In 1993, a group of psychologists asked people to submerge a hand in painfully cold water, twice. In the first trial, the water sat at 14 degrees Celsius for 60 seconds, then the hand came out. In the second, the same 60 seconds at the same temperature, followed by an extra 30 seconds during which the water was allowed to warm slightly, to 15 degrees, still cold, just a little less so. Afterwards, most people chose to repeat the second trial if they had to do it again.
That’s a strange thing to prefer. The second trial contains everything unpleasant about the first, plus 30 more seconds of discomfort. By any reasonable accounting, it is the worse experience. But it doesn’t feel that way in memory, because it ends on a slightly better note, and it turns out endings carry a weight in memory that has very little to do with arithmetic.
The experiment that started it
The study belonged to Daniel Kahneman, working with Barbara Fredrickson, Christopher Schreiber and Donald Redelmeier, and it gave a name to something psychologists now call duration neglect. When people are asked to judge how unpleasant, or how enjoyable, a past experience was, they don’t average their feelings across its length. They rely on something closer to a snapshot: the single most intense moment, and whatever was happening right at the end. How long the whole thing lasted barely factors in. The pattern is usually described as the peak-end rule, and Fredrickson and Kahneman’s original paper on it showed the same distortion in something as passive as watching a short film clip. People weren’t reliving the whole clip in memory. They were reconstructing it from a couple of representative frames.
It isn’t limited to a lab
Cold water and film clips are one thing. What made the finding harder to dismiss was that it showed up in a setting nobody would call artificial. In the mid 1990s, Redelmeier and Kahneman found the same pattern in patients recalling colonoscopies, a procedure that, before routine sedation became standard, was genuinely unpleasant and gave researchers a real, memorable discomfort to track. Patients whose procedures happened to end on a less painful note remembered the whole thing as less unpleasant, regardless of how long the procedure had actually run.
A later randomised trial pushed the question further, because a correlation in memory is one thing and a change in future behaviour is another. Redelmeier, Katz and Kahneman deliberately extended some patients’ colonoscopies by a few minutes of mild, low intensity discomfort at the very end, while leaving everything else the same. Over more than five years of follow up, the patients who’d had that softer ending returned for their next scheduled screening at a meaningfully higher rate than those who hadn’t. Nobody had reduced anyone’s actual pain. They had only changed the last few minutes of it, and that alone measurably changed what people were willing to do afterwards.
The same rule works in reverse
None of this is specific to suffering. Researchers at the University of Canterbury in New Zealand later tested whether pleasant experiences worked the same way, asking people to rate a holiday some time after it ended. The peak-end pattern showed up again, though it wasn’t quite the whole story. An even better predictor turned out to be how the single most memorable or unusual day of the trip had gone, wherever in the trip it happened to fall. What barely mattered, in either version of the analysis, was how long the holiday had actually lasted. A short trip with one extraordinary day and a long one with the same extraordinary day tended to be remembered about the same.
What this means for a whole life
The most unsettling extension of the idea concerns not a holiday or a medical procedure but a life. Ed Diener, Derrick Wirtz and Shigehiro Oishi asked people to compare two imagined biographies: one where a person lived a genuinely happy life and then died suddenly, and another where the same happy life continued for several more years at a milder, still perfectly pleasant level, before ending the same way. People rated the shorter life as the better one. Adding years of real, if modest, happiness to the end of a life made the whole life sound worse in the retelling, because those milder years dragged the ending down from where the peak had been.
I think this is the part of the finding that’s actually about happiness, rather than about memory as a curiosity. Kahneman has drawn a distinction between what he calls the experiencing self, the one having the feelings moment to moment, and the remembering self, the one who gets asked afterwards how it all went and has to produce a verdict. Most of the time, when someone asks whether we’re happy, they’re really asking the remembering self, and the remembering self is a strange, selective narrator. It isn’t lying. It just isn’t counting the way we assume it’s counting.
What I’d take from this
I wrote recently about the World Happiness Report, and the fact that its ranking measures one considered judgement about your life rather than how cheerful you tend to feel day to day. Reading the peak-end research alongside that made something click for me. Those are two different instruments measuring two different things, and neither one is wrong, they’re just answering different questions. Asking how happy a period of your life was and asking how it felt to live through is not the same question, even though we tend to use the same word, happiness, for both answers.