On Choice and Preference
Revealed Preference Theory in the light of Davidson s theory of action

Erik Angner
History and Philosophy of Science Dept.
University of Pittsburgh


While the likes of Amartya Sen have called for its rejection, Revealed
Preference Theory (RPT) remains widely popular among economists.
Articles relying on the theory are still accepted for publication in the
most prestigious journals, textbooks keep treating it as the theory of
preference, and the theory remains taught in graduate seminars.
Moreover, RPT has been imported into philosophy by philosophers of
science like Brian Skyrms (1998), and ethicists like Ken Binmore (1994).

I take RPT to be an attempt to provide an operational definition of the
notion of preference by identifying it with the notion of choice. In the
present paper, I use Davidson
s theory of action to argue that the
theory of revealed preference so understood is based on a simple, but
devastating, philosophical error. While any acceptable notion of
preference is intensional, I claim, the notion of choice presupposed by
RPT must be extensional. Since nothing can be both intensional and
extensional, I conclude that it is impossible to identify preference
with choice. Incidentally, my argument also shows what is wrong with an
inference from a sentence of the form 
P chooses A over B
 to a sentence
of the form 
P prefers A to B
 (as in 
We know people want this product
since they choose to buy it
).

Finally, I attempt to explain how the mistake could arise in the first
place. In my view the mistake is an instructive one, and I try to say
why. Psychologists and economists working in the rational choice
tradition tend to assume that there are privileged descriptions of the
alternatives facing the subjects, and of their actions in decision
situations. However, the assumption that there are privileged
descriptions of alternatives and actions is unjustified. How the
observer-theorist should describe the alternatives facing subjects in
decision situations, and the subjects
 actions, is an important, and
largely open, question in the methodology of psychology and the social
sciences.