Toward Non-Vacuous Economics

The Economist:

To say that A brings greater utility than B just is to say that it is preferred. We infer preference, and therefore utility, by observing what people have done. If I have in fact chosen A, then I preferred to it B, or —to say the same thing—A's utility to me is greater than B's.

That's it. It is not a part of the model that choosing A will in any sense make me feel better than choosing B. Subjective states are simply one more type of thing I may have preferences over. It is a fairly reliable generalisation that people prefer pleasure over pain. But it is is not a part of the model. People do run marathons and watch Jim Carrey movies, after all. It is also not a part of the model that people prefer more money to less, unless it is added as an assumption.

The trouble is that the formal model strictly construed fails to predict nothing, since it is completely vacuous. It simply says that whatever has been done—whether burning a billion dollars in a bonfire, or stabbing yourself in the thigh—in fact maximized utility.

In order to rule anything out, empirical generalisations about preferences must be added. But economists are at no special advantage at detecting these patterns by inferring preferences from behavior. Each action can be described any number of ways. And any number of preferences could motivate each action.

If Lois Lane kisses Superman, she has kissed Clark Kent, but we cannot infer inconsistent preferences when she refuses to kiss the bespectacled reporter. Preferences are, as philosophers of language put it, intensional —they refer to choices in certain guises or under certain modes of presentation. To use the example Mr Carroll discusses in his discussion of non-linearity, if I buy a bottle of wine at a restaurant for a price I would refuse in a shop, that likely goes to show that I have represented the options before me in different ways in the two cases. I have not chosen differently, given the same option. The options, as I see them, are different.

In order to be usefully predictive, what economics needs is not so much better mathematical tricks, but, first, a better grasp of the nature of mental representation.