One of the traditional arguments for the existence of God – the cosmological argument (CA) – has seen a resurgence of interest over the last 40 years. Ever since Al Plantinga and Saul Kripke pioneered ground breaking work in modal logic by arguing persuasively for the legitimacy of the category of broad logical necessity, philosophers of various ilks such as Bruce Reichenbach, Alexander Pruss, Stephen T. Davis, and Richard Gale have put forward their own versions of the CA, based on the assumption that (a) modality is a real feature of the world, and that (b) human faculties like “conceivability,” “rational reflection,” or something approximating to those, can provide us with reliable (though defeasible) grounds for making claims about possibility and necessity.
I have a preferred way of formulating the CA which, I wager, approximates what these theists have been getting at.
- If something contingent exists, then it can have an explanation for its existence.
- If something can have an explanation of its existence, and there are gains (in terms of explanatory power or prior probability) to be had in postulating an explanation for it, then it probably has an explanation.
- The aggregate of all contingently existing things is itself contingent
- The aggregate can have an explanation (from 1), and there are explanatory gains in positing an explanation for that aggregate
- Therefore, that aggregate probably has an explanation
- The explanation of that aggregate would be a necessary being.
- Therefore, it is probable that a necessary being explains why that aggregate exists
Now, (1) can be defended on the grounds that there is a possible world (we can “conceive” of it) in which the reason why something contingent exists is because God chooses not to annihilate it. This possibility is compatible with something’s existing as a “brute fact,” but it still follows that it could have had an explanation, even if it actually doesn’t.
Premise (2) is a modest version of the Principle of Sufficient Reason which seems to be presupposed by abductive reasoning, or inferences to the best explanation. According to Richard Swinburne, an abductive account is complete if positing a further explanation of that account would add no explanatory power or prior probability to the case in hand. That is where abduction stops and posits a terminus to the chain of explaining. Surely, one can insist on stopping the chain earlier, but this is not how abduction works: note, most of our progress of knowledge in areas like science and philosophy depends on positing explanations (even “unknown” ones) when such gains can be had.
The classic objection to (3) comes form Hume, who thought that, even if an aggregate exists contingently, if one part explains another part, which explains another part, and so one, then as long as all the parts are individually explained, we have no need for an explanation of the whole aggregate itself. But Hume’s reasoning fails on two counts. First, even if he is correct that no explanation is needed, it does not follow that an explanation isn’t better than no explanation. Indeed, if (2) is true, then the aggregate probably has an explanation even if one isn’t needed. Second, Hume’s conjecture is false for aggregates in which parts explain other parts. Suppose a computer virus has always been replicating itself in a beginningless past and will always replicate itself in an infinite future. Even if each copy of the virus is individually explained by another copy prior to it, we still have not explained why an infinite series of viruses exists at all. By analogy then, even if the existence of every contingent thing is explained by other contingents, we have not provided an explanation of the whole. Perhaps Hume would revise his position and say that explanations of each individual are sufficient to explain to whole, but the computer virus analogy undercuts that manoeuvre anyway.
But why not stop the chain of explanation at the aggregate, and deny the second part of premise (4)? Well, we could, but this is implausible if the conditions in premise (2) are satisfied. Are these conditions satisfied? Well, the aggregate can have an explanation. Moreover, positing the existence of a Necessary Being (since it cannot be contingent, otherwise it would be part of the aggregate!), means there is at least a chance that something contingent would exist. It at least gives us some reason to think that contingents would exist rather than none, and therefore raises the probability of their existence, and therefore slightly raises the probability that a contingent material universe would exist. [Note: the existence of a necessarily existing personal cause raises the probability that a complex and orderly material world would exist, but that’s another discussion]. Now consider the alternatives. Naturalism (or some version of the “contingents only” view) gives us no reason at all to think that any contingents would exist rather than not: their existence is just a brute unexplained fact. So, it can be concluded, explanatory gains are secured by positing a Necessary Being.
Finally, there could be several necessary beings, but premise (2) suggests that, all other things being equal, one is enough to do the explaining because no explanatory gains are to be had by positing many.
Anyway, that’s one way to state the CA. Perhaps I’ve skipped some steps with the premises and some tweaking needs to be done, but do you find this version plausible at all? Clearly, the conclusion of the argument need not be the kind of God posited by any revealed religion, such as the God of “Abraham, Isaac, and Jacob” but it does conclude with some form of Ultimate Explanation of the universe, indeed of all contingent things. That would be serious progress in natural theology.