If you're seeing this message, it means we're having trouble loading external resources on our website.

If you're behind a web filter, please make sure that the domains *.kastatic.org and *.kasandbox.org are unblocked.

Main content

Cognitive Biases: Peak-End Effect

In this video, Laurie Santos (Yale University) explains why our memories of good and bad events are a biased. Specifically, she explains how our retrospective evaluations fall prey to the peak-end effect— a bias in which we overweight the peak and end of our everyday experiences— and how this bias leads us to ignore other features of the event like its duration.

Speaker: Dr. Laurie Santos, Associate Professor of Psychology, Yale University.

Want to join the conversation?

  • mr pants teal style avatar for user Anthony Natoli
    How exactly did Kahneman and the doctors working with him "(they) extended the length of the painful procedure so that it worked like this: patients had to be in pain for overall longer ...", as the speaker says at ? Intentionally extending a painful procedure on humans sounds unethical. Did the patients get paid for these extended-painful clinical trials? Did they give INFORMED consent, and not merely blanket consent to a procedure? If the patients were informed that the procedure would be extra painful or intentionally painful for a longer duration, that might negate the results of the "experiment". Patients might think they were experiencing more pain during the procedure, and if not, be relieved at the end, which might not represent what actually would have happened without informed consent. Dunking people's hands in water is not so unethical, but what were these "painful procedures" that "doctors" were subjecting their patients to as mentioned at ?
    (6 votes)
    Default Khan Academy avatar avatar for user
    • female robot grace style avatar for user Inger Hohler
      You can read about the original experiment here https://www.amherst.edu/system/files/media/0678/colonoscopy%202.pdf The patients did not get paid. They were referred to the hospital for real examinations. A pain level of 1.7 (the average score for the "extra pain" group is so low that many people will not take pain medication for it even if it lasts for an hour. In the study the extra duration was for one minute, and the maximum discomfort was 3.7 for that minute. A pain level of 4 for a minute, well, if you are not used to it you definitely think it's quite bad. It cannot be ignored, you probably clench your muscles and it affects your ability to concentrate. On the other hand, if you are used to level 4 pain (and worse) you chuckle at the thought of just one minute of it, and you can relax in spite of it, at least if you are allowed to lie down and work on relaxation. So I'm not surprised this study was accepted by the ethical committee. Many medical procedures or visits to the dentist that we undergo regularly and voluntarily are no less painful and may last longer. The graph shown in the video does not actually represent what happened in the trial. The examination did not end at peak pain for any of the participants, and the difference between peak and end pain was greater.
      (4 votes)
  • male robot hal style avatar for user Wudaifu
    So, it sounds like this video is making the argument that it is better to painfully rip off the dressing over a wound slowly over a longer period of time than to do it very quickly, like my mother used to do, to get the pain over with as fast as possible. Is the slower technique actually preferable?
    (2 votes)
    Default Khan Academy avatar avatar for user
  • blobby green style avatar for user P vinod Menon
    How different is this from writing it down and have a opening balance and spent money and the closing balance-it may not be possible to do all accounting in head -so to say.thanks
    (1 vote)
    Default Khan Academy avatar avatar for user
  • aqualine ultimate style avatar for user nachie14
    Can there be several peaks throughout an event? And if so how would we measure that event in our minds in a case where there were both good and bad peaks?
    (1 vote)
    Default Khan Academy avatar avatar for user
  • male robot hal style avatar for user Wudaifu
    Was anyone else amused that the restaurant "Report Card" at had the category "Desert" listed, which is actually a dry area of land? The sweet treat that we enjoy at the end of a meal is actually called a "Dessert", with 2 "S's".
    (0 votes)
    Default Khan Academy avatar avatar for user

Video transcript

(intro music) My name is Laurie Santos. I teach psychology at Yale University, and today I want to talk to you about peak-end effects. This lecture is part of a series on cognitive biases. I want you to think back to the last time you went to a restaurant and had an expensive meal. Now remember all the different parts of that event. Done? Well, then I want you to answer a very simple question: how was it? No matter whether you said it was just so-so or the best meal ever, I bet you experienced the act of answering this question as relatively simple. Deciding whether some event was good or bad is something we find pretty easy. We can quickly say whether we liked a particular movie, or how much fun we had on vacation, or even how our last job interview went. But making this kind of judgment should involve some pretty complicated computations. To judge how your last restaurant meal was, you might have to computationally combine how you rated a whole bunch of different points in the experience, and you have to judge how each of them fits together, and come up with an overall rating. You also have to decide how many time points you're going to sample to figure out your rating. When you thought about the restaurant, did your mind give you a rating to the experience across every single minute, or every second, or even every millisecond? It turns out that our minds figure out this complicated problem by using a surprisingly streamline strategy. In fact, when we remember an event, we tend to ignore lots and lots of the parts of that event. In fact, we seem to make an evaluation based on only two parts of our experience: the peak (that is, the part of the experience that's most extreme, either good or bad) and the end (that is, how it ended). All of our memories, both good ones and bad ones, are often remembered as such because of the action of these two points. To study this effect, the psychologist Danny Kahneman and his colleagues presented participants with a painful event. They had to stick their hand in really, really cold, fourteen degree water for sixty seconds. After that, the subjects were given a second event. It started out exactly the same as the first. Their hands were placed in really, really cold, fourteen degree water for sixty seconds. But then the second painful event continued. Participants kept their hand in the water for thirty more seconds, as the temperature was increased to a still painful, but not as painful, fifteen degrees. Kahneman and colleagues then asked participants which event they liked better, or which they would prefer to experience again. Surprisingly, participants overwhelmingly preferred the second event. Even though they were in pain for longer, the second event ended a bit better. And when you average the peak and end point of each event, the second one is better, even though it involves overall more pain. Kahneman and colleagues wondered whether the same peak-end effect would work for people's memories in situations that we experience in the real world. They teamed up with doctors who were giving patients a painful medical procedure. Usually, the procedure went like this. It started out as not so painful, and gradually got more and more painful, ending in a moment of maximum pain. Kahneman and his colleagues thought that this procedure would be remembered as particularly awful, not because it was a painful procedure, but because the peak and end point of this procedure were particularly bad. To test this assumption, Kahneman and his colleagues ran a clinical trial. They extended the length of the painful procedure so it worked like this. Patients had to be in pain for overall longer, but the procedure ended on a slightly less painful note. Kahneman and colleagues found that people reported liking the second procedure much better. They reported not feeling as much pain, and were even willing to come back for more repeated procedures. Kahneman and colleagues' clinical trial highlights one of the most surprising things about our peak-end heuristic, namely, the information that it doesn't take into account, and that's how long the event happens to be. Patient would rather be in pain for twice as long, so long as the average of the peak and end is more pleasant. Since our minds only take into account the peak and the end when remembering an event, there's no way to tell how long a particular event was. We just kind of ignore that part of the event in our evaluations. This phenomena is known as "duration neglect." Our duration neglect means that wonderful experiences can stay just as wonderful even when you keep them really short. It also means that you can make somewhat annoying experiences longer, and you won't even remember them that way. The peak-end effect also gives us some hints about how to make an event especially awesome. Since each event's goodness involves only its peak and its end point, we can make events particularly fun by making them end their best point possible. In fact, lots of the events that we enjoy most involve ends that are also peaks. >From the final winning goal to firework finales, saving the best for last seems to make it an event better. So next time you're having a really nice meal, be sure to save some room for dessert Subtitles by the Amara.org community