Is divine command theory either false or circular?
June 15, 2004 — 10:41

Author: David Efird  Category: Religion and Life  Comments: 25

One of the standard objections to divine command theory (DCT) can be framed in terms of the following argument.
(1) Necessarily, if God commands that subject S commit action A, then it is ethically right for S to commit A.
(2) Possibly, God commands subject S to commit an act of wanton cruelty.
So, (3) Possibly, it is ethically right for S to commit an act of wanton cruelty.
How should the DCTer respond to this argument?

Now I think that the DCTer has to accept, or, hedging my bets, ought to accept (2). And the DCTer can’t deny outright (1). So, it looks like the DCTer has to accept (3). That’s Ockhamist way, and it’s a hard road to follow. If the DCTer has to accept (3), then I think her theory is false.
But there is another, more subtle manoeuvre available to the DCTer, one that is suggested by Robert Adams’s ‘A Modified Divine Command Theory of Ethical Wrongness’. (Note: the following is not consistent with Adams’s revised theory in ‘Divine Command Metaethics Modified Again’. I think that the revised version loses a crucial insight of the first modified DCT.)
The trick to defeating the argument is to modify (1) so that the necessity is not an absolutely unrestricted necessity, but rather a necessity relative to God’s loving his creatures. So (1) then becomes:
(1*) Given that God loves his creatures, necessarily, if God commands that subject S commit action A, then it is ethically right for S to commit A.
The thought motivating (1*) is that ethical claims have truth value only in worlds in which God loves his creatures. And (1*) together with (2) does not entail (3) because there is no world in which both God loves his creatures and he commands an indiviudal to commit an act of wanton cruelty; consequently, the claim that it is ethically right for S to commit an act of wanton cruelty is either false (in all worlds in which God loves his creatures) or it is without truth value (in all worlds in which God does not love his creatures). Ockhamism is thereby avoided consistently with DCT.
But there is a problem, I think. It is that this manoeuvre seems to render DCT circular. A necessary condition on this version of DCT is that the restriction on the necessity operator in (1) not presuppose any ethical concept. This is so because (1) is meant to characterise ethical rightness on the basis of God’s commands, and if that’s our aim, we can’t use ethical concepts in characterising God’s commands. But it seems that the restriction ‘God loves his creatures’ on the necessity does employ ethical concepts. For to love a person is to value that person, to promote her well being and her worthwhile projects, and, in general, to see that person as a source of moral considerability. All this seems highly ethical to me. And so it seems to me that this way of restricting the necessity in (1) renders DCT circular.
Now I wonder if there’s another way of restricting the necessity in (1) that renders the claim that it is ethically right for S to commit an act of wanton cruelty either false or without truth value in the way that ‘God loves his creatures’ does without rendering DCT circular. I can’t think of any right now. Suggestions would be much appreciated.

  • christian

    Im not sure why we can’t deny 2. God would only command what he loved and would not command anything inconsistent with his nature. Facts about what God commands determine moral facts. Facts about what God loves determines what he commands. Nothing determines what God loves other than God’s nature. In some sense, God’s nature grounds all moral claims and his commands make it clear what morality requires.
    However, one would have thought 2 is true because God is omnipotent and accepting 2 requires us to revise our concept of omnipotence. We may have to do this already to reconcile foreknowledge with freedom though. If this is true, then the cost to omnipotence isn’t as heavy, it’s already motivated.

    June 15, 2004 — 12:44
  • Mike

    I agree. Why suppose there is a world in which God commands wanton cruelty (WC)?
    Ans. It is consistent with God’s goodness to make such a command (given (1)), so there is no moral reason against it.
    Rpl. It is consistent with God’s goodness not to make such a command. So there is no moral reason for it.
    Ans. If there is no such world, then God cannot freely command WC. That’s a limit on God’s omnipotence.
    Rpl. If there is such a world, then God cannot freely countermand WC in every world. That’s a limit on God’s omnipotence.
    Ans. The proposition ‘God commands WC’ appears consistent and so is true in some world.
    Rpl. The proposition ‘God countermands WC in every world’ appears consistent, and so true in some world.
    …and so on.
    So I can’t see how affirming (2) is any more reasonable than denying it.

    June 15, 2004 — 13:01
  • Chris

    Me too–deny (2).

    June 15, 2004 — 13:33
  • deny (2).”
    The problem is that premise (2) clearly is a possibility under DCT, which makes no reference to God’s “nature.”
    Moreoever, and analogous to what I argue here, any statement about God’s nature perforce will be normative. Thus, to modify DCT with a premise about God’s nature will be to institute a normative premise without any appeal to DCT. Therefore, DCT is superfluous as a normative theory.

    June 15, 2004 — 15:07
  • Mike

    _The problem is that premise (2) clearly is a possibility under DCT, which makes no reference to God’s “nature.”_
    You’re confusing conceivability with possibility. It is conceivable that (2) is true along with DCT, no doubt. But so what? That is nowhere near showing that (2) is true. It is very old news, for instance, that water is conceivably not H2O, but water is not possibly not H2O.

    June 15, 2004 — 16:42
  • Mike

    On conceivable worlds, as noted above, both of the following are conceivable.
    1.God commands subject S to commit an act of wanton cruelty.
    1a. In every world God commands S not to commit an act of wanton cruelty.
    Neither is inconsistent with DCT or internally inconsistent. If conceivability were a reliable guide to what is possible, we could derive the following from (1) and (1a),
    2. Possibly, God commands subject S to commit an act of wanton cruelty.
    2a. Possibly, in every world God commands S not to commit an act of wanton cruelty.
    Taking the logic of necessity to be at least as strong as S5–as nearly all do–it follows from (2a) that (3) is true.
    3. Necessarily, God commands S not to commit an act of wanton cruelty.
    But (3) and (2) are logically inconsistent. They cannot both be true. Moral of the story: conceivability is not a reliable guide to what is possible. Therefore the fact that it is conceivable that God commands S to commit an act of wanton cruelty is not good evidence that (2) is true.

    June 15, 2004 — 17:17
  • My first thought on reading this post this morning was that Alston answers exactly this worry. I haven’t had a chance to read through the comments carefully yet, but I don’t think any of them addresses the circularity worry, which is one of Alston’s two main interrelated targets. I didn’t get enough time today to review more than a few pages of his paper, though, so I can’t offer much help yet.

    June 15, 2004 — 17:47
  • I guess I should say that my memory has Alston grounding God’s moral goodness in some prior notion of metaphysical goodness to avoid the circularity, but I don’t remember if he answers the worry that there’s nothing grounding the metaphysical goodness. Both would obviously have to be necessarily true, but something necessarily true can still be constituted by something else or explained in terms of something else. I’m wondering if you can keep calling for such an explanation. Doesn’t it have to stop somewhere?

    June 15, 2004 — 17:49
  • Mike

    This is a nice idea. And Alston did propose something like what you note. There is the alternative too of having moral requirements supervene on God’s commands or having those requirements supervene on God’s commands plus non-moral facts, or having them supervene on such facts without identity with such facts (in contrast to the “being gold = having atomic number 79” relation or the “being water = being H2O” relation) and so on. I agree entirely that there’s lots to say here in development of DCT.
    I simply want to point up that the initial problem is *much* more difficult to generate than is even hinted at in the PHI 101 version of the problem.

    June 15, 2004 — 19:48
  • In virtue of what nonnormative property would a divine command to engage in wanton cruelty be a metaphysical impossibility?

    June 16, 2004 — 1:15
  • Mike

    It need not be entailed by any nonnormative (or normative) property of God. It might be true that in every world in which God exists he fails to make such a command. This is perfectly consistent with all of the assumptions of the case–including DCT. There is nothing to be said in defense of premise (2) other than ‘it is not inconceivable that God makes such a command’. Right, it isn’t. But it is also not inconceivable that there is no world in which God makes such a command. Let W be the set of possible worlds containing God. It is consistent with those who defend (2) that God fails to command wc is every world except w. So in [W-w] God fails to command wc. Can God also choose not to command wc in w? Those who defend (2) must say no, it is impossible that God fail to command wc in w. I urge that it is consistent with all of our assumptions that God fails to command wc in w AND it is also consistent with all of our assumptions that God commands wc in w. So those who assert (2) are in no better position than those who deny it.

    June 16, 2004 — 8:28
  • Mike,
    The problem with this distinction between possibility and conceivability as you draw it is that it is trivial: You can be drawn it in all cases of nonactual possibility. Thus, for example,

    (1)It’s possible that I might pilfer sweets from the corner shop.


    (2) Possibly, in no possible world is it the case that I pilfer sweets from the corner shop.

    Thus (from your reading of S5),

    (3) Necessarily, I don’t pilfer sweets from the corner shop.

    Since metaphysical possibility is only of interest in cases of nonactuality, I think the above shows the distinction between conceivability and possibility can’t do any meaningful work in salvaging DCT.

    June 16, 2004 — 11:10
  • Mike

    Much of the work on this issue is collected in John Hawthorne and T. Gendler’s _Conceivability and Possibility_(Oxford, 2002). The relation is not trivial. There are times when conceivability fairly reliably indicates possibility. These are cases concerning (what Peter van Inwagen calls) nearby possibilities. I can conceive of my desk being closer to the north wall of the house. Since this case is so close to home, conceivability is a good indication that this is possible. I conclude (reliably) that it is possible that my desk is so located. But when it comes to more remote possibilities it is not reliable at all. Consider for instance unproven theorems in mathematics or set theory. In most cases, we can conceive of them as being true AND conceive of them as being false: but of course, if true then they are necessarily true and if false then they are necessarily false. So conceivability indicates nothing here. And when it comes what God *can* command, that seems a question about a distant possibility. I can’t see why conceivability would provide a good indication of what is possible here.
    Let me go to your specific case. It seems clear to me that (3) is false. But how? This does not seem like such a remote metaphysical question. Human beings are prone to temptation, akrasia, moral misinformation (maybe in some world you believe that pilfering is right), lesser-of-two-evil situations (maybe in some world pilfering IS right). And of course there are many other familiar situations in which someone steals the candy. But this is not the case at all for God. In the case we are discussing, we can’t imagine what might bring it about that God chooses one way rather than another. In fact, after all of the information is in, there simply is nothing that would have him choose one way rather than another. This is exactly the problem: one series of choices seems just as possible as another.

    June 16, 2004 — 13:17
  • If “reliability” is going to mediate possibility, then that mediating concept needs to be unpacked. Until it is, it merely begs the question.
    But let me make an observation in plain talk: It seems rather remarkable to say that something like pilfering sweets is (1) possible for a minipotent being like me and yet (2) impossible for a maxipotent being like God. One quite naturally wants to ask: “In virtue of what is it impossible for God?” And it seems a very unnatural sort of response to climb up the ladder of formal, modal abstractions and say the question doesn’t need answering.

    June 16, 2004 — 19:44
  • Mike

    This last post on omnipotence is entirely beside the point. Look, while we’re speaking plainly and noting the “remarkable”, it seems rather remarkable that you haven’t seen or addressed the point that one series of choices is (given our information) as possible as another. That is (once again) the very point. I suppose I’d recommend that you try reading Hawthorne and Gendler. This might get us into a contemporary discussion.

    June 16, 2004 — 20:44
  • I’m not familiar with Hawthorne and Gendler, but then I can only proceed on arguments that are actually put in issue.
    1. Equiprobability. The claim is that “one series of choices is (given our information) as possible as another.” I don’t think they are.
    But if they are, then so are my propositions (1) and (3), above: It is loading the dice to state that the conceivability of my willing evil is a more “reliable” indicator of possibility than the conceivability of God’s.
    Now, if one wants to argue that said conceivability is a “reliable” indicator of possibility in the case of humans because humans have certain properties (e.g., being “prone to temptation, akrasia, moral misinformation, lesser-of-two-evil situations”), then it seems to me the balanced approach is to say what divine properties make like conceivability an unreliable indicator of possibility in the case of God.
    2. Omnipotence. In addition, the issue of divine potence seems entirely relevant. God is taken to be have maximal practical ability across all domains of being and action. ‘Ability’ entails possibility–unless one wants to classify as coherent statements like “I am able to do x, but it’s impossible for me to do x.”
    On this standard conception of ability, then, God is either able to will evil or he is not. If he is able, then by semantic entailment under the term ‘ability’ there is a possible world in which he does so, in which case the standard objection to DCT goes through.
    On the other hand, if God is unable to will evil, then there is a conative limitation on him, and this must be in light of some property–or it just is a property. What is that property? Perhaps that’s unclear. But any property that prevents a being from willing evil in specie by definition has a normative dimension. Thus, if evil is conatively closed to God, it is so closed in light of some property having a normative dimension. In which case…the standard objection to DCT goes through.

    June 16, 2004 — 22:30
  • Mike

    Let’s get directly to the problem. (1) seems to me possible. As far as I can tell, you think that (1) is not possible.
    1.God wills that the rule “do no commit acts of wanton cruelty” be a necessary law.
    Isn’t that where things stand? I would add only that I don’t know why it is obvious to you that (1) is not possible.

    June 17, 2004 — 8:16
  • No, I don’t think it’s possible. Let me see if I can explain why.
    There is a natural doxastic asymmetry between possibility, on the one hand, and impossibility and necessity on the other. Our standard presumption is to favor possibility over the other two. So when someone says, “X is possible,” we generally grant it, unless we have a reason to deny it.
    Conversely, if someone says, “X is impossible,” or “X is necessary, not merely possible,” we generally don’t grant it unless we have a reason to grant it. This is asymmetry is a concession to our inelimable epistemic ignorance and fallibility. (We sense that all other things being equal, the chances are against us if we say something is necessarily so or not so.)
    Now it might look like this is an argument in favor of (1*),

    (1*) (1) is possible.

    since (1*) is prima facie a statement about possibility. But that’s illusory. (1*) is a possibility that if actual is necessary. That kind of nested possibility does not dovetail with our natural presumption to grant what I’ll call first-order possibility, which one can define as those possibilia that if actual are not necessary.
    Thus, since if (1) is true it is necessarily true, it is not a first-order possibility and requires independent justification.

    June 17, 2004 — 9:25
  • Mike

    I have no idea what this “natural presumption” talk is supposed to mean. There is just no reason offered here why God CANNOT or is UNABLE to will that ‘no one commits acts of wanton cruelty’ is a necessary law. Since God can will in whatever way he wants, it is reasonable to believe that he can so will. Or do you want to say that this something that an omnipotent being cannot do?

    June 17, 2004 — 9:47
  • Absent some counterargument, and all other things being equal, we take first-order possibilities to be true. That is, per the conversational norms I sketched, the “natural presumption.”
    To deny

    (1) Possibly, God commands subject S to commit an act of wanton cruelty

    is to deny a first-order possibility. Such a denial needs further justification.
    In contrast, no justification is needed to deny

    (2) Possibly, God wills that the rule “do no commit acts of wanton cruelty” be a necessary law

    because it is a second-order possibility. Second-order possibilities foreclose first-order possibilities, and our default preference–our “natural presumption”–is in favor of the former.
    SIDE NOTE: Of course I’m saying that S5 is not rationally normative on this point. Undifferentiated possibility statements derived using S5 are inadequate to motivate the defense of DCT you outlined precisely because S5 doesn’t account for the default preference I earlier argued govern standard argument situations. For S5 operations to be normative on this score, S5 at the least would have to model all the rules for modal argument we abide by in standard discourse. It would appear S5 fails to model those rules, at least if my description of our standard attitudes toward possibility, impossibility and necessity is correct.

    June 17, 2004 — 12:30
  • Mike

    You say this:
    _In contrast, no justification is needed to deny
    (2) Possibly, God wills that the rule “do no commit acts of wanton cruelty” be a necessary law_
    But so far I’m offered nothing but red herrings. Are you now claiming that this is something that God–an omnipotent being–cannot do? You realize that there is no contradiction in his so willing. Are you further suggesting that you NEED NO JUSTIFICATION for claiming that this is something that an omnipotent being cannot do?! Please, don’t tell me that some rules of conversation make it permissible for you to offer no justification at all for why an omnipotent being has just the limits you suggest. I mean, honestly, only someone directly off the farm could believe that.

    June 17, 2004 — 15:53
  • Let me put it this way: Given our default preferences in conventional argument, we count the conceivability of statements like (1) to be much more a reliable indicator of possibility than we do statements like (2).
    Let me put it another way: Given the way we routinely argue about possibility, impossibility and necessity, (1) and (2) are both conceivable states of affairs, but (1) is more likely true.
    It follows from either formulation that a stated preference for (2) would require some kind of suitable justification.
    Does that help?

    June 17, 2004 — 16:19
  • Mike

    That helps a little: of course, it does not answer the question. Two observations: (1) you don’t want to infer the probability of p from the reasons to believe p. If we have no better reasons to believe p than to believe ~p, it does not follow that we should assign probability p = ~p = .5. This was Laplace’s mistake (long ago). Bertrand’s paradox shows this sort of reasoning quickly leads to contradiction. (2) Given God’s omnipotence, I don’t agree that the reasons to believe (2) are any greater than the reasons to deny it. It is too obvious to say, but conversational rules simply cannot settle complex metaphysical questions. In any case, I guess I’m prepared to leave it at that.

    June 17, 2004 — 17:09
  • Your second point first: I called it a conversational rule but I also called it an argumentative norm, and I also argued this norm reflects a rational preference for first-order over second-order possibility.
    I definitely think rationally justified norms of argument can help settle complex questions. (Indeed, without them I don’t know what else would.)
    Your first point: I was not infering probability from reasons for belief. I was arguing that as between two kinds of statement, one was more probable than the other given our rational bias for first-order possibility. (I left my a subjective estimate of prior odds out of it.)
    Thanks for an interesting discussion.

    June 17, 2004 — 21:11
  • Back to the interesting issue here, which is about whether the introduction of God’s loving nature renders DCT circular. I don’t think it’s enough for the circularity charge that there are many morally right things that a loving person would do. The question is how to characterize the nature of love itself, and whether one can do so without employing any moral concepts of duty, obligation, and the like.
    Here’s a proposal: identify love with some combination of actions, intentions, and emotional states (the right combination here is hard to specify, but that issue doesn’t touch on the circularity charge) directed toward what’s best for another. This account clarifies love in value-theoretic terms, not in terms of the theory of obligation.
    David, you might still object. In fact, I would too, at least initially. But here are some reasons that the objection is not easy. First, you’d need to adopt a moral theory on which value theory and the theory of obligation are related. Second, the relation would have to be meaning-theoretic. If it isn’t, then the account above avoids the circularity charge, since no language of obligation is involved in the characterization. Third, the relation between value theory and the theory of obligation can’t be simply this: the right action is the one that maximizes value. That ignores the category of the supererogatory, and if “promoting the best for another person” always and everywhere involved the category of the supererogatory, the circularity charge would not be sustained very easily either.

    June 27, 2004 — 17:02