Ben R. Newell and David R. Shanks in the MIT Press Reader:
A famous experimental example of this too-much-thinking effect involves strawberry jam. The setup was as follows. Participants were brought into a laboratory and were asked to taste five different jams lined up on a table in front of them. After tasting the jams, they were asked to rate how much they liked each one. However, before making the ratings, half of the participants were asked to write down their reasons for liking or disliking each of the jams, while the other half (a control group) listed reasons for a completely unrelated decision: choosing their university major. The experimenters had carefully selected the jams to be representative of a spectrum of quality according to trained experts. The crucial question was whether the participants who were asked to think and provide reasons for their preferences or those who judged without justification ended up with ratings more similar to those of the experts.
The results were intriguing.
More here.