{{ summaryTitle }}
{{ summaryValue }}

{{ summaryLine }}

{{ interpretationText }}

{{ stageLabel }} · {{ displayLength }} digits
+
Expected{{ digitText(feedbackRow.expected) }}
{{ cell.actual || '—' }}
Direction:
Forward: type digits in the order shown; Backward: type them in reverse.
{{ settings.rate_ms }} ms
Leave blank for one digit per second; 400–3000 ms includes a 200 ms blank gap.
ms
{{ settings.start_length }} digits
Leave blank for 3 Forward or 2 Backward; override with 2–12 digits.
{{ settings.misses_allowed }}
Leave blank for 1 retry; choose 0–3 before the final miss ends the session.
Saved sessions:
The last 20 session summaries stay in this browser. Raw trials and seeds stay in this tab until exported; results never enter a URL or leave through a network service. Browser storage can be cleared or unavailable. Compare sessions with matching direction, rate, start and allowance.
{{ history.length }} saved summaries

{{ storageMessage }}

#LengthAttemptShownExpectedYour answerResultErrorsCopy
{{ row.trial }} {{ row.length }} {{ row.attempt }} {{ digitText(row.shown) }} {{ digitText(row.expected) }} {{ digitText(row.response) }} {{ resultLabel(row.result) }} {{ errorsText(row) || '—' }}

{{ rulesNote }}

No counted trials to plot.

Loading length chart.

Recall positionTestedCorrectSubstitutedOmittedTransposedIntrusions before positionCopy
{{ row.tested ? row.position : 'After last' }} {{ row.tested }}{{ row.correct }}{{ row.substituted }}{{ row.omitted }}{{ row.transposed }}{{ row.intrusions }}
Started (UTC)DirectionRate (ms)Start lengthMisses allowedSpan (digits)Correct / countedEnded byCopy
{{ directionName(row.direction) }}{{ row.rate_ms }}{{ row.start_length }}{{ row.misses_allowed }}{{ row.span === null ? '—' : row.span }}{{ row.correct }} / {{ row.counted }}{{ row.ending.replace('Ended after ', '') }}

Similar tools

Keeping a verification code in mind while switching screens requires remembering both the digits and their order. A digit span task examines immediate serial recall by presenting a sequence for someone to reproduce from memory. As sequences become longer, an exact answer requires retaining more digits without losing their positions.

Short-term memory refers to keeping information available briefly. Working memory also involves using or manipulating that information while holding it in mind. Forward recall emphasizes preserving the original sequence. Backward recall adds the requirement to reverse its order. These are related tasks with different demands, so their results should be considered separately.

Recall direction Digits presented Correct response
Forward 5 0 7 5 0 7
Backward 5 0 7 7 0 5

Span is commonly reported as the longest sequence length recalled correctly. That maximum describes a successful attempt under a particular procedure. It does not show how consistently the same length can be recalled. Procedures differ in how many attempts they allow, how they increase length and when they stop, so two results labelled “digit span” may not be directly comparable.

How the digits are remembered matters. Rehearsal means mentally repeating the sequence. Chunking means organizing several digits into a group, such as remembering 5 0 7 8 as 50 and 78. Repetition, grouping and learned associations can change performance without changing the number of digits in the answer. Practice can also develop skill at this particular task; a longer span alone does not establish a general improvement in memory.

Presentation pace changes the opportunity to take in and rehearse each digit. A slower sequence also lasts longer, so its earliest digits must be retained for more time. Distractions can compete for attention, and sleep loss can affect digit recall. Fatigue and reduced alertness are therefore reasons to be cautious when comparing sessions, rather than explanations that can be read directly from a score.

Digit-span practice is useful for exploring recall strategies and comparing your own sessions under similar conditions. A practice result is not a clinical or normed score.

How to Use This Tool:

Choose a recall direction before starting. For useful comparisons, keep the setup and your approach to remembering the digits consistent across sessions.

  1. Choose Forward to enter the original order or Backward to enter the reverse order. With Advanced fields blank, the session starts at 3 digits Forward or 2 digits Backward, uses 1000 ms per digit and allows one retry at each length.
  2. Set any changes under Advanced before starting. Time per digit accepts 400 to 3000 ms, Starting length accepts 2 to 12 digits, and Misses allowed per length accepts 0 to 3. All three require whole numbers.
    If Check settings appears and Start is disabled, correct the out-of-range or fractional value, or clear that field to restore its default.
  3. Select Start and watch the fixation cross followed by the digits appearing one at a time. The session settings remain fixed until the session ends.
    If fixation or presentation is interrupted by a hidden tab, loss of window focus or a detected scheduling gap, the trial is marked Voided. Select Continue for a fresh sequence at the same length and attempt number. There is no Pause control.
  4. Enter your answer when Recall appears and the digits are masked. Digits pressed during presentation are ignored. Use the keypad or keyboard digits, including the number pad, and Backspace to correct the last digit. Recall is untimed. A nonempty answer may be shorter than the sequence but cannot be longer; leading zeros count. Select Submit, or press Enter while the recall area has focus.
  5. Check the expected sequence and the marked differences in your answer, then select Next when ready. A fresh Enter or Space press can also advance while the recall area has focus. Use Stop to finish early and retain a partial result; Stop is unavailable while a voided trial is waiting for Continue.
  6. Read the final span together with the ending reason. Use Session comparison to find previous sessions with the same direction, rate, starting length and miss allowance before judging a change.

Interpreting Results:

The displayed span is the longest length recalled exactly in that session. One successful trial establishes a length, even if an earlier attempt at that length was missed. Correctly remembered individual digits do not earn partial credit toward span.

  • No span yet means there has been no exact response. It is not a zero-digit score. If the starting length was too demanding, shorter lengths remain untested.
  • Ceiling reached means 20 digits were recalled correctly. The session supplies no evidence about longer sequences.
  • Stopped early: partial session preserves the results already submitted. The recorded span may be below a length you could have reached by continuing.

The Length vs correct chart shows counts of correct and missed trials at each tested length, not accuracy percentages. Voided trials are excluded. Because one success advances to the next length, each completed length can have only one correct trial. A higher overall correct-to-counted ratio can reflect a different starting length or retry allowance, rather than better recall.

In Error positions, positions follow the expected answer order. Position 1 in Backward recall corresponds to the last digit shown. Compare errors with the Tested count, since later positions occur only in longer trials and may have fewer observations. The error labels describe how two digit strings align; they cannot distinguish forgetting from a typing slip.

Session comparison can contain rows with different settings and early stopping. Match the four setup values and check the ending reason, then consider several sessions using the same device, browser and strategy. A larger span after slowing presentation or adding retries does not isolate a change in recall skill.

Technical Details:

Exact serial recall requires the right digits in the right order. The scoring decision compares the whole answer with the expected sequence. Error alignment is a separate description of the differences and does not soften the exact-match requirement.

Formula Core

For a session with at least one exact response, span is the maximum length among the correctly recalled trials.

S = max { Li : Ai = Ei }

S is the span in digits, Li is the length of submitted trial i, Ai is its answer and Ei is its expected sequence. For Forward recall, the expected sequence is the order shown; for Backward recall, it is the reversed order. If there is no exact response, the maximum is undefined and no span is assigned. Lengths and trial totals are whole-number counts with no rounding.

Each submitted answer counts as either correct or missed. A void contributes only to the separate voided total. After each miss, the miss count at the current length increases before it is compared with the allowance.

Trial outcome Consequence
Exact recall below 20 digits Establish that span, increase the next length by one digit and reset the miss count.
Exact recall at 20 digits End the session at its ceiling.
Miss count at most the allowance Allow another attempt at the same length with a fresh sequence.
Miss count greater than the allowance End the session and retain the longest earlier exact recall, if any.
Voided trial Exclude it from scored trials and position tallies. It consumes no attempt or miss allowance.

An allowance of 0 ends the session on the first miss at a length. Allowances of 1, 2 and 3 end it on the second, third and fourth miss respectively. A success at any permitted attempt moves immediately to the next length. Stopping with an answer still unsubmitted records that trial as voided and preserves previous counted trials.

Presentation timing

Every trial begins with a planned 500 ms fixation. The selected interval runs from one digit onset to the next and includes both the visible digit and a fixed 200 ms blank. With interval I and visible time V both expressed in milliseconds:

V = I − 200

At the default 1000 ms interval, the planned display is 800 ms of digit followed by 200 ms of blank. The final digit also has this blank interval before recall begins. Entering the answer has no time limit and does not produce a response-speed score.

These values describe the requested schedule, not a calibrated measurement of screen exposure. Browser rendering and display refresh can change actual presentation timing. Skipped presentation phases can void a trial, but that check does not detect every timing deviation. Focus and visibility checks apply during fixation and presentation, not during untimed recall.

Digit selection

Sequences use digits 0 to 9, and a leading zero is valid. Within each trial, digits are selected without repetition until all ten have been used, after which selection starts from a fresh set of ten. Sequences up to ten digits therefore contain no repeats; longer sequences can repeat digits across sets. Adjacent identical digits and three-digit runs rising or falling by one, such as 123 or 321, are excluded. Consecutive trials do not reuse the same complete sequence.

A new random seed starts each session, and the seed together with the progression determines the sequences. The sequence restrictions affect the task and should be considered when comparing it with procedures that allow unrestricted digit repetition.

Error alignment

A missing digit need not make every later digit wrong. Optimal string alignment, or OSA, finds a minimum-cost alignment between the expected sequence and the submitted answer using the following operations.

Mark Meaning Edit cost
✓ Correct The aligned digit matches. 0
× Substituted A different digit occupies the aligned position. 1
− Missing An expected digit is omitted. 1
+ Extra A response digit is inserted before an expected position or after the end. 1
↔ Swapped Two adjacent digits are transposed. 1

OSA does not allow the same part of a sequence to be edited more than once. When equally cheap alignments exist, the choice is resolved from the end of the strings, preferring a match, then a swap, substitution, omission and intrusion. This alignment rule is the same in both recall directions.

A swap has one edit cost but marks two recall positions as transposed. At each expected position, the tested total consists of correct, substituted, omitted and transposed outcomes. Intrusions are counted separately at their insertion location. These categories describe a consistent alignment of the answer, not a reconstruction of the person's memory process.

Privacy Notes:

Digit generation, response checking and scoring take place in your browser without sending trial answers to a scoring service. The recall task needs no camera, microphone or location permission. Displaying the chart may download external resources.

  • The latest 20 session summaries, including stopped sessions, can be saved in this browser. They retain the setup, time, span, totals and ending information, rather than the full shown sequences and answers.
  • The detailed trial record belongs to the current session. Starting a new session or reloading discards it. After the session ends or you select Stop, use Download session JSON to retain the complete record and its random seed before leaving or starting again.
  • Only setup settings enter the URL. Sharing the page address does not share the trial answers or result.
  • If browser storage is unavailable, comparisons remain in the current tab only. Clear saved sessions asks for confirmation and leaves the current result intact. Check the storage message after clearing, since a failed deletion may leave saved summaries available again after reload.

Worked Examples:

A backward session with a retry and an interruption

Set Backward recall, a starting length of 3 and a miss allowance of 1. Suppose the session has an exact three-digit answer, a miss at four digits, then an exact four-digit answer on the retry. At five digits, the first answer is missed, the next presentation is voided by an interruption, and the fresh sequence after Continue is also missed. The result is a span of 4 digits with 2 of 5 counted trials correct. Six trials appear in the ledger because the void is recorded separately. The session ends after two misses at five digits; the interruption did not use the remaining attempt.

An extra digit in an answer of the right length

For a Forward trial showing 5 0 7 8, submitting 9 5 0 7 is a miss even though four digits were entered. The alignment identifies an extra 9 before recall position 1, matches 5 0 7, and marks the final 8 as missing. Its edit cost is 2: one intrusion and one omission. This explains how an extra digit can appear in feedback despite the answer-length limit. Review the marked differences before the next fresh trial rather than treating all four positions as substitutions.

References: