I find Frank Jackson’s arguments against non-naturalist realism in his From Metaphysics to Ethics fascinating. So I think that, over a longer period of time, I would like to post a couple of things about them. Today, I am puzzled about the one that has received most attention. I wonder whether this argument is available for Jackson himself.
Very roughly, as I understand it, this argument begins from the premise that, given the truth-aptness of moral claims and global moral superveniece, for every moral property, there must be a naturalist property that is necessarily co-extensive with it. This property can perhaps, at worst, be a very disjunctive property consisting of full descriptions of every words of which the given moral term can be truthfully used.
The second premise of the argument is that no necessarily co-extensive properties can be distinct ones. Necessary co-extensiveness must imply identity. From these two premises the wanted conclusion follows; truth-aptness and supervenience imply that moral properties cannot be non-natural properties. Rather, they must be natural properties.
The latter, essential second-premise is becoming much debated. Russ Shafer-Landau (in his Moral Realism) and Brad Majors (in his “Moral Discourse and Descriptive Properties” in Philosophical Quarterly, 2005) present a variety of counter-examples to the principle according to which necessary co-extensiveness guarantees identity. Many of these examples are discussed and undermined in Jackson’s original presentation and in Bart Streumer’s forthcoming “Are There Irreducibly Normative Properties?” (in Australasian Journal of Philosophy). It would be nice to talk about these cases too but I’ll leave it to some other time.
What puzzles me most about the argument is that Jackson himself, as far as I can tell, offers a perfectly good counter-example to his own principle. This seems to happen in a paper he wrote with Elizabeth Prior and Robert Pargetter called “Three Theses about Dispositions” (American Phil. Quarterly 1982) and also in the Chapter 4 about colours in From Metaphysics to Ethics.
In these works, Jackson argues that dispositional properties must have categorical properties as their causal bases. That is, something that is fragile must be fragile because it has some molecular structure that causes it to shatter when dropped. Now, Jackson accepts that fragility can be multiply realisable. A fragile object can have the molecular structure alpha, or beta, and so on so long as these structures are such that the object shatters when dropped.
Now, it seems that there must be a disjunctive property consisting of all the causal bases for fragility (Jackson accepts that a similar property for the causal bases for the property of looking yellow on page 106 of fMtE where he wants to identify the colour yellow with this disjunctive property). It also looks like this disjunctive property must be necessarily coextensive with fragility.
But, now, Jackson himself argues that this disjunctive property is not identical with the dispositional property. He argues that, given that the disjunctive causal bases property of the dispositional property is not excessively disjunctive, it has causal powers – it causes the causal manifestations of the relevant disposition. However, dispositional properties are, he argues, causally impotent. Fragility does not cause the glass to break, its molecular structure does and the forces upon it.
This difference between the causal powers of dispositional properties and the disjunctive properties that are the causal bases of such properties seem to imply by the law of indiscernibility of identicals that these two properties are distinct – as Jackson seems to explicitly accept. But, these properties are necessarily coextensive. Thus, there must be distinct necessarily coextensive properties. So, it looks to me that either Jackson must give up his view about dispositions or the crucial premise in his argument against non-naturalism. But, I do trust Jackson more than myself so I must have missed something.
Interesting.
But, hang on, dispositions don’t supervene on their categorical bases alone, do they? Don’t we have to include the laws? (Then we get global supervenience.) So, no categorical property (no matter how disjunctive) is identical with a disposition, since it can occur in a world with different laws and so with different dispositions.
Of course, Lewis wouldn’t buy this, but I don’t think Jackson accepts the strong Humean supervenience that Lewis did.
Jamie,
that’s a good point. He could and perhaps should say that. At a quick glance I don’t find him saying anything about the matter here. He does say about the fragile case that ‘*all the causal work* is done by bonding B in concert with the dropping’. This sounds as if *the actual laws* are doing no work in the picture anyway in which case dispositions could supervene on their categorical bases.
I’m also currently Galen Strawson’s forthcoming paper on dispositions where he writes that:
“I very much doubt that the idea that X can retain its intrinsic nature or basic categorical being unchanged across different nomic environments is even coherent, if laws are understood not as human linguistic or conceptual creations, but as non-linguistic objective principles of working. For I very much doubt that laws so understood can be properly thought to be in any way independent of, rather than essentially constitutive of, or part of, the (categorical, intrinsic) nature of matter; only a bad ‘separatist’ habit of thought can make this seem initially plausible’.
If Strawson is right about this, then the example would still seem to stand.
Also come to think of it (and I’m not really sure about this), one could make the disjunctive, categorical property on which the disposition supervenes more complicated and disjunctive. So you might that there is a disjunctive categorical property of being F-or-G-etc. in nomic environment 1 or H-or-I-etc. in nomic environment 2, and so on. This property would be necessarily coextensive with the dispositional property, but still it seems that Jackson would need to say that it has causal powers whereas the dispositional property does not.
Just quickly:
Strawson may be right, but Jackson does say, I think, that he doesn’t agree. If the properties that figure in the laws had the laws essential to them, then (Jackson thinks) it would be very mysterious what the relata are in the relations that the laws impose.
Second, if you add the nomic environment to the supervenience base (which I think is a good idea), then it isn’t so clear to me that Jackson would reject the identification of certain (incredibly complex) base properties with the dispositions.
Jussi,
From these two premises the wanted conclusion follows; truth-aptness and supervenience imply that moral properties cannot be non-natural properties. Rather, they must be natural properties
This seems to me to be a quick move. For wouldn’t this imply, also, that all natural properties are moral properties? How does Jackson get the identity in one direction only?
This difference between the causal powers of dispositional properties and the disjunctive properties that are the causal bases of such properties seem to imply by the law of indiscernibility of identicals that these two properties are distinct.
I think you can raise this worry with a simpler example. If supervenience is true, any moral property M supervenes on some natural fact F. Thus, M has the property of supervening upon F. F does not have this property. So M and F are not identical.
Jamie,
about the second point – would be nice to actually know what he would say. Maybe someone could ask him. He does say that ‘the dispositional property of being fragile is the second-order property of having some first-order property or other … that is responsible for the glass being such as to break when dropped’. That seems something different from even complex base properties.
Christian,
I must be missing something but I don’t see the first problem. It doesn’t seem to follow from all Fs are Gs that all Gs are Fs. I guess he could say that only those natural properties that are referred to by moral terms are moral properties. But, you are right. Mark Schroeder’s new book has a great section on how the talk about identity relations is misleading and how it would be better to talk about constitution relations that are asymmetric.
Also, I’m not sure about the second problem either. Doesn’t everything supervene on itself? No difference in Fness of an object without a difference in its Fness seems trivially true.
Jussi,
Right, that (the second order property) is a different property from the complex disjunction. I’m going to think about this for a few days, and also bring it up in my seminar (by coincidence we are hitting exactly this section of From Metaphysics to Ethics on Thursday). Then if we’re still wondering I will e-mail Frank.
Christian,
Supervenience is asymmetric, so the premises do not imply the converse (as Jussi notes). And properly speaking, supervenience (at least as I understand it) is not a relation between individual properties but rather between families of properties, classes closed under Boolean combination. So, no property has the property of supervening on itself. And every family does supervene on itself — so I’m agreeing with the spirit of Jussi’s last remark.
Jussi,
It doesn’t seem to follow from all Fs are Gs that all Gs are Fs.
That’s right. The thought was that when Jackson claims Necessary co-extensiveness must imply identity, then by ‘identity’ we get that the subvening natural properties are moral properties iff the supervening moral properties are natural. I can imagine contexts in which pointing at some candidate natural property (instance) and saying, “shame on it”, is going to be odd.
it would be better to talk about constitution relations that are asymmetric
That’s basically what I’m getting at. I’m wondering why Jackson is wanting identity between dispositions and their bases, and moral properties and their bases. Why not deny identity here to accept the asymmetry in both cases? Does he think constitution-like relations are bad?
Jamie,
And every family does supervene on itself
Not my family! Okay, I just need to do homework. Thanks for the link.
It’s always nice to forget the ‘not’. Please correct this appropriately…”I’m wondering why Jackson is […] wanting identity between dispositions and their bases…”
Christian,
That’s kind of what I’m after too. In the case of dispositions the identification (or even constitution in the ontologically innocent sense) is ruled out by the difference in causal efficacy. The base properties cause things whereas the dispositional properties do not. I guess the challenge for the non-naturalist would be to come up with a similar property that one of the necessarily coextensive properties have and the other one doesn’t. Unless this challenge is satisfied then I think the argument goes through that there is no reason not to identify the properties with one another (or accept an ontologically innocent asymmetric constitution relation).
I’m not too convinced by the shame case. It could be that we are happy with saying that of one property under one mode of presentation but not the other. So, the moral and natural terms can present the same properties in different ways and maybe we naturally react to them only under some ways of being presented to us.
Jussi,
About the “shame” example. The idea is that there are natural facts, say, facts about brainstates, that will be (I suppose) the subvening base of something bad, e.g., a vice, or a disposition to cause pain for no good reason. We don’t say the brainstate is bad, we say the disposition it gives rise to is bad. We don’t say the cluster of particles is bad, we say what supervenes on it is bad. Thus, one is an object for criticism and disapproval and the other is not. This would be a reason to think the moral and natural properties are different making the appeal to modes unavailable.
Well, no. If we don’t know yet that Superman is Clark Kent (or just have that vividly in mind), we would say that Superman flies wheras Clark Kent does not, that Clark Kent works with Lois Lane whereas Superman does not, and so on. We might even admire Superman but not the nerdy Clark Kent. All of these are attitudes towards the same facts differently presented to us.
I can see how the MOP view is attractive in the Superman case. But it has a different motivation. We move from intuitions of sameness of referent to difference in senses, as we seem to have different thoughts about the same referent in conditions of ignorance. In the natural property = moral property case, it is up for debate whether there is sameness of referent. And we cannot, it seems to me, appeal to the Superman case to motivate sameness of referent since we are assuming there is no ignorance at all. We have all of the info, but then we ask: Do the moral terms and natural terms, in the “brainstate” (or, whatever) case co-refer given this information we have?
Right. I think I agree with that. The point I thought I was trying to make was that one cannot argue directly from the difference in our attitudes towards moral properties and the natural properties to the conclusion that these are different properties as long as it is up for a debate (as you suggest) that there is sameness of reference (but a difference in meaning). I wasn’t trying to use the Superman case to argue that there is a sameness of reference either. If the debate is still up for graps about the reference of moral terms, then it seems to be that the relevant ignorance about the reference has to be allowed in.
Jussi, here’s a strategy that seems to me to be a reasonable one.
First. Feign agnosticism about whether x = y (where x is the instantiation of a natural property, y the instantiation of a moral one). You think there are good arguments for and others against.
Second. Then ask whether it seems that x has a property that y lacks? If so, you think it would be reasonable to believe they are not the same.
Third. Then consider the property of causing a certain kind of reaction of disapproval, or being deserving of disapproval. Does x have it and y lack it?
Fourth. If you answer yes to (3), I do, then you (me) have some reason to treat x and y as distinct.
Fifth. Finally, you retreat from agnosticism, bring the original arguments to bear, the explanations of error that may or may not involve MOPs and then weigh the balance of reasons.
That’s what I mean to do. And I think the different reactions we take towards certain instantiations of moral and natural properties, count, provide at least one reason, to believe they are distinct (contra Jackson I guess).
Jussi,
If you’re making a study of this issue, then maybe you should also glance at chapter six of Graham Oddie’s _Value, Reality, and Desire_, where he argues for non-reduction and non-identity on the basis of a “convexity” condition for properties in general.
The idea is that since for any moral property (which are convex) there will (probably) be a no convex descriptive property it supervenes on.
Thanks Dan. I’ll check that out. ‘Convexity’ sounds interesting even though I don’t yet know what it really would mean here.
Jussi,
I’ve come upon this thread only now. It’s not clear to me in which sense the disjunctive property consisting of a disjunction of the categorical bases of fragility has causal powers. Suppose we are told that something fragile has broken, and that we know that, since it’s fragile then its molecular structure is x or y or z. Still we don’t know what caused it to crack, though we have a list of candidates: but a list of candidate causes is not a cause. What has causal powers is (the singular instantiations of) a certain molecular structure, and this obviously is not necessarily coextensive with fragility. So, Jackson could just admit a slip, and claim that fragility and the disjunctive property are necessarily coextensive, ergo identical. Missing something?
Well, I wonder if a couple of things would follow. First, he would have to give up his theory of dispositions as higher order properties as higher order properties presumably are not the same as disjunctive properties (or maybe they are if they are necessarily coextensive).
Second (and you know much more about this than I do), I thought that Jackson’s theory about program explanations is just meant to be a view on which the right kind of disjunctive properties (boiling of the water and a disjunction of molecular movements explaining shattering of the glass comes to mind) can have causal powers. Yablo also argues for similar conclusions.
I don’t know. It just still seems to me that not everything in vast area of the Jackson theorising fits as a glove.
I should have another look at Jackson’s theory of dispositions. However, as far as program explanations go, what does the causal work are particular molecular configurations, not the disjunctive property of having this or that or…molecular configuration. This disjunctive property can at most provide a program explanation of why (eg) the glass shatters, as indeed does the property of the water being at 100°.
There’s of course a question (for Jackson as well as for others) whether the disjunctive property and its necessarily co-exemplified property (fragility, water being at 100°, rightness) have the same explanatory power: if they don’t, may this be a good reason to keep them distinct?
That’s right. Sorry. The idea is that the program-explaining properties are causally ‘relevant’ (can be used in causal explanations?) even when they are not causally ‘efficacious’ (lacking in causal ‘oomph’ as the technical term goes).