The Colonoscopy pain-memory study was a research project designed by Daniel Kahneman and Don Redelmeier to compare pain as it was experienced moment by moment with pain as it was remembered afterward. Its decisive finding was that retrospective evaluations followed the peak-end rule and ignored duration: patients’ memories were shaped disproportionately by the worst moment and the final moment, rather than by the total amount of pain represented by the full procedure. The result made memory, rather than immediate experience, a central force in later judgment and choice.

Design and Measurement

The study was carried out by Don Redelmeier, a physician and researcher at the University of Toronto, in the early 1990s, when anesthetic and amnesic drugs were less widely used for colonoscopy than they are now. A total of 154 patients took part. During each procedure, patients were prompted every 60 seconds to report the intensity of their pain on a scale from zero, meaning no pain at all, to 10, meaning intolerable pain.

The procedures varied greatly in length. One illustrated patient’s colonoscopy lasted eight minutes, while another’s lasted 24 minutes; across the full sample, the shortest procedure lasted four minutes and the longest 69 minutes. These repeated momentary reports made it possible to construct a duration-weighted measure of pain: the hedonimeter total, or the area under the curve. On that measure, every moment counted, so two minutes of pain at level 9 represented twice as much pain as one minute at the same intensity.

The Reversal in Retrospective Judgment

From the perspective of the moment-by-moment reports, patient B clearly suffered more than patient A. Patient B spent at least as much time as patient A at every level of pain, and the longer procedure produced the larger area under the curve. If total experienced pain were the basis of memory, patient B should therefore have remembered the procedure as worse.

The patients’ retrospective judgments did not reproduce those totals. After the procedure, each participant was asked to rate the total amount of pain experienced. The statistical analysis instead found that the global rating was well predicted by the average of the pain at the worst moment and the pain at the end. This became the peak-end rule. A second finding, duration neglect, showed that the length of the procedure had no effect on the patients’ ratings of total pain.

The contrast between the two illustrated patients made the reversal especially clear. Both reached a peak pain rating of 8. Patient A’s last rating before the end was 7, producing a peak-end average of 7.5. Patient B’s last rating was only 1, producing an average of 4.5. Patient A consequently retained the worse memory, even though patient B had endured the longer and objectively greater quantity of pain. The ending transformed the remembered meaning of the episode: patient A’s procedure ended at a bad moment, while patient B’s ended after the pain had eased.

Experience, Memory, and Choice

The experiment distinguished two measures of experienced utility. The hedonimeter total was calculated from reports of successive moments and was duration-weighted; the retrospective assessment was formed afterward and gave exceptional weight to the peak and the ending. These measures were not merely different ways of expressing the same experience. They systematically produced different evaluations of what had happened.

This difference became more consequential in the related cold-hand experiment, which was designed to place the experiencing self and the remembering self in direct conflict. Participants held a hand in painfully cold water while continuously reporting the pain they were enduring. The short episode lasted 60 seconds at 14 degrees Celsius. The long episode began identically but lasted 90 seconds, with slightly warmer water flowing in during the final 30 seconds, enough for most participants to notice a small reduction in pain.

The long episode was plainly worse for the experiencing self because it contained all 60 seconds of the short episode plus another 30 seconds of pain. Yet 80 percent of participants who reported that the pain diminished during the final phase chose to repeat the long episode when offered a choice between repeating one of the two experiences. Their decision was governed by the less aversive memory created by the gentler ending, while duration neglect caused them to discount the additional 30 seconds. The colonoscopy findings supplied the central logic for this choice: people often select the experience they remember more favorably rather than the one that was less painful while it was occurring.

Consequences for Medical Practice

The study’s implications extended beyond the description of memory. If the goal is to reduce patients’ memories of pain, lowering the peak intensity may matter more than minimizing the procedure’s duration. The same reasoning suggests that gradual relief may be preferable to abrupt relief, because a relatively mild ending can leave a better retrospective impression.

This recommendation does not mean that duration is irrelevant to the patient’s immediate suffering. The momentary reports and their duration-weighted totals preserve the fact that longer pain is generally worse for the experiencing self. Rather, the study revealed that medical procedures have at least two outcomes: what the patient feels during the procedure and what the patient later remembers and uses in future decisions. A treatment can therefore improve remembered pain without reducing the total pain actually endured, or reduce immediate suffering without producing the best retrospective memory if it ends badly.

From Pain Memory to Well-Being

The colonoscopy study became a foundation for a broader distinction between the experiencing self and the remembering self. Confusing an experience with the memory of it creates a powerful cognitive illusion: the remembering self keeps score, determines what is learned from the past, and helps govern future choices, even though it can misrepresent the duration and totality of what the experiencing self endured.

That distinction shaped the later study of experienced well-being. Daniel Kahneman became suspicious of global life-satisfaction questions because they ask the remembering self to evaluate life as a whole. He instead focused on the well-being of the experiencing self, defining happiness in terms of how much time a person spends in activities she would rather continue, how little time she spends wishing to escape, and how little of life is spent in an emotionally neutral state.

To measure such experience, Kahneman and a team of psychologists and an economist developed experience-sampling procedures and the Day Reconstruction Method. Experience sampling interrupts people at random intervals through a mobile phone and asks what they are doing, who is present, and how intensely they feel happiness, tension, anger, worry, engagement, or physical pain. The Day Reconstruction Method asks participants to reconstruct the previous day as a sequence of episodes, report their activities and social contacts, identify what received their attention, and rate feelings on separate zero-to-six scales. Because episodes have beginnings and endings, the method can calculate duration-weighted daily affect while also collecting life-satisfaction judgments associated with the remembering self.

The resulting research separated two aspects of well-being: the well-being people experience as they live and the judgment they make when evaluating their lives. The colonoscopy patients provided the clearest early demonstration of why those measures should not be assumed to coincide. A life, a medical treatment, or a painful episode can be lived in one temporal form and remembered in another, with the peak and the ending exerting a power that the full duration does not possess.