Sunday, December 05, 2010

Liberal Naturalism and the Defeat of the Theistic Hypothesis


I think there is a version of naturalism that seems to explain the relevant range of data better than theism. To be a tad more precise: there is a prima facie viable version of naturalism that (a) explains the data appealed to in theistic arguments at least as well as theism, and (b) there is a range of other data that is better explained by this version of naturalism than by theism. Below I will provide a brief sketch of the sort of view I have in mind, as well as some considerations in its favor vis-a-vis theism.

Thus, consider the following hypothesis, which I'll call 'Chalmersian Liberal Naturalism' (in honor of the contemporary philosopher David Chalmers, who appears to accept a view somewhat similar to it. Call the view 'CLN' for short):

(CLN) There are both abstract objects and concrete objects. The abstract objects are eternal, necessary beings. All concrete objects are composed of just one kind of substance, and its essence has both physical and phenomenal or proto-phenomenal (or at least representational or proto-representational) attributes as a part of its essence (alternatively, the one kind of substance is neither physical nor mental, but the physical and mental are composed of it). Furthermore, this kind of substance is factually or metaphysically necessary. It is also eternal, and comprises a multiverse.

It seems to me that CLN can explain all the data appealed to by the standard arguments of natural theology: we'd expect fine-tuning if for every possible combination of fundamental constants, there is a universe that instantiates it -- indeed, a finely-tuned universe is inevitable on such a hypothesis; we'd expect consciousness in animals and humans if proto-phenomenal states are a part of the essence of concrete substance, since consciousness logically supervenes on structures composed of such a substance when it is suitably complex, and such complexity is accounted for in terms of mutation and natural selection; we'd expect abstract objects if they were eternal, necessary beings; we'd expect moral properties if they logically supervene on certain states of affairs, the latter of which are abstract, necessary beings that contingently obtain or fail to obtain; and the contingency of objects in the world is explained in terms of the factually or metaphysically necessary stuff of which it's composed.[1]

Furthermore, it seems to me that CLN explains a wide range of other data better than the hypothesis of theism. Thus, if CLN were true, then we'd expect the data of huge amounts of prima facie gratuitous human and animal suffering; we'd expect the data of divine hiddenness; we'd expect the data of radical religious diversity; we'd expect the data of scientific studies involving double-blind experiments indicating the ineffectivenss of prayer; and we'd expect the religious demographics data that we actually have. However, we wouldn't expect such data if theism were true.

Thus, it seems to me that CLN explains not only all the data appealed to in theistic arguments at least as well as theism, but it also better explains a wide range of data that is only awkwardly explained if explained at all by the hypothesis of theism. But CLN is a version of naturalism. Therefore, I conclude that naturalism is a better explanation of the range of relevant data than theism.

Some objections and replies:

Objection 1: "CLN is too weird to be true!"
Reply: True, CLN is weird. However, I don't know how to validly argue from "x is weird" to "x is false". A theory accrues support in virtue of embodying various theoretical virtues (simplicity, explantory scope, explanatory power, etc.), and so the theory stands or falls on that basis and that basis alone. Furthermore, CLN is certainly no weirder than the hypothesis of theism. For compare CLN to theism:

T: The world is composed of two really distinct kinds of substance: purely physical substances and purely immaterial substances. Furthermore, these two sorts of substances are capable of interacting with one another. In addition, there are both finite and infinite immaterial substances -- human (and perhaps animal) souls and God -- and the infinite immaterial substance created the finite immaterial substances (and perhaps the material ones, too), and created them without pre-existing materials (i.e., out of nothing).

Things get even more exotic if we move to specifically Christian theism, with its additional doctrines of the trinity and the incarnation. But the point is that both hypotheses -- theism and CLN -- include odd and problematic theses, and when one does a "cost-benefit analysis" of the two views, comparing the oddities and problematic features of the two hypotheses, it seems to me to be, at best, a wash.

In any case, it's a mistake to think that one must be a Liberal Naturalist to accept the conclusions here. One could be a Conservative or Moderate Naturalist -- or even a skeptic or agnostic -- and yet still properly accept the crucial claim here, viz., that whether it's the actual explanation of the relevant data or not, it's a better explanation of the data than theism -- or at the very least: as good an explanation of the data as theism --, in which case the data doesn't favor theism over naturalism.

Objection 2: CLN is too complex to be plausible.
Reply: Two points. First, CLN posits two sorts of entities -- abstract and concrete -- and they require separate treatment. As to the former: Since the abstract objects are posited as necessary beings, they need no explanation. That leaves us with the realm of concrete objects, and here we have postulated one type of substance, which in turn gives rise to a multiverse. Is this hypothesis complex?

Well, it's complex in one sense; in another it's not. The objector mistakenly assumes that there is only one kind of theoretical parsimony, viz., *quantitative* parsimony (i.e., the explanation postulates fewer entities). However, as David Lewis has taught us, another type is *qualitative* parsimony (i.e.,the explanation postulates fewer *kinds* of entities). And while the theistic hypothesis is a much more *quantitatively* parsimonious explanation of the data (it explains all of the data in terms of just one entity, viz., a god), the CLN multiverse hypothesis is a more *qualitatively* parsimonious explanation of the data (since it explains all of the data solely in terms of one *kind* of entity, viz., Chalmersian panprotopsychist substance). And it's not clear which type of theoretical parsimony is more important here.

[1] Objection: "but I can imagine the fundamental stuff failing to exist. And since conceivability is sufficient evidence for possibility, it's possible for the fundamental stuff posited by CLN to fail to exist, in which case we have reason to doubt that such stuff is metaphysically necessary, in which case it can't explain the data of contingency." Reply: Either conceivability is sufficient evidence of possibility or it isn't. If it isn't, then of course the data of the conceivable non-existence of a Chalmersian multiverse isn't sufficient evidence of its possible non-existence, in which case the objection fails. On the other hand, suppose conceivability is sufficient evidence of possibility. Then since it's conceivable that both God and the Chalmersian multiverse fail to exist, then there's sufficient evidence that it's possble that both God and the Chalmersian multiverse fail to exist, in which case it looks as though no being of the relevant sort could be metaphysically necessary, in which case the jig is up for arguments from contingency, in which case contingency falls out of the range of data that needs explaining. Either way, then, the objection fails.


Matt McCormick said...

This is interesting, Ex. Nice and tight, and clear. My own suspicion is that you can explain all of that while dispensing with the weird phenomenal properties that Chalmers thinks supervene on physical matter, but I'll play along.

I suspect that what the theist who understands your argument might say is that there are other things you haven't mentioned that are not explained or not explained as well without supernaturalism. I don't know what they'd put on this list, but maybe moral truths, or maybe they'd insist the consciousness properties need explaining.

And unfortunately, what's on your grocery list of all that needs to be explained doesn't really admit of clear resolution when people differ. And whether or not some X is adequately explained also tends to be something of a Bayesian judgment call, if you know what I mean. Don't get me wrong, I think your argument has the right approach. But I suspect it will fail to convince because of alleged incompleteness in its explanatory power. Make sense? Keep up the excellent work.

Matt McCormick

exapologist said...

Hi Matt,

Thanks very much for your kind words and commments!

Perhaps I should've emphasized a bit more in the post what I only mentioned in passing, viz., that while I think CLN is actually a better explanation of the relevant data than theism, I'm happy if others find CLN to be at least as good an explanation of the data. For then the data, when considered as a whole, doesn't favor theism over naturalism. And the latter, weaker claim is all that is required for CLN to function as a defeater for theism (although here I'm of course leaving aside consideration of candidates for independent, prima facie justification of theism, such as those discussed in the reformed epistemology literature).


Mike Almeida said...

Hi EA,

There are two competing hypotheses, right? Theism and CLN each offer possible explanations, and CLN is the better explanation. But this argument invites reductio. CLAIM: If the theistic hypothesis offers one (not so great) explanation, then it offers the only explanation. Assume for reductio that CLN is true. Since there are infinitely many universes in the multiverse, there is at least one universe in which God exists. If God exists in one universe he exists in them all. Therefore the CLN hypothesis is false and CLAIM is true. The only available response that I can see is to deny the initial assumption in the argument: viz. that there are two possible explanations for everytying there is: God and CLN. If God is a possible explanation whose probability of explaining the universe is (say) barely non-zero, then he's the actual explanation. You might argue that the God explanation has zero probability of being right. But that's another argument altogether.

exapologist said...

Hi Mike!

A couple of clarifications:

First, I'm taking the two competing hypotheses to be possible in the epistemic sense, not the metaphysical sense (and I deny that all epistemic possibilities are metaphysical possibilities).

Second, on my construal of CLN, there are infinitely many universes. But that is not to say that CLN contains all possible universes (including universes with God). Rather, on my CLN, each universe contains no concrete objects beyond the Chalmersian panprotopsychist substance(s) (here I'd like to bracket discussion of dispositional essentialism and related matters re: panprotopsychist universes with different laws).


Mike Almeida said...

we'd expect fine-tuning if for every possible combination of fundamental constants, there is a universe that instantiates it -- indeed, a finely-tuned universe is inevitable on such a hypothesis

This generates a problem for the fine-tuning claim above. If you have infinitely many universes, but not every possible universe, there is no assurance you have every possible combination of fundamental constants. So it is definitely not inevitable that you do.

Second, if there are universes with God that are not part of the multiverse, then God is in the multiverse too. Has to be.

Last, for the reason you gave, epistemic possibilities are not assured positive probability. (setting aside my view that epistemic possiblities just are genuine possibilities). But if for all you know the theistic hypothesis has probability 0, then it is not a competing explanatory hypothesis. In short, any candidate for such an hypothesis has to be minimally (metaphysically) possible.

Site Meter