Let me articulate what I find problematic about mainstream contemporary metaphysics from the point of view of philosophy that wishes to be scientifically informed and open to learning from and be surprised by science. I will discuss one of the central player of that scene, Ted Sider, and point to its debt in David Lewis.
Sider writes this:
"The ontological realist draws the line in a certain place: part of the world’s distinguished structure is its quanticational [sic--it's a pre-print so maybe not in final version]structure. Those who regard ontological realism as “overly metaphysical” should remember that they too must draw a line.
And in fact, the ontological realist can give a pretty convincing argument
for his choice of where to draw the line. Quine’s (1948) criterion for ontological commitment is good as far as it goes: believe in those entities that your best theory says exists. But in trying to decide how much structure there is in the world, I can think of no better strategy than this extension of Quine’s criterion: believe in as much structure as your best theory of the world posits. The structure posited by a theory corresponds to its primitive notions—its “ideology” in Quine’s (1951) terminology—which includes its logical notions as well as its predicates.
This criterion is as vague..."
http://tedsider.org/papers/ontological_realism.pdf [It's in the Chalmers, Manley volume discussed on this blog recently]
So, the story go like this: in Quine (first order) logic is a) a tool in regimenting our scientific language so that we can generate a perspicuous scheme to articulate or read off our ontology. Moreover, given Quine's linguistic idealism [Jody Azzouni's characterization] (I mean, of course, his semantic holism), logic plays a second role of b) holding together the core of the whole scheme. It's this two-fold function that we may call its "ideology." In principle, Quine is open to whole-scale revolutions of the general web, including giving up standard logic, but these are extremely rare occurrences in principle. (No Carnapian principle of tolerance.)
Now, Sider comes along and says, why separate the underlying logic (quantificational structure) from the scientific theories? All scientific theories have an underlying logic (see the second role of logic in Quine), and we are committed to this. Moreover, this underlying logic captures what David Lewis calls 'natural properties.' So, once one gives up on Quine-ean holism and returns to realism, then there really is no reason to call the underlying logic/quantificational structure an ideology--it just is the way nature is carved up, etc. (Moreover, Seider also seems to argue that having an underlying logic is indispensable to science--I found this argument a bit hazy.)
Sider relies on David Lewis' argument in favor of natural properties, which is really a kind of transcendental argument--if reference is possible the world must have natural properties; there is reference, ergo, future/final scientific language must include natural properties (and, thus, have a classical logic).
If I get this right, then this would explain the general hostility among 'core' metaphysicians to non-classical logics (as I was surprised to discover this during my stay at Syracuse); if pluralism about logic gets a beach-head, then claims about natural properties and shared underlying logic (or a general quantificational structure) don't get off the ground.
So, what's wrong with Sider's view? I can think of six proposals:
i) it throws away Occam's razor unnecessarily. Even if there is an underlying logic, quantificational structure really isn't indispensable in science or in justifying science;
ii) it is dogmatic about logic (in this sense of, can't take alternative logics seriously)
iii) it is dogmatic about (future) science; it can't take conventionalism in science seriously, (and takes the mere possibility of quantification way too seriously);
iva) quantificational structure is too coarse-grained to have any connection with the content of the sciences.
IVb) the sciences don't and won't exhibit natural properties. (Here I am with Ladyman & Ross!).
[iva-b, alternatively claim that Sider's metaphysics is constructed in such a way as to make appeals from science unrewarding.]
ivc) There is a conflict between first-order and second-order axiomatizers, which is the preferred one? [It sometimes seems that Sider's quantificational structures are part of a future, unspeakable language.]
V) Why can't there be non trivial different axiomatizations (constructive, non constructive) of particular sciences? (Start making Cartwright/Dupre style noises.)
VI) Deny that the sciences refer in the way this is understood by Lewis/Sider--argue that science is operationalism all the way down.
No doubt not all of these proposals are equally promising or popular. No doubt Sider and his fans have responses...I want to thank Erik Curiel, Jody Azzouni, Matt Frank, Graham Priest, and Mark Barber for discussing some of these reflections in emails with me this week.