The following three steps are fairly standard (I’ve seen the third step in a talk this year by Wes Morriston–does anybody have an earlier source?).
Step 1: Consider the conditional:
- Even if God had commanded it, you shouldn’t torture the innocent.
From (1), we can argue that in the possible world where God commands torture of the innocent, it is still wrong, and hence it is not the case that right and wrong are defined by what God commands. (Minor issue: The antecedent of the conditional really perhaps say “if God commanded it and did not forbid it”.)
Step 2: Because of God’s nature, God cannot command torture of the innocent.
Step 3: Let’s grant this. Still:
- Claim (1) is a non-trivially true per impossibile counterfactual.
- From (1) and (2) it follows that right and wrong are not defined by God’s commands.
Now here is where I want to add a new step to the dialectics:
Step 4: One should deny the conjunction of (2) and (3). The first approach is this. Consider the statement:
- Even if it were right, you still shouldn’t torture the innocent.
I think that the intuitions that pull us to affirm (2) equally pull us to affirm (4). But (4) is, on reflection, absurd. And in any case, (4) should not make us deny that what is right is right! But maybe you’re not convinced. Maybe you find (4) ridiculous. Fine. Take whatever metaethical theory you think is right. For concreteness, suppose it’s Kantianism. Consider:
- Even if the categorical imperatives required it, you still shouldn’t torture the innocent.
I think (5) is as plausible as (1), and analogues to (2) and (3) where (5) replaces (1) are just as plausible as the originals. Hence, if the argument in Step 3 is a good argument against divine command metaethics, it is a good argument against every non-trivial metaethical theory, and if I am right about the plausibility of (4) it might even be a good argument against the trivial metaethical theory (what is right is right). Hence, the argument in Step 3 is not a good argument against divine command metaethics. Whether the problem is with (2) or with (3) is something I do not know.
Final remark: I find myself with some intellectual akrasia here. I still find (1) a plausible argument against divine command metaethics, despite the criticism. This suggests that there is something about (1) that I am not managing to capture here.
James Rachels ‘God and Human Attitudes’ in Paul Helm (ed) Divine Commands and Morality (OUP, ’81) offered the following argument for the incompatibility of there existing both a being that is worthy of worship and autonomous moral agents.
1. Necessarily, if God exists then He is worthy of worship.
2. It is impossible that some being is worthy of worship.
3. Therefore, it is impossible that God exists.
Phil Quinn, ‘Religious Obedience and Moral Autonomy’ (also in P. Helm) submits his doubts about premise (2) and (more or less) plausibly reconstructs Rachels’ argument for (2) in this way.
4. It is impossible that some being is worthy of worship and there are some moral agents.
5. Necessarily there are some moral agents.
2. Therefore, it is impossible that some being is worthy of worship.
In defense of (4) Rachels commits himself to a Kantian or neo-Kantian conception of moral agency. He says, for instance,
“to be a moral agent is to be an autonomous or self-directed agent . . . The virtuous man is therefore identified with the man of integrity, that is the man who acts according to precepts which he can, on reflection, conscientiously approve in his own heart” (43).
If (4) is true then a genuinely self-directed moral agent could not co-exist with a being that is worthy of worship. More on this puzzling premise in a moment.
In his Divine Commands and Moral Requirements (Oxford: Clarendon Press, 1978), Philip Quinn terms the claim:
If God did not exist, then everything would be permitted.
‘Karamazov’s Thesis’ (KT). Are divine command theorists committed to KT? And are divine command theorists then committed to saying that if God did not exist, stealing this diamond ring would be permitted?