Divine attitude theory elucidated

As is well known, many philosophers who believe that God is the source of moral requirements have embraced one or another version of divine command theory (DCT). An action is morally wrong, according to divine command theorists, if it is forbidden by God; morally required if it is commanded by God; morally optional if it is neither forbidden nor commanded by God; and supererogatory if it ‘goes beyond’ the commands of God in a praiseworthy way.Footnote ¹ I do not believe that this is the best account a theist can offer. More plausible, by my lights, is a theory I will call ‘divine attitude theory’ (DAT). In the first part of this essay, I will sketch the contours of DAT by showing how it differs in important respects from DCT. In the second part, I will argue that DAT, but not DCT, can explain the modal status of fundamental moral truths. This gives the theist good reason to prefer DAT over DCT.

My goals in what follows, therefore, are quite limited. Among other things, I will not attempt to defend either DCT or DAT from the standard challenges raised against theistic metaethical theories. I will also take it for granted that theism is true and that it is not implausible to think of the divine being as experiencing affective states that are recognizably similar to our own. Of course, this is not to deny that the standard challenges to theistic metaethical theories are important, nor is it to deny that my assumptions are contentious ones. Let it suffice to say that these issues, although of great philosophical interest, are issues that space prevents us from addressing here.Footnote ²

With these remarks in place, we should begin by noting that DAT is not offered as a complete account of all the moral facts there are. DAT acknowledges a distinction between evaluative moral facts (e.g. it is good to be honest) and deontic moral facts (e.g. persons are morally obliged to be honest), and purports merely to be an account of the latter. The moral wrongness of an action, on this account, consists in God's being displeased with an agent's performance of that action. DAT holds that an action is morally required if God would be displeased with an agent's failure to perform it; morally optional if God would be neither pleased nor displeased with an agent's performing it or failing to perform it; and supererogatory if God would be pleased with an agent's performing it but not displeased with an agent's omission of it.Footnote ³

It should also be noted that DAT makes an exclusively metaphysical claim about the nature of deontic properties. It is neither an epistemological thesis about the way in which we come to know moral facts nor a semantic thesis about the meanings of moral terms. To say that moral wrongness consists in divine displeasure is not to say that one must know (de dicto) what displeases God in order to know what is morally wrong, nor is it to say that one must believe (de dicto) that an action is displeasing to God in order to believe that an action is morally wrong. To the contrary, and like the version of DCT that has been defended in recent years by Robert Adams,Footnote ⁴ DAT is committed to the denial of these claims. The identification of moral wrongness with the property being-such-as-to-entail-God's-being-displeased-with-the-act involves a claim similar in relevant ways to the claim that water is H₂O. For our purposes here, the most relevant similarity is this: just as a person need not have any de dicto beliefs at all about the chemical structure of the stuff in the glass in order to know that the stuff is water, so the moral agent need not have any de dicto beliefs about God in order to know that some action is morally wrong. Thus one seemingly obvious objection to DAT – expressed, perhaps, by the rhetorical question ‘if moral wrongness is a matter of being displeasing to God, then how can an atheist have any beliefs about moral wrongness at all?’ – can be dismissed at once.

Deeper worries lurk nearby, however. After all, the identification of water with H₂O is plausible in large part because we can see how our standard epistemic practices concerning water are the right kind of practices to get us in touch with H₂O. If the ways in which we came to know about water were radically different from the ways we come to know about H₂O, then we would have some reason to doubt the plausibility of that identification. One might wonder whether the identification of moral facts with facts about divine attitudes admits of the same kind of plausibility; even if it is granted that there are good reasons for theists to look to divine attitudes in an effort to explain the nature of moral properties, it is far from obvious that our standard practices for acquiring moral truths are the right sort of practices for getting in touch with facts about divine attitudes. If no connection between these two kinds of practices exists, then the plausibility of the central thesis of DAT may be greatly diminished. Fortunately for the divine attitude theorist, at least two replies can be made to this challenge.

The first reply is one that is available to theistic metaethicists of all stripes. It is a disappointingly uninformative reply, but it does succeed in rebutting the challenge that our moral epistemic practices are too different from our theological epistemic practices for the identification of moral facts with theological facts to be plausible. The basic idea is very simple. If we assume that God intends for human beings to have cognitive access to moral facts (and there seems to be no good reason for a theist to reject this assumption), then it is reasonable to infer that, whatever the correct account of moral epistemology turns out to be, God has provided us with cognitive equipment that is up to the task of providing us with justified moral beliefs. If we have reason to believe that moral facts are identical to facts about (for example) divine attitudes,Footnote ⁵ then we have reason to believe that our best moral doxastic practices are the right kind of practices to get us in touch with facts about divine attitudes. This account is uninformative in that it provides no details about what those best practices might look like, but it is sufficient for explaining what the connection is between those practices and facts about divine attitudes, which is what matters here. Our best moral doxastic practices give us access to facts about divine attitudes because God intends for us to be the kind of creatures whose best moral doxastic practices give us access to facts about divine attitudes. The relevant connection is explained by appeal to the intentions of a creator who means for such a connection to exist and who has the power to bring it about that those intentions are realized.

The second reply to this challenge is a bit more informative, but it is not available to the divine command theorist. The crux of the reply is this: at least one prominent stream in contemporary moral philosophy suggests an obvious and satisfying harmony between our best moral doxastic practices and DAT. In recent decades, moral theories that appeal to the beliefs or attitudes of an idealized observer or idealized moral agent have become increasingly popular. R. M. Hare's famous utilitarian archangel is perhaps the first clear example of this approach, but philosophers favouring a variety of moral theories have adopted relevantly similar accounts: David Lewis, Peter Railton, and Michael Smith are among the most prominent.Footnote ⁶ Noncognitivists who exhort us to think of moral evaluations in terms of second-order endorsements operate in a closely related vein, and virtue theorists in the Aristotelian tradition typically appeal to the characteristic actions and dispositions of a virtuous phronimos in order to give an account of right action.Footnote ⁷ Without overstating the similarities between these approaches, what all of these disparate thinkers have in common is a tendency to maintain, in one way or another, that what it is right for me to do in such-and-such circumstances is connected in some significant way to what an ideal agent (or desirer, version of myself, etc.) would do (or think, desire, etc.) in those circumstances. This claim is not always intended to give us a decision-making procedure for determining what ought to be done (Smith, for example, emphasizes that he takes moral facts to be constituted by facts about what an ideally rational agent would desire), but the popularity of this kind of approach at least suggests that the imaginative exercise of thinking about how an agent who manifests various kinds of excellences would deliberate about what to do in some situation is among our very best tools for determining what we ought to do in that situation. If this is true, then the connection between our best moral doxastic practices and facts about divine attitudes becomes considerably more obvious: in imagining how an excellent-in-various-ways agent would think/act/feel in some scenario, we at least begin to approximate how God, understood by traditional theists as a maximally excellent being, would think/act/feel in that scenario. On this picture, there is nothing mysterious about the connection between our moral doxastic practices and facts about divine attitudes. Worries about whether the identification of water with H₂O is problematically disanalogous to the identification of moral properties with divine attitude properties are therefore allayed.

Another feature of DAT that should be noted is its ability to make normative claims that are more nuanced than the claims made by some of its rivals. For example, DAT can offer an attractively subtle understanding of moral wrongness. It is natural to suppose that divine displeasure, like human displeasure, admits of degrees, and thus it is natural for a divine attitude theorist to maintain that moral wrongness admits of degrees as well. DAT thereby provides us with the conceptual resources to explain why murder generally seems to be more wrong than lying. Not all moral theories will be able to offer such an explanation; DCT, for instance, seems to imply that moral wrongness is a non-degreed property, since an action either violates God's commands or it does not. This is not, by my lights, a decisive reason for preferring DAT to DCT (since we can distinguish between an action's wrongness and its badness, divine command theorists may maintain that murder is typically much worse, or a more serious wrong, than lying), but the ease with which DAT accommodates the commonsensical notion that some actions are more wrong than others is at the very least an appealing feature of the account.

A second respect in which DAT allows for more nuanced deontic assessments than some of its rivals appears when we consider actions that are morally ‘mixed’. To take a standard case, imagine a businessperson who donates a large sum of money to a worthy charity but does so merely because she believes that the positive publicity generated by her donation is likely to yield financial dividends over the long haul, not because she properly assesses the worthiness of the charity. It seems correct to say that such a person has done no wrong in supporting the charity, but she has done wrong in how she deliberated and in what she took to be sufficient reasons for acting. On DAT, it is easy to explain why this is so. Since God can be pleased with an action in some respects (in this case, its consequences) while being displeased with it in other respects (e.g., the motivation behind it), it is not surprising to find that our actions are occasionally praiseworthy in some ways but condemnable in others.

A third way in which DAT allows for nuanced moral judgements becomes apparent in light of its implications concerning the existence of moral dilemmas. An agent faces a moral dilemma when she is in a situation such that, no matter what she does, the action she takes will be morally wrong. Many philosophers believe that there can be no genuine moral dilemmas, because a person who must choose one from among an array of prima facie wrong actions, and successfully chooses the least of all the possible evils, thereby acts rightly. But it is not always easy to see how this claim can be justified. DCT, for example, seems to imply that a person could often be in a situation in which any action she takes will violate some divine command or other. Standard illustrations include cases in which obligations of promissory fidelity and beneficence come into conflict. If God has commanded us to keep our promises and also to help those who are in need, then we may sometimes find ourselves in situations in which we must violate a divine command: as, for instance, when one must choose between keeping an appointment and helping a stranded motorist. Keeping the appointment means not assisting the motorist, and hence failing to help a person in need; helping means missing the appointment, and hence failing to keep a promise. Either way a divine command is violated and a wrong action is performed.

Divine command theorists may respond to this challenge in a variety of ways. One option is simply to bite the bullet and claim that there are indeed moral dilemmas. Another option is to propose a hierarchical ranking of moral principles in order to resolve such conflicts. A third alternative is to maintain that while God commands that certain rules ought to be followed ceteris paribus, there can be extenuating circumstances in which a person is not commanded to perform a particular action that would seem to follow from the general rule. Whether any or all of these suggestions are tenable, it should be clear that DAT can offer a simpler (and, in my view, more appealing) solution to the problem. DAT tells us that in any situation, an agent acts wrongly if and only if God is displeased with the agent's action. On the reasonable assumption that God would not be displeased with a person's choosing the least bad of several unattractive options, it follows that a person who chooses the least bad of several unattractive options does not act wrongly.

DAT, DCT, and the ‘trilemma’ argument

Having established with some clarity what DAT is, we are now in a position to consider in more detail why a theist might prefer it over DCT. A few considerations of this sort have already been noted, but for the reasons discussed above, none of these are decisive. In the remainder of this essay, I will present an argument that does, in my view, provide compelling reason for theists to abandon DCT in favour of DAT. This argument extends a discussion between Mark Murphy and Michael Almeida in Religious Studies and shows that divine command theorists, but not divine attitude theorists, face a dilemma generated by the modal status of moral truths. Whether or not God is significantly free with respect to the content of the commands heFootnote ⁸ issues, I will argue, DCT is unable to account for the fact that fundamental moral truths are necessary truths.

If any philosophical thesis is uncontroversial, it is this one:

To put it another way: if an action A is morally wrong and an action B is morally right, there must be some non-moral difference between A and B that explains this fact.Footnote ¹⁰ Moral properties do not vary independently of non-moral properties. To affirm this thesis is to embrace the doctrine of global moral supervenience. Murphy (Reference Murphy2002) argues that the most popular contemporary version of DCT, when combined with a widely accepted claim about God's freedom in acting, is incompatible with moral supervenience. DCT is less plausible than either the claim about divine freedom or the doctrine of moral supervenience, so it should be rejected. Murphy's argument goes as follows.

According to contemporary divine command theorists like Robert Adams, an action's being morally obligatory depends upon its being commanded by God. This thesis is not proposed as a definition or as a conceptual truth, but as an a posteriori identity claim that is discovered through a process of investigation, similar in relevant respects to the empirical investigation by which we come to learn that the property being H ₂O is identical to the property being water. The relationship between these properties, of course, is not merely contingent but is a matter of metaphysical necessity. Anything in any possible world that is H₂O is water, and anything that is water is H₂O. Likewise, according to divine command theorists, any action in any possible world that is morally obligatory is commanded by God, and any action that is commanded by God is morally obligatory.

In addition to this claim about the nature of moral properties, Murphy expects that the divine command theorist will want to embrace an independent claim about God's freedom. ‘On most views’, Murphy writes,

It is worth noting that Murphy does not attempt to argue for these claims, and his statement that ‘on all views’ God could have created a somewhat different world is surely false. Thomas Aquinas, for example, found it sufficiently contentious that he devoted several chapters of the Summa Contra Gentiles to explaining why it must be that God does not will the things he wills necessarily,Footnote ¹¹ and Gottfried Wilhelm Leibniz rather notoriously held that God's perfect wisdom and perfect goodness together entail both that God creates the best possible world (if there is such a world) and that ‘if there were not the best among all possible worlds, God would not have produced any’ (Leibniz (Reference Leibniz and Farrer1710), 128). Although it is safe to say that relatively few philosophers have been committed to this view, Murphy's suggestion that no-one has held it is mistaken.Footnote ¹²

This qualification aside, however, the really important point is one Murphy rightly emphasizes: any philosophical theory that entails that God's commands are ‘entirely fixed by the way the world otherwise is’ ipso facto comes at a cost – a cost many theists will be unwilling to pay. The suggestion that God's freedom might be radically constrained in this way is uncomfortably reminiscent of attempts to defuse the problem of evil by arguing that our conceptual categories cannot be applied to the divine, so that no tension of any kind exists between the claim that God is perfectly loving and compassionate and the claim that God could easily prevent the suffering of innocents but chooses not to. The logic of the strategy is impeccable; nevertheless, it may be a way of winning the battle by losing the war.

The doctrine of moral supervenience, of course, tells us that moral facts are unlike facts about God's free actions in that they are entirely fixed by the way the world otherwise (i.e. non-morally) is. It is this fact that generates an apparent problem for divine command theorists. Let us give a somewhat more rigorous statement of the doctrine of moral supervenience:

• (S) Necessarily, for any action ϕ and any moral property M, if ϕ has M, then there exists a non-moral property N such that ϕ has N, and necessarily, if any action Ψ has N, then Ψ has M.Footnote ¹³

It should be noted that this is a statement of strong moral supervenience rather than weak moral supervenience; a statement of weak supervenience would omit the second occurrence of ‘necessarily’. The argument that follows will require strong rather than weak moral supervenience, but this is not a commitment that threatens to undermine the argument. In Frank Jackson's words, the ‘least controversial part of folk moral theory is that moral properties supervene on descriptive properties, that the ethical way things are supervenes on the descriptive way things are’, and it in no way violates our common-sense intuitions to stipulate that the supervenience in play is strong supervenience, obtaining across possible worlds as well as in the actual world (Jackson (Reference Jackson1998), 118). Just the contrary; strong moral supervenience is taken by many to be a platitude of moral discourse, so that a person who uses moral terms in a way that violates (S) thereby demonstrates that he or she has an inadequate grasp of the concepts expressed by those terms. Nick Zangwill makes the point nicely: ‘This modal doctrine concerning morality is not a rejectable piece of dubious philosophical speculation; it is an assumption built into moralizing. To moralize is to make cross-world commitments. Quit modalizing and you quit moralizing. Moralizing is a modal pastime’ (Zangwill (Reference Zangwill1995), 259).

In addition to (S), divine command theorists like Adams are committed to the proposition that:

• (C) Necessarily, for any action ϕ and any moral property M, if ϕ has M, then there exists a divine command property C such that ϕ has C.

Together, (S) and (C) entail:

1. (1) Necessarily, if any action ϕ has N, then ϕ has C.

The purported problem for DCT is generated by the fact that many theists will take the following claim about God's freedom in issuing commands to be just as uncontroversial as (S):

• (F) For some action ϕ that has non-moral property N and divine command property C (e.g. being forbidden by God), it is possible for ϕ to fail to have C,

or more simply,

1. (2) Possibly, there is an action ϕ that has N but not C.

Obviously, (1) and (2) are contradictory, so it is impossible for (S), (C), and (F) to all be true. Given the platitudinous status of (S) and the uncontroversial nature (for many theists) of (F), it appears that (C) should be rejected. (C), however, is a straightforward implication of DCT, so any philosopher who is committed to (S) and (F) must reject DCT.

There are interesting questions to ask here about the best way to understand the alleged relationship between moral properties and divine command properties, as well as the nature of the supervenience relation. Michael Almeida, for example, has argued that it is possible for a property to supervene on itself and that this fact undermines Murphy's challenge; if moral properties are identical to divine command properties, and if a property can supervene on itself, then divine command properties can be part of the subvenient base on which moral properties supervene.Footnote ¹⁴ This assumption, however, is quite contentious, and even if it turns out to be defensible, Murphy believes that it misses the point:

Surely Murphy is correct in his suggestions about how most of us would respond to these questions. I do not wish to dismiss his claims and will return to them shortly. For now, however, I will note that Murphy seems to have missed a very important point concerning the motivation for property identity DCT, a point whose relevance to this discussion is missed by Almeida as well.

Much of the attractiveness of property identity DCT stems from its significant explanatory power. Space precludes defending this claim in detail here, but it is reasonable to think of DCT as being motivated, in large measure, by a desire to answer the question what must the world be like in order for anything to be morally right or morally wrong? Of course, no-one who is in a position to ask this question can be wholly ignorant of the details of our moral discourse and practice. The question can be meaningfully posed only by someone who has some understanding of how moral language is used and what kinds of things are, or at least are widely thought to be, morally right and morally wrong.

With this background information in place, the enquirer is in a position to consider what properties might play the roles delineated by our use of moral terms. For example, what sort of thing might moral wrongness turn out to be? Given that it is fitting to feel guilty when one does something wrong, that performing a wrong act makes it appropriate to seek forgiveness, that moral considerations are at least potentially motivating, and so on, we can investigate whether there might be some shared feature of paradigmatically wrong acts in virtue of which they can all be said to be wrong. DCT, of course, tells us that there is: wrong actions are all actions that are forbidden by God. DCT is thus an explanatorily powerful theory in at least two important ways: it provides an account of why morally wrong acts have the features they do, and it unifies the class of wrong actions in a systematic and satisfying way.Footnote ¹⁵

If this is so, then the divine command theorist is in a position to maintain that the considerations of ‘ex ante plausibility’ identified by Murphy are simply too weak to pose a threat to versions of DCT that are developed in this way. In response to Murphy's charge that we are not likely to think that ‘the property being obligatory might be instantiated, or not instantiated, while every other distinct property instantiated remains the same’, the savvy divine command theorist should stand her ground. Once the case has been made for DCT, once we appreciate the explanatory power of that theory and the reasons that are adduced for identifying moral properties with divine command properties, this claim of Murphy's loses its bite. From the perspective of the informed divine command theorist, there should be nothing worrisome about the fact that her theory implies that ‘the property being obligatory might be instantiated, or not instantiated, while every other distinct property instantiated remains the same’. This is a straightforward implication of her position and – so long as we recognize that quite a lot is built into the term ‘distinct property’ in the above quote from Murphy – it is an implication that she ought to embrace. It is not the sort of anomalous consequence that should lead the divine command theorist to an ad hoc adjustment of her theory, nor is it a consequence that should be listed among the theory's ‘theoretical costs’. If God does not issue a command forbidding some action, then if DCT is true, that action is not morally wrong. This is hardly a surprising result.

So it appears that Murphy's argument against DCT is not as compelling as it may have initially seemed. Nevertheless, it is worth dwelling a bit on the tension that he has pointed out. Murphy has put us in a good position to see how deeply contingent morality appears to be for the divine command theorist, and – perhaps more importantly – how odd that contingency is. According to DCT, the wrongness of an action hinges upon whether God forbids it or not. Far from being necessary truths, fundamental (and all other) moral principles turn out to be contingent on God's decision to require or forbid certain kinds of actions. Even if we assume that the divine nature establishes certain parameters concerning which commands God will issue, there still seem to be multitudinous commands from among which God may choose. For example, consider the moral prohibition against cruelty. That cruelty is wrong is a good candidate for a moral truth if anything is. Of course, if we assume with traditional theists that God is perfectly loving, we may safely assume that God would not issue a command requiring cruel behaviour. It would be a necessary truth that God will not act cruelly himself, and indeed that God will allow others to act cruelly only if he has a good reason for doing so – a reason that could figure in the deliberations of a perfectly loving agent. All of this seems unproblematic. But is it necessarily true that a loving being will issue a command governing cruelty? More broadly, is it necessarily true that, for every fundamental moral principle, God will issue a command concerning that principle? To assert that the answer is ‘yes’ is to make a rather dubious claim. At the very least, such an assertion would run afoul of the enormously plausible principle (F) above; God's nature may require him to act in certain ways and to refrain from acting in others, but it seems to be a gross and ad hoc limitation on divine freedom to claim that God's lovingness makes it necessarily false that God refrains from issuing a command forbidding cruelty.Footnote ¹⁶ It would be quite odd to maintain that, although a perfectly loving God may allow some of his beloved creatures to be treated cruelly, he must – as a matter of metaphysical necessity – tell their tormentors not to do so. This seems exactly backwards. Having a perfectly loving nature may indeed limit the number of choices a being has in any particular situation, perhaps even requiring that such a being attempt to help tormentors appreciate the badness of cruelty. The choice to declare or not to declare ‘Thou shalt not!’, however, does not seem to be among the choices that are constrained in this way. To appreciate the force of this point, and to see how it fits into the case for DAT (about which I will say more below), one need only consider whether it seems possible for there to be any action such that God might choose not to forbid it, yet would be displeased with a person for performing it. It is very difficult to resist the suggestion that this is indeed possible. At a minimum, the burden of proof lies on anyone who believes otherwise to explain why it is impossible.

In the absence of such an explanation, it is undeniable that the divine command theorist must affirm that at least some deontic truths are contingent. But it might seem that this is an unproblematic result; one might suppose that the very same response I suggested on behalf of the divine command theorist in response to Murphy's claim about ‘ex ante plausibility’ could just as easily be employed here: viz. that although this is a somewhat surprising and not especially desirable consequence of DCT, it is a consequence nevertheless, and its undesirability is greatly outweighed by its ability to give an explanatorily powerful and theoretically unified account of the nature of moral properties. This response is not objectionable in and of itself. But if there is a rival theory that can lay claim to theoretical virtues of the same kind and of comparable strength, and that is not subject to the sort of challenges we have been discussing, then that rival should be preferred. My suggestion is that DAT is such a rival, and rather obviously so. Again, space precludes any detailed discussion of the explanatory power of DCT or DAT here, but it seems safe to say that the same facts about our use of moral terms – e.g. the fittingness of feelings of guilt when one has acted wrongly, the potentially motivating character of moral judgements – that make divine command properties good candidates to play the role of moral properties may also be appealed to on behalf of DAT. Therefore the divine command theorist cannot appeal to explanatory power as a reason for preferring her account to DAT.

For our purposes here, the really important thing to notice is that facts about what is pleasing and displeasing to God, unlike facts about what God has commanded or required, are not plausibly understood as contingent facts. In the words of Alvin Plantinga,

It is metaphysically impossible that God would be cruel, devoid of compassion, or deceitful. In colloquial terms, God could not avoid being the kind of person he is. And if we grant that character traits like these have essentially affective components – e.g. to be cruel is not merely to act in ways that are harmful, but to have a certain pro-attitude toward the suffering of others – we must conclude that God is such that, necessarily, God is pleased by some kinds of actions and displeased by others. Therefore DAT implies that if cruelty is wrong, then cruelty is necessarily wrong, precisely what reflection on Murphy's argument leads us to expect.Footnote ¹⁷

This represents one horn of a dilemma: if the divine command theorist concedes that God has significant freedom in deciding which commands to issue, then the divine command theorist must deny the necessity of moral truths. Therefore we should consider the possibility that God does not have such freedom; if we deny (F), then we can embrace both DCT and the necessity of moral truths. And as was noted above, there is at least some reason for theists seriously to consider rejecting (F). If God is omniscient and perfectly good, the thinking goes, he always knows what is best and he always does what is best. God creates because it is best that he do so, and of the many possible worlds God could create, he actualizes the best one among them. Among the many features of this best possible world are the commands that God himself issues within it; if it is best that God forbid ϕ, then God forbids ϕ; if it is best that God require ϕ, then God requires ϕ; and so on. Since God's own metaphysically necessary nature dictates that God do what is best, the content of divine commands turns out not to be contingent.

The other horn of the dilemma begins to emerge, however, when we recognize the following: the best possible world might be one in which we reach the decision to avoid an action in the absence of an explicit command to refrain from performing that action. There is value in persons coming to ‘see for themselves’ certain reasons to act or to refrain from acting. This is something we acknowledge in our everyday moral lives. One can easily imagine, for example, a parent who chooses not to forbid her son from engaging in some behaviour of which she disapproves precisely because she thinks it would be better for him to decide on his own not to engage in it. We can imagine God deliberating in similar fashion about what commands to give to his creatures. Consider slavery, an issue whose moral status continues to be debated by theologians of various stripes. No-one, so far as I know, claims that a slave owner may permissibly be cruel to his slaves, nor do any theologians of whom I am aware promote the reintroduction of slavery into contemporary society. Whether the owning of slaves is itself morally objectionable is nevertheless a matter of dispute. After all, the Jewish and Christian scriptures make reference to the proper treatment of slaves without ever explicitly condemning slavery as an institution, and this might lead a Jewish or Christian theologian to believe that God has not forbidden it. Yet – at the risk of begging the question against divine command theory – it is difficult to suppose that this observation could be sufficient to end the dispute over the moral status of slavery. For example, many Christians believe that the biblical doctrine of the imago Dei, according to which human beings are made ‘in the image of God’, implies that each individual human being ought to be respected as a person. Since slavery seems to be a straightforward way of violating an individual's personhood, we have compelling reason to believe that it is morally wrong, even if it is not strictly forbidden by God.

The divine command theorist, of course, cannot say this. If the best possible world is one in which humans choose not to own slaves without being commanded not to own slaves, and if God always does what is best, then God does not forbid humans from owning slaves. That is to say, if it would be better overall for us not to be forbidden from owning slaves, then the divine command theorist is forced to deny the wrongness of slavery in the best possible world (which is also the actual world).Footnote ¹⁸ This seems bizarre. DAT, in contrast, brings with it no such untoward consequences. If it is indeed best for God to refrain from forbidding slavery, and if God always does what is best, then God refrains from forbidding slavery. If we have reason to believe that God is nevertheless displeased with slavery, then the divine attitude theorist is justified in claiming that slavery is morally wrong even if God does not, for whatever reason, forbid it. DAT implies that the wrongness of slavery – or anything else – does not hinge on the value or disvalue of God issuing a command prohibiting it. This is exactly as it should be.

More importantly, we are now in a position to see that even a divine command theorist who is willing to deny that God has significant freedom concerning which commands he will issue will still be committed to an unattractive position concerning the modal status of moral truths. The reason for this has to do with the differences between possible worlds: depending on the specific circumstances that obtain in those worlds, divine commands that God issues in one possible world may not be issued in another; there might be some possible worlds in which God's omniscience and goodness entail that he forbids ϕ, and other possible worlds in which they entail that he does not forbid ϕ (because it would be better for the persons in those worlds to choose to refrain from ϕ-ing on their own). That is to say, even the necessitarian divine command theorist must countenance the existence of possible worlds in which an action ϕ is not morally wrong, even though ϕ is wrong in the actual world. The sting of this consequence is diminished somewhat by the supposition that it is necessarily true that God creates the best possible world – and thus (de re) necessarily true that God issues the commands that he issues – but the problem for DCT has not disappeared: the necessity seems to be ‘misplaced’. The strategy under consideration here allows divine command theorists to say that it is necessarily true that, e.g., cruelty is wrong, but the necessity of this truth does not stem from features of cruelty itself, but rather from the fact that the best possible world is a world in which it is better for us to be commanded not to be cruel than to decide for ourselves that we will refrain from cruelty. This is the other horn of our dilemma: if the divine command theorist denies (F), then she must accept that the modal status of moral truths is determined by something other than the content of those truths. Again, DAT does not encounter this sort of problem. Since God's nature is fixed across all possible worlds, God is displeased by cruelty in all possible worlds, and cruelty is wrong in all possible worlds. This is precisely what a satisfying moral theory should imply. Even theists who reject (F) and thereby avoid the first horn of the dilemma have compelling reason to reject DCT in favour of DAT.Footnote ¹⁹