Decision Making
Peak-End Task
Also called: peak-end rule, peak-end effect, duration neglect, remembered utility, retrospective evaluation, adding a better end
Measure how the memory of an episode follows its worst and last moments: participants rate rounds in which points are taken away moment by moment and choose which round to repeat, scored as the end effect.
5-7 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
- 88
- Evidence grade
- D
- Registered clinical studies
- 0
What it measures
Measures how memory of an episode follows its worst and last moments. In 12 short rounds, points are taken away a few at a time while the participant rates each moment on a slider; after each round they rate the whole round, and after each pair they choose which round to do again. The pairs differ in their ending, their peak, their length, or a milder stretch added at the end. The main score is the end effect: how much worse a round is rated when the same moments end at the worst one.
Constructs
- Peak-end rule
- Duration neglect
- Remembered utility
- Retrospective evaluation
Research uses
- Judgment and decision making research
- Research on memory for experiences
- Behavioral economics research
- Within-study group comparison
How the paradigm works
The Peak-End Task measures how the memory of an episode follows its worst and last moments. Kahneman, Fredrickson, Schreiber and Redelmeier (1993) had people hold one hand in water at 14 °C for 60 seconds, and the other hand for the same 60 seconds and 30 seconds more while the water warmed to 15 °C, still unpleasant but milder; asked which trial to repeat, a significant majority chose the longer one. Whole-episode ratings of film clips then turned out to be a weighted average of moments, with little effect of how long the clips lasted (Fredrickson and Kahneman, 1993), and with unpleasant sounds, people strongly preferred episodes improved by a milder stretch at the end (Schreiber and Kahneman, 2000). With painful heat and pressure, the last intensity and the trend in the second half carried the most weight, and rating each moment as it happened changed the later evaluation (Ariely, 1998). Do, Rupert and Wolford (2008) found the rule held for material goods as well as pain, and when episodes are lists of numbers people neglect their length, as they do not with graphs (Liersch and McKenzie, 2009).
More about the paradigm
Here the episodes are rounds in which points are taken away a few at a time: mild, and never pain. Each moment shows its loss, 10 to 90 points, for 1.25 seconds. With the standard setting the participant keeps a slider at how bad each moment feels, and the slider at the end of each moment is that moment's rating. After each round the participant rates the whole round on a bar from not bad at all to very bad, and after each pair of rounds chooses which round to do again. The 6 pairs each differ in one way: added end (a round, then the same round with three milder moments added at the end, twice), order (the same six moments rising or falling, twice), peak (six moments of 50, or a spike of 90 with the same total and end) and duration (four moments, or the same four each shown twice). The session starts with 5,000 points, and one choice, drawn at random, is played again at the end with its points taken too. A practice round comes first.
The main score is the end effect: the rating of a round that ends at its worst moment minus the rating of the same moments ending at the mildest, averaged over the two order pairs. The peak effect, the duration effect (0 means the length did not change the rating, duration neglect) and the added-end effect follow, with the shares of choices that took the longer added-end round, the round that ended mildest, the round with the lower average of its worst and last moments, and the round with fewer points taken. Two correlations show whether the ratings followed the average of the worst and last moments or the points taken. Scores support research and do not diagnose.
Scores
- End Effectscore
- Mean over the two order pairs of the rating of the round ending at its worst moment minus the rating of the same moments ending at the mildest, on the 0 to 100 bar.
- Positive: the end pulled the rating; 0: the order made no difference
- Peak Effectscore
- Rating of the round with a spike of 90 points minus the flat round with the same total, length and end.
- Positive: the worst moment pulled the rating
- Duration Effectscore
- Rating of the round shown at twice the length minus the short round: the same worst, last and average moments over twice the points.
- 0: the length made no difference (duration neglect); positive: the longer round seemed worse
- Added-End Effectscore
- Mean over the two added-end pairs of the short round's rating minus the rating of the same round with three milder moments added at the end.
- Positive: the longer round, with more points taken, seemed less bad
- Longer Round Chosen%
- Share of the added-end pairs in which the longer round was chosen to do again.
- Informational
- Milder End Chosen%
- Share of the order pairs in which the round ending at its mildest moment was chosen to do again.
- Informational
- Lower Peak-End Chosen%
- Share of choices that took the round with the lower average of its worst and last moments, over the pairs where the averages differ.
- Informational
- Fewer Points Chosen%
- Share of choices that took the round with fewer points taken, over the pairs where the totals differ.
- Informational
- Ratings and Peak-Endscore
- Pearson correlation over the rated rounds between the ratings and the average of each round's worst and last moments.
- Informational
- Ratings and Points Takenscore
- Pearson correlation over the rated rounds between the ratings and the points each round took.
- Informational
- Moment Ratings and Pointsscore
- Pearson correlation between the slider at the end of each moment and that moment's points.
- 0 or below is flagged
- Slider Movescount
- Times the moment slider changed during the task rounds.
- 0 is flagged
- Mean Round Ratingscore
- Mean rating of the rounds on the 0 to 100 bar.
- Higher means the rounds seemed worse
- Median Rating Timems
- Median time from a rating screen appearing to Confirm.
- Informational
- Anticipationscount
- Ratings and choices made within 100 ms of their screen appearing; such ratings are left out.
- 0 expected
- Rounds Ratedtrials
- Task rounds rated, out of 12.
- Informational
- Points Takenpoints
- Points taken in the task rounds and the round done again; the practice round's do not count.
- Informational
Use and evidence
88 papers from 2021 to 2025 name this task in their title or abstract, up 21% compared with 2016 to 2020. That ranks it 231 of 495 test families in our census. See the counts for every test.
Show the numbers
| Year | Papers |
|---|---|
| 2025 | 18 |
| 2024 | 21 |
| 2023 | 19 |
| 2022 | 16 |
| 2021 | 14 |
| 2020 | 16 |
| 2019 | 18 |
| 2018 | 12 |
| 2017 | 13 |
| 2016 | 14 |
| 2015 | 6 |
| 2014 | 6 |
| 2013 | 7 |
| 2012 | 6 |
| 2011 | 6 |
| 2010 | 6 |
| 2009 | 6 |
| 2008 | 9 |
| 2007 | 2 |
| 2006 | 2 |
| 2005 | 1 |
| 2004 | 2 |
| 2003 | 0 |
| 2002 | 0 |
| 2001 | 0 |
| 2000 | 2 |
Use grade D
30 to 99 papers from 2021 to 2025.
Evidence grade D
Little published measurement evidence found.
- Papers on test-retest reliability
- 1
- Papers on reliability or validity
- 15
- 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.
- Rounds
- Points are taken away a few at a time: each moment shows its loss, 10 to 90 points, for 1.25 seconds, then a 0.25-second gap. A session starts with 5,000 points.
- Moment ratings
- With the standard setting, the participant keeps a slider at how bad each moment feels, and its position at the end of each moment is that moment's rating. A setting leaves the slider out.
- Ratings and choices
- After each round, a rating of the whole round on a bar from not bad at all to very bad, with no numbers shown. After each pair of rounds, a choice of which round to do again; one choice, drawn at random, is played again at the end, and its points are taken too.
- Pairs
- 6 pairs in an order drawn from the session's seed, each differing in one way: a round and the same round with 3 milder moments added at the end (twice), the same 6 moments rising or falling (twice), 6 moments of 50 or a spike of 90 with the same total and end, and 4 moments or the same 4 each shown twice. In 3 pairs, drawn from the seed, the shorter, worsening or flat round comes first.
- Practice
- One practice round of 4 moments, with its rating and a note on whether the slider moved. Its points do not count.
- Scores
- The end effect: the rating of the round ending at its worst moment minus the rating of the same moments ending at the mildest, over the two order pairs. Also the peak, duration and added-end effects, the shares of choices that took the longer added-end round, the milder end, the lower peak-end average and fewer points, and how the ratings correlate with the peak-end average and with the points taken.
- Checks
- Fewer than half the rounds rated, or the same rating on every round, is invalid. A slider that never moved, a slider that did not follow the points, and choices always in the same position are flagged and never exclude.
The classic studies used cold water, films and unpleasant sounds; here the episodes are points taken away, mild and never pain, so the effects may differ in size. Each effect rests on one or two pairs of rounds, so compare groups or conditions. Points are a setting: the payment line says whether the study pays for them. These are research measures of how episodes are remembered, 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: peak_end_rating from the scored run, one rating of each whole round from 0 (not bad at all) to 100 (very bad), with no correct answer, so accuracy is not judged against chance. Fewer than half the planned rounds rated, or the same rating on every round, invalidate, and any other incomplete run is flagged. A moment slider that never moved, a slider that did not rise with the points of each moment, and choices that always picked round A or always round B are flagged for review. Ratings confirmed within 200 ms of the rating screen appearing on at least half the rounds invalidate.
Why these rules
There is no correct option to compare with chance, so accuracy is not judged against chance. The scores are differences between rounds, so the same exact rating for rounds that take 190 to 540 points separates nothing and the session is invalid. A slider left still or moved against the points can be honest, since some people find every loss of points equally mild, and the round ratings do not depend on it, so it only flags the session. Choices in one position throughout may still be real preferences, so they flag and never exclude. 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 |
|---|---|---|
| Moment ratings | Slider during the rounds (standard) | Slider during the rounds (standard), No moment ratings |
| 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 Peak-End Task in a web browser (ConductCognition task version 1.0.0, scoring version 1.0.0; Kahneman D, Fredrickson BL, Schreiber CA, Redelmeier DA, 1993), with the standard settings (moment ratings: Slider during the rounds; payment line: Hypothetical points). The main score was end effect (mean over the two order pairs of the rating of the round ending at its worst moment minus the rating of the same moments ending at the mildest, on the 0 to 100 bar).
ConductCognition versions: task version 1.0.0, scoring version 1.0.0. Every export records the versions each session used, for a methods section.
- Kahneman, D., Fredrickson, B. L., Schreiber, C. A., & Redelmeier, D. A. (1993). When more pain is preferred to less: Adding a better end. Psychological Science, 4(6), 401-405. https://doi.org/10.1111/j.1467-9280.1993.tb00589.xA short episode and the same episode with a milder end added, and the choice of which to repeat.
- Fredrickson, B. L., & Kahneman, D. (1993). Duration neglect in retrospective evaluations of affective episodes. Journal of Personality and Social Psychology, 65(1), 45-55. https://doi.org/10.1037/0022-3514.65.1.45Ratings of each moment during an episode and of the whole episode afterwards; duration neglect.
- Schreiber, C. A., & Kahneman, D. (2000). Determinants of the remembered utility of aversive sounds. Journal of Experimental Psychology: General, 129(1), 27-42. https://doi.org/10.1037/0096-3445.129.1.27Moment ratings that track intensity, and the preference for episodes improved by a milder end.
- Ariely, D. (1998). Combining experiences over time: The effects of duration, intensity changes and on-line measurements on retrospective pain evaluations. Journal of Behavioral Decision Making, 11(1), 19-45. https://doi.org/10.1002/(SICI)1099-0771(199803)11:1<19::AID-BDM277>3.0.CO;2-BThe weight of the end and the trend, and why moment ratings are a setting.
- Do, A. M., Rupert, A. V., & Wolford, G. (2008). Evaluations of pleasurable experiences: The peak-end rule. Psychonomic Bulletin & Review, 15(1), 96-98. https://doi.org/10.3758/PBR.15.1.96
- Liersch, M. J., & McKenzie, C. R. M. (2009). Duration neglect by numbers: And its elimination by graphs. Organizational Behavior and Human Decision Processes, 108(2), 303-314. https://doi.org/10.1016/j.obhdp.2008.07.001Episodes shown as numbers, one moment at a time here.
- Alaybek, B., Dalal, R. S., Fyffe, S., Aitken, J. A., Zhou, Y., Qu, X., Roman, A., & Baines, J. I. (2022). All's well that ends (and peaks) well? A meta-analysis of the peak-end rule and duration neglect. Organizational Behavior and Human Decision Processes, 170, 104149. https://doi.org/10.1016/j.obhdp.2022.104149
Source papers
- Kahneman, Fredrickson, Schreiber and Redelmeier (1993): a significant majority chose to repeat the longer cold-water trial. Fredrickson and Kahneman (1993): with 32 participants rating film clips, the effects of film duration on the retrospective evaluations were small.Paradigm source, Adults
Questions
How long does the Peak-End Task take?
About 5 to 7 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?
End Effect, Peak Effect, Duration Effect, Added-End Effect, Longer Round Chosen, Milder End Chosen, Lower Peak-End Chosen, Fewer Points Chosen, Ratings and Peak-End, Ratings and Points Taken, Moment Ratings and Points, Slider Moves, Mean Round Rating, Median Rating Time, Anticipations, Rounds Rated and Points Taken. Each session is also marked valid, needs review or invalid, with the reason.
How widely is the Peak-End Task used in research?
88 papers from 2021 to 2025 name it in their title or abstract, up 21% compared with 2016 to 2020. That ranks it 231 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 Peak-End Task 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?
Moment ratings (Slider during the rounds (standard), No moment ratings) 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.