Saturday, April 12, 2014

Defeasible Reasoning in Cosmological Arguments for Theism and for Naturalism

(Revised)

Perhaps one of the most important recent innovations with respect to cosmological arguments is the application of groundbreaking work in defeasible reasoning to the formulation of their core explanatory principles (Cf. Joshua Rasmussen and Robert Koons). Defeasible causal and explanatory principles have the following basic form:
(DP) Normally, entities of type T have a cause or explanation.
The "normally" operator indicates that it is a default explanatory principle. Principles of this sort may or may not have exceptions; they are general rules of thumb. From a dialectical point of view, default principles may be preferable to ordinary principles for at least two reasons: (i) they're more modest than unqualified principles, and thus easier to justify, and (ii) when justified, one can't reject the implication that they apply in the given case at stake unless one has a good reason to think that they don't in that very case;  unlike unqualified principles, not any old counterexample is going to do the trick. In this way, they shift a very strong burden of proof onto the "skeptic".

As mentioned above, both Koons and Rasmussen have proposed default causal/explanatory principles in their versions of the cosmological argument. Here they are:
(DPK) Normally, a wholly contingent situation has a cause. 
(DPR) Normally, for any intrinsic property p that (i) can begin to be exemplified and (ii) can be exemplified by something that has a cause, there can be a cause of p’s beginning to be exemplified.
It seems to me that there are arguments for naturalism that should likewise be revised in light of these advances. Here's an example. On previous occasions, I have appealed to a principle of material causality to argue that classical theism is false:
(PMC) All concrete objects with originating or sustaining causes have material causes of their existence.
So far as I know, PMC has no clear counterexample. However, the sorts of considerations sketched above lead me to think that I need not appeal to a causal or explanatory principle anywhere near as strong as that. Rather, I can just rely on a weaker default version of the principle:
(DPMC) Normally, objects with originating or sustaining causes have material causes of their existence.
Then if we replace DPMC for PMC in the previous version of my argument, we get:
1. If classical theism is true, then the universe has an originating or sustaining cause (or both) without a material cause of its existence.
2. Normally, concrete objects (or collections of such) that have an originating or sustaining cause (or both) have a material cause of their existence.
3. The universe is a concrete object (or collection of such).
4. Therefore, classical theism is false.
Assume DPMC is warranted, along with the other premises. Then as with the cosmological arguments that deploy DPK and DPR, the only way for one to rationally resist the conclusion is to give an adequate reason for thinking the principle does not hold in the crucial case in play, viz. the origin or existence of the universe. Not any old counterexample is going to cut it. 

2 comments:

Obsidian said...

Craig actully gives a response to a similar objection by Morriston to the CP in the Blackwell Companion to Natural theology .

exapologist said...

He does, but most of them are irrelevant. Remember that my causal premise is restricted to just concrete objects (as defined in the post, "Theism and Material Causality"). As such, the counterexamples he proposes that don't refer to concrete objects as defined are irrelevant.

But the main reason why his proposed counterexamples are irrelevant to the argument is that most of them are attempted counterexamples to an unrestricted principle of material causality. But the beauty of a default causal/explanatory principle is precisely that it can grant counterexamples. With such principles, the only sort of counterexample that's relevant is one that applies in the type of case at stake -- in this case, the origin or sustenance of the universe without a material cause. On the nature of default reasoning, the SEP entry on default reasoning, as well as the papers by Koons and Rasmussen, linked to in the post above.

Having said all of that, it's instructive to see that his counterexamples aren't even convincing against an unrestricted version of the principle of material causality. So, for example, his example of models of the universe that construe it created from zero energy (in the sense that positive and negative energy are perfectly balanced) is just a semantic trick (that's not nothing in the relevant sense); his example of space cloning itself isn't a counterexample at all, as it involves creation from prior space (and he notes parenthetically that it's currently speculative how space actually clones itself (and would require a substantivalist construal of space to even make the case relevant to the causal principle at issue, which is controversial).

What about his account of libertarian agent causation? As I pointed out in the post alluded to above, this will only be persuasive to those who are accept libertarian agent causal views of the self (a tiny minority of philosophers), the bulk of which are Christian theists. As such, this sort of counter-example is question-begging at best, implausible at worst. In any case, it's not clear that libertarian agent causation requires creation ex nihilo, but only transfer from energy outside the physical causal system to inside it. But that's just a case of material causality.

Finally, what about his appeal to the Kalam argument's case for a necessary beginning of the universe? As myself and others (esp. Wes Morriston) have argued, all of his philosophical arguments on this score are subject to undercutting defeaters, and his appeal to the BGV theorem is shaky at best. For the latter to suffice to ground resisting the default causal principle of material causality, the evidence for the former would have to be at least as strong as the grounds for the latter, which it clearly doesn't (or so it seems to me).

Site Meter