Confusion of tongues: A theory of normative language by Stephen Finlay (review)

We begin with good. Good, Finlay suggests, is an incomplete predicate. When we say something is good we are saying something of this form.

(1) It is good for e if p.

Where p is some proposition (for good, he argues, is fundamentally expressive of a propositional operator), and e is some end salient in the context. Such salience is a slippery thing. Sometimes Drinking a lot of gin is good might be relative to the end of having fun. But we can still ask Drinking a lot of gin may be fun, but is it good? to raise the question of its conduciveness to some other end comparable in eligibility for salience, such as health perhaps. So we have a form of reductive naturalism that offers to defuse the open question argument. Or at least we do when we complete the reduction in the way F proposes.

(2) ‘It is good for e if p’ means that p increases the probability of e.

Or more precisely:

(3) ‘It is good for e if p’ means that pr(e/p & b) > pr(e/~p & b).

Where b specifies certain background conditions.

Along the way in discussing good, F has a swipe at the Aristotelian moral functionalist. A sentence saying X is a good K, where K is some functional kind, has both a functional reading, where the end served by the state of affairs denoted by p in which X features as a K is that end e_K specified by the function of K; and a nonfunctional reading, where X is something that happens to be a K featuring in some state of affairs good for some contextually salient e. F now claims we can distinguish these readings in the following way. On a nonfunctional reading, X being a good (or bad) K entails X being a K. On a functional reading, it does not. (This test is credited to Shyam Nair.) To be a good (or bad) person, he then proposes, entails being a person. Likewise, to be a good (or bad) human being entails being a good human being. Goodbye moral functionalism. It would be good to see this very quick argument slowed down a bit. As it stands, I am not quite clear what the rules would be for seeking counterexamples. An ice-axe is not a weapon, but it may be a good weapon. Only, if it is a good weapon, it surely is a weapon. Do we take weapon here to mean something made for a martial purpose or something serviceable to one? On the latter understanding, weapon surely still counts as a functional kind, but it is hard here to see how a good weapon could fail to be a weapon. Maybe things get clearer if we accentuate the negative. A postage stamp is a really bad weapon. Indeed, it is (partly for that reason) not a weapon at all.

We turn next to ought. F focuses on the instrumental conditional.

(4) If Max wants to evade arrest, then he ought to mingle with the crowd.

F tackles this in two steps. The first offers an analysis of the more straightforward:

(5) If Max is going to evade arrest, then he has to mingle with the crowd.

This F takes to mean that, in all possibilities consistent with certain background conditions in which Max will evade arrest, Max will mingle with the crowd. F then offers the neat proposal that we read the original wants to as a relevance or ‘biscuit’ conditional akin to 6, which is supposedly elliptical for 7.

(6) If you want biscuits, there are some on the table.

(7) If you want biscuits, [then this will be relevant information:] there are some on the table.

Likewise, for 8 we get the preliminary analysis proposed in 9.

(8) If Max wants to evade arrest, then he ought to mingle with the crowd.

(9) If Max wants to evade arrest, [then...