Social Cognition
Third-Party Punishment Game
Also called: third-party punishment task, altruistic punishment game, third-party sanction game
Measure costly punishment by an unaffected observer: participants may spend points to deduct from programmed allocators who split 100 points, scored by spending on unfair splits.
3-6 minUntimed
Free to start: 10 free participants per account, then $0.99 per participant for academic research, or any yearly plan. Preview any test free after you sign up.
- Papers, 2021 to 2025
- 256
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures costly punishment by an unaffected observer: programmed allocators split 100 points with programmed receivers, and the participant may spend points to deduct three times as many from the allocator.
Constructs
- Costly punishment
- Fairness norms
- Social preferences
- Social decision making
Research uses
- Behavioral economics research
- Fairness and social norm studies
- Social decision-making research in clinical groups
- Within-study group comparison
How the paradigm works
The Third-Party Punishment Game measures whether people pay to sanction unfair behavior that does not affect them. In each round an allocator has 100 points and gives 0, 10, 20, 30, 40 or 50 of them to a receiver who starts with none. The participant watches as a third party with 50 points and may spend any of them on deductions: each point spent removes 3 points from the allocator, and the participant keeps the points not spent. Allocators and receivers are computer programs that follow set rules, and by default the instructions say so.
More about the paradigm
The task follows Fehr and Fischbacher (2004): the same points, splits and deduction rate, and instructions that speak of deducting points rather than punishing. As standard each split appears twice, in blocks of the six splits shuffled from the session's seed, each round with a new allocator and receiver, for 12 rounds. A comprehension question with feedback comes first. Researchers can set 1 to 4 rounds per split, use the source's strategy method (one answer for each split before the allocator's choice is known, with a split drawn from the seed picking the answer that counts), use a neutral wording for the allocator and receiver, and choose the payment line.
Scores
- Points Spent on Unfair Splitspoints
- Points spent on deductions when the allocator gave 0 to 40 of the 100 points, averaged over those decisions.
- Higher means more costly punishment of unfair splits
- Points Spent When Nothing Was Givenpoints
- Points spent on deductions when the allocator gave nothing. Means are also reported for 10, 20, 30 and 40 points given.
- Higher means more punishment
- Points Spent on Even Splitspoints
- Points spent on deductions when the allocator gave 50 of the 100 points, an even split.
- Informational
- Spending Rate%
- Share of decisions with any points spent.
- Informational
- Spending Slopescore
- Least-squares slope of the points spent on how far the amount given falls short of 50 points.
- Higher means spending rises more as splits get less even
- Comprehension Check%
- Share of the comprehension questions answered correctly.
- 1 supports interpretation
Use and evidence
256 papers from 2021 to 2025 name this task in their title or abstract, up 72% compared with 2016 to 2020. That ranks it 97 of 395 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 74 |
| 2024 | 54 |
| 2023 | 43 |
| 2022 | 45 |
| 2021 | 40 |
| 2020 | 43 |
| 2019 | 38 |
| 2018 | 23 |
| 2017 | 18 |
| 2016 | 27 |
| 2015 | 23 |
| 2014 | 39 |
| 2013 | 14 |
| 2012 | 18 |
| 2011 | 14 |
| 2010 | 16 |
| 2009 | 16 |
| 2008 | 10 |
| 2007 | 8 |
| 2006 | 3 |
| 2005 | 6 |
| 2004 | 4 |
| 2003 | 0 |
| 2002 | 0 |
| 2001 | 0 |
| 2000 | 0 |
Use grade C
100 to 299 papers from 2021 to 2025.
Evidence grade D
Little published measurement evidence found.
- Papers on test-retest reliability
- 1
- Papers on reliability or validity
- 6
- 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 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
Task design
The trial sequence as the test runs in ConductCognition.
- Roles
- An allocator splits 100 points with a receiver who starts with none, and the participant decides as a third party. Allocators and receivers are computer programs that follow set rules; the standard instructions say so, and a neutral wording leaves that line out.
- Splits
- The allocator gives 0, 10, 20, 30, 40 or 50 points.
- Deductions
- 50 points for the participant in each round. Each point spent removes 3 points from the allocator, and points not spent are kept. Deductions do not change the receiver's points.
- Rounds
- Each split 2 times (standard; 1 to 4), in blocks of the six splits shuffled from the session's seed, each with a new allocator and receiver: 12 rounds.
- Strategy method
- A setting asks once for each split, in ascending order, before the allocator's choice is known. A split drawn from the seed picks the answer that counts.
- Check
- One question with feedback before the first round: the allocator's points after an example deduction drawn from the seed.
- Money
- Points are hypothetical by default; a setting changes the instruction line to say the study pays.
Fehr and Fischbacher (2004) used the strategy method with paid participants in all three roles. Here the allocators and receivers are computer programs, the standard version shows each split directly, and the points are hypothetical unless the study pays.
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 decisions. No amount is correct, so accuracy is not judged against chance, and the same amount on every decision is accepted. A wrong answer to the comprehension question is flagged for review. Decisions confirmed under 200 ms on at least half the decisions invalidate.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. Spending nothing, or the same amount every time, is a stable preference, so repeated amounts are not checked. A participant who misreads how a deduction works may choose the opposite of what they intend, so a wrong answer to the comprehension question is flagged for review and never invalidates. 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 |
|---|---|---|
| Method | Direct: one decision per split shown (standard) | Direct: one decision per split shown (standard), Strategy method: one answer for each possible split |
| Rounds per split | 2 | 1 to 4 |
| Allocator and receiver wording | Computer programs (standard) | Computer programs (standard), Neutral: the allocator and the receiver |
| Payment line | Hypothetical points (standard) | Hypothetical points (standard), The study pays for points |
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 Third-Party Punishment Game in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Fehr E, Fischbacher U, 2004), with the standard settings (method: Direct: one decision per split shown; rounds per split: 2; allocator and receiver wording: Computer programs; payment line: Hypothetical points). The main score was points spent on unfair splits (points spent on deductions when the allocator gave 0 to 40 of the 100 points, averaged over those decisions).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Fehr, E., & Fischbacher, U. (2004). Third-party punishment and social norms. Evolution and Human Behavior, 25(2), 63-87. https://doi.org/10.1016/S1090-5138(04)00005-4
- Henrich, J., McElreath, R., Barr, A., Ensminger, J., Barrett, C., Bolyanatz, A., Cardenas, J. C., Gurven, M., Gwako, E., Henrich, N., Lesorogol, C., Marlowe, F., Tracer, D., & Ziker, J. (2006). Costly punishment across human societies. Science, 312(5781), 1767-1770. https://doi.org/10.1126/science.1127333Costly punishment across 15 populations.
Source papers
- Fehr and Fischbacher (2004): the third-party punishment game with these points and deduction rate. Henrich and colleagues (2006): costly punishment across 15 populations.Paradigm source, Adults
Questions
How long does the Third-Party Punishment Game take?
About 3 to 6 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?
Points Spent on Unfair Splits, Points Spent When Nothing Was Given, Points Spent on Even Splits, Spending Rate, Spending Slope and Comprehension Check. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Third-Party Punishment Game used in research?
256 papers from 2021 to 2025 name it in their title or abstract, up 72% compared with 2016 to 2020. That ranks it 97 of 395 test families in the ConductCognition test census (use grade C). Source: ConductCognition test census 0.14.0, counted 31 October 2026 from OpenAlex and ClinicalTrials.gov.
How much reliability evidence does the Third-Party Punishment Game have?
In the published research, 1 paper reports 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?
Method (Direct: one decision per split shown (standard), Strategy method: one answer for each possible split), rounds per split (1 to 4), allocator and receiver wording (Computer programs (standard), Neutral: the allocator and the receiver) and payment line (Hypothetical points (standard), The study pays for points). 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.