Jackson and Pettit REPLY to Van Roojen

(The following constitutes our first published reply to a BEARS review. Longer versions of both are forthcoming in Philosophical Quarterly. The Van Roojen review was posted here on 3/8/95.)


Posted September 18, 1995

"Comments On Van Roojen"

By Frank Jackson (Australian National University) (fcj@coombs.anu.edu.au) and Philip Pettit (Australian National University) (pnp@coombs.anu.edu.au)

Mark van Roojen in his brief BEARS review of our (1995) and again in his longer 'Moral Functionalism and Moral Reduction', forthcoming in Philosophical Quarterly, raises interesting questions. We address them in appropriate detail in our response (forthcoming, we understand, in the same issue of Philosophical Quarterly). Here we do no more than outline our response to the two main points he raises.

1. Van Roojen rightly observes that we defend an account of ethical terms as theoretical terms in the Lewis style, and that we hold that evaluative properties are natural properties. However he seems, wrongly, to think that we view the first as leading to the second. But first we should say something about his (proper) worry about how to identify the natural properties in these debates.

Natural properties cannot mean properties distinct from evaluative properties, for that definition would contradict the conclusion that evaluative properties are natural properties. Natural properties cannot mean properties of things and happenings in the natural world, for that would make G. E. Moore (!) an ethical naturalist. He thought that goodness was a property of happenings in the natural world. Natural properties cannot mean physical properties in the philosophy of mind sense. For on that account the ethical does not a priori supervene on the natural. We might have identity in all respects that are counted as naturalistic in the philosophy of mind but still have a moral difference on account of differences in the kinds of properties dualists believe in-enough agony realised in cartesian stuff will change moral nature.

We think it is best to mean by natural properties the properties ascribed by certain descriptive predicates. The thesis that evaluative properties are natural properties is then the thesis that the natural properties are a proper sub-set of the properties ascribed by descriptive language. For this reason it is probably best to avoid the term 'natural property', and indeed it nowhere appears in our (1995).

Of course, how to characterise descriptive language is itself an issue in the debate. Perhaps the best approach is to start with the ethical and value terms and specify the descriptive terms as those that are left after culling the ethical and evaluative ones. This approach makes transparent the disagreement with Moore. For it is clear that he denied that the properties ascribed by ethical terms like 'right' and 'bad' are one and the same as those ascribed by some combination or other of non-ethical terms. (Note that we say ascribed, not denoted. It is common ground that descriptive terms can denote ethical properties. For example, 'the property Moore most discusses in his first book' denotes the property of being good-that is, denotes the property ascribed by 'is good'.) Also, this indirect approach highlights the line of thought that makes the famous supervenience of the ethical on the descriptive so persuasive. As we put it in (1995), ethical ascriptions are answerable to how things are as expressed in non-ethical, that is, descriptive terms.

2. Van Roojen wonders if our reason for supposing that the evaluative properties are one and all natural properties is a conjunction of folk morality with naturalism. However in fact we hold that folk morality by itself and without recourse to metaphysical doctrines concerning what the world is like, lends support to the identification of evaluative properties with natural ones.

We took it (from our cognitivist perspective -- non-cognitivists cannot take literally the talk that follows of worlds being alike ethically) that it is part of folk moral theory that the ethical supervenes on the descriptive in the following sense:

(S) For all w and w', if w and w' are exactly alike descriptively, they are exactly alike ethically.

(S) is a global supervenience thesis, one that quantifies over complete ways things might be. It is not the kind of intra-world supervenience thesis often discussed in connection with universalisability in ethics. (S) is of course compatible with the view that ethical nature, the ethical way things are, is in part determined by facts about our (and other sentient creatures') responses and attitudes. For included in the global descriptive supervenience base will be facts about responses, both actual and hypothetical, and both first and higher order.

If (S) is true, any ethical sentence in the sense of any sentence about ethical nature, is logically equivalent to some purely descriptive sentence. Let E be a sentence about ethical nature in the following sense: a) E is framed in ethical terms and descriptive terms; b) every world at which E is true has some ethical nature; and c) for all w and w' if E is true at w and false at w', then w and w' differ ethically. Intuitively, the idea is that E counts as being about ethical nature by virtue of the fact that there must be some ethical nature for it to be true, together with the fact that the only way to change its truth value is by changing ethical nature; the worlds must, that is, differ somehow in the distribution of ethical properties and relations. Now each world at which E is true will have some descriptive nature: ethical nature without descriptive nature is impossible. And for each such world there will be a sentence containing only descriptive terms that gives that nature in full. Now let w1, w2, etc. be the worlds where E is true, and let D1, D2, etc. be purely descriptive sentences true at w1, w2, etc., respectively, which give the full descriptive nature of w1, w2, etc. Then the disjunction of D1, D2 etc., will also be a purely descriptive sentence, call it D. But then E entails and is entailed by D. For every world where E is true is a world where one or other of the Di is true, so E entails D. Moreover, every world where one or other of the Di are true is a world where E is true, as otherwise we would have a violation of (S): we would have descriptively exactly alike worlds differing in ethical nature. Therefore, D entails E.

What (S) means, therefore, is that any claim about how things are ethically is equivalent to some claim about how things are frameable in purely descriptive terms. This is not a proof that ethical properties are one and all descriptive properties. But it does, we submit, establish a very strong case that they are. Despite the notoriously a posteriori nature of 'x is water iff x is H2O', few think that water and H2O are distinct kinds. True, some think that although 'x is an equilateral triangle iff x is an equiangular triangle' is necessarily true, the property of being an equiangular triangle is distinct from that of being an equilateral one. But on examination, their arguments bear not so much on properties in our sense of ways things might be, but on a notion more closely tied to the concepts involved.

Jackson, F. and Pettit, P. 1995: 'Moral Functionalism and Moral Motivation', Philosophical Quarterly, 45, pp. 20-39.