Previous research in psychology, neuroscience, and behavioral economics has shown that people have difficulty predicting what they will like and how much they will like it, which leads them to make a wide variety of poor decisions. Interventions aimed at improving the accuracy with which people imagine future events have been generally unsuccessful.via ScienceDaily
So rather than trying to improve human imagination, Gilbert and his colleagues sought to eliminate it from the equation by asking people to predict how much they would enjoy a future event about which they knew absolutely nothing -- except how much a total stranger had enjoyed it. Amazingly enough, those people made extremely accurate predictions.
In one experiment, women predicted how much they would enjoy a "speed date" with a man. Some women read the man's personal profile and saw his photograph, and other women learned nothing whatsoever about the man, but did learn how much another woman (whom they had never met) had enjoyed dating him. Women who learned about a previous woman's experience did a much better job of predicting their own enjoyment of the speed date than did woman who studied the man's profile and photograph.
Interestingly, both groups of women mistakenly expected the profile and photo to lead to greater accuracy, and after the experiment was over both groups said they would strongly prefer to have the profile and photograph of their next date.
In the second experiment, two groups of participants were asked to predict how they would feel if they received negative personality feedback from a peer. Some participants were shown a complete written copy of the feedback. Other were shown nothing, and learned only how a total stranger had felt upon receiving the feedback. The latter group more accurately predicted their own reactions to the negative feedback. Once again, participants mistakenly guessed that a written copy of the feedback would be more informative than knowledge of a total stranger's experience.