Remember Remember...

If we are rational creatures then we need to be able to remember accurately. Interestingly, however, our memory is as complex as a Rubik's cube in a maze.

One rule of thumb initially discovered by Daniel Kahneman is the peak-end rule. He got patients undergoing a painful operation to note down the amount of pain (out of ten) they were in, minute by minute. Thus he was able to create graphs of the sum total of pain patients experienced. For example (my data):

The patients were then asked after the operation to remember how much pain they went through. To be consistent they should have given the average amount of pain they recorded during the operation (here 5.5). But they didn't. Instead they tended to report the average of two points: the most memorable (the peak) and the end of the operation (here 5).

Thus our memory tricks us. We do not remember a how a whole experience (good or bad) was, instead averaging the best/worst bit and the end. This is a challenge to those who defend human rationality. If we do not remember accurately, then how can we make decisions that maximise our personal benefit/satisfaction/goals?

Kahneman then decided to prove the peak end rule (using, in my opinion, ethically dubious methods). He asked the surgeons to extend the operations subjects went through by leaving their instruments in the patient at the end for a couple of minutes. This increased the length of the operation. It also meant that the pain experienced in the final couple of minutes of the operation fell. In accordance with the theory, although the total amount of pain experienced went up, the amount of pain patients remembered fell. Gruesome but true.