Don't Press People for Their Reasons

An interesting cognitive psychology result from the 1970s onwards is that if you ask people about their reasons for liking or disliking something, you cannot trust the answers. This can be demonstrated by subliminally priming subjects to like or dislike a neutral stimulus, ask them for their evaluation of that stimulus, and then ask them for their reasons. People give all kinds of reasons, but not the priming - unsurprisingly, given that it was subliminal.

A lesson from this research is that the valuation of a stimulus we experience can be based on reasons unknown to us. A corollary, I think, is that pressing people for their reasons will often be unwise. If a person says "I don't know why I like this", that is probably an accurate answer. If you get a different one after repeating your question, the reply is unlikely to be a reflection of the reasons behind the valuation - what you get is probably an answer that the person thinks is socially acceptable and/or easily understood, so that it's likely to shut you up; also, easily accessible patterns of explanation are more likely to be delivered than less accessible ones. You have hence measured something, but you don't know what it is, and it likely isn't what you intended to measure. If you take these answsers at face value, you are probably going to become stupider: Replacing uncertainty with certainty about an inaccurate assumption removes you from the truth by about as much as replacing certainty about an accurate assumption with uncertainty.

Jonathan Haidt uses an argument roughly along these lines to point out that much of the research on peoples' moral reasoning is misleading; this is also true of everyday conversations involving people who want to know why.