Many writers and readers alike say that the most important part of a story is the ending. When asked why, they say things like “an ending allows you to form a complete opinion about the work as a whole,” or “the ending is the payoff of what the writer builds during the story.”

These are probably part of the answer, but if the last 30 years of experimental psychology have taught us anything, it’s that we are surprisingly ignorant of why we form beliefs and opinions. A series of experiments by psychologist Daniel Kahneman–one of my personal heroes–and his colleagues Donald A. Redelmeier and Joel Katz suggest a simpler reason why endings are so important, and why as a writer it’s more vital for you to deliver a satisfying ending than almost any other factor.

It starts with a group of men. More specifically, their colons.

The colonoscopy procedure is about as much fun as it sounds. It is painful, intrusive, and embarrassing enough that many people skip the procedure as often as possible even when it’s vital to their long-term health. In their experiment, Kahneman et al. took a group of men who were coming in for a medically needed colonoscopy and divided them into two groups.Both groups were given the procedure and asked to rate their pain and discomfort during the process, and then once again immediately after it was finished.

The first group, the control group, was given the colonoscopy as normal. The second group, the experimental group, was given the colonoscopy as normal, but then the apparatus was left in for a short period after the actual procedure was done. They did not know that this extra period wasn’t part of the procedure, but they did know that the last few minutes of their procedure was considerably less painful than the bulk of it.

During the procedure, the experimental group experienced more total discomfort than the control group, because they had a full-length full-discomfort procedure followed an extra period of reduced discomfort that was still much less fun than less irritations like, for example, getting a root canal. However, when asked to rate the overall unpleasantness of the experience, the experimental group rated it as significantly less bad than the control group. Even though they experienced more total pain, the reduced pain at the end made the entire procedure seem less bad. Not only that, the people in the experimental group wer more likely to come back for a repeat procedure.

In other words, when they thought back on the experience, they mostly remembered the ending. The last thing that happened to them had a higher impact on their impression of the experience than the much longer period that preceded it. Even though their total discomfort was higher. It didn’t matter. The taste it left in their mouth, to use a metaphor that we should all agree not to think about too much in this context, was based on the ending.

These findings have been reproduced in other experiments under different conditions. Kahneman and Redelmeier, their sadistic lust for psychological insight not sated by even the most unpleasant medically necessary procedures, went on to blast people’s ears with unpleasant noises and submerge their hands in freezing water. In both cases, subjects rated the overall experience as less unpleasant as long as it ended more gently, even when the gentle-ending condition was twice as long. When asked which condition they wanted to repeat,the overwhelming majority of subjects chose to repeat the double-length torture period over the shorter one, as long as the long one had a milder finish.

This data stacks up with a great deal of other data from other directions to tell a specific story about the human mind: we massively overvalue whatever happens last in an experience. A fantastic vacation with a crappy final day will be remembered less fondly than a mediocre vacation that went out with a bang, according to survey data. A great dessert will save an otherwise mediocre restaurant meal.

So it may be, then, that the major contributing factor for why endings are so important in stories is simply that they happen last. No matter how carefully you craft the beginning and middle of your story, it just won’t have the same lasting psychological impact as those last few scenes. Another implication here, and I admit I am speculating, is that it might matter less than you think how well the ending ties everything together. It might be more important that the ending is enjoyable for its own sake.

Someone clever writer is probably going to have to torment some hapless readers to find out.