Episodic Memory
Zeigarnik Effect Task
Also called: interrupted task paradigm, Zeigarnik paradigm, memory for interrupted tasks, recall of unfinished tasks, Zeigarnik quotient
Do picture puzzles, some stopped by the program before they are finished, then type the ones you remember. Measures whether stopped puzzles are remembered better, a classic result that has proved hard to replicate.
8-11 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
- 36
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures whether puzzles stopped before they are finished are remembered better than finished ones, a classic result that has proved hard to replicate. Participants do a series of picture puzzles, half of them stopped by the program, then type the puzzles they remember. The main score is the share of the recall that went to stopped puzzles.
Constructs
- Memory for interrupted tasks
- Unfinished intentions
- Free recall
- Task involvement
Research uses
- Memory and motivation research
- Research on interruptions and unfinished goals
- Replication research
- Within-study group comparison
How the paradigm works
The Zeigarnik Effect Task tests a classic claim about memory: that tasks stopped before they are finished are remembered better than finished ones. Zeigarnik (1927) gave 32 adults 22 tasks of a few minutes each, let half of them run to the end and stopped the other half, and found that the stopped tasks were recalled about twice as often (MacLeod 2020 describes her studies). The classic result has proved hard to replicate. Van Bergen (1968) found no advantage in close replications, and a meta-analysis of 38 publications found that interrupted and finished tasks were recalled about equally often overall, with a ratio of 0.99 (Ghibellini and Meier 2025). The effect seems to depend on the situation and on the person, such as how involved people are in the tasks. This task gives researchers a standard way to study when it appears.
More about the paradigm
The participant does a series of picture puzzles. Each puzzle is a picture cut into 3 by 3 tiles that start in the wrong places; tapping or clicking two tiles swaps them, and a puzzle is finished when every tile is back in its place. The puzzles come in pairs, 6 in the standard session (4 to 8). In each pair the program stops one puzzle as soon as only one swap is left and lets the other finish, and which one is stopped is drawn for each session, so each picture is stopped in about half the sessions. A stopped puzzle ends when it is nearly done and has been on screen almost as long as a finished one, which reduces a known weakness of the paradigm: finished tasks are usually seen for longer. The instructions say that some puzzles will be stopped, and a line on screen says so at each stop. Nothing says that memory will be tested. A puzzle not finished in 3 minutes ends unsolved and is left out of both kinds.
After the last puzzle, the participant types the puzzles they remember, by what their pictures showed, one at a time, for up to 2 minutes (60 to 300 seconds). Entries are matched to the pictures' names in the session's language with the rule of the word tests: case, accents and articles are ignored, plurals and other common names of the picture count, and a one-letter slip is accepted where it names one picture only. The end screen explains that the stops and the recall were part of the task.
The main score is the share of the recall that went to interrupted puzzles: the recall rate of interrupted puzzles divided by the sum of the two recall rates, so 0.5 means no difference. With equal numbers of each kind this is Marrow's proportion, the measure Ghibellini and Meier recommend. The recall rate of each kind, their difference and Zeigarnik's quotient (interrupted over finished recall) are also reported, with whether the first puzzle typed was an interrupted one and how long each kind was on screen. Scores support research and do not diagnose.
Scores
- Interrupted Share of Recall%
- Recall rate of interrupted puzzles divided by the sum of the interrupted and finished recall rates.
- 0.5 is no difference; above 0.5 interrupted puzzles were recalled more; no better direction
- Interrupted Puzzles Recalled%
- Share of the stopped puzzles recalled.
- Informational
- Finished Puzzles Recalled%
- Share of the finished puzzles recalled.
- Informational
- Recall Difference%
- Interrupted recall rate minus finished recall rate.
- Positive when interrupted puzzles were recalled more
- Zeigarnik Quotientscore
- Interrupted recall rate divided by finished recall rate; null when no finished puzzle was recalled.
- 1 is no difference; sensitive to small counts
- Puzzles Recalledcount
- The session's puzzles recalled, timed-out ones included.
- Informational
- First Recalled Was Interrupted0/1
- 1 when the first puzzle typed was an interrupted one, 0 when it was finished.
- Read across a group as a share
- Practice Puzzle Recalled0/1
- 1 when the practice puzzle was typed.
- Not part of the score
- Other Entriescount
- Entries that name no puzzle of the session.
- Context for the recall
- Repeated Entriescount
- Entries naming a puzzle already typed.
- Informational
- Entries Typedcount
- Entries typed in the recall.
- Informational
- Paste Attemptscount
- Attempts to paste text into the recall, which are blocked.
- Any is flagged
- Time on Interrupted Puzzlesseconds
- Median time from the start of an interrupted puzzle to its stop.
- Exposure, compared with the finished puzzles
- Time on Finished Puzzlesseconds
- Median time from the start of a finished puzzle to its last swap.
- Exposure, compared with the interrupted puzzles
- Puzzles Timed Outcount
- Puzzles that reached the 3-minute limit; they count as neither kind.
- Any is flagged
- Puzzles Recordedcount
- Puzzles recorded in the scored run.
- Informational
Use and evidence
36 papers from 2021 to 2025 name this task in their title or abstract, up 44% compared with 2016 to 2020. That ranks it 336 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 13 |
| 2024 | 11 |
| 2023 | 4 |
| 2022 | 2 |
| 2021 | 6 |
| 2020 | 7 |
| 2019 | 3 |
| 2018 | 2 |
| 2017 | 7 |
| 2016 | 6 |
| 2015 | 6 |
| 2014 | 0 |
| 2013 | 2 |
| 2012 | 0 |
| 2011 | 1 |
| 2010 | 4 |
| 2009 | 2 |
| 2008 | 0 |
| 2007 | 0 |
| 2006 | 0 |
| 2005 | 0 |
| 2004 | 1 |
| 2003 | 1 |
| 2002 | 2 |
| 2001 | 2 |
| 2000 | 1 |
Use grade D
30 to 99 papers from 2021 to 2025.
Evidence grade D
Little published measurement evidence found.
- Papers on test-retest reliability
- 0
- Papers on reliability or validity
- 4
- Registered clinical studies using it as an outcome
- 0
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 3 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Puzzles
- Each puzzle is a Noto Emoji picture in a dark frame, cut into 3 by 3 tiles that start in the wrong places and need 7 swaps to solve. Tapping or clicking one tile and then another swaps them. A puzzle not finished in 3 minutes ends unsolved and counts as neither kind.
- Stops
- The puzzles come in 6 pairs (standard; 4 to 8). In each pair the program stops one puzzle as soon as 1 swap is left and lets the other finish, and the session's seed decides which, so each picture is stopped in about half the sessions. The instructions say that some puzzles will be stopped.
- Practice
- One puzzle needing 5 swaps, which is never stopped.
- Recall
- After the last puzzle, with no warning beforehand, the participant types what the pictures showed, one puzzle at a time, for up to 120 seconds (standard; 60 to 300). Entries are read against the pictures' names in the session's language, with plurals, other common names and one-letter slips accepted.
- Scores
- The interrupted share: the recall rate of stopped puzzles divided by the sum of the two recall rates, 0.5 for no difference. Also each recall rate, their difference, Zeigarnik's quotient and how long each kind of puzzle was on screen.
- Debrief
- The end screen says that the stops and the recall were part of the task, not something the participant did.
Zeigarnik (1927) reported that interrupted tasks were recalled about twice as often as finished ones, but the classic result has proved hard to replicate: a meta-analysis of 38 publications found interrupted and finished tasks recalled about equally often (Ghibellini and Meier, 2025), and the effect seems to depend on the situation and the person. Stopping each puzzle one swap from the end keeps the two kinds on screen for almost the same time. The recall is a surprise only once, so a person's later sessions are flagged. The puzzles, the frame and the stops are our own. These are research measures, read across groups; no result describes a person, and they do not diagnose.
Quality checks
Each session is marked valid, needs review or invalid, with the reason, before you see its scores.
Scored rows: the zeigarnik_effect_recall row of the scored run, typed recall of the puzzles, with that run's zeigarnik_effect_puzzle rows. Typed recall has no chance level and no response is required, so accuracy is not judged against chance. Recalling no puzzle, more entries naming no puzzle than puzzles recalled (with at least 6 of them), blocked paste attempts, puzzles that reached the time limit, fewer puzzles than planned and an earlier session of this test are flagged; the last two withhold the percentile. A hidden tab is flagged.
Why these rules
Recall has no options to guess between, and recalling nothing can be a real result, so it is flagged for review rather than invalidated. Many entries that name no puzzle suggest the instructions were not followed, and pasting could bring in words from elsewhere. A puzzle stopped by the time limit was neither interrupted by design nor finished. The recall is a surprise only once: after one session a person expects it and may rehearse the puzzles as they go, which changes what the test measures. There is no correct option to compare with chance, so accuracy is not judged against chance.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Pairs of puzzles | 6 | 4 to 8 |
| Recall time (seconds) | 120 | 60 to 300 |
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
- 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.
- No percentile. Scores and quality flags only. How norms work
- Recruiting online? See the setup guides for Prolific, Sona Systems, CloudResearch Connect and Qualtrics.
Cite and describe
Participants completed the Zeigarnik Effect Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Zeigarnik B, 1927; Ghibellini R, Meier B, 2025), with the standard settings (pairs of puzzles: 6; recall time (seconds): 120). The main score was interrupted share of recall (recall rate of interrupted puzzles divided by the sum of the interrupted and finished recall rates).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Ghibellini, R., & Meier, B. (2025). Interruption, recall and resumption: A meta-analysis of the Zeigarnik and Ovsiankina effects. Humanities and Social Sciences Communications, 12, 962. https://doi.org/10.1057/s41599-025-05000-wThe meta-analysis, and Marrow's proportion of interrupted among all tasks recalled as the recommended measure.
- MacLeod, C. M. (2020). Zeigarnik and von Restorff: The memory effects and the stories behind them. Memory & Cognition, 48(6), 1073-1088. https://doi.org/10.3758/s13421-020-01033-5Zeigarnik's experiments and the replications that followed.
- Zeigarnik, B. (1927). Das Behalten erledigter und unerledigter Handlungen. Psychologische Forschung, 9(1), 1-85. https://doi.org/10.1007/BF02409755The interrupted task paradigm: half the tasks stopped, then a surprise recall.
- Butterfield, E. C. (1964). The interruption of tasks: Methodological, factual, and theoretical issues. Psychological Bulletin, 62(5), 309-322. https://doi.org/10.1037/h0039873A review of the paradigm, which notes that it does not separate how well tasks were learned from how well they were kept.
Source papers
- Ghibellini R, Meier B (2025). Interruption, recall and resumption: a meta-analysis of the Zeigarnik and Ovsiankina effects. Humanities and Social Sciences Communications, 12:962.Paradigm source
- MacLeod CM (2020). Zeigarnik and von Restorff: The memory effects and the stories behind them. Memory & Cognition, 48(6):1073-1088.Paradigm source
Questions
How long does the Zeigarnik Effect Task take?
About 8 to 11 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?
Interrupted Share of Recall, Interrupted Puzzles Recalled, Finished Puzzles Recalled, Recall Difference, Zeigarnik Quotient, Puzzles Recalled, First Recalled Was Interrupted, Practice Puzzle Recalled, Other Entries, Repeated Entries, Entries Typed, Paste Attempts, Time on Interrupted Puzzles, Time on Finished Puzzles, Puzzles Timed Out and Puzzles Recorded. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Zeigarnik Effect Task used in research?
36 papers from 2021 to 2025 name it in their title or abstract, up 44% compared with 2016 to 2020. That ranks it 336 of 495 test families in the ConductCognition test census (use grade D). Source: ConductCognition test census 0.22.0, counted 3 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Zeigarnik Effect Task have?
In the published research, 0 papers report its test-retest reliability and 0 registered clinical studies use it as an outcome (evidence grade D in the ConductCognition test census).
Does it report a percentile?
No. Scores and quality flags only.
Which settings can a researcher change?
Pairs of puzzles (4 to 8) and recall time (seconds) (60 to 300). 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.