This post completes my discussion of the deductive cosmological argument from contingency. In my previous post, I considered a set of objections to the argument that didn't seem to be persuasive. The moral of that discussion seemed to be that the argument stands or falls with the viability of PSR.
Here, I offer objections to PSR that seem to have some force. These criticisms aren’t original with me, but rather are standard objections (except perhaps the last one, although it's based on ideas of other authors). Furthermore, I don’t mean to imply that there aren’t other versions of the argument from contingency that may avoid these criticisms. However, they do seem to apply to the variants of the argument that one finds in standard “intermediate-level” apologetics books. The criticisms can be divided into two broad categories: (i) those that undercut the reasons offered for accepting PSR, and (ii) those that indicate that PSR is positively false or unreasonable.
1. Type-(i) Criticisms:
1.1 Contrary to what its proponents often assert, PSR does not seem to be supported by reflection on cases. Rather what such reflections support is the weaker principle that objects and events are explained in terms of antecedent causes and conditions. In actual practice, ordinary individuals and scientists explain the existence of objects and events in terms of antecedent causes and conditions, provisionally taking the latter things to be brute facts unless or until they, too, can be further explained. But the prinicple implicit in this sort of search for explanations isn't sufficient to generate the need for an explanation of the universe as a whole in terms of a necessary being.
1.2 Contrary to what some of its proponents assert, PSR does not seem to be self-evident. For what makes a proposition self-evident is that grasping its meaning is sufficient for seeing that it’s true. Consider the two standard categories of self-evident propositions: analytic a priori propositions and synthetic a priori propositions. Both sorts of propositions are knowable independently of empirical investigation of the world. But they differ in that the former (analytic a priori propositions) are tautologous and uninformative, while the latter are not. So, for example, "All bachelors are unmarried" is an analytic a priori proposition, while "Nothing can be red all over and green all over at the same time" is arguably a synthetic a priori proposition.
Now consider PSR: (a) For every object, there is a sufficient reason for why it exists; (b) for every positive state of affairs, there is a sufficient reason for why it obtains. This isn't a tautology; so it's not analytic a priori. Furthermore, although it's a substantive claim, its truth or falsity is not evident merely by reflecting on its constituent conceps. Thus, it doesn't seem to be synthetic a priori, either. Perhaps there is another category of self-evident propositions, but if so, PSR seems not to belong to it. For what makes a proposition self-evident is that one can see that it's true merely be reflecting on its contituent concepts, and we have seen that PSR doesn't safisfy this condition.
1.3 Even if PSR were a presupposition of reason, it wouldn’t follow that it would then be true. But in any case, PSR does not seem to be a presupposition of reason. Rather, again, reason only seems to demand that the existence of each object or fact is explained in terms of antecedent causes and conditions, which are provisionally taken as brute facts unless or until they, in turn, can be explained. Reason does not seem to require anything beyond this.
2. Type-(ii) Criticisms:
2.1 PSR absurdly entails that everything obtains of necessity. The argument for this can be stated as follows. Consider the conjunction of all contingent facts (CCF). By PSR, there is a sufficient reason for CCF. Now the sufficient reason for CCF is itself either contingent or necessary. But it can’t be contingent, because then it would represent a contingent fact, in which case it would itself be a part of the CCF. But contingent facts don’t contain within themselves the sufficient reason for why they obtain – let alone the sufficient reason for why the CCF obtains. Thus, the sufficient reason for CCF must be necessary. But whatever is entailed by a necessary truth is itself necessary, in which case all truths would be necessary truths, and the referents they represent would obtain of necessity. But this is absurd. Therefore, PSR is false. 
2.2 The following scenario is prima facie possible: there are just two kinds of beings that exist: contingent-and-dependent beings (e.g., rocks, trees, planets, galaxies, you and me) and contingent-yet-independent, “free-standing” beings, out of which all contingent-and-dependent beings are made (perhaps matter-energy is like this). If so, then even though there are possible worlds at which the contingent-yet-independent beings don’t exist, they are eternal and indestructible at all possible worlds in which they do exist (interestingly, some theists -- e.g., Richard Swinburne -- take God to be just such a being). On this account, then, there are contingent beings that come to be and pass away – viz., the contingent-and-dependent beings. But the beings out of which they’re made – i.e., the contingent-yet-independent beings -- do not; nor can they. This scenario seems possible. But if so, then since PSR entails that such a state of affairs is impossible, then so much the worse for PSR.
The basic point here is that PSR assumes that dependent beings must have their ultimate explanation in terms of necessarily existent independent beings (beings who exist in all possible worlds), when in fact essentially independent beings (beings that are independent at all possible worlds in which they exist) are all that are needed to do the requisite explanatory work. PSR entails that this isn't enough: if there are any essentially independent, indestructible, free-standing beings, then these must be further explained in terms of a necessarily existent being. But surely this is explanatory overkill, and since PSR entails that such further explanations are required, this implication undercuts any prima facie plausibility PSR may seem to have had.
These criticisms have varying degrees of force. However, it seems to me that criticism 2.2 is an undercutting defeater for PSR, and that criticism 2.1 is a rebutting defeater of PSR. But if these things are so, then the argument from contingency is defeated.
Appendix: Recent Defenses of PSR
A number of philosophers have attempted to revive the Leibnizian cosmological argument in recent years by advancing a weaker version of PSR. According to their version of PSR, every contingent being has a possible explanation in terms of something else. That is, every contingent being is such that there is at least one possible world at which it has an explanation for why it exists. Call this version of PSR, 'Modal PSR'.
Now some authors, such as Garrett DeWeese and Joshua Rasmussen -- offer an argument for Modal PSR. Now I think their argument has a couple of problems, but here I just want to mention one that I think is decisive: The argument uses Modal PSR as a premise to derive the standard version of PSR we discussed above. But this premise is implausible at best, and outright false at worst. For unless they just beg the question and assume that there are no possible beings that lack a sufficient reason, then they must be claiming that, even if there are possible worlds at which a given contingent beings lacks a sufficient reason, there are *other* possible worlds at which it does. But this is implausible, For It seems to me that the only way to accept Modal PSR is to reject origin essentialism. Allow me to me unpack and explain this criticism below:
Suppose origin essentialism is true, and suppose we've got our hands on a universe, and we give it a Kripkean baptism: (pointing to the universe) "Let that be called 'Uni'. 'Uni' is now a Kripkean rigid designator -- it refers to that universe in all possible worlds in which it exists.
So now we have a way to hold Uni fixed, so we can start considering modal claims about it. Given this, there are two relevant possibilities for us to consider: (i) Uni has its origin in the causal power of a divine being, and (ii) Uni has no origin. If (i) is true, then, by origin essentialism, this is an essential property of Uni, in which case there is no possible world in which Uni lacks such an origin. On the other hand, if (ii) is true, then Uni lacks an origin in the causal activity of a divine being, and so this fact about Uni is essential to it, in which case there is no possible world in which it has an origin in the causal activity of a divine being.
The moral, then, is that if we accept origin essentialism like good Kripkeans, then whether a universe has an explanation in terms of a divine being doesn't vary from world to world. But if so, then Modal PSR is of no help unless we know beforehand whether our universe has its origin in the causal activity of a divine being. But if we already knew that, then the contingency argument would be superfluous.
Of course, one could always reject origin essentialism, or restrict its scope in a way favorable to the argument, but then the audience for the argument shrinks considerably.
 This is a rough paraphrase of one of J.L. Mackie’s objections in The Miracle of Theism (Oxford: Oxford UP, 1983), pp. 84-87.
 See ibid.
 This objection is a rough paraphrase of one of Peter Van Inwagen’s objections in his textbook, Metaphysics, 2nd edition (Boulder: Westview Press, 2002), pp. 119-122.
 Here are a couple of possible ways this could be:
(i) The Strong Version: at least one possible independent being x has indestructibility as an essential property: x is indestructible at all possible worlds in which x exists. This is because nothing has what it takes to destroy x at each possible world in which it exists. Now consider the set S of worlds at which x exists. Suppose at least one member of S -- call it 'W' -- is such that x never began to exist in W, and that. It follows from this and the above-mentioned properties that x is independent, indestructible and everlasting in W. However, since x isn't a metaphysically necessary being, there are possible worlds at which x doesn't exist. X is therefore a contingent-yet-independent being of the requisite sort.
(ii) A Weaker Version: there is a being y like x, except that y's indestructibility is world-indexed: it has the property of being indestructible-at-W, where 'W' denotes a possible world. How can y's indestructibility be indexed to W? Because nothing in W has what it takes to destroy y (although things may well have such an ability at other worlds in which y exists). Thus, y is not destroyed at W. This W-indexed fact about y then grounds the fact that W is not destroyed at all worlds counterfactual to W. And if that's right, then y is indestructible in W. Now suppose that, at W, Y never began to exist. Then y is independent, indestructible, and everlasting at W. Y is therefore a contingent-yet-independent being of the requisite sort.
 If one is skeptical that this is possible, one could re-construe it as an epistemic possibility, and thus re-categorize it as an undercutting (instead of a rebutting) defeater of PSR along with the other type-(i) criticisms.
 See, for example: Gale, Richard and Pruss, Alexander. "A New Cosmological Argument", Religious Studies 35 (1999), pp. 461–476; “Hume and the Kalam Cosmological Argument”, in In Defense of Natural Theology: A Post-Humean Reassessment, ed. Douglas Groothuis and James Sennett (IVP, 2005)
 “Hume and the Kalam Cosmological Argument”.