I accept a more or less functionalist account of the mind, according to which mental states are individuated by their functional role in the economy of cognition and behavior. I also believe in the possibility of what is sometimes called the “Cartesian Fallacy,” the assumption that our own mental states are transparently accessible to consciousness. Functionalism together with anti-Cartsianism about introspective access imply that we may not know what words in our language mean, even if we use them correcly, and that we may have false beliefs about what we believe.
I’ve brought this up before in an earlier discussion of “meta-atheism,” [pdf] roughly the idea that people may sincerely believe they believe in God when they do not in fact. The disposition to avow a belief that P is neither necessary nor sufficient for believing that P. Actually believing P requires that one is generally disposed not only to say that one does, but that one is disposed to make certain inferences, to behave in certain ways, and more. This raises the possibilty that people may sincerely believe that they are happy or unhappy, when they actually aren’t.
It is hard to believe that one could make a mistake about whether one was in a state of pain, say. But happiness probably isn’t like that. If happiness is a complex, partly historically and socially constructed condition composed of dispositions to experience certain basic emotions and moods in a distinctive combination, togteher with dispositions to have certain thoughts, and to behave in certain ways, then it may be pretty plausible that we could just be wrong about whether we are happy, or about how happy we are.
I don’t think I want to press this view very hard, but it strikes me as a real possibility, and another reason why self-report is not the most promising technique of measurement.