The peak-end rule is the tendency to judge a past experience largely by how it felt at its most intense moment and at its end, while barely counting how long it lasted. It governs what you remember of an experience, not what it was like to live through — which is where the interest lies.
Daniel Kahneman and colleagues, whose 1990s experiments on remembered pain produced both this rule and its companion finding, duration neglect.
// Where it sits
Memory isn't a recording. It's a summary built from the peak and the exit.
What it really means
The experiment that establishes this is genuinely startling, because it makes people choose more pain on purpose.
Subjects held one hand in 14°C water for sixty seconds. Then, separately, they did the same sixty seconds plus thirty more, during which the temperature crept up by a single degree — still unpleasant, slightly less so. Asked which trial they'd rather repeat, most chose the longer one. It contained strictly more total discomfort. It simply ended better.
What memory keeps, it turns out, is roughly two numbers: the peak and the end. How long the experience lasted is discarded almost entirely, which is why the extra thirty seconds cost nothing in the ledger and the improved final moment counted for everything.
The applications are obvious and mostly benign — encores, desserts, the last email in a support thread. But there's a version of this that has drawn published ethical analysis, and it's worth stating rather than skipping. In a randomised medical trial, extending a procedure by up to three less-painful minutes left patients rating the whole thing as less unpleasant and returning for repeat screening more often. That is a real clinical benefit, achieved by giving someone more discomfort than they needed to leave them with a better memory of it. Whether the patient should be told is not a design question. It's a consent question, and the peak-end rule is what makes it one.
Where it comes from
The name is the finding: memory weights two moments — the peak and the end — and largely ignores the rest.
Myths & misconceptions
People remember the average of an experience.
They don't, and the gap is large enough to invert preferences. Duration barely registers at all — this is duration neglect. A trial containing strictly more total pain was preferred, because it ended better.
It's a harmless tip for designing nice endings.
It has an uncomfortable edge, which is why it's now the subject of published medical ethics analysis. If a clinician can lower remembered pain by deliberately extending a procedure, they're trading real experienced discomfort for a better memory — without the patient's knowledge.
Compare & contrast
Peak-end rule vs duration neglect
Two halves of one result rather than rivals. The peak-end rule names what memory keeps — the worst moment and the last moment. Duration neglect names what it discards — how long the whole thing went on. Neither makes sense without the other, and they came from the same experiments.
| // | Peak-end rule | Duration neglect |
|---|---|---|
| Describes | What's kept | What's dropped |
| Data points | Two | Zero for length |
| Discovered | Together | Together |
| Consequence | Endings matter | Length doesn't |
How it connects
- Recency Biasthe end weighs heavily in both — one about sequences, one about experiences.
- Negativity Biasthe worst moment carries weight the rest of the experience does not.
Tell it apart
Questions people ask
- Why do I remember a bad holiday as good?
- Because a single peak and a strong finish can carry the whole memory. Ten mediocre days plus one spectacular final evening usually beats twelve pleasant, uneventful ones in recall — and the extra two days contribute almost nothing to the verdict.
- Should I design services around the ending?
- It's a high-leverage moment — the checkout, the discharge, the last email. But note the honest limit: you're improving the *remembered* experience, and the lived one is a separate thing. A queue made pleasant only at the exit was still a long queue.
- Does it work for good experiences too?
- Yes, which is why concerts save a signature song for the encore and tasting menus put a strong dessert last. The limit worth knowing is that the ending is only half of it: a strong finish is working alongside whatever the worst moment was, not instead of it.
Sources
- Armitage, R. C., 'Prolonging Medical Procedures to Exploit the Peak-End Rule: An Ethical Analysis' — Journal of Evaluation in Clinical Practice 31(6), 2025; an ethical analysis of the Redelmeier/Katz/Kahneman colonoscopy trial, via PMC.
- All's Bad That Ends Bad: There Is a Peak-End Memory Bias in Anxiety (Frontiers in Psychology) — PMC.