This is even simpler than the meta-preference model proposed by Patrick, and even has some basis in psychology (habituation). A model like this can accommodate an intrinsic (state-independent) preference for one brand over another as well as brand addictions (past consumption increases marginal utility of consuming the same thing). So before going to multiple selves, I’d want to look at simple dynamic preferences of this sort.

The water buffalo thanks you. Much clearer. You have a gift for exposition. Thanks.

So this is, it sounds like, a difference that makes no difference. What concerns me is that the same study performed at only marginally different times could catch the consumer in preference A or B and the outcome would be skewed. And here the difference would a difference.

And there is a problem here to do with the architecture of knowledge. If the model climbs ever upwards to protect its assumptions, it begins to take itself out of real explanatory range. Increasingly, it is only good at explaining the general and less and less good at explaining the particular.

This is not to say that I disagree with you. It is a very interesting comment and a useful part of this water buffalo’s education.

Thanks again, Grant

I want to point out a couple things about Professor Boudreaux’s example and your use of it. The first deals with that old economist chestnut: ceteris parabis, and the conditional tense. The second, related, point is a little more exploratory, but I think it’s germane to the question at hand, the choice between possible worlds.

First, the example Boudreaux presents is a simplification of the actual principle of transitivity. The important simplication is in the tense. He says “If John prefers apples to bananas,,” A more rigorous formulation might be:

When faced with a given decision problem, at a given time, with a given fixed existential situation (all else equal), John _would prefer_ apples to bananas. In the _exact same situation_, with the sole exception of the choice presented, John _would prefer_ bananas to cantaloupe. The axiom of preference trasitivity then leads us to concludes that if John had been faced with that same situation but with the choice between apples and cantaloupe, he _would prefer_ apples.

Now this may seem like nit-picking, and in empirical work we loosen a lot of these assumptions, but it becomes really important when trying to theorize. Because how often is a person faced with the exact same set of initial conditions? In it’s strongest form, the axiom of preference transitivity is just that, an axiom, because it is completely unfalsifiable. An argument against axioms because come as an attack of the conclusions that follow from them, and that seems to be one of the things that you’re up to in this piece.

It is important, then to be clear what exactly follows from the axiom. It does not, for example, follow that if you prefer Molson to Keith’s today and Keith’s to Beck’s tomorrow, that you’ll prefer Molson to Beck’s on Friday. If I were better educated I’d know who said that you can never step over the same stream twice, but my ignorance doesn’t make it any less true. Likewise, I doubt you ever make the same decision twice.

So what good is a theory that can’t make simple predictions like beer preferences? It’s not to great by itself. But when combined with other assumptions economists usually make, you can build up a theory of decisions that is faily robust and helpful in understanding a lot of situations. I’ll let the pudding of economics be my proof in this instance.. now is not the time for this argument.

So what, then, of possible worlds? Well, a better way to understand the dificult, complex, decisions the seem to characterize the modern world might be to throw out all this talk of apples and bananas completely, and move to a higher level of analysis. I think it’s a mistake to drop the useful model of preference maximization, we don’t want to throw the baby out with the fruit. Instead of choosing between an apple and a banana, think instead of choosing between a world with a certain characteristics, including a banana in my pocket (world A) and a world with identical characteristics , except with me having an apple in my pocket (world B).

The different (higher?/broader?) perspective gained by forming the question in this way avoids confusing today’s decision with tomorrow’s. After all I’m now choosing between two completely different worlds: world C (with banana) and world D (with apple). There will certainly be similarities between the A/B choice and the C/D choice, but now no one would say they are identical. (We could further theorize, of course, about what sort of comparison rules you can make. Can you, for example, drop common characteristics to simplify the decisions? Maybe even to compare A and C?)

Applying this perspective to your friend’s decision problem allows us to view her preference structure as consistent, if exceedingly complicated. Surely she builds little models to help with these decisions, instead of surveying every aspect of the possible worlds, but these models are heuristics with the aim of approximating a (world)preference maximizing decision.

Alas, I seemed to have wandered fairly far afield this time. Well, see what you can make out of what I’ve said already.. give me some comments back and maybe I can try to give a little more on the problematic parts.