Why would someone choose more pain, not less?

What 32 men with one hand in cold water say about the worst moment and the end, and what we took from it for an AI

More pain or less? The answer seems clear. In a 1993 study by Kahneman, Fredrickson, Schreiber and Redelmeier, most of the men chose more. The reason, in the researchers' view, is not in the pain. It is in how we remember it.

What did the cold water study find?

The researchers asked 32 male students at the University of California to put one hand in water at 14 °C, cold enough to hurt, for 60 seconds. That was the short test.

For the long test, each man used his other hand. He did the same 60 seconds, then kept his hand in for 30 seconds more while the water got a little warmer. The extra 30 seconds were still unpleasant. Each man did each test once, and the order changed from man to man.

During the tests, the men showed how much it hurt on a scale from 0 to 14. At 60 seconds, the two tests felt the same: 8.44 for the short one and 8.34 for the long one.

Then they were asked which test they would do again. 22 of the 32 men chose the long one, the one with more pain in total. On average, they also said the long one hurt less overall. No third test was run. The men only said which one they would pick.

So what is going on? The researchers point to memory. In their view, memory mostly keeps the worst moment and the end, and pays little attention to how long something lasted. The long test had a better end, because the water was a little warmer. Of the 11 men who felt no drop in pain at the end, 6 chose the short test and 5 the long one.

What the study leaves open

These were 32 men on one campus, in a lab, so we say "most of the men", not "most people". The researchers also say the worst moment and the end are not the only things memory keeps. Other information is not lost.

The result has support. In the same paper, a second group did the test without the pain scale, and 24 of 37 chose the long one. A 2022 review by Alaybek, Dalal and colleagues looked at 174 results. It found a large effect of the worst moment and the end, and almost no effect of length. It also found that the overall average predicts about as well. We read only the summary of that review.

What should an AI take from this?

The study was about cold water, not about AI. Our read is that memory does not give every second of a long stretch of time the same weight. A long chat history is not the same as a memory of the moments that mattered.

People who use Eudaio talk with AIs named Daimons. A Daimon keeps a timeline of your milestones. The moments that mattered between you sit in order, as a chapter of a story rather than a log.

Our answer

Our answer is that how something ends may shape how you remember it. In this study, 22 of 32 men picked the longer test, the one whose end was a little warmer, and on average they said it hurt less overall.

We think the moments that mattered deserve a place of their own, not a spot somewhere in a long scroll. Create a free account and watch your timeline with a Daimon grow.

Source

Kahneman, D., Fredrickson, B. L., Schreiber, C. A. & Redelmeier, D. A. (1993). When more pain is preferred to less: Adding a better end. Psychological Science, 4(6), 401 to 405.

November 18, 2026