Wright on Disjunctive Objects
In The Conceivability of Naturalism, Crispin Wright notes:
When we disjoin or existentially generalise on names, the results -- for instance, "Tom or John was to blame", "Someone was to blame"-- had better not be conceived as forms of expression involving reference to disjunctive, or existentially general objects. There are no such objects.
What does he mean? Is his point merely the semantical hypothesis that "Tom or John" and "someone" should not be treated as refering expressions? It is probably easy to create a semantics where they are assigned a reference. I'm not even sure (though I believe it) that such a semantics would be perverse, given that lots of people have argued that expressions like "Tom and John" should be assigned some kind of reference to account for sentences like "Tom and John ate the cake". But at any rate, this presumably isn't Wright's point. For even if there was a disjunctive object consisting of Tom and John, it doesn't follow that "Tom or John" must be interpreted as refering to it. So the converse is also invalid: it doesn't follow from the fact that "Tom or John" doesn't refer that there are no disjunctive objects. The passage rather sounds like Wright has independent reason to believe in the non-existence of disjunctive and existentially general objects; a reason that merely gives further support to the semantic claim that "Tom or John" and "someone" don't refer.
What could the independent reason be? The problem is that I have no idea what "disjunctive object" and "existentially general object" is supposed to mean. Hence I also have no idea whether I believe in these things, and what could possibly be a reason not to believe in them. To be sure, there are candidate objects satisfying the predicates "disjunctive" and "existentially general". Those objects are linguistic expressions, for example the expression "red or green". But certainly such expressions do exist. Some people deny this, I know. But it would be bizarre even by philosophers' standards to hold that disjunctive expressions in particular don't exist, while, say, conjunctive ones do. So these expressions can hardly be what Wright has in mind. There is also a fairly widespread use of "disjunctive" in connection with properties: grue for instance is said to be disjunctive. This usage can be explained (without returning to predicates) by postulating a class of relatively basic properties and saying that disjunctive properties are somehow disjunctively built out of more basic ones. Correspondingly, one could postulate a class of relatively basic objects and introduce disjunctive objects as objects that are disjunctively built out of more basic ones. Still, I have no idea what this disjunctive mode of construction is supposed to be. Anyway, I think this is also not what Wright has in mind. What then?
I don't know if this is what Wright has in mind, but here is what I think of when I think of why "disjunctive objects" are a bad idea.
We'll say that c is the disjunction of the objects a and b just when for each F, Fc iff (Fa or Fb). Then there cannot be any disjunctive objects on pain of inconsistency because provided that for some F, Fa and ~Fb, (which seems required if a is not equal to b!) then Fc, since (Fa or Fb) and ~Fc, since (~Fa or ~Fb). So we have Fc & ~Fc, a contradiction.
So, if by the "disjunctive object" of a and b you mean something satisfying F iff either a does or b does (for each F), this disjunctive object is inconsistent if a and b are distinct.
(I suppose that you could avoid the problem by restricting the domain of quantification over properties, or by rejecting the characterisation of disjunctive objects entirely, or by accepting some contradictions anyway, or perhaps by other means. None of this seems particularly appealing in this context, however.)