt does not run digit-span tests on users. The numbers here are public-domain norms with sources.
The data
Forward digit span by adult age group
Forward digit span: hear a sequence of digits, repeat them back in the same order. The score is the longest sequence you can correctly repeat (a value typically between 4 and 9).
| Age group | Typical average | Strong performers |
|---|---|---|
| 20-39 | 6-7 items | 8-9 items |
| 40-59 | 6-7 items | 7-8 items |
| 60-69 | 5-7 items | 7-8 items |
| 70+ | 5-6 items | 6-7 items |
Backward digit span by adult age group
Backward digit span: hear a sequence of digits, repeat them back in reverse order. Harder than forward, because it requires actively manipulating the sequence in working memory rather than just storing and retrieving it.
| Age group | Typical average | Strong performers |
|---|---|---|
| 20-39 | 5-6 items | 7-8 items |
| 40-59 | 4-6 items | 6-7 items |
| 60-69 | 4-5 items | 6-7 items |
| 70+ | 3-5 items | 5-6 items |
What these numbers mean in context
The classic "seven plus or minus two" estimate from George Miller's 1956 paper is a useful rule of thumb but a slight overestimate for true working-memory tasks. More recent research, including the norms above, puts the typical capacity closer to four to seven items, depending on the specific task and the population.
Three observations worth holding:
Working memory is finite and small. Even strong performers max out around 8-9 items on forward span and 6-7 on backward. This is not a limit you can train past in any general, lasting way — the Stanford-organized scientific consensus on brain training is specifically skeptical of "increase your working memory capacity" claims.
Decline with age is modest for forward span, more visible for backward span. This pattern is consistent with the broader cognitive-aging literature: simple storage holds up better than active manipulation as we age.
Individual variation is wide. Two adults at the same age can differ by 2-3 items, which is a large difference relative to the cohort range. Use the cohort-level numbers as orientation; don't read too much into individual scores.
What changes working-memory performance
Variables with the largest within-individual effect:
- Sleep. Working-memory performance is one of the most sleep-sensitive cognitive measures. A short night can drop digit span by 1-2 items.
- Distraction. Working memory is heavily affected by competing attentional load. Testing in a noisy environment lowers measured span.
- Anxiety. Test anxiety, evaluation pressure, and general arousal can each impair span by 1-2 items in many adults.
- Practice with the specific test. Span on the digit-span test improves with practice on the test itself; this is the standard "practice effect" that any cognitive assessment accounts for in clinical use.
The practice effect is precisely the reason brain-training apps reliably improve performance on the games inside the app — and precisely the reason that improvement does not generally transfer to unrelated tasks.
Methodology note
The ranges on this page are drawn from published WAIS-IV adult norms and from associated clinical neuropsychology literature. They are not the result of a Senwitt study. We present them as ranges rather than point estimates because clinical norms themselves are presented with confidence intervals; a single number would imply more precision than the underlying data supports.
What this is not
This data page is not a working-memory assessment. It is not a substitute for the clinical use of the digit-span test, which has clinical context, trained interpretation, and an assessment battery around it.
If you've noticed working-memory changes that worry you — sustained difficulty holding instructions in mind, repeated need to ask for information to be repeated, difficulty following multi-step conversations — that is a doctor conversation, not a webpage benchmark.
