Wednesday, March 10, 2010

Factually Necessary Beings, Modal Epistemology, and the Leibnizian Cosmological Argument

Hi gang,

I'm still really busy, but an objection occurred to me regarding Craig's recent defense of the Leibnizian cosmological argument, and I'd like to get some feedback to see whether I'm on to something. If I am, then I'd like to add the point to a paper I'm working on. Here is a first pass at the criticism. Feel free to have at it while I'm away!


On another occasion, I argued that Craig has so far failed to justify a seemingly crucial claim in his revived Leibnizian cosmological argument, viz.,

PQFE: It’s possible that the fundamental constituents of material reality (quarks, say) fail to exist.

Here is a more worrisome objection. Even if Craig were to justify PQFE, it wouldn't help him infer

QNEG: The fundamental constituents of reality (quarks, say) have a necessary being as their explanatory ground.

Indeed, I will demonstrate the stronger claim that even if Craig were to justify PQFE, it would fail to provide evidence that would even slightly favor QNEG over at least one epistemically possible naturalistic rival hypothesis.

The Leibnizian cosmological argument is often presented in a way that suggests that there are only two possible sorts of beings:

(i) contingent, dependent beings
(ii) necessary, independent beings

If one could justify such a categorization, then one could properly conclude that all contingent beings are dependent, in which case one will have gone a considerable distance toward justifying the inference to a necessary, independent being as their explanatory ground.

However, this categorization of beings is dubious, for two reasons. First, necessary dependent beings seem epistemically possible. Indeed, Christian philosophers often take the second person of the trinity to be just such a being. On the sort of account I have in mind, God the father is a necessary being, and he necessarily and eternally wills the existence of the Son as an act of essence. On this account, then, the Son exists in all possible worlds, and is thus a necessary being. However, despite this, his existence is dependent on the causal activity of at least one other being, viz., the Father. Therefore, on this account, God the Son is a necessary yet dependent being.

Second, contingent independent beings seem epistemically possible. Indeed, if Richard Swinburne is right, the first person of the trinity is just such a being. For on his account, God the father is a merely factually necessary being. Thus, God the father is a being that fails to exist in at least some possible worlds. However, he is an existentially independent, free-standing being who is everlasting and indestructible at all the worlds in which he does exist. Therefore, on Swinburne's account, God the Father is a contingent yet independent being.

In light of the preceding, then, it appears that the previous categorization of types of beings is inadequate, and that a more neutral way of carving up epistemically possible space would look like this:

(i) contingent, dependent beings
(ii) contingent, independent beings
(iii) necessary, dependent beings
(iv) necessary, independent beings

One implication of this categorization will prove important for our purposes: one can’t automatically infer “dependent” from “contingent” without further argument.

So that’s the setup. Here’s the punch line: It’s epistemically possible that the fundamental constituents of matter (quarks, say) are contingent yet independent beings. But if so, then it's epistemically possible that such beings fail to exist in at least some metaphysically possible worlds, and yet they lack an explanation for their existence beyond the de facto lack of things that can annihilate them in the actual world and the relevant counterfactual worlds. And if that's right, then even if Craig's account of conceivability is a good guide to metaphysical possibility, and even if we can properly conceive of the non-existence of quarks, such modal evidence fails to justify the claim that quarks are dependent beings requiring a necessary being for their explanatory ground. For while such conceivability evidence is just what one would expect if quarks were contingent dependent beings, it’s also just what one would expect if quarks were contingent independent beings. But while the former is a theistic hypothesis, the latter is a naturalistic hypothesis. But if so, then the evidence from conceivability doesn't favor the theistic hypothesis over the naturalistic hypothesis.


Marc said...


I'm attracted to discussions surrounding the PSR and the Leibnizian cosmological argument, so I appreciate your efforts here. I'm not terribly acquainted with these discussions, unfortunately, so I hope that the following comments don't make this fact especially obvious.

As an Anselmian Trinitarian, I don't think I'd be inclined to accept your Trinitarian motivations for expanding the "types of beings" category.

Regarding the first motivation, very briefly, it seems to me that this particular depiction of the Son's generation (or begottenness) engenders some unpalatable consequences for the essential coequality and consubstantiality of the divine persons. To illustrate one of these consequences, suppose the Father's essence E contains the property being the generator of a divine person. Call this property P. Most (perhaps all) proponents of the doctrine of the Son's generation maintain that the Father communicates the whole divine essence to the Son, which, presumably, is E. If the Father has communicated E to the Son, that entails that Son's essence is also E. But then that entails that the Son, exemplifying P, should be generating another divine person, and that this new person should be generating yet another divine person, and so forth ad infinitum. But this is absurd: there aren't an actually infinite number of divine persons in the Godhead, so something must be wrong with the doctrine in question.

Regarding the second motivation, the Anselmian theist, as you'd expect, wouldn't endorse Swinburne's supposition that God (the Father) is a brute contingent -- or, as you put it, a contingent independent being. In my judgment, it seems more plausible to hold that a maximally great God exists than to hold that a God* exists who exemplifies every excellent-making property except for one, and, furthermore, that this God's* existence inexplicably obtains in some worlds and inexplicably fails to obtain in others. Ironically, in Philosophical Foundations for a Christian Worldview, Craig and Moreland note a contention from Swinburne which may be incompatible with his belief that God (the Father) is a brute contingent. Swinburne's contention is as follows: "it is simpler (or perhaps less ad hoc) to posit either zero or infinity as the measure of a degreed property than to posit some inexplicably finite measure" (498). Perhaps this consideration doesn't apply, however, to the excellent-making property at issue. But I'm guessing it can be appropriately adjusted so as to be more pertinent.

Are there additional motivations regarding the expansion of the "types of beings" category to which an Anselmian Trinitarian might be more sympathetic?


-- Marc

TaiChi said...

Another candidate for a necessary and dependent being: the world. Suppose, prior to creation, God and only God existed. He creates the world for a reason, and this reason is a necessary one (as the PSR seems to indicate). So the world, as a consequence of a necessary being having his necessary reason, is itself necessary. But the world is dependent on God's act of creation. Thus we have a necessary dependent.

exapologist said...


Thanks for your comments.

Sorry for not being clearer. My aim is to provide a neutral categorization of types of beings. As such, I include possibilities not ruled out by by antecedent large-scale worldview commitments, such as Anselmian theism (or naturalism, for that matter. This is one reason why I chose pro-theistic illustrations). As such, I'm considering the space of candidate possibilities not ruled out prior to evaluating the evidence for and against theism, naturalism, etc.

@TaiChi: Nice example.

Marc said...


Thanks for the helpful clarifications. In light of them, I hope you don't mind tolerating some further comments.

The first concerns necessary, dependent beings (NDBs). Even if we admit NDBs into our ontology, I'm worried about the extent to which they facilitate opposition to the Leibnizian argument. Insofar as the opponent wishes to avoid commitment to the world's being contingent and dependent, the appeal to NDBs seems to afford no support, even if they're epistemically possible. As the their name entails, a precondition of NDBs is a dependence relation in which one of the relatum is a necessary being. A contingent, independent being (CIB) would be more availing, the type of being which your particular criticism endorses.

My second comment concerns CIBs. Of such beings, you said that they're "existentially independent," "free-standing," and "everlasting and indestructible" in all of the worlds in which they exist. Letting P be the property being indestructible, I'm curious about what accounts for CIBs' exemplifying P.

Regarding this issue, you noted that CIBs "lack an explanation for their existence beyond the de facto lack of things that can annihilate them in the actual world and the relevant counterfactual worlds." So, at least one of the reasons why CIBs have P is explained by de facto absence of things which can annihilate them. This appears to imply that there's nothing intrinsic to CIBs which account for their having P. In other words, it's not the case that CIBs are essentially indestructible. Rather, in the worlds in which some CIB exists, as it happens, there simply isn't anything which can destroy it. The de facto absence of suitable CIB-destroyers, then, seems to entail that P is an accidental property of CIBs. But if P is accidental property of CIBs, then there's at least one possible world W in which a CIB exists and a CIB-destroyer exists. This invites the question: how do we know we're not in W? Unless P is an essential property of CIBs, this particular line of opposition to the Leibnizian argument seems vulnerable to this modal worry. If this is right, I suspect the opponent will be inclined to claim something stronger about CIBs than that their indestructibility is the contingent result of there being no CIB-destroyers.

Do you foresee any problems associated with claiming that P is an essential property of CIBs? I need to give it more thought. Perhaps such a claim accords with what you had in mind from the beginning.

Here's a final thought. Revisiting your categorization of beings,

(i) contingent, dependent beings
(ii) contingent, independent beings
(iii) necessary, dependent beings
(iv) necessary, independent beings

the opponent's project looks more promising if, as you've suggested, CIBs are epistemically possible. Not so with respect to types (i), (iii), and (iv), as they seemingly favor the theist's project. Thus, if three out of four being-types favor the theist, and only one out of four being-types favor the opponent, does this render the theist's project more inductively probable? Put differently, given any being B, is it more probable that B is of type (i), (iii), or (iv) than of type (ii)?


-- Marc

Site Meter