Peak-End Rule

The Peak-End Rule is a principle of retrospective evaluation: when people judge an experience as a whole, they rely mainly on its emotional peak and its final moments rather than integrating every moment or weighting the episode by its duration. Its importance lies in the gap it exposes between what the experiencing self undergoes and what the remembering self later recalls, learns from, and uses to make decisions.

Definition and Measurement

The rule emerged from a comparison between two ways of measuring experienced utility. A duration-weighted measure treats an episode as an area under a curve: every moment contributes to the total, so a longer period at a given level of pain or pleasure counts more than a shorter period at the same intensity. Retrospective evaluation works differently. The global assessment is well predicted by the average of the experience’s most painful or pleasurable moment and its intensity at the end. Duration, by contrast, may have little or no influence on the remembered evaluation.

This distinction separates the experiencing self, which registers whether something hurts or feels good now, from the remembering self, which answers how the episode was on the whole. The remembering self does not preserve a complete record of successive moments. It constructs a summary organized around a few especially salient points, with the peak and ending exerting an influence far greater than their share of the episode’s actual duration.

Colonoscopy Evidence

The clearest demonstration came from a study designed by Daniel Kahneman and Don Redelmeier and carried out by Redelmeier, a physician and researcher at the University of Toronto, in the early 1990s. Participants undergoing painful colonoscopies reported their pain every 60 seconds on a scale from zero, meaning no pain, to 10, meaning intolerable pain. The study included 154 patients. Procedures varied greatly, from four minutes to 69 minutes; the two profiles used to illustrate the findings lasted eight minutes for patient A and 24 minutes for patient B.

A duration-weighted calculation indicated that patient B suffered more. Patient B spent at least as much time as patient A at every level of pain, and the longer procedure therefore produced the larger area under the curve. When the procedures ended, however, participants were asked to rate the total amount of pain they remembered. Their judgments did not reproduce the moment-by-moment totals. Instead, the statistical analysis showed the Peak-End Rule together with duration neglect: retrospective ratings were predicted by the average of the worst moment and the ending, while the length of the procedure had no effect.

The contrast between the two patients made the mechanism visible. Both reached a worst rating of 8. Patient A’s final reading was 7, whereas patient B’s final reading was only 1. Their peak-end averages were therefore 7.5 for A and 4.5 for B. Patient A, whose procedure was much shorter, retained the worse memory because the episode ended at a bad moment. Patient B endured far longer pain but received a gentler ending, and that ending lowered the remembered severity of the entire event.

The result creates two systematically different measures of utility. The hedonimeter total records the actual sequence of reported moments and gives duration its proper weight. The retrospective assessment compresses the episode, giving exceptional weight to two singular moments. The difference is not merely a technical discrepancy: it determines which aspect of an experience a person later treats as the experience itself.

Cold-Hand Experiment and Choice

The rule also predicts behavior, not merely verbal reports. In the cold-hand experiment, participants immersed a hand to the wrist in painfully cold water while continuously recording their momentary pain with their free hand. Each person endured two episodes. The short episode lasted 60 seconds in water at 14 degrees Celsius, after which the participant was told to remove the hand and was offered a warm towel. The long episode began identically but lasted 90 seconds; during its final 30 seconds, slightly warmer water flowed into the tub, reducing the pain just enough for most participants to notice.

From the standpoint of the experiencing self, the long episode was plainly worse because it contained the same first 60 seconds plus another 30 seconds of discomfort. The Peak-End Rule and duration neglect nevertheless predicted that the long episode could be remembered more favorably: its ending was less painful, and the additional duration would be largely ignored. The prediction was confirmed. Among participants who reported that their pain diminished during the final phase, 80 percent chose to repeat the longer episode when later offered a choice between repeating the short or the long experience.

The choice exposed a conflict between experienced utility and decision utility. Participants knew which episode had lasted longer and would have selected the shorter option if the alternatives had been described directly as a 90-second immersion versus only its first 60 seconds. Instead, they chose the episode they remembered as less aversive. The rule governing the remembered evaluation therefore governed the later decision, leading them to volunteer for 30 seconds of needless pain.

The participants were not seeking pain. Their mistake arose because memory supplied the values used in choice. The remembering self treated the more tolerable ending as evidence that the longer episode had been better, even though the experiencing self had endured strictly more discomfort. The experiment thus shows how an evaluation rule can transform a retrospective distortion into a future action.

Consequences for Experience and Well-Being

The Peak-End Rule has practical consequences whenever the aim is to improve either actual experience or its later memory. If physicians want to reduce patients’ memory of pain, lowering the peak intensity may matter more than shortening the procedure, and gradual relief may be preferable to abrupt relief because a mild ending leaves a better retrospective impression. If the aim is instead to reduce pain actually experienced, a swift procedure may be preferable even when it produces a higher peak and a terrible final moment. The two objectives can point in opposite directions.

The rule also explains why an unpleasant ending can seem to have ruined an otherwise good episode. A listener once described a long symphony that was scratched near the end and produced a shocking sound; the ending, he said, had ruined the whole experience. Yet the earlier musical enjoyment had already occurred and could not literally be undone. What the bad ending damaged was the memory and retrospective grade, not the preceding experience. Because the experiencing self has no enduring voice, the remembering self keeps the score and determines what the person learns from the episode.

This distinction shaped later work on experienced well-being. Daniel Kahneman initially questioned global life-satisfaction judgments because they address the remembering self, then turned toward measures of the experiencing self. Experience sampling interrupts people at random points during the day to record what they are doing, who is present, and the intensity of feelings such as happiness, tension, anger, worry, engagement, and physical pain. The Day Reconstruction Method offers a more practical alternative: participants reconstruct the previous day as a sequence of episodes, rate feelings for each episode on separate zero-to-six scales, identify their activities and companions, and report when each episode began and ended.

These methods preserve a crucial contrast. Episode-by-episode reports can be combined into duration-weighted measures of daily affect, whereas a global life evaluation remains a judgment by the remembering self. Large surveys, including the Gallup World Poll, consequently compare the well-being people experience while living their lives with the judgment they make when evaluating life as a whole. The Peak-End Rule belongs to the broader account of why those two perspectives cannot simply be assumed to coincide.

Significance and Limits

The rule does not claim that every judgment is determined by only two instants, nor that duration is irrelevant to lived experience. The colonoscopy profiles show that duration contributes substantially to the hedonimeter total. The claim is narrower and more consequential: when people remember and evaluate an episode globally, the peak and the ending receive disproportionate psychological weight, while duration can disappear from the summary.

Its deeper significance is that memory is not a neutral archive of experience. It is a selective construction that compresses extended events into evaluations usable for later choice. This compression can be adaptive when salient moments genuinely summarize an episode, but it can also produce systematic errors, as in the cold-hand participants’ decision to repeat the longer exposure. The remembered past then becomes a guide to the future, even when the guide misrepresents what actually happened moment by moment.