Preference semantics

This is a taxonomy of human preferences, which are human mental features. In theories of such features, it is fundamentally important to distinguish the uncontroversial mentalistic attributions from those attributions that are more contested. For instance, when I was talking about whether an observed action is emotional, I gave the example that in some cases it is uncontroversial to say that crying is an emotion-displaying behavior, whereas some other attributions can be more contested, such as in a dispute whether a deserting soldier was moved by overwhelming fear or by a more calculated plan to preserve himself. In this theory, this fundamental distinction features as the distinction between enacted and unenacted preferences, where enacted preferences are the uncontroversially attributed ones.

Enacted preferences are subdivided between revealed preferences (what Rothbard called demonstrated preferences) and bound preferences. Revealed preferences are uncontroversial because they are nothing other than the action itself; as long as an action was truly an action in the sense of something done consciously by the agent (a question about which there is usually little doubt), then certainly everyone agrees that in some sense the agent wanted to do the action, although it may be a rather weak sense. Out of all attributions of preferences, revealed preferences are the least empty but most blind, having the best semantics but the worst predictive power. Bound preference is my own name for a preference which is codified into a contract. Although all language is vague, indeed even the language of contracts, what happens in contracts is that the question of what the language means about human preferences is submitted to the judge’s interpretation, so that it is the convention that any hard dispute is settled one way or the other by the judge, so that, even though there may be different views on what precisely the parties had intended to commit themselves to, the only view that matters is the judge’s, so that their preference is bound to his interpretation of their language.

Unenacted preferences are those that are inferred to exist beneath an action that does not uncontroversially display it, for instance, if Alice wears a band T-shirt, Bob may infer that Alice is a fan of the band, and possibly further infer that Alice enjoys spending her time listening to music more generally. Although all preferences are in some sense inferred, we reserve the name of inferred preference (simply speaking) for those that are inferred from something other than verbal language, since there is no better name for those, and we should like to distinguish them from stated preferences, which are inferred from verbal language. A further kind of unenacted preference is what I’ll call postulated preference, which is where you say something is in someone’s true interest in spite of what everything about their behavior seems to communicate that they want. For instance, a parent may enforce a kid’s bedtime in spite of the kid’s protestations because the parent thinks the kid actually wants to feel well-rested the next day, and simply has a poor grasp of the causal connection with regard to their bedtime.

So these are the kinds of preferences:

• Enacted preference
• Revealed preference (shown by action)
• Bound preference (contracts)
• Unenacted preference
• Stated preference (verbal)
• Inferred preference (nonverbal)
• Postulated preference (“true interests”)