Bob --
> You wrote -- "On the contrary, if rational behavior is defined, following
> Aristotle,
> as behavior for which an actor is able to provide reasons when
> questioned, then I should think this definition is quite operational.
> The definition says nothing about the nature, adequacy, optimality, or
> other properties of the reasons given or of the action-option selected,
> only that reasons can be provided."
>
> Many economists don't find "that reasons can be provided" as
operational. After all, I may say "I heard a voice tell me to do
that" and who can deny that I didn't?
But if I ask you:
"Why did you do a certain action?"
and you answer
"Because a voice told me to do that."
then a reason has been given for your action. This is an
operationalization of the definition of rationality I gave in my earlier
posting. As I said before, this definition does not address the value
or validity or any other property of the reasons given, only that a
reason can be given.
> What is operational is whether a decision theory can predict the
> action events, not on whether it can predict unobservable
> rationalizations.
Nice use of the word "rationalization" in Aristotle's sense! :-)
However, this rationalization is not unobservable at all. If you answer
my question about your reasons, your answer is observed by me. Whether
your answer provides the real reasons for your action is another matter
entirely, and (I stress again) not part of the definition of the word.
I think we are now disagreeing over the use of the adjective
"operational". I was applying it to a definition, not to a theory. In
fact, the property of theories you mention, that they "can predict the
action events", would not usually be considered operationalizability,
but predictive effectiveness. And, as Milton Friedman argued 50 years
ago, theories can be effective predictors of phenomena while being based
on false or even unfalsifiable assumptions.
-- Peter
