In what is meant to be “a contribution of major importance to a unified theory of probability and utility” Jeffrey (The Logic of Decision) says about Bayesian decision theory that
Indeed, it is because logic and decision theory are woefully inadequate as descriptions that they are of interest as norms. (p.167)
Now, here’s a worry I presented yesterday in the seminar and that I’d like to present again, so that other people may consider it and that the ones that heard it can see why it’s worrisome. There are, at least, two questions the claim above prompts:
1) If theory T is woefully inadequate as a description of phenomena F, and yet it is meant to be a normative theory of F, couldn’t it be that it makes absurd demands about F?
2) If it is in virtue of theory T’s woeful descriptive inadequacy towards F that T is an interesting normative theory of F, wouldn’t it be the case that false descriptive theories turn out to be interesting normative theories?
Here are some examples that illustrate both worries. Consider the following theory about human decision making behavior called ‘T1’
T1: For any subject and any decision D at any time t, subject S should fly up into the sky at time t2 if S reaches decision D at time t1.
T1 is woefully inadequate as a description of human decision behavior. If we accept Jeffrey’s diagnosis about Bayesian Decision Theory, it is in virtue of this inadequacy that T1 is an interesting normative theory of decision making! But you may object, of course, that we have independent reasons to reject T1. The demands that it presents just don’t make sense at all! Point taken. But it will not get us out of trouble.
Consider the ‘ought’ implies ‘can’ principle which seems, to my mind, to offer a very sensible limitation on ANY normative theory: if subject S ought to do X in context C it must be the case that S is able in C to do X. Now suppose that, for some or other reason, whenever humans are in real decision making contexts (e.g., when they are decision where to go for dinner, or whether to watch a movie or do a reading) as opposed to consciously working out in a formal epistemology seminar, they cannot, because of their cognitive architecture, compute an algebra. Decision theory will still claim that humans should distribute their credences over some or other algebra. So it seems that Decision Theory will be claiming that humans ought to do what they cannot do. That seems absurd. Decision Theory should be a bit more informed about human psychology.
So far so good for the first worry. The second one is just as bad (if not worse). Suppose we accept Jeffrey’s claim that it is precisely because a theory T is woefully inadequate as a description that it is interesting as a normative theory. If so, then we seem to get the following argument:
P1 Theory T is woefully inadequate as a description of F.
P2 If T is woefully inadequate as a description of F then T is an interesting normative theory of F.
C Theory T is an interesting normative theory of F.
Now consider theories that give woefully inadequate descriptions of the phenomena they intend to theorize about. Here’s a big group of such theories: the false ones. Consider, in particular, Aristotle’s physics and astronomy. They are both woefully inadequate theories of the behavior of physical objects and planets. So we may run the argument:
P1 Aristotle’s astronomy is woefully inadequate as a theory of the behavior of planets.
P2 If Aristotle’s astronomy is woefully inadequate as a theory of the behavior of the planets, then it is an interesting normative theory of the behavior of the planets.
C Aristotle’s theory is an interesting normative theory of the behavior of the planets.
Now that does seem like a bad result, doesn’t it? That’s why we should worry about Jeffrey’s claim that
Indeed, it is because logic and decision theory are woefully inadequate as descriptions that they are of interest as norms.
The fact that a given theory T about phenomena F delivers a woefully inadequate description of F gives no good reason to take it to be an interesting normative theory of F. Jeffrey is simply wrong about this. And if he is right about decision theory being so woefully inadequate, then we should worry a bit about its theoretical value. In general, I think, all normative theories should worry about being so woefully inadequate!