Thursday, October 22, 2009

Vértigos a la Gibbard

A week ago Allan was asking whether the following argument gave a valid non-normative reduction of moral notions into psychological ones.

1) The principles by means of which our moral judgment works yield rejection of X.
-----
C) It is wrong to do X.

As it stands, the argument is clearly invalid. But the discussion had a context. Allan was offering the following Rawlsian (since it isn't Rawls') definition of right/wrong:

Definition:
X is right/wrong = the principles by means of which our moral judgment works yield acceptance/rejection of X.

I pointed out to Allan that the argument is valid if we consider it to be enthymematic with respect to the definition above (which Allan had given before). The result is a clearly valid argument:


1) The principles by means of which our moral judgment works yield rejection of X.
2) X is wrong if and only if the principles by means of which our moral judgment works yield rejection of X.
-----
C) It is wrong to do X.

Allan retorted then by claiming that this second argument, though valid, involved an unnacceptable trick: it is no longer non-normative since premise 2) includes the use of a normative term; i.e., 'wrong' on the left hand side of the biconditional.

That seems strange though. It seemed clear to me that the use of terms (of any given discipline) within definitions are not properly classified as uses. They seem to be more like mentions. So I replied to Allan with the following new definition.

Definition 2:

Definition:
'right'/'wrong' applies to X = the principles by means of which our moral judgment works yield acceptance/rejection of X.

That definition, and the corresponding premise, certainly makes no use of moral terms. It makes use of terms like ' 'right' ' not of terms like 'right'. The distinction is ridiculous, to some, but so was the worry (to me). Allan replied by pointing out that if I went "metalinguistic" then we couldn't know what I meant by 'right' and that the only way to fix this would be for us to assume that 'right' means "right", which would involve a normative premise.

I think there's something wrong with this reasoning. If I am defining the term, then certainly I don't want you to presuppose any content for it. That's exactly my goal: to deliver the content you were looking for. If you accept the definition, then you have all you need to make the argument work. That seems clear.

But there's something else that worried me: the assumption that no adequate definition of a moral term can do without normative uses of those terms. That is tantamount to assuming that no non-normative definitions can be given. That's a big bullet to bite. This, to my mind, involves some nauseating form of argument that Pereda classifies as "prescriptivist vertigo" which consists in the insistent projection of prescriptive content where there is none.

If, as Allan claimed, we cannot define any term X but by using it, and using it presupposes competent use of this term's content, then it seems we cannot define anything whatsoever. Competent use of terms becomes some kind of magical ability we all have which, unfortunately, we cannot describe in any non-question begging way. We cannot understand the notion of 'chemical structure' but in chemical ways, which presuppose that we understand chemical notions. And there's no way around it.

Either the use of terms in definitions is more like a mention of the term (or a metalinguistic use of the term) or it is completely kosher to define terms by using them. Otherwise, language use and concept use become magical, unexplainable feats that we humans somehow achieve.