The 'what if God commanded something horrific?' objection to DCT and W.L. Craig's moral argument

What if God commanded something horrific? What if, for example, God commanded you to torture another person? Would it be all right if you obeyed him? Would it be obligatory for you to obey him? Maybe you don't think torture is categorically wrong, if so replace it with something you do think is categorically wrong. If you don't think anything is categorically wrong—that every action has some condition under which it's the right thing to do—just relativize the command to fail that condition. What if God commanded you to torture another person for the fun of it, or for the express purpose of causing them bodily harm? Surely then you should not obey him.

And yet on divine command theory you should obey him; you have a moral obligation to obey him. This is because, on divine command theory, obligations simply are divine commands. In other words, facts about how we ought to act reduce to facts about what God has commanded us to do. This is in the same way that facts about the morning star reduce to facts about the evening star; they are one in the same thing.

This seems like a vivid counterexample to divine command theory, and yet many are not persuaded by it. Take William Lane Craig, for instance, who responds to this objection by pointing out that, on some theories of counterfactual semantics, a counterfactual with an impossible antecedent is, if true at all, only vacuously so. That counterpossibles (which are what such counterfactuals are called) cannot be informatively true. On this account an impossible antecedent entails everything, and so while it's true that “if God commanded torture, we ought to torture”, it's likewise true that “if God commanded torture, we ought not torture”. So, Craig thinks, however he answers the question, whether it follows that we should or should not commit such horrific acts, it doesn't matter to the success of divine command theory.

This is similar to how material implication with an impossible (or, in this context, inconsistent) antecedent entails anything. That is, if you assume a contradiction you can derive literally any proposition you'd like. But there is one crucial difference between material implication and counterfactual statements. Material implication is a function of its primitive propositions, where as a counterfactual is not. In other words a conditional statement is true in virtue of not having a true antecedent and a false conclusion, the content of the antecedent and conclusion are irrelevant. But for a counterfactual the content of the antecedent and conclusion are relevant to the truth of the statement. The counterfactual is, after all, true in virtue of accurately describing some aspect of reality, and not in virtue of its components. It is obviously not the case that “if it never rained in Vancouver, we would still have beautifully lush forests”, even though the antecedent is false.

Craig's answer might appear attractive at first, but it has unforeseen consequences. If counterpossibles are at best uninformatively true, then any argument which relies on them will be rendered useless. This is ironic because Craig tries to use divine command theory to defend his moral argument, which itself contains counterpossible premises;

1. If God did not exist, then objective moral values and duties would not exist
2. If God did exist, then there would be a solid ontological foundation for objective moral values and duties

And since objective moral values and duties do in fact exist, he concludes that God must as well. Strangely Craig runs two arguments at the same time; statement (1) being the premise of a modus tollens argument, and statement (2) being the premise of an inference to the best explanation. Maybe his intention is that, if the deductive argument fails, he always has the abductive argument to fall back on or vice versa.

Nevertheless, as you I'm sure can already see, either statement (1) or (2) must be a counterpossible. For if God exists then, at least on Anselmian theology, he cannot possibly fail to exist. And, likewise, if he does not exist, then he cannot possibly exist. Craig, being an Anselmian theist, must then regard statement (1) as uninformative, since by his view of counterpossibles he's committed to believing that “if God did not exist, objective moral values and duties would in fact exist”. And so statement (1) cannot factor into his reasoning for believing God exists—the modus tollens moral argument is rendered useless.

Of course at the same time Craig cannot expect an atheist to accept statement (2) without begging the question, for within the atheistic world view God is impossible. And so (assuming Craig's view of counterpossibles) it's also the case that, within the atheistic world view, “if God existed, then there wouldn't be any ontic foundation for objective moral values and duties”. Whatever explanatory advantage theism would seem to have goes out the door, and with it Craig's abductive argument.

Now what is interesting is how this bleeds into the rest of natural theology. While there are many arguments for God that don't explicitly rely on counterpossibles, it would seem they implicitly all do. It's hard to see how “God is the best explanation for the origin of the universe” could be true, when “if God were to to not exist the universe still would” is true as well. Or note that “The probability of a life permitting universe, on theism, is high” seems to really just mean “If God existed, then a life permitting universe would probably exist”. But then it's also the case that (on atheism, at least), “if God existed, then a life permitting universe probably wouldn't exist”. And so every interesting argument for God would seem to be rendered question begging or uninformative, or some mix of the two.

It has been suggested to me that, while “God does not exist” is merely metaphysically impossible, “God commands torture” is logically impossible. The distinction being that, for whatever reason, logical impossibility renders a counterpossible vacuous where metaphysical possibility does not. There is a ring of plausibility to this. It's agreeable that, whether our intuitions were the product of natural evolutionary pressures or divine action, they were probably not made to be reliable in inconsistent scenarios. And so if the thought experiment I've contrived is indeed logically impossible, we should take whatever intuitive force it has to offer with a grain of salt.

But is “God commands torture” really logically impossible? I don't see why it would be—it doesn't seem to entail any contradictions. Maybe the reader wants to say that God wouldn't do such a thing, but then that's adding a conjunct. If something is logically impossible then it must entail a contradiction on its own, without the aid of further conjuncts (or else it's really just the conjunction which is logically impossible).

Maybe the reader has a more broad sense of logical impossibility in mind, such as involving some sort of conceptual inconsistency. That, by conceptual analysis, we can tease a contradiction out of “God commands torture”. God is defined as being morally perfect, so to say God was of any less than impeccable virtue would be supposing a conceptual absurdity. But is commanding torture and being morally perfect in conceptual tension? This could only be if the what we meant when we spoke about moral perfection was, among other things, a disposition to not command the torture of others. But this cannot be, for there are people—mainly young children—who are so innocent of evil deeds that they have no concept of torture (and many other immoral acts), and yet still surely understand the concept of moral perfection. Such a person, while fully understanding the definition of 'God' and of 'morally perfect', could not simply reduce “God commands torture” to absurdity by conceptual analysis.

Some might think; but of course it's conceptually absurd to think God could command evil! Yes, I agree. The problem is that torture isn't simply evil by definition, and so it doesn't follow by sheer conceptual analysis that a morally perfect person would have a negative attitude or disposition towards torture.

Of course it still seems true that God, were he to exist, would not issue such horrific commands. But it cannot be true simply in virtue of what the word 'God' means. And so “God commands torture” must be metaphysically impossible, if impossible at all.