Executive Function
Water Jars Task
Also called: Luchins water jar task, water jug problem, water jar test, Einstellung task, mental set task
Measure whether a method practised on five water jar problems carries over to problems that a shorter method also solves, read from every fill, pour and empty step.
8-12 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
- 15
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures whether a method practised on earlier problems carries over to later problems that a shorter method also solves (Luchins, 1942). Participants fill, pour and empty three jars with buttons to get an exact amount of water. Five problems are solved only by one long method, then four test problems also allow a shorter one, with one problem in the middle that only the shorter method solves. Every step is recorded, and the method is read from the steps. The main score is the share of solved test problems solved with the long method.
Constructs
- Mental set (Einstellung)
- Problem solving
- Cognitive flexibility
- Strategy use
- Executive function
Research uses
- Problem solving and mental set research
- Learning and expertise research
- Time pressure and stress research
- Within-study group comparison
How the paradigm works
The water jars task measures whether a method practised on earlier problems carries over to later problems that a shorter method also solves, which Luchins (1942) called Einstellung. Each problem shows three jars of different sizes and an amount of water to get in one of them, with as much water as needed. Under each jar, buttons fill it to the top, empty it, or pour it into another jar until the first is empty or the second is full; a button that would change nothing is disabled. A problem ends when a jar holds the amount, or after 180 seconds (standard, 60 to 300).
More about the paradigm
A practice problem with two jars comes first, with a hint and then its solution. Five set problems follow that only one long method solves: fill jar B, then pour from it into A once and into C twice, B - A - 2C. Then come four test problems that this method solves but a shorter one, A - C or A + C, solves too, with an extinction problem after the first two that only the shorter method solves. The jar sizes were written for this test by a rule that a search over every route of up to five steps checks, and none is one of Luchins' problems. A setting runs the control version without the set problems, as Luchins' control groups had.
Every step is saved with its time. Each problem is scored by replaying its steps, with the water in each jar tracked as a sum of the jar sizes, so the route that made the amount shows the method. The main score is the share of solved test problems solved with the long method. Scores also give that share before and after the extinction problem, whether the extinction problem was solved and how long it took, the set and test problems solved, solutions by any other route, the median time to solve, the time to the first step and the steps made.
In Luchins' first experiment, as Binz and Schulz (2023) reconstruct it from his tables, 128 of 155 people who had worked the set problems used the long method on the first two test problems and 100 on the last two, against none of 155 in the control group. Working with real jars at a sink lowered this carry-over compared with problems on paper (Vallée-Tourangeau, Euden and Hearn, 2011), so jars worked with buttons may show less of it than Luchins' written problems. Scores support research and do not diagnose.
Scores
- Long Method on Test Problems%
- Share of the solved test problems solved with the long method practised on the set problems, B - A - 2C.
- Higher means more use of the practised method
- Long Method, First Two Test Problems%
- The same share for the two test problems before the extinction problem.
- Higher means more use of the practised method
- Long Method, Last Two Test Problems%
- The same share for the two test problems after the extinction problem.
- Higher means more use of the practised method
- Extinction Problem Solvedflag
- 1 when the problem that only the shorter method solves was solved in its time, 0 when not.
- Informational
- Extinction Problem Timems
- Time from the extinction problem appearing to the step that solved it; empty when unsolved.
- Informational
- Test Problems Solvedcount
- Test problems solved in their time, of four.
- Informational
- Set Problems Solvedcount
- Set problems solved in their time, of five; empty in the control version.
- Informational
- Long Method on Set Problems%
- Share of the solved set problems solved with the long method; empty in the control version.
- Informational
- Other Routescount
- Solved problems whose amount came by neither the long method nor a shorter one.
- Informational
- Problems Solvedcount
- Problems solved in their time.
- Informational
- Timeoutscount
- Problems ended by their time limit.
- Informational
- Time to Solvems
- Median time from a problem appearing to the step that solved it, over the solved problems.
- Informational
- Time to First Stepms
- Median time from a problem appearing to its first step. A first step under 100 ms is left out.
- Informational
- Stepscount
- Fill, empty and pour steps made over the problems.
- Informational
- Anticipationscount
- First steps under 100 ms after a problem appeared.
- Lower is better
- Completed Problemscount
- Problems with readable jars and a replayed step log.
- Fewer than half the planned problems invalidates the session
Use and evidence
15 papers from 2021 to 2025 name this task in their title or abstract, up 7% compared with 2016 to 2020. That ranks it 429 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 3 |
| 2024 | 3 |
| 2023 | 5 |
| 2022 | 2 |
| 2021 | 2 |
| 2020 | 1 |
| 2019 | 5 |
| 2018 | 4 |
| 2017 | 1 |
| 2016 | 3 |
| 2015 | 3 |
| 2014 | 1 |
| 2013 | 6 |
| 2012 | 1 |
| 2011 | 4 |
| 2010 | 1 |
| 2009 | 2 |
| 2008 | 0 |
| 2007 | 2 |
| 2006 | 0 |
| 2005 | 0 |
| 2004 | 1 |
| 2003 | 0 |
| 2002 | 1 |
| 2001 | 1 |
| 2000 | 1 |
Use grade E
Fewer than 30 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 2 November 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Problem
- Three jars of different sizes and an amount to get in one jar, with as much water as needed. A step fills a jar, empties it, or pours it into another until the first is empty or the second is full.
- Order
- 5 set problems that only B - A - 2C solves, then 2 test problems that A - C or A + C also solves, 1 extinction problem that only the shorter method solves, and 2 more test problems.
- Jars
- Written for this test by a rule that a search over every route of up to 5 steps checks. None is one of Luchins' problems.
- Control
- A setting leaves out the 5 set problems, as Luchins' control groups did.
- Limit
- A problem ends at the goal or after 180 seconds (standard, 60 to 300). A button that would change nothing is disabled.
- Practice
- 1 problem with 2 jars, with a hint, then its solution.
- Scoring
- Every step is saved with its time. The method of each solution is read from its replayed steps, with each jar's water tracked as a sum of the jar sizes.
Luchins' participants wrote their answers on paper. Here the jars are worked with buttons, and working real jars lowered the carry-over in one study, so this test's shares are not on Luchins' scale. The 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: water_jars_problem from the scored run, one per problem, each replayed from its steps. A run with fewer than half its planned problems is invalid and any other shortfall is flagged. No step on at least half the problems invalidates, and on any problem is flagged. First steps under 200 ms on at least half the problems invalidate. Fewer than 3 of the five set problems solved, and fewer than 2 test problems solved, are flagged.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. Both methods solve the test problems, so the method used is a choice and is not checked against an answer. The long method share compares people who practised it, so a run where most set problems went unsolved may not show its pull, and a share from one test problem is coarse. Making no step at all is not an attempt. 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. The time limit keeps running while the tab is hidden, so a percentile is withheld after a hidden tab.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Problems | Set problems first (standard) | Set problems first (standard), Control: test problems only |
| Time limit per problem (seconds) | 180 | 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 Water Jars Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Luchins, 1942), with the standard settings (problems: Set problems first; time limit per problem (seconds): 180). The main score was long method on test problems (share of the solved test problems solved with the long method practised on the set problems, B - A - 2C).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Luchins, A. S. (1942). Mechanization in problem solving: The effect of Einstellung. Psychological Monographs, 54(6), i-95. https://doi.org/10.1037/h0093502The water jar problems, their order and the control group.
- Binz, M., & Schulz, E. (2023). Reconstructing the Einstellung effect. Computational Brain & Behavior, 6(3), 526-542. https://doi.org/10.1007/s42113-022-00161-2Luchins' results reanalysed from his tables, with the share of the long method in each of his conditions.
- Vallée-Tourangeau, F., Euden, G., & Hearn, V. (2011). Einstellung defused: Interactivity and mental set. Quarterly Journal of Experimental Psychology, 64(10), 1889-1895. https://doi.org/10.1080/17470218.2011.605151Less carry-over with real jars at a sink than with the problems on paper.
Source papers
- Luchins (1942), as Binz and Schulz (2023) reconstruct his first experiment: 128 of 155 people who worked the set problems used the long method on the first two test problems and 100 on the last two; none of 155 in the control group did.Paradigm source, School and college students
Questions
How long does the Water Jars Task take?
About 8 to 12 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?
Long Method on Test Problems, Long Method, First Two Test Problems, Long Method, Last Two Test Problems, Extinction Problem Solved, Extinction Problem Time, Test Problems Solved, Set Problems Solved, Long Method on Set Problems, Other Routes, Problems Solved, Timeouts, Time to Solve, Time to First Step, Steps, Anticipations and Completed Problems. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Water Jars Task used in research?
15 papers from 2021 to 2025 name it in their title or abstract, up 7% compared with 2016 to 2020. That ranks it 429 of 495 test families in the ConductCognition test census (use grade E). Source: ConductCognition test census 0.22.0, counted 2 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Water Jars 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?
Problems (Set problems first (standard), Control: test problems only) and time limit per problem (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. 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.