Reasoning
Rule Discovery Task
Also called: 2-4-6 task, Wason 2-4-6 task, Wason's rule discovery task, number triples task, triples task, hypothesis testing task
Measure how people test their ideas: participants find the rule behind 2 4 6 by testing triples of their own, saying before each test whether they expect it to fit.
5-8 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
- 10
- Evidence grade
- D
- Registered clinical studies
- 1
What it measures
Measures how people test their ideas about a hidden rule. Participants learn that 2 4 6 fits a simple rule about three numbers, then type triples of their own; before each test they say whether they expect it to fit, and they then see whether it does. When sure of the rule they sort 8 new triples. The main score is the share of tests expected to fit, the pull toward confirming tests; the scores also say whether the rule was found.
Constructs
- Hypothesis testing
- Positive test strategy
- Inductive reasoning
- Rule discovery
Research uses
- Reasoning and decision-making research
- Hypothesis testing and confirmation research
- Scientific thinking and education research
- Within-study group comparison
How the paradigm works
The rule discovery task, often called the 2-4-6 task, is Wason's (1960) test of how people test their ideas. Participants are told that 2 4 6 fits a simple rule and find the rule by proposing triples of their own, hearing after each whether it fits. The rule is three numbers in increasing order, but 2 4 6 suggests narrower ideas, such as numbers going up by 2 or even numbers in a row. Klayman and Ha (1987) call a triple a person expects to fit a positive test and one they expect not to fit a negative test. Every positive test of an idea narrower than the rule fits the rule, so positive tests can never show such an idea to be wrong; only a negative test can. People rely mostly on positive tests, and first announcements of the rule are right for about one person in five, as Branchini and colleagues (2023) summarize earlier studies. Prompts that point to opposites raise success: in Branchini and colleagues' (2025) study, 78% of people trained to think in opposites found the rule within three announcements, against 37% of controls.
More about the paradigm
Here, the screen says that 2 4 6 fits a simple rule about how three numbers relate, not about how large they are, after Wason's instructions. The participant types three whole numbers from 0 to 999 and, before each test, presses one of two buttons to say whether they expect the triple to fit; the next screen says whether it fit, and the triples tested so far stay listed with their results. The participant presses "I know the rule" when sure, and testing ends after 15 triples otherwise (5 to 30 as a setting). Instead of announcing the rule in words, the participant then sorts 8 new triples, one at a time, as fitting the rule or not, with no feedback. The 8 triples were written for this task so that the hidden rule and the common narrower ideas sort them differently; only the hidden rule sorts all 8 rightly. The rule is never shown. There is no time limit.
The main score is the share of tested triples the participant expected to fit, Klayman and Ha's positive tests. The scores also give whether all 8 check triples were sorted as the rule sorts them, how many were, the triples tested, how many did not fit, and how many results differed from what the participant expected, each of which disproved the idea behind it. After Branchini and colleagues (2023, 2025) they count the triples going down and give the position of the first, and count the triples with two different steps and with an odd number. Scores describe how a person tested ideas in this task; they are research measures and do not diagnose.
Scores
- Positive Test Rate%
- The share of tested triples the participant expected to fit the rule, said before each test.
- Higher means more tests of triples expected to fit; neither direction is better
- Rule Found0/1
- 1 when all 8 check triples were sorted as the hidden rule sorts them, 0 when not.
- 1 means the participant's rule sorted every check triple as the hidden rule does
- Check Triples Sorted Rightlycount
- Check triples sorted as fitting or not as the hidden rule sorts them, out of 8.
- Higher means a rule closer to the hidden one
- Triples Testedcount
- Triples tested before the check.
- Informational
- Triples That Did Not Fitcount
- Tested triples that did not fit the hidden rule.
- Higher means more tests outside the rule
- Unexpected Resultscount
- Tested triples whose result was not the one the participant expected. Each one disproved the idea behind the test.
- Informational
- Descending Triplescount
- Tested triples whose numbers go down, the test of the rule's direction.
- Informational
- First Descending Triplecount
- The position of the first tested triple whose numbers go down; empty when none was tested.
- Lower means the rule's direction was tested sooner
- Unequal-Step Triplescount
- Tested triples whose two steps differ, unlike 2 4 6.
- Informational
- Triples With an Odd Numbercount
- Tested triples with at least one odd number, unlike 2 4 6.
- Informational
- Median Test Timems
- Median time from a test screen appearing to the expectation button, over tests at or after 100 ms.
- Describes pace; the test has no time limit
- Median Check Timems
- Median time from a check triple appearing to its answer, over answers at or after 100 ms.
- Describes pace; the test has no time limit
- Check Answers Under 100 mscount
- Check answers given less than 100 ms after the triple appeared. They still count.
- Expected to be 0
- Check Triples Answeredcount
- Check triples answered in the scored run.
- 8 in a full session
Use and evidence
10 papers from 2021 to 2025 name this task in their title or abstract. That ranks it 456 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 1 |
| 2024 | 2 |
| 2023 | 3 |
| 2022 | 2 |
| 2021 | 2 |
| 2020 | 2 |
| 2019 | 3 |
| 2018 | 2 |
| 2017 | 2 |
| 2016 | 0 |
| 2015 | 3 |
| 2014 | 3 |
| 2013 | 5 |
| 2012 | 5 |
| 2011 | 2 |
| 2010 | 2 |
| 2009 | 4 |
| 2008 | 4 |
| 2007 | 0 |
| 2006 | 2 |
| 2005 | 4 |
| 2004 | 3 |
| 2003 | 5 |
| 2002 | 1 |
| 2001 | 0 |
| 2000 | 0 |
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
- 0
- Registered clinical studies using it as an outcome
- 1
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.
- Rule
- Three numbers in increasing order, Wason's rule. The screen says only that 2 4 6 fits a simple rule about how three numbers relate, not about how large they are.
- Tests
- Three whole numbers from 0 to 999, typed in three boxes, then one of two buttons: I expect they fit, or I expect they do not fit. The next screen says whether the triple fit, and the triples tested so far stay listed with their results.
- End of testing
- The participant presses I know the rule, from the second test on, or testing ends after 15 triples (5 to 30 as a setting).
- Check
- 8 triples written for this task, sorted one at a time as fitting the rule or not, in an order drawn from the session's seed, with no feedback. Only the hidden rule sorts all 8 rightly; going up by 2, even numbers, equal steps and other narrower ideas each miss some. The rule is never shown.
- Practice
- None. A practice rule or a worked triple would suggest ideas, or a balance of tests, which is what the task measures.
- Scores
- The share of tests expected to fit, Klayman and Ha's positive tests; whether all 8 check triples were sorted rightly, and how many; triples tested, triples that did not fit and results that differed from the expectation; triples going down and the position of the first, and triples with unequal steps or an odd number.
The task follows Wason (1960), with two changes. Whether the rule was found is read from how 8 check triples are sorted, not from a rule written in words, and each test carries the participant's own expectation, so Klayman and Ha's positive and negative tests are counted without coding written reasons. Numbers are whole numbers from 0 to 999, so ideas about negative numbers or fractions cannot be tested. 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: the scored run's sorts of the 8 check triples, each as fitting the rule or not. Accuracy is not judged against chance, and the same answer throughout is accepted. No triple tested invalidates and fewer than 3 is flagged. A run with fewer than half its check triples answered is invalid and any other shortfall is flagged. Answers under 200 ms on at least half the check triples invalidate.
Why these rules
Most people settle on a narrower rule than the hidden one, so they sort some check triples wrongly, and a narrow rule can give the same answer to nearly every check triple: a low or one-sided result is what the task measures, not a sign of careless answering. The tests themselves have no right answer. 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.
No percentile. Scores and quality flags only.
Running this test
Settings you can change
| Setting | Standard | Allowed |
|---|---|---|
| Most triples to test | 15 | 5 to 30 |
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 Rule Discovery Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Wason, 1960; Klayman & Ha, 1987; Branchini, Burro & Bianchi, 2023), with the standard settings (most triples to test: 15). The main score was positive test rate (the share of tested triples the participant expected to fit the rule, said before each test).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Wason, P. C. (1960). On the failure to eliminate hypotheses in a conceptual task. Quarterly Journal of Experimental Psychology, 12(3), 129-140. https://doi.org/10.1080/17470216008416717
- Klayman, J., & Ha, Y.-W. (1987). Confirmation, disconfirmation, and information in hypothesis testing. Psychological Review, 94(2), 211-228. https://doi.org/10.1037/0033-295X.94.2.211Positive and negative tests, and why positive tests cannot disprove a rule narrower than the real one.
- Branchini, E., Burro, R., & Bianchi, I. (2023). Training people to think in opposites facilitates the falsification process in Wason's rule discovery task. Journal of Intelligence, 11(5), 91. https://doi.org/10.3390/jintelligence11050091Triples going down as the test of the rule's direction.
- Branchini, E., Bianchi, I., & Burro, R. (2025). Thinking in opposites improves hypothesis testing performance in Wason's rule-discovery task. Memory & Cognition, 53(6), 1927-1943. https://doi.org/10.3758/s13421-025-01691-3Triples with unequal steps or odd numbers, and how often people found the rule with and without training.
Source papers
- Branchini E, Bianchi I, Burro R (2025). Thinking in opposites improves hypothesis testing performance in Wason's rule-discovery task. Memory & Cognition, 53(6), 1927-1943.Paradigm source, Adults
Questions
How long does the Rule Discovery Task take?
About 5 to 8 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?
Positive Test Rate, Rule Found, Check Triples Sorted Rightly, Triples Tested, Triples That Did Not Fit, Unexpected Results, Descending Triples, First Descending Triple, Unequal-Step Triples, Triples With an Odd Number, Median Test Time, Median Check Time, Check Answers Under 100 ms and Check Triples Answered. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Rule Discovery Task used in research?
10 papers from 2021 to 2025 name it in their title or abstract. That ranks it 456 of 495 test families in the ConductCognition test census (use grade E). Source: ConductCognition test census 0.22.0, counted 3 November 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Rule Discovery Task have?
In the published research, 0 papers report its test-retest reliability and 1 registered clinical study uses 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?
Most triples to test (5 to 30). 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.
Related tests