Working Memory
Dual N-Back Task
Also called: dual-task n-back, audiovisual n-back, position and sound n-back
Measure working-memory updating for two streams at once: a square's place and a spoken digit, each compared with n steps back, at a level set before the session.
6-8 minTimed: desktop or laptop with a mouse or trackpad
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
- 64
- Evidence grade
- C
- Registered clinical studies
- 6
What it measures
Measures working memory updating for two streams at once (Jaeggi and colleagues, 2003). A square appears in one of eight places while a computer voice says a digit, one pair every 3 seconds; participants press A when the place is the same as n steps back and L when the digit is. The level is set before the session and never changes. Scores are the sensitivity (d prime) for places and for digits; results are research measures.
Constructs
- Working memory updating
- Divided attention
- Spatial working memory
- Auditory working memory
Research uses
- Brain imaging and EEG studies of working memory load
- Dual-task and divided attention research
- Within-study group comparison
How the paradigm works
The dual n-back task measures working memory updating for two streams at once: where a square was and which number was spoken, both kept track of as new ones arrive. A white square appears in one of eight places around a cross while a computer voice says a digit, one pair every 3 seconds; the square shows for half a second. The participant presses A when the place is the same as n steps back and L when the digit is, both keys when both match, and nothing otherwise.
More about the paradigm
Jaeggi and colleagues (2003) ran n-back tasks of spoken and seen material alone and as dual tasks at set levels from 0-back to 3-back, each stimulus for 500 ms and the next 3 seconds after it, and found prefrontal activity rising with load in both. The streams and blocks here follow the dual n-back the same group used as a measure (Jaeggi et al. 2010): a square in one of eight places with a spoken item, one key for places and another for sounds, and blocks of 20 + n steps with 6 matches in each stream, 2 of them at once. The level is set before the session (2-back as standard; 1 to 4) and never changes with performance: there is no staircase and no training mode. The dual n-back is best known from training studies (Jaeggi et al. 2008), and a placebo-controlled study found no transfer of such training to other tests (Redick et al. 2013); this test measures performance in one session and claims no improvement.
A practice of 10 + n steps with feedback after each comes first, then three blocks (2 to 4 as a setting), each drawn from the session's seed. The digits are 1, 2, 3, 5, 6, 8 and 9, spoken by a synthetic voice in the participant's language: zero and seven have two syllables in English, and a speech recognizer misheard the Japanese four, so they are left out. Each digit plays from the browser's audio clock and the square appears when it is heard. The sound plays through speakers or headphones after a loudness step. Desktop or laptop with a keyboard; about 6 to 8 minutes.
The main score is d-prime averaged over the two streams, as Jaeggi and colleagues (2010) averaged their measure over both modalities. d-prime for places and for digits, hit and false alarm rates, hits minus false alarms (their measure), median response times to matches and anticipations (presses under 100 ms) are also given. Scores support research and do not diagnose.
Scores
- d-Primed'
- The mean of the place and digit d-primes.
- Higher is better
- Place d-Primed'
- Sensitivity to place matches: z(hit rate) minus z(false alarm rate) for the A key, each rate with the log-linear correction.
- Higher is better
- Digit d-Primed'
- Sensitivity to digit matches: z(hit rate) minus z(false alarm rate) for the L key, each rate with the log-linear correction.
- Higher is better
- Place Hit Rate%
- Place matches pressed with A at or after 100 ms, out of all place matches.
- Higher is better
- Place False Alarm Rate%
- Steps with no place match pressed with A, out of those steps.
- Lower is better
- Digit Hit Rate%
- Digit matches pressed with L at or after 100 ms, out of all digit matches.
- Higher is better
- Digit False Alarm Rate%
- Steps with no digit match pressed with L, out of those steps.
- Lower is better
- Hits Minus False Alarms%
- Hit rate minus false alarm rate, averaged over places and digits, the measure of Jaeggi and colleagues (2010).
- Higher is better
- Median Place Response Timems
- Median time from the square to A on the place matches pressed.
- Informational
- Median Digit Response Timems
- Median time from the start of the spoken digit to L on the digit matches pressed.
- Informational
- Anticipationscount
- Presses under 100 ms in either stream, counted as neither hits nor false alarms.
- Lower is better
- Blocks Completedcount
- Test blocks with steps recorded in the scored run.
- Informational
- Steps Showntrials
- Test steps recorded in the scored run.
- Under 90% of the planned steps is invalid
- Steps Plannedtrials
- Test steps the run set out to show.
- Informational
Use and evidence
64 papers from 2021 to 2025 name this task in their title or abstract, down 18% compared with 2016 to 2020. That ranks it 275 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 10 |
| 2024 | 12 |
| 2023 | 10 |
| 2022 | 15 |
| 2021 | 17 |
| 2020 | 21 |
| 2019 | 17 |
| 2018 | 8 |
| 2017 | 14 |
| 2016 | 18 |
| 2015 | 7 |
| 2014 | 4 |
| 2013 | 10 |
| 2012 | 9 |
| 2011 | 2 |
| 2010 | 2 |
| 2009 | 1 |
| 2008 | 0 |
| 2007 | 0 |
| 2006 | 0 |
| 2005 | 0 |
| 2004 | 0 |
| 2003 | 1 |
| 2002 | 0 |
| 2001 | 0 |
| 2000 | 0 |
Use grade D
30 to 99 papers from 2021 to 2025.
Evidence grade C
At least 2 test-retest papers, 30 reliability or validity papers or 3 registered clinical studies.
- Papers on test-retest reliability
- 0
- Papers on reliability or validity
- 8
- Registered clinical studies using it as an outcome
- 6
Note on the counts: includes training studies.
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 0.22.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Streams
- A white square in one of eight places around a central cross, and at the same moment a digit spoken by a synthetic voice in the participant's language: 1, 2, 3, 5, 6, 8 or 9.
- Timing
- Each square for 500 ms; the next pair starts 3,000 ms after it. The digit plays from the browser's audio clock and the square appears when it is heard. Presses count from then to the end of the step.
- Rule
- Press A when the place is the same as n steps back and L when the digit is; both keys when both match, nothing otherwise.
- Level
- 2-back (standard; 1 to 4), set before the session. It never changes with performance: no staircase and no training mode.
- Blocks
- Three blocks (standard; two to four), each of 20 + n steps with 6 place matches and 6 digit matches, 2 of them at once, at seeded steps after the first n.
- Practice
- 10 + n steps with 3 matches in each stream, 1 at once, each followed by 1,200 ms of feedback on both streams. Each test block ends with the share of matches pressed for.
- Sound
- A loudness step comes first, with no headphone check, so speakers or headphones both work. Digits play at -12 dB re the loudness tone.
- Score
- d-prime averaged over places and digits, from hits and false alarms with the log-linear correction. d-prime, hit and false alarm rates and median hit time for each stream, and hits minus false alarms.
The dual n-back is best known as a training task. This version measures one session at one level set beforehand and claims no improvement. The digits are spoken by synthetic voices, so no published norms apply. Presses under 100 ms count as anticipations, neither hits nor false alarms. Scores support research and do not diagnose.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: dual_nback_test from the scored run, each counted once for places (press A on the 6 of every 20 + n whose place is the one n back) and once for digits (L, 6 more, 2 of them on the same trials), withholding otherwise. Fewer than 90% of the planned trials invalidates. Responding mainly on non-matches invalidates; not separating matches is flagged. Missed matches are the measure: only a run of misses covering half the matches (minimum 8) is flagged and withholds the percentile. Responses under 100 ms follow the timed-test rule.
Why these rules
The hit rate is half of d-prime, the primary measure. Responses concentrated on trials that needed none (one-sided Fisher exact test, p < .05) mean the instructions were reversed and invalidate; responses that do not separate the two trial types are flagged. Missed targets are what this test measures, so missed responses never invalidate on their own. A session with no responses is invalid. An unbroken run of misses covering at least half the response trials (minimum 8) may mean the participant stopped, so the session is flagged and its percentile withheld. Responses under 100 ms follow the timed-test rule. The 90% completion limit is a ConductScience choice.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| N-back level | 2 | 1 to 4 |
| Blocks | 3 | 2 to 4 |
Yearly plans can change settings within these limits. Free and pay-as-you-go studies use the standard settings.
Need a setting this test does not have?
Devices and languages
- Timed test: participants need a desktop or laptop with a mouse or trackpad. Participant devices
- Languages: English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese.
- Available on: Every plan, including the free plan.
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Dual N-Back Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Jaeggi et al., 2003; Jaeggi et al., 2010), with the standard settings (n-back level: 2; blocks: 3). The main score was d-prime (the mean of the place and digit d-primes).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Jaeggi, S. M., Seewer, R., Nirkko, A. C., Eckstein, D., Schroth, G., Groner, R., & Gutbrod, K. (2003). Does excessive memory load attenuate activation in the prefrontal cortex? Load-dependent processing in single and dual tasks: Functional magnetic resonance imaging study. NeuroImage, 19(2), 210-225. https://doi.org/10.1016/S1053-8119(03)00098-3Dual n-back tasks at set levels, each stimulus 500 ms and the next 3 seconds after it.
- Jaeggi, S. M., Studer-Luethi, B., Buschkuehl, M., Su, Y.-F., Jonides, J., & Perrig, W. J. (2010). The relationship between n-back performance and matrix reasoning: Implications for training and transfer. Intelligence, 38(6), 625-635. https://doi.org/10.1016/j.intell.2010.09.001The streams and blocks: eight places and a spoken item, a key for each, 20 + n steps with 6 matches in each stream, 2 at once.
- Jaeggi, S. M., Buschkuehl, M., Jonides, J., & Perrig, W. J. (2008). Improving fluid intelligence with training on working memory. Proceedings of the National Academy of Sciences, 105(19), 6829-6833. https://doi.org/10.1073/pnas.0801268105Background: the training study that made the task known, and the A and L keys.
- Redick, T. S., Shipstead, Z., Harrison, T. L., Hicks, K. L., Fried, D. E., Hambrick, D. Z., Kane, M. J., & Engle, R. W. (2013). No evidence of intelligence improvement after working memory training: A randomized, placebo-controlled study. Journal of Experimental Psychology: General, 142(2), 359-379. https://doi.org/10.1037/a0029082Background: no transfer of dual n-back training to other tests.
- Kirchner, W. K. (1958). Age differences in short-term retention of rapidly changing information. Journal of Experimental Psychology, 55(4), 352-358. https://doi.org/10.1037/h0043688The n-back task.
- Hautus, M. J. (1995). Corrections for extreme proportions and their biasing effects on estimated values of d'. Behavior Research Methods, Instruments, & Computers, 27(1), 46-51. https://doi.org/10.3758/BF03203619The log-linear correction.
Source papers
- Jaeggi SM, Seewer R, Nirkko AC, Eckstein D, Schroth G, Groner R, Gutbrod K (2003). Does excessive memory load attenuate activation in the prefrontal cortex? Load-dependent processing in single and dual tasks: functional magnetic resonance imaging study. NeuroImage, 19(2):210-225.Paradigm source, Young adults
- Jaeggi SM, Studer-Luethi B, Buschkuehl M, Su YF, Jonides J, Perrig WJ (2010). The relationship between n-back performance and matrix reasoning: implications for training and transfer. Intelligence, 38(6):625-635.Paradigm source, Young adults
Questions
How long does the Dual N-Back Task take?
About 6 to 8 minutes, as listed in the test library.
What device does a participant need?
Timed test: participants need a desktop or laptop with a mouse or trackpad. The browser must be Chrome, Edge, Firefox or Safari.
What scores does it produce?
d-Prime, Place d-Prime, Digit d-Prime, Place Hit Rate, Place False Alarm Rate, Digit Hit Rate, Digit False Alarm Rate, Hits Minus False Alarms, Median Place Response Time, Median Digit Response Time, Anticipations, Blocks Completed, Steps Shown and Steps Planned. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Dual N-Back Task used in research?
64 papers from 2021 to 2025 name it in their title or abstract, down 18% compared with 2016 to 2020. That ranks it 275 of 495 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.22.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Dual N-Back Task have?
In the published research, 0 papers report its test-retest reliability and 6 registered clinical studies use it as an outcome (evidence grade C in the ConductCognition test census).
Which settings can a researcher change?
N-back level (1 to 4) and blocks (2 to 4). Settings can be changed on yearly plans; free and pay-as-you-go studies use the standard settings.
Which languages can participants use?
Participants can take this test in English, Spanish, French, German, Italian, Portuguese (Brazil), Chinese (Simplified) and Japanese, with its words in the same language. 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