Working Memory
Operation Span Task
Measure complex working memory by combining simple processing judgments with serial letter recall.
15-25 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Try it here first, with no account.
- Papers, 2021 to 2025
- 237
- Evidence grade
- B
- Registered clinical studies
- 24
What it measures
Measures working memory capacity with the automated operation span procedure. Participants check simple math problems while remembering letters, then recall the letters in order.
Constructs
- Working memory capacity
- Executive attention
- Serial recall
- Processing under time pressure
Research uses
- Individual-difference studies of working memory capacity
- Cognitive training and intervention outcome studies
- Studies relating working memory to attention control and reasoning
- Remote research batteries with mouse or touch input
How the paradigm works
Operation Span measures complex working memory, the ability to keep letters in mind while solving math problems. On each item the participant solves a short problem, judges whether a shown answer is true or false, and then sees a letter to remember. After each set, the participant clicks the letters in the order shown on a 4 by 3 letter grid, using a Blank button for any letter they forgot.
More about the paradigm
The task follows the automated operation span procedure of Unsworth et al. (2005). Practice has three parts: letters only, math only, and both together. The math practice sets each participant's time limit for later problems at their mean solving time plus 2.5 standard deviations. The test has 15 sets, three each of 3 to 7 letters in random order, for 75 letters and 75 problems. Each letter shows for 800 ms. A problem that runs past the time limit counts as a speed error and the task moves on to the letter.
Scoring reports the partial span score (letters recalled in the correct position, 0 to 75), the absolute span score (letters from sets recalled perfectly), and math errors split into speed and accuracy errors. Sessions with math accuracy below 85% are flagged.
Scores
- Partial Span Scorecount
- Letters recalled in the correct position across all 15 sets (0 to 75).
- Higher is better
- Absolute Span Scorecount
- Sum of set sizes for sets recalled perfectly (0 to 75).
- Higher is better
- Partial-Credit Unit Score%
- Mean proportion of letters recalled in the correct position per set.
- Higher is better
- Math Accuracy%
- Problems answered correctly within the time limit, out of all 75 problems.
- Below 0.85 flags the session
- Speed Errorscount
- Problems not solved within the participant's time limit.
- Lower is better
- Accuracy Errorscount
- Problems where the true or false judgment was wrong.
- Lower is better
- Low Math Accuracyflag
- 1 when math accuracy is below 85%; review before analysis.
- Informational
- Math Time Limitms
- The participant's time limit per problem: mean practice solving time plus 2.5 SD.
- Informational
Use and evidence
237 papers from 2021 to 2025 name this task in their title or abstract, down 1% compared with 2016 to 2020. That ranks it 142 of 711 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 62 |
| 2024 | 50 |
| 2023 | 40 |
| 2022 | 41 |
| 2021 | 44 |
| 2020 | 43 |
| 2019 | 47 |
| 2018 | 57 |
| 2017 | 40 |
| 2016 | 53 |
| 2015 | 48 |
| 2014 | 41 |
| 2013 | 49 |
| 2012 | 34 |
| 2011 | 29 |
| 2010 | 34 |
| 2009 | 16 |
| 2008 | 21 |
| 2007 | 19 |
| 2006 | 14 |
| 2005 | 14 |
| 2004 | 17 |
| 2003 | 7 |
| 2002 | 7 |
| 2001 | 7 |
| 2000 | 4 |
Use grade C
100 to 299 papers from 2021 to 2025.
Evidence grade B
At least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers.
- Papers on test-retest reliability
- 8
- Papers on reliability or validity
- 52
- Registered clinical studies using it as an outcome
- 24
Measured on ConductCognition
Pooled across the studies that feed our platform norms, by device group. A figure needs 50 or more people.
No pooled figure yet.
How the grades work
Use: A is 1,000 or more papers from 2021 to 2025, B 300 to 999, C 100 to 299, D 30 to 99 and E fewer than 30. Evidence: A means at least 25 papers report its test-retest reliability and at least 25 registered clinical studies use it as an outcome; B, at least 8 test-retest papers, and either 5 registered clinical studies or 150 reliability or validity papers; C, at least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies; D, little published measurement evidence found. The counts measure use. They include every paper that names the task in its title or abstract. Grades describe the published research on this task, not ConductCognition's version of it.
Source: ConductCognition test census 0.58.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Letters
- F H J K L N P Q R S T Y, each shown for 800 ms.
- Sets
- 15 sets, three each of 3 to 7 letters in random order: 75 letters and 75 problems.
- Problems
- For example (1 × 2) + 1 = ?, then a true or false judgment of a shown answer.
- Time limit
- Each participant's mean solving time in math practice plus 2.5 standard deviations.
- Recall
- Click the letters in order on a 4 by 3 grid, with a Blank button for a forgotten letter.
- Practice
- Letters only, math only, then both together.
- Math criterion
- Sessions below 85% math accuracy are flagged.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: test-phase TRUE or FALSE answers (chance 50%, half true) and speed errors. Math accuracy below chance invalidates and not above chance is flagged. One button on 90% of answers invalidates when accuracy is not above chance. Speed errors are missed responses: more than max(3, 25%) invalidate, any is flagged. Answers under 200 ms on at least half the problems invalidate. Letter recall is not judged against chance.
Why these rules
Accuracy significantly below chance (one-sided exact test, p < .05) cannot come from guessing; it means the response mapping was reversed or the task was not followed (Slick, Sherman and Iverson 1999 treat below-chance forced-choice performance as definite response bias). Chance-level accuracy can be a real result in impaired participants, so it is only flagged. One option on at least 90% of answered trials, when the correct options are balanced and accuracy is not above chance, means the answers did not depend on the items (the task counterpart of long-string responding in surveys; Curran 2016). Typical participants make about one speed error (Unsworth et al. 2005, M 1.15), so more than a quarter signals non-engagement. An answer under 200 ms is faster than a simple visual reaction, so it cannot reflect the item. Following response-time effort screening for untimed tests (Wise and Kong 2005), the session is invalid only when at least half the answers are that fast. A hidden tab lets the study, delay or response timers run without the participant, so a percentile is withheld after one.
Percentiles against a published norm, on paid plans and pay-as-you-go studies.
Running this test
Settings you can change
This test runs with fixed settings.
Need a different version of this test?
Devices and languages
- Untimed test: runs on desktops, laptops, tablets and phones. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- Percentiles against a published norm, on paid plans and pay-as-you-go studies. How norms work
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Operation Span task in a web browser (ConductCognition task version 2.0.0, scoring version 2.0.0; Unsworth, Heitz, Schrock & Engle, 2005; Turner & Engle, 1989; Conway et al., 2005). The main score was partial span score (letters recalled in the correct position across all 15 sets (0 to 75)).
ConductCognition versions: task version 2.0.0, scoring version 2.0.0. Every export records the versions each session used, for a methods section.
- Unsworth, N., Heitz, R. P., Schrock, J. C., & Engle, R. W. (2005). An automated version of the operation span task. Behavior Research Methods, 37(3), 498-505. https://doi.org/10.3758/BF03192720
- Turner, M. L., & Engle, R. W. (1989). Is working memory capacity task dependent? Journal of Memory and Language, 28(2), 127-154. https://doi.org/10.1016/0749-596X(89)90040-5
Source papers
- Unsworth, Heitz, Schrock & Engle (2005). An automated version of the operation span task. Behavior Research Methods, 37(3), 498-505.N 252, 18-35
Questions
How long does the Operation Span Task take?
About 15 to 25 minutes, as listed in the test library.
What device does a participant need?
Untimed test: runs on desktops, laptops, tablets and phones. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
Partial Span Score, Absolute Span Score, Partial-Credit Unit Score, Math Accuracy, Speed Errors, Accuracy Errors, Low Math Accuracy and Math Time Limit. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Operation Span Task used in research?
237 papers from 2021 to 2025 name it in their title or abstract, down 1% compared with 2016 to 2020. That ranks it 142 of 711 test families in the ConductCognition test census (use grade C). Source: ConductCognition test census 0.58.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Operation Span Task have?
In the published research, 8 papers report its test-retest reliability and 24 registered clinical studies use it as an outcome (evidence grade B in the ConductCognition test census).
Does it report a percentile?
Yes. Percentiles against a published norm, on paid plans and pay-as-you-go studies. The session needs standard settings, the participant's age and no quality warnings.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese. English is on every plan, and the other languages come with yearly plans. Have a native speaker read the machine-translated text before launch.
Is this a clinical test?
No. ConductCognition is for research use only. It runs, scores and exports the test; clinical interpretation is outside the platform.
Every test: the full test list.
Run this test in your next study
Start free with 10 free participants on any test, then use participant credits or a yearly plan.
Related tests