In a recent paper in PPR, Gene D. Witner, William Butchard and Kelly Trogdon (WBT) have a paper called “Intrinsicality without Naturalness”. (It doesn’t seem to be available anywhere online, hence the lack of link.) As the title suggests, the paper is an attempt to offer an analysis of intrinsicness without appeal to Lewisian naturalness, which they argue is too mysterious to use as the basis of a usable analysis. I don’t quite understand this motivation; the difference between grue-like and natural properties is about as clear to me as any distinction in philosophy. Be that as it may, WBT’s analysis is interesting in its own right.

They propose that in place of Lewis’s naturalness, we should take as primitive the idea that some properties are had _in virtue of_ other properties. They fret a little about how well understood this might be, but I don’t want to press that point. (I couldn’t really, given that I use the same concept in “Morality, Fiction and Possibility”:http://www.philosophersimprint.org/004003/.) Rather, I’m not sure that their analysis works even given that concept as a primitive.

Like Langton and Lewis’s analysis, WBT’s starts with independence from accompaniment. They extend that idea to an analysis the following way.

(I1) F is an intrinsic property iff for all a, if a is F, a has F intrinsically
(I2) a has F intrinsically, then for any G such that a if F in virtue of being G, G is independent of accompaniment.

They also defend, in the course of defending these claims, (I3).

(I3) If a is F in virtue of being G, and G is extrinsic, then a does not have F intrinsically.

I think (I1) is correct. But I think that the conjunction of (I1) with either (I2) or (I3) leads to counterexamples.

WBT acknowledge that this analysis won’t work for impure properties, like _being identical with a_, where a is something that could exist unaccompanied. So it is only proposed as an analysis of pure properties. And it doesn’t work for necessary properties, so it is only proposed as an analysis of contingent properties.

The biggest problem concerns a property they discuss in the paper, Ted Sider’s example R = _being a rock_. As Ted points out, R is extrinsic because it is maximal. But it is independent of accompaniment. So WBT have to find a property that is not independent of accompaniment in virtue of which something is a rock.

Their proposal (roughly) is that any given rock is a rock in virtue of having some intrinsic properties and being embedded in something that is quite different intrinsically. This proposal has an odd implication. Consider some rock a in any everyday environment, and a duplicate rock b that is in a void, i.e. surrounded by nothing. WBT are committed to the odd view that a is a rock in virtue of very different properties to b, because b is _not_ embedded in something that differs from it intrinsically, since it is not embedded in anything at all. This seems wrong. To use some of the paraphrases of this idea that WBT propose, what makes it the case that a is a rock is very different to what makes it the case that b is a rock. The metaphysical basis of a’s status as a rock is very different ot the metaphysical basis of b’s status as a rock. This all seems wrong.

The metaphysically natural thing to say is that something is a rock in virtue of having the right intrinsic properties and not being embedded in a sufficiently similar environment. This is a property that a and b share. Unfortunately for WBT, it is independent of accompaniment.

Consider now a variant of an example of John Hawthorne’s that WBT also discuss, T = attending to something that is could have been one of your parts. So if I’m attending to one of my actual parts I have this property, as I do if I’m attending to a piece of my hair that was just cut off, or a pacemaker that I could have installed. But if I could not have a planet as a part (as I imagine), then I don’t have it if I’m attending exclusively to planet earth.

Clearly T is independent of accompaniment, since a lonely person attending to his foot has the property, so it is compatible with loneliness, and the other three cases are actually satisfied. But it is also clearly extrinsic. So WBT has to find a property that is independent of accompaniment in virtue of which something has T. The proposal they come up with (roughly, remember this is a variant on a case they discuss) is that some things have T in virtue of attending to something wholly distinct from them, and others have T in virtue of attending to one of their parts. The first of these properties is not compatible with loneliness, so it works for their purposes.

Again, the problem is that this does not seem to get the metaphysical basis of my having T right. Imagine that I’m looking at a particularly long hair. I’m attending only to it, because I’m concentrating on cutting it off without doing myself a mischief. I keep attending to it, and it alone, while I cut it off. All through this time I have T, since I’m attending to this hair and it could be, indeed was, a part of me. I think the properties in virtue of which I have T don’t change before and after the hair ceases to be a part of me. (Maybe I should pluck the hair rather than cut it iff I want it to cease being part of me. If that’s not too painful to contemplate for you, let that be the example. It is too painful for me, so I’ll stick with cutting.)

Now I think it’s fairly simple to say which property I have T in virtue of before and after the cutting. I have the property of attening to _that very hair_. In general, a person who has T generally has so because there is some a she’s attending to, and she has T in virtue of _attending to a_. But since _a_ could have been one of her parts, that property is compatible with loneliness. So I’m sceptical that there’s any property not independent of accompaniment that I have T in virtue of.

Finally, I want to raise a worry about (I3). Consider D = being a duplicate of a possible human. Since this is a property that never differs between duplicates, it is an intrinsic property. But I think some things have it in virtue of extrinsic properties. In particular, I have this property in virtue of _being a human_, and hence, because I’m by necessity a self-duplicate, being a duplicate of a possible person. But _being a human_ is maximal and hence an extrinsic property.