Thursday, December 14, 2006

Perspectives and magnets

As Brian Weatherson notes, the new Philosophical Perspectives is now out. This includes a paper of mine called "Illusions of gunk". The paper defends mereological nihlism (the view that no complex objects exist) against a certain type of worry: (1) that mereological nihlism is necessary, if true; and (2) that "gunk-worlds" (worlds apparently containing no non-complex objects) are possible. (See this paper of Ted Sider's for the worry) I advise the merelogical nihilist to reject (2). There are various possibilities that the nihilist can admit, that plausibly explain the illusion that gunk is possible.

The volume looks to be full of interesting papers, but there's one in particular I've read before, so I'll write a little about that right now.

The paper is Brian Weatherson's "Asymmetric Magnets Problem". The puzzle he sets out is based on a well-entrenched link between intrinsicality and duplication: a property is intrinsic iff necessarily, it is shared among duplicate objects. Weatherson examines an application of this principle to a case where some of the features of the objects we consider are vectorial.

In particular, consider an asymmetric magnet M: one which has a pointy-bit at one end, and is such that the north pole of the magnet "points out" of the pointy end. Intuitively, the following is a duplicate of another magnet M*: one with the same shape, but simply rotated by 180 degrees so that both the north pole and the pointy end are both orientated in the opposite direction to M. (Weatherson has some nice pictures, if you want to be clear about the situation).

Though M and M* seem to be duplicates, their vectorial features differ: M has its north pole pointing in one direction, M* has its north pole pointing in the opposite direction. Moral: given the link, we can't take vectorial properties "as a whole" (i.e. building in their directions) as intrinsic, for they differ between duplicates.

What if we think that only the magnitude of a vectorial feature is intrinsic? Then we get a different problem: for their are pointy magnets whose north pole is directed out of the non-pointy end. Call one of these M**. But in shape properties, and so on, it matches M and M*. And ex hypothesi, in all intrinsic respects, their vectorial features are the same. So M, M* and M** all count as duplicates. But that's intuitively wrong (it's claimed).

Such is the asymmetric magnets problem. The challenge is to say something precise about how to think about the duplication of things with vectorial features, that'd preserve both intuitions and the duplication-intrinsicality link.

Weatherson's response is to take a certain relationship between parts of the pointy magnet its vectorial feature, as intrinsic to the magnet. In effect, he takes the relative orientation of the north-pole vector, and a line connecting certain points within the magnet, as intrinsic.

Ok, that's Weatherson's line in super-quick summary, as I read him. Here are some thoughts.

First thing to note: the asymmetric magnets problem looks like a special case of a more general issue. Suppose point particles a, b, c each have two fundamental vectoral features F and G, with the same magnitude in each case. Suppose in a's case they point in different directions, whereas in b and c's cases they point in the same direction (in b's case they both point north, in c's case they both point south). The intuitive verdict is that a and b are not duplicates, but b and c are. But, if you just demand that duplicates preserve the magnitudes of the quantities, you'll get a, b, and c as duplicates of one another; and if you demand that duplicates preserve direction of vectoral quantities, you'll get none of them as duplicates. That sounds just like the asymmetric magnets problem all over again. Let me call it the vector-pair problem.

What's the natural Weathersonian thought about the vector-pair problem? The natural line is to take the relative orientation ("angle") between the instances of F and G as a perfectly natural relation. (I think that Weatherson might go for this line now: see his comment here).

It seemed to me that a natural response to the problem just posed might be this: require that the magnitude of any quantities is invariant under duplication; also that the *relative orientation* of vectoral properties be invariant under duplication. Thus we build into the definition of duplication the requirement that any angles between vectors are preserved. There's thus no easy answer to the question of whether vectorial features of objects are intrinsic: we can only say that their magnitudes and relative orientations are, but their absolute orientation is not.

This leads to a couple of natural questions:

(A) Why do we demand absolute sameness of magnitude, and only relative sameness of direction, when defining what it takes for something to be a duplicate of something else?

I'm tempted to think that there's no deep answer to this question. In particular, consider a possible world with an "objective centre", and where various natural laws are formulated in terms of whether objects have properties "pointing towards" the centre or away from it. E.g. suppose two objects both with instantaneous velocity towards the centre will repel each other with a force proportional to the inverse of their separation; while two objects both with instantaneous velocity away from the centre will attract each other with a similar force (or something like that: I'm sure we can cook something up that’ll make the case work). Anyway, since the behaviour of objects depends on the "direction in which they're pointing", I no longer have strong intuitions that particles like b and c should count as duplicates (with that world considered counteractually).

I find it harder to imagine worlds where only relative magnitudes matter to physical laws, but I suspect that with ingenuity one could describe such a case: and maybe (considering such a scenario counteractually again) we'd be happier to demand only relative sameness of magnitudes, in addition to relative sameness of orientation of vectoral properties, among duplicates.

(B) The above proposal demands invariance of relative orientation of vectoral properties among duplicate entities. But that doesn't straightaway deal with the original asymmetric magnet case. For there we had the orientation of the shape-properties of the object to consider, not just the orientation of the vectoral quantities that the (parts of) the object has.

I'm tempted by the following way of subsuming the original problem under the more general treatment just given: say that some perfectly natural spatial properties are actually vectoral in character. E.g. the spatial property that holds between my hand and my foot is not simply "being separated by 1m" but rather "being separated by 1m downwards" (with, of course, the converse relation holding in the other direction). After all, if in giving the spatial properties that I currently have, we just list the spatial separations of my parts, we leave something out: my orientation. And that is a spatial property that I have (and is coded into the usual representations of location, e.g. Cartesian or polar coordinates. Of course, such representations are all relative to a choice of axes, just as the representation of spatial separation is relative to a choice of unit.)

Now, there might be ways of getting this result without saying that spatial-temporal relations among particulars are fundamentally vectorial. But I'm not seeing exactly how this would work.

(Incidentally, if we do allow fundamentally vectorial spatio-temporal relations, then it's not clear that we need to appeal to spatio-temporal relations among parts of an object to solve the asymmetric magnets problem: appealing to the angle between the "north pole" and the (vectorial) spatio-temporal properties of the pointy magnet may be enough to get the intuitive duplication verdicts. If so, then the Weathersonian solution can be extended to the case where the magnets are extended simples, which is (a) a case he claims not to be able to handle (b) a case he claims to be impossible. But I disagree with (b), so from my perspective (a) looks like a serious worry!)

(x-posted on theories n things)

Thursday, December 07, 2006

CMM workshop

This is just a heads-up to the world that there will be a CMM workshop at Leeds on March 10th. The topic is 'Structure in Metaphysics', and Julian Dodd (Manchester), Katherine Hawley (St Andrews), Kris McDaniel (Syracuse) and Leeds' own Robbie Williams are signed up to speak.

Detials of paper titles/registration details etc will follow sooner or later - but for now: keep your diaries free.

The kinds of topic the workshop will deal with are (i) mereology, (ii) grades of being/existence, (iii) ontological dependence, (iv) facts, states of affairs, propositions, etc.

A preliminary webpage is up at

Wednesday, December 06, 2006

Identity of Indiscernibles and Haecceities

According to the identity of indiscernibles, as a matter of necessity, no two things can share all their properties. If a and b share all their properties then they are not two things, but one: they are numerically identical. You can’t have qualitative identity without thereby having numerical identity.

Why think the identity of indiscernibles is true? Well, suppose it’s false; in that case there is a possible world containing two entities, a and b, which share all their properties. And the uncomfortable thought is meant to be that in that world there is nothing that makes these entities distinct.

This thought proceeds from the idea that, for every object o, something grounds the identity of o: there is a metaphysical explanation for o’s being that very object, and not some other thing. What could do this grounding other than some subset of o’s properties? But if o is the thing it is in virtue of having some subset of its properties, then any thing which had all and only o’s properties would have its identity determined in the same way as o, and hence would be numerically identical to o.

That, I take it, is the intuitive thought behind the identity of indiscernibles. Any defender of that thought has to say something about Max Black’s world consisting solely of the two homogenous iron spheres Castor and Pollux. One thing to say is that the objects have different haecceitistic properties. Castor is distinct from Pollux because one has the property being identical to Castor and the other the property being identical to Pollux. Something seems unsatisfying about that. John Heil describes it somewhere as “the sort of move that gives philosophy a bad name.” But what’s wrong with it?

Firstly, you might not like the admission of such a ‘property’ into your ontology. Our grasp of what properties are seems to be tied to qualitative ways for things to be – such as being red or being round etc; the introduction of mere haecceitistic properties might be thought to have stretched the concept of property beyond breaking point.

But suppose you accept the existence of properties of the form being identical to x; there is a deeper problem. We are told that Castor and Pollux are distinct because Castor has the property being identical to Castor and Pollux has the property being identical to Pollux. But that only serves to ground the distinctness of Castor and Pollux if these properties are themselves distinct. If ‘the property being identical to Castor’ and ‘the property being identical to Pollux’ are two names for the same property then the fact that Castor has the former property and Pollux the latter doesn’t serve to ground their distinctness. So we need to ensure that these haecceitistic properties are distinct if their admission into our ontology is going to help us with Black’s example. But then, by the same reasoning that led us to the identity of indiscernibles in the first place, there must be something that grounds the distinctness of these properties. Now what do we usually think are the individuation conditions for properties? One might try: P and Q are the same property iff they make the same contribution to the qualitative nature of their bearer. But of course, this will only do for qualitative properties; applying it to mere haecceitistic properties would make them all identical, since mere haecceitistic properties don’t make any contribution to the qualitative nature of their bearers (that’s precisely why it rankles to call them ‘properties’). So for mere haecceitistic properties it seems we should have instead: P and Q are the same haecceitistic property iff they belong to the same thing.

But if that’s right then being identical to Castor is distinct from being identical to Pollux in virtue of the bearer of the former being distinct from the bearer of the latter. In which case we can hardly appeal to the distinctness of the properties to ground the distinctness of the bearers.

One who invokes mere haecceitistic properties to deal with Black’s world wants to hold that [Castor is distinct from Pollux] in virtue of the distinctness of being identical to Castor and being identical to Pollux; but it seems that they should be committed to [being identical to Castor is distinct from being identical to Pollux] holding true in virtue of the distinctness of Castor and Pollux. But these ‘in virtue of’ claims can’t both be true, since in-virtue-of is asymmetric. So the introduction of mere haecceitistic properties doesn’t seem to help: to ground the distinctness of Castor and Pollux the distinctness of the properties would apparently need to be taken as brute; and if we’re prepared to do that, why not just accept the distinctness of Castor and Pollux as brute in the first place?