Definition of divine command metaethics

In this section, I will illuminate what divine command metaethics is. In the course of this article, I will refine this definition. For a start, I will define divine command metaethics in the following way:

• DCM:

  1. (a) It is metaphysically necessary that, for every act x, x is morally obligatory if and only if God commands x.

  2. (b) It is metaphysically necessary that, for every act x, if x is morally obligatory, then x is morally obligatory because God commanded x.

I would like to clarify three aspects of this definition. First, please note that part (a) merely claims that the property of being morally obligatory and the property of being commanded by God are co-instantiated across possible worlds. Part (b) goes beyond such a merely modal characterization by claiming that divine commands provide a suitable explanation for moral obligations. Second, I take it for granted that divine command metaethics is formulated in terms of moral obligations and not in terms of any other normative statuses (e.g. moral goodness or wrongness). Presumably, this is not very important for the discussion of the objection of horrible commands because one could easily reformulate the objection in terms of other normative statuses. Third, I recognize that it might not be entirely lucid how the ‘because’ relation used in part (b) of my definition is supposed to be understood.Footnote ¹ However, I would like to point out that this is not important for the purpose of understanding the objection of horrible commands either. In the next section, we will see that part (a) of my definition is already sufficient to get the objection of horrible commands off the ground.

Simple version of the objection of horrible commands

In this section, I will depict what Adams calls the ‘gravest objection’ (Adams (Reference Adams1987), 98) against DCM, viz. the so-called ‘objection of horrible commands’.

The objection is based on the intuition that there are acts that are not morally obligatory as a matter of metaphysical necessity. To illustrate this point, I will henceforth use Morriston's example of ‘the gruesome and painful sacrifice of randomly selected ten-year-old children’ (Morriston (Reference Morriston2009), 250). Presumably, it is metaphysically necessary that this act is not morally obligatory. Following Davis and Franks (Reference Davis and Franks2015, 2), I will call this act a ‘sacrificial scenario’. If the reader does not share my intuition that this act cannot be morally obligatory under any circumstances, then he or she is free to exchange this example for something that he or she finds even more appalling.Footnote ²

I believe that the objection of horrible commands is best construed as a reductio argument with three premises. To understand the first premise, remember that we chose the example of a sacrificial scenario such that we may take the following premise for granted:

1. (1) It is metaphysically necessary that a sacrificial scenario is not morally obligatory (premise).

Second, virtually every theist believes that God is omnipotent.Footnote ³ Furthermore, an omnipotent being has the power to command a sacrificial scenario. For this reason, we may take the following premise for granted:

1. (2) It is metaphysically possible that God commands a sacrificial scenario (premise).

Third, let us assume the truth of DCM for reductio. Remember that, according to the first component of DCM, morally obligatory acts and acts that have been commanded by God are co-extensional as a matter of metaphysical necessity. Therefore, the following counterfactual proposition follows:

1. (3) If God were to command a sacrificial scenario, then a sacrificial scenario would be morally obligatory (from (a) of DCM).Footnote ⁴

Fourth, the following proposition follows from (2) and (3):Footnote ⁵

1. (4) It is metaphysically possible that a sacrificial scenario is morally obligatory (from (2) and (3)).

This is a problem. There is an obvious contradiction between (4) and (1). According to (4), it is metaphysically possible that a sacrificial scenario is morally obligatory. According to (1), this is metaphysically impossible. However, both cannot be true. Therefore, unless we are willing to give up premise (1) or (2), we may draw the following conclusion:

1. (5) DCM is false (by reductio from (1) and (4)).

This is what the objection of horrible commands was supposed to show.

Let me anticipate one important objection. Often, it is claimed that premise (1) is not as innocent as it seems. Thus, one might be inclined to try to defend DCM by giving up premise (1). However, denying (1) commits one to the position that it is metaphysically possible for a sacrificial scenario to be morally obligatory. Ockham is usually considered as the most famous proponent of this defence of DCM. However, I doubt that this manoeuvre is particularly attractive because it runs into conflict with the widely shared assumption ‘that there is no metaphysically possible world where the basic norms are different’ (Enoch (Reference Enoch2011), 146). Adams (Reference Adams1987, 99–100) points out that especially Jewish and Christian believers tend to emphasize the importance of this assumption.Footnote ⁶ Therefore, the Ockhamist attempt to evade the objection of horrible commands by dropping premise (1) is of no avail. The objection of horrible commands against DCM remains in full force.

Perfect being theology as a defence of divine command metaethics

In this section, I will illuminate how DCM can be defended against the objection of horrible commands by insisting that it is metaphysically necessary for God to be perfectly morally good. This claim follows from the assumptions of perfect being theology that have a long history in Western philosophy of religion (cf. Leftow (Reference Leftow1989)). The backbone of this proposal is the following claim:

• NECESSARY GOODNESS: It is metaphysically necessary that God is perfectly morally good.Footnote ⁷

There is a de re and de dicto reading of NECESSARY GOODNESS. If we understand it in terms of de dicto metaphysical necessity, it merely claims that being perfectly morally good is a precondition for bearing the title ‘God’. On this understanding, the proposition ‘God is perfectly morally good’ is true in every possible world. However, on this understanding, nothing prevents the entity that bears the title ‘God’ from ceasing to be perfectly morally good and, thus, losing the title. If we understand NECESSARY GOODNESS in terms of de re metaphysical necessity, it claims that the entity that bears the title ‘God’ is perfectly morally good in every possible world. Hence, on this understanding, the entity that bears the title ‘God’ cannot cease to be perfectly morally good. I will take for granted the latter de re reading of NECESSARY GOODNESS. At this point, let me introduce another term: I will use the term ‘DCM*’ for the conjunction of DCM and NECESSARY GOODNESS.

Let me explain why the objection of horrible commands does not threaten to refute DCM*. We saw that the objection relies on premise (2), which claims that it is metaphysically possible for God to command a sacrificial scenario. Now, to command a sacrificial scenario is without a doubt a gross moral imperfection. However, according to NECESSARY GOODNESS, God is perfectly morally good as a matter of metaphysical necessity. If God is necessarily perfectly morally good, God necessarily refrains from commanding a sacrificial scenario because this would be a moral shortcoming. For this reason, a proponent of DCM* can no longer sustain premise (2). Therefore, a proponent of DCM* is no longer committed to (4). Thus, the contradiction between (1) and (4) no longer arises. For this reason, DCM* is not vulnerable to the objection of horrible commands.

Of course, one might argue that this cure is not better than the disease because NECESSARY GOODNESS is at odds with other assumptions that theists tend to hold dear. For instance, there has been significant debate whether NECESSARY GOODNESS is compatible with (a) God's omnipotence (Morriston (Reference Morriston2001), Mawson (Reference Mawson2002), Funkhouser (Reference Funkhouser2006), Carey (Reference Carey2017)), (b) God's perfect freedom (Rowe (Reference Rowe2004), Bergmann & Cover (Reference Bergmann and Cover2006), Timpe (Reference Timpe and McCann2017)), and (c) God's praiseworthiness (Maitzen (Reference Maitzen2004)). If any of these worries turned out to be justified, then even the simple version of the objection of horrible commands that I just depicted might successfully refute DCM after all. However, to discuss all these objections against NECESSARY GOODNESS exhaustively would exceed the scope of this article.

Counterpossible version of the objection of horrible commands

In this section, I will depict an improved version of the objection of horrible commands that threatens to refute even the conjunction of DCM and NECESSARY GOODNESS (which I called DCM*). Recently, Morriston (Reference Morriston2009, Reference Morriston2015), Wielenberg (Reference Wielenberg2005), and Sinnott-Armstrong (Reference Sinnott-Armstrong, Garcia and King2009) have argued that NECESSARY GOODNESS fails to disarm the objection of horrible commands because this objection can be reformulated in terms of counterfactual propositions with metaphysically impossible antecedents, so-called ‘counterpossibles’:

In other words, Morriston, Wielenberg, and Sinnott-Armstrong are happy to give up premise (2) but believe that we can easily reformulate the objection from horrible commands in a way that does not require the truth of (2). The new argument is much simpler than before and relies on a simple application of modus tollens. Morriston, Wielenberg, and Sinnott-Armstrong point out that DCM* entails proposition (3): ‘If God were to command a sacrificial scenario, then a sacrificial scenario would be morally obligatory.’ Furthermore, even if the impossible were to become actual and God were to command a sacrificial scenario, a sacrificial scenario would still not be morally obligatory. For this reason, they claim that the proposition (3) bears the truth-value ‘false’. Now, since DCM* entails (3) and (3) is false, we may infer that DCM* must be false as well. Therefore, Morriston, Wielenberg, and Sinnott-Armstrong argue that even if we endorse NECESSARY GOODNESS, the objection of horrible commands remains in full force.

Semantic preconditions of the counterpossible version of the objection of horrible commands

At this point, I would like to elaborate on the semantic preconditions of Morriston's, Wielenberg's, and Sinnott-Armstrong's ‘improved’ counterpossible version of the objection of horrible commands. We saw that their argument hinges on the assumption that the counterfactual proposition (3) bears the truth-value ‘false’. This assumption is at odds with ‘vacuist’ semantics of counterfactuals. In general, vacuism holds that every counterfactual with a metaphysically impossible antecedent proposition is true irrespectively of what consequent proposition is attached to it. In other words, according to vacuism, all counterpossibles are true. For instance, Stalnaker (Reference Stalnaker and Rescher1968) and Lewis (Reference Lewis1973) have famously defended vacuism. Let me point out how a vacuist semantics would undermine counterpossible version of the objection of horrible commands. If we take for granted NECESSARY GOODNESS, then the antecedent of (3) is metaphysically impossible. Thus, (3) is a counterpossible. For this reason, a vacuist semantics of counterfactuals assigns the truth-value ‘true’ to (3).Footnote ⁸ Furthermore, we saw that the counterpossible version of the objection of horrible commands held that if (3) is false, then we may infer by modus tollens that DCM* is false as well. Thus, if vacuist semantics is correct and (3) is true, we can no longer maintain this objection. Thus, it seems as if we can defend DCM* against the counterpossible version of the objection of horrible commands by endorsing vacuist semantics.

However, Morriston has anticipated this manoeuvre on behalf of DCM*. He argues that there must be something fishy about vacuist semantics because it is easy to come up with counterexamples of false counterpossibles. In a footnote, Morriston mentions the following counterpossible that he takes to be false even if we assume that it is a ‘metaphysically necessary truth that God is good’ (Morriston (Reference Morriston2009), 266n.)

1. (6) ‘[I]f (per impossibile) God were evil, He would be good’ (ibid.).

Morriston points out that since we have the intuition that at least (6) is false, no sensible semantic theory can hold that literally all counterpossibles are true. Therefore, it seems to be a bad idea to defend DCM* by endorsing vacuist semantics.Footnote ⁹

Pruss's reductio argument against the counterpossible version of the objection of horrible commands

Before I turn to my own pragmatics-based attempt to defend DCM* against the counterpossible version of the objection of horrible commands, I will discuss Pruss's (Reference Pruss2009) popular reductio defence of DCM*.

The underlying idea of Pruss's defence of DCM* is the following: Pruss argues that the counterpossible version of the objection of horrible commands ‘proves too much – it equally applies against every substantive metaethical theory’ (Pruss (Reference Pruss2009), 434). So, Pruss argues that the intuition that drives Morriston's, Wielenberg's, and Sinnott-Armstrong's objection against DCM* commits them to believe that all substantive metaethical theories are false. Pruss believes that this commitment is sufficient to render the underlying intuition of the counterpossible version of the objection of horrible commands ad absurdum.

To understand Pruss's argument in more detail, we need to get clear on what he means by a ‘substantive metaethical theory’. If we generalize the structure of DCM, we get a good sense of what Pruss takes to be a metaethical theory in general:

• A metaethical theory is a theory that posits that there is a characteristic property F that satisfies the following two conditions:

  1. (a) It is metaphysically necessary that, for every act x: x is morally obligatory if and only if x is F.

  2. (b) It is metaphysically necessary that, for every act x: if x is morally obligatory, then x is morally obligatory because x is F.

Pruss claims that a metaethical theory needs to meet two further criteria in order to count as substantive. On the one hand, a substantive metaethical theory cannot define the property F ‘by giving an infinite list of all the obligatory and wrong actions’ (Pruss (Reference Pruss2009), 434). On the other hand, the property F should serve as a suitable non-circular definition of the property of being morally obligatory (ibid.). Thus, the definition of F cannot involve the property of being morally obligatory again. Pruss (ibid., 437) explicitly mentions that if we believe that Moore's (Reference Moore1993) ‘open question argument’ can be applied to moral obligations,Footnote ¹⁰ then we cannot provide a metaethical theory that satisfies the latter criterion. Moore's argument is supposed to show that there is only one correct way to define the property F: namely, as the property of being morally obligatory. This, however, leads to an obviously circular definition of the property of being morally obligatory. According to Moore's argument, every other attempt to define F would leave an ‘open question’ whether F really is a suitable definition of the property of being morally obligatory. For this reason, Moore dismisses all non-circular metaethical theories and, thus, all metaethical theories that are substantive in Pruss's sense.

Obviously, DCM* and DCM are examples of substantive metaethical theories. In the case of DCM* and DCM, F is the property of being commanded by God. Furthermore, Pruss mentions ‘Kantianism’ as another example for a substantive metaethical theory. In the latter case, F is the property of being ‘consistently required by both categorical imperatives’ (Pruss (Reference Pruss2009), 435).

Let me turn to the core of Pruss's argument: Pruss argues that we can construct an objection against every substantive metaethical theory that has the very same structure as the counterpossible version of the objection of horrible commands of Morriston, Wielenberg, and Sinnott-Armstrong. Again, the argument presupposes that there is an act that is not morally obligatory as a matter of metaphysical necessity. Again, I will assume that the example of a sacrificial scenario fits this bill. Now, let T be a substantive metaethical theory as defined above. To ensure that T is subject to the same modal intuitions as DCM*, we need to make the further assumption that it is metaphysically impossible for a sacrificial scenario to be F – this assumption plays the same role as NECESSARY GOODNESS in the counterpossible version of the objection of horrible commands. This assumption is innocent in the case of most substantive metaethical theories.Footnote ¹¹ For instance, it appears uncontroversial that it is metaphysically impossible for a sacrificial scenario to be required by all versions of Kant's categorical imperative. Henceforth, let T* be the conjunction of T and the assumption that it is metaphysically impossible for a sacrificial scenario to be F. I hope that it is easy to see that DCM* is an instance of T*.

Pruss's argument starts by assuming the truth of T* for reductio. Remember that, according to part (a) of the definition of a metaethical theory, it is metaphysically necessary that every act that is morally obligatory also is F and vice versa. Therefore, T* entails the following counterfactual proposition:

1. (7) If a sacrificial scenario were to be F, then a sacrificial scenario would be morally obligatory (from (a) of T*).

(7) is the equivalent to (3) in the counterpossible version of horrible commands. Now, because T* comprises the assumption that it is metaphysically impossible for a sacrificial scenario to be F, (7) is a counterpossible just like (3). Arguably, the same intuition that makes Morriston, Wielenberg, and Sinnott-Armstrong believe that (3) is a false counterpossible commits them to believe that (7) is a false counterpossible as well. They may be expected to hold that even if the impossible were to become actual and a sacrificial scenario were to become F, then a sacrificial scenario would still not be morally obligatory. This is exactly what they argued for in the case of DCM*. Now, since T* entails (7) and (7) is false, we may infer by modus tollens that T* must be false as well.

The structure of this argument mirrors exactly that of the counterpossible version of the objection of horrible commands. Therefore, Pruss claims to have shown that the underlying modal intuition of the counterpossible version of the objection of horrible commands renders all substantive metaethical theories equally untenable. Pruss claims that this is a devastating consequence. For this reason, Pruss concludes that something about the counterpossible version of the objection of horrible commands must be fishy.

Problems

I believe that Pruss's reductio argument suffers from two weaknesses.

First, Pruss's aim was to point out that there is a certain asymmetry with respect to our treatment of objections that have a similar structure to the counterpossible version of the objection from horrible commands. Apparently, these objections seem to refute DCM* but they fail to harm other substantive metaethical theories. This is inconsistent, according to Pruss. In order to restore consistency, Pruss recommends that one withdraws one's support for all objections that have the same structure as the counterpossible version of the objection from horrible commands. I agree that this is a way to restore consistency but I have doubts whether we are forced to draw this conclusion. Rather, a proponent of the counterpossible version of the objection of horrible commands might equally well withdraw support for the very project of substantive metaethics. This manoeuvre would restore consistency equally well. Pruss explicitly admits that this is a loophole in his argument when he concedes that ‘there is always the possibility that someone will embrace this line of reasoning as an argument against the very possibility of a substantive meta-ethics, along the lines of Moore's open question argument’ (Pruss (Reference Pruss2009), 437).

This criticism can be strengthened by pointing out that the very idea of a substantive metaethical theory is more controversial than it might seem at first sight. Keep in mind that, according to Pruss, a metaethical theory only counts as substantive if its characteristic property F serves as a non-circular definition of the property of being morally obligatory.Footnote ¹² However, for instance, proponents of the increasingly popular position of non-theistic non-naturalist moral realism doubt that such a property exists (e.g. Moore (Reference Moore1993), Shafer-Landau (Reference Shafer-Landau2003), Wielenberg (Reference Wielenberg2009), Enoch (Reference Enoch2011), Parfit (Reference Parfit2011), Cuneo & Shafer-Landau (Reference Cuneo and Shafer-Landau2014)). They argue that there is no property that serves as a non-circular definition of the property of being morally obligatory and that is not ultimately defined in terms of the property of being morally obligatory already. In this vein, for instance, Wielenberg (who incidentally is one of Pruss's main dialectical opponents) claims that ‘[e]thical properties [e.g. the property of being morally obligatory] are sui generis properties that are not reducible to other kinds of properties’ (Wielenberg (Reference Wielenberg2009), 25). Of course, it is an option for non-theistic non-naturalist moral realists to endorse a metaethical theory as long as it is not substantive – thus, they merely are committed to assume that its characteristic property F can only be defined in terms that involve the property of being morally obligatory. For this reason, a proponent of the counterpossible version of the objection of horrible commands can evade Pruss's reductio argument by endorsing non-theistic non-naturalist moral realism. Thus, Pruss's defence of DCM* is dialectically inefficient against a major strand of contemporary metaethics. This is a significant loophole of Pruss's reductio argument.

Second, Pruss himself admits that his argument is ‘unsatisfactory since it does not show what is fishy about [arguments that have the same structure as the counterpossible version of the objection of horrible commands]’ (Pruss (Reference Pruss2009), 437). Maybe, this weakness can be overcome. Pruss discusses three attempts to locate the underlying mistake:

1. (a) Pruss observes that ‘when we are very strongly sure that something is true and its truth is very important to us [e.g. that a sacrificial scenario is not morally obligatory], we have a tendency to carry it over into counterfactual situations’ (ibid., 438). We might simply reject this assumption as unwarranted and claim that it is metaphysically possible for a sacrificial scenario to be morally obligatory. In other words, we might follow the lead of Ockham and jettison the belief that certain acts are opposed to moral obligation as a matter of metaphysical necessity. However, this proposal would boil down to rejecting premise (1) again. We have already pointed out that this strategy is of no avail because it runs into conflict with the (especially among Jewish and Christian believers) widespread assumption that ‘(basic) moral and other normative norms . . . are norms with maximal, and even modally maximal, jurisdiction’ (Enoch (Reference Enoch2011), 145).

2. (b) Pruss argues that we could simply endorse vacuist semantics. We saw that Pruss's argument relies on the semantic assumption that certain counterpossibles are false (in particular, (7) in general and (3) in the special case of DCT*). Therefore, if we endorse vacuist semantics, (7) and (3) would turn out to be true and the worrisome consequences depicted by Pruss would no longer arise. However, when we try to discuss the prospects of vacuist semantics, we enter a controversial and increasingly complex debate. Morriston's counterexample (6) is only one problem that a defence of vacuist semantics would have to deal with. There are numerous other arguments that need to be taken into account.Footnote ¹³ For this reason, a lot more work would need to be done in order to prove that the underlying mistake of the counterpossible version of horrible commands is its tacit commitment to non-vacuist semantics.

3. (c) Pruss considers a suggestion by C. S. Evans. Evans claims that we tend to assume that the falsity of the counterpossible (3) threatens to refute DCM* because of ‘our ignorance of the nature of God’ (Pruss (Reference Pruss2009), 438). In this vein, Evans claims that ‘[i]f we really knew God's perfectly good nature, we would not take [(3)] seriously’ (ibid.). This line of argument can be generalized to all substantive metaethical theories: one might argue that once we properly understood a substantive metaethical theory T*, we would not take the falsity of (7) to be a worrisome consequence that threatens to refute T*. However, I fail to see how the problems for T* would vanish once (7) turned out to be less worthy of attention. To ignore a problem (unfortunately) does not make it less of a problem. Surely, an educated proponent of T* might not be preoccupied with metaphysically impossible scenarios in which a sacrificial scenario somehow ends up being F. However, once far-fetched scenarios of this kind are brought up, it becomes apparent that the beliefs of the proponent of T* are inconsistent (at least, if we suppose that the counterpossible version of the objection of horrible commands as well as Pruss's argument are correct). For this reason, I doubt that Evans's suggestion manages to show what exactly has gone awry with the counterpossible version of the objection of horrible commands.

Let us take stock. We saw that Pruss's reductio argument suffers from two weaknesses. On the one hand, Pruss's argument only manages to impress those who endorse the very idea of substantive metaethics in the first place. Of course, this does not mean that Pruss's argument is useless. Pruss's argument indeed forces those who buy into the project of substantive metaethics to withdraw support for the counterpossible version of the objection of horrible commands. On the other hand, Pruss admits that his argument is ‘unsatisfactory’ (ibid., 437) because it fails to explain what exactly has gone wrong with the counterpossible version of the objection of horrible commands. Maybe, this weakness can be eliminated, say, by providing a full-blown defence of vacuist semantics. However, a lot more work would be required to show that this can be done. Again, this aspect of Pruss's argument does not make it less forceful because one does not need to show why a theory fails in order to show that it fails. However, I believe that it is a legitimate desire to try to answer the question why the counterpossible version of the objection of horrible commands fails. If this question turns out to be impossible to answer, then the suspicion arises that Pruss's argument ultimately relies on nothing but a logical trick. Thus, I believe that it would be preferable to defend DCM* in a way that does not suffer from this weakness.

A new defence of divine command metaethics: pragmatic debunking

In this section, I will depict a new way to defend DCM* against the counterpossible version of the objection of horrible commands.

We saw that Morriston, Wielenberg, and Sinnott-Armstrong try to explain our intuition that the counterpossible (3) is false in a semantic way – by claiming that (3) must in fact bear the truth-value ‘false’. In contrast, I believe that we should provide a pragmatic explanation for this intuition while leaving intact the semantic assumption that (3) bears the truth-value ‘true’. More precisely, I believe that the intuition that (3) is false arises as a result of a conversational implicature. If this strategy is successful, then we can no longer infer the falsity of DCM* from the falsity of (3) and the counterpossible version of the objection of horrible commands loses its edge.Footnote ¹⁴

I would like to highlight that this proposal is different from simply endorsing vacuist semantics: to defeat the counterpossible version of the objection of horrible commands, we do not need to endorse the vacuist claim that literally all counterpossibles are true. Rather, my proposal merely aims to argue for the much more modest claim that (3) is true.

At first, I would like to clarify a couple of concepts that are involved in my argument. Famously, Grice has pointed out that we usually assume that conversational participants obey the so-called ‘Cooperative Principle’:

In particular, Grice observes that sensible conversations are governed by the so-called ‘maxim of quantity’ which advises us to be ‘as informative as is required (for the current purposes of the exchange)’ (ibid.). Let me briefly illustrate this maxim of Grice's ‘Cooperative Principle’. For example, take Esther's assertion ‘If I perish, I perish’ (Esther 4:16). If we assume that Esther's assertion has to be interpreted literally, then it expresses nothing but a mere tautology. Therefore, on such an interpretation, Esther's assertion fails to convey any information. Therefore, Esther's assertion would violate the maxim of quantity of Grice's ‘Cooperative Principle’.

Nevertheless, Esther's assertion seems perfectly reasonable to us. The reason is that the meaning of Esther's assertion is affected by a so-called ‘conversational implicature’. In general, we say that a speaker has ‘conversationally implicated’ a proposition p by an utterance U if and only if the application of Grice's ‘Cooperative Principle’ allows a sensible hearer to infer p from U (Birner (Reference Birner2013), 43–44). For instance, Esther may assume that a sensible hearer would not believe that she was trying to convey an uninformative tautology. Rather, Esther may assume that a sensible hearer would be able to make use of Grice's ‘Cooperative Principle’ to infer that she meant to express something more informative like ‘I am willing to risk my life’. In other words, we may say that Esther conversationally implicated that she is willing to risk her life.

Now, I believe that we have all the conceptual ingredients to understand my pragmatics-based explanation for our inclination to take (3) to be false. At this point, it is important to keep in mind what (3) said:

1. (3) If God were to command a sacrificial scenario, then a sacrificial scenario would be morally obligatory.

I believe that (3), on a literal reading, provides more information than we find appropriate in this context. Suppose we are in a context in which NECESSARY GOODNESS is part of the common ground.Footnote ¹⁵ For our purposes, only these contexts matter because we are discussing DCM against the backdrop of NECESSARY GOODNESS. Now, in these contexts, it is known that it is metaphysically impossible for God to command a sacrificial scenario. Now, suppose we interpret (3) in a literal way and, thus, bracket the effects of potential conversational implicatures for a moment. Then, I believe that it is very hard to see on which grounds a reasonable speaker might ever be entitled to assert a counterpossible like (3). Since the antecedent of (3) is known to be metaphysically impossible, it does not seem to make any difference what exactly the consequent of (3) is. A literal interpretation of (3) seems to convey more information than we find appropriate in such a context. For this reason, I believe that asserting (3) in this context would violate the maxim of quantity of Grice's ‘Cooperative Principle’.Footnote ¹⁶ Incidentally, already Aristotle seems to express a similar thought when he writes that ‘no one deliberates about what cannot be otherwise’ (Aristotle (Reference Aristotle and Crisp2004), 1139a).

Therefore, in the aforementioned context, a sensible hearer who takes for granted Grice's ‘Cooperative Principle’ may be expected to assume that the utterer of (3) meant to convey a proposition that is more informative than a literal interpretation of (3). I believe that we obtain the most plausible candidate for a more informative interpretation of (3) if we take for granted a less sophisticated understanding of God. Henceforth, I will use the term ‘God^–’ to refer to a being that is just like God with the only difference that it is metaphysically possible for it to exhibit moral shortcomings. For this reason, NECESSARY GOODNESS only applies to God, but not to God^–. Now, I believe that a sensible hearer of (3) who takes for granted Grice's ‘Cooperative Principle’ will assume that a speaker of (3) meant to conversationally implicate the following proposition (3)^–:

1. (3)^– If God^– were to command a sacrificial scenario, then a sacrificial scenario would be morally obligatory.

Let me briefly explain why (3)^– actually bears the truth-value ‘false’. We know that God^– (in contrast to God) is not perfectly morally good as a matter of metaphysical necessity. For this reason, (3)^– is an ordinary counterfactual with a metaphysically possible antecedent and we may use conventional semantic methods to assign a truth-value to (3)^– (e.g. Stalnaker's and Lewis's orthodox semantics of counterfactuals). Keep in mind that the very starting point for our discussion was the intuition that a sacrificial scenario can under no circumstances be morally obligatory. Therefore, we may conclude that even if God^– were to command a sacrificial scenario, it would still not be morally obligatory. Therefore, (3)^– is false.

The crucial step in my argument is the claim that our truth-value intuitions about (3) stem from our inclination to confuse it with (3)^–. In other words, I believe that we may assume that a speaker of (3) meant to conversationally implicate (3)^– because asserting (3) would violate Grice's ‘Cooperative Principle’ and (3)^– is very similar to (3) but yet actually conveys an appropriate amount of information. In other words, I believe that we tend to lapse back to a less sophisticated understanding of God if this is a way to make sense of a particular assertion about God. I believe that it is plausible to assume that a speaker of (3) meant to conversationally implicate (3)^– because we obtain (3)^– if we take (3) and bracket the information that is responsible for the fact that (3) violates Grice's ‘Cooperative Principle’, viz. the information that it is metaphysically impossible for God to command a sacrificial scenario.

So, we have seen that (3)^– bears the truth-value ‘false’ and that we tend to confuse (3) with (3)^– for pragmatic reasons. This explains why we tend to believe that (3) bears the truth-value ‘false’. However, since (3) and (3)^– are entirely different propositions, there is no need to assume that the counterpossible (3) actually bears the truth-value ‘false’. Rather, my pragmatic explanation is perfectly compatible with the semantic assumption that (3) is true. I conclude that drawing attention to our intuition to take (3) to be false does not threaten to refute DCM* because it is not clear how this intuition is to be explained best. The problems for DCM* vanish if we explain this intuition in a pragmatic way. For this reason, I conclude that being committed to the counterpossible (3) does not suffice to refute DCM*. Thus, Morriston's, Wielenberg's, and Sinnott-Armstrong's counterpossible version of the objection of horrible commands did not turn out to be successful. For this reason, both the simple and the counterpossible versions of the objection of horrible commands against DCM* turn out to be unsuccessful.

I believe that my pragmatics-based defence of DCM* does not suffer from the two weaknesses of Pruss's defence. First, we saw that Pruss's defence has the dialectical disadvantage that it only manages to impress those who endorse the very idea of substantive metaethics in the first place. In contrast, my defence of DCM* is entirely independent of what one thinks of the very project of substantive metaethics. Second, we saw that Pruss fails to explain exactly what has gone wrong with the counterpossible version of the objection of horrible commands. In contrast, my defence of DCM* offers a clear explanation in terms of pragmatics. Furthermore, a third advantage of my pragmatics-based defence of DCM* is that it enables us to explain why Pruss's defence of DCM* works: namely, it explains the fundamental asymmetry with respect to our tendency to treat objections that have the same structure as the counterpossible version of the objection from horrible commands. Let me elaborate. We saw that Pruss's argument is based on the assumption that we are reluctant to be convinced by the analogue of the counterpossible version of the objection of horrible commands in cases of non-DCM* substantive metaethical theories. Presumably, the reason is that DCM* is special insofar as its version of (7) employs a notoriously controversial concept (‘God’) that has a different modal profile depending on one's preferred account of it. In contrast, the version of (7) that is endorsed by other substantive metaethical theories (e.g. Kantianism) does not contain any concepts that are equally contested. Thus, in the case of non-DCM* substantive metaethical theories, we are not at risk of lapsing back to a less sophisticated understanding of the concepts involved in (7) in order to come up with a reasonable explanation of why someone might want to assert this proposition. Presumably, this is why in these cases we do not tend to assume that the speaker of (7) meant to assert a proposition that differs from (7) by means of a conversational implicature. Therefore, in the case of non-DCM* substantive metaethical theories, the intuition that (7) cannot be asserted anyway is more likely to arise and, thus, we tend to believe that it is of no importance what truth-value we assign to it. This might explain why we are reluctant to believe that (7) must be false in the case of non-DCM* substantive metaethical theories. However, as we have seen, this intuition is required in order to get the analogue of the counterpossible version of the objection of horrible commands off the ground.Footnote ¹⁷

Please note that even if my defence of DCM* is correct, this does not mean that DCM* is correct. The objection of horrible commands is merely one objection that DCM* has to face. So, even if this objection turns out to be harmless, some other objection might pose a more serious challenge for DCM*.

Objections

Of course, one may argue that the God of (3) is fundamentally different from the God^– of (3)^–. Therefore, it might seem implausible to try to explain our truth-value intuitions about (3) by our inclination to confuse it with (3)^–.Footnote ¹⁸

In response, I would like to make two points.

First, I would like to highlight that I do not recommend that we interpret (3) as (3)^–. I believe that replacing God for God^– is an erroneous (yet tempting) way to interpret (3). DCM* is formulated in terms of God and not God^–. However, by definition, God (who is subject to NECESSARY GOODNESS) is not the same as God^– (who is not subject to NECESSARY GOODNESS). Thus, one can neither replace God by God^– in the definition of DCM* nor in any proposition that follows from DCM* – in particular, not in (3). For this reason, we are well advised not to confuse (3) with (3)^–.

Second, I believe that it is nevertheless tempting to confuse the God of (3) with the God^– of (3)^– when we are supposed to make sense of (3). I would like to offer two additional arguments to back up this claim.

First, I think that it might be helpful to think about the meaning of the term ‘God’ for a moment. For instance, Oppy (Reference Oppy2014, 1–29) has put forward an account of the term that relies on the distinction between ‘concepts’ and ‘conceptions’ of God. On the one hand, Oppy suggests that the ’concept’ of God is a minimal definition of the term ‘God’ that everyone has to accept in order to count as a competent user of this term. Oppy claims that the concept of God is best formulated in the following way:

If Oppy's proposal is correct, then one cannot be considered as a competent user of the term ‘God’ if, say, one claims that God has no power over the natural world. In contrast, a ‘conception’ of God is an elaborate proposal about how God is best understood. Typically, we arrive at a conception of God as the result of a lengthy investigation of the divine nature. For example, Oppy mentions Swinburne's proposal to understand God as ‘a person without a body – i.e. a spirit – who is eternal, perfectly free, omnipotent, omniscient, perfectly good and the creator of all things’ (Swinburne (Reference Swinburne2004), 8). If Oppy's proposal is correct, then one still counts as a competent user of the term ‘God’ if, say, contra Swinburne, one claims that God has a body. Now, the details of Oppy's proposal might be controversial but that is not important for the purposes of this article. It seems that Oppy is right to claim that NECESSARY GOODNESS is not part of the concept of God. Thus, we may deny that God is perfectly morally good as a matter of metaphysical necessity and still count as competent users of the term ‘God’. Nevertheless, NECESSARY GOODNESS is part of some conceptions of God. Presumably, many theologians will agree that we can only get to know NECESSARY GOODNESS by means of ‘empirical’ investigation (e.g. by consulting or interpreting scripture or tradition). For this reason, a hearer still counts as a competent user of the term ‘God’ if he or she confuses the God of (3) (who is subject to NECESSARY GOODNESS) with the God^– of (3)^– (who is not subject to NECESSARY GOODNESS). Therefore, I do not believe that the God of (3) is so fundamentally different from the God^– of (3)^– that a confusion of both is unlikely to occur, as the objection claimed.

Second, Mugg (Reference Mugg2016) assembles psychological evidence that strengthens the assumption that we tend to treat (3) as (3)^–. Mugg points out that contemporary research relating to the phenomenon of ‘cognitive decoupling’ (e.g. Leslie (Reference Leslie1987); Nichols & Stich (Reference Nichols and Stich2003), 16–59) suggests the following picture: when we are asked to assign a truth-value to a counterpossible without a glaring contradiction in the antecedent, we tend to confine ourselves with assessing ‘the closest (partial) state of affairs’ (Mugg (Reference Mugg2016), 450) in which the antecedent is true. Mugg argues that we obtain this ‘closest (partial) state of affairs’ by pretending that the antecedent is true while ‘screen[ing] out those beliefs that (with the antecedent of the counterfactual) imply contradictions’ (ibid., 449). Thus, this supports the assumption that we tend to ‘screen out’ the belief that NECESSARY GOODNESS holds when we are asked to assign a truth-value to (3) because NECESSARY GOODNESS gives rise to a contradiction in conjunction with the antecedent of (3). Now, if we suspend our belief that NECESSARY GOODNESS holds, (3) turns into (3)^–. Therefore, if Mugg's assessment of the psychological evidence is correct, we have an empirically informed reason to believe that we tend to treat (3) like (3)^–.Footnote ¹⁹ Thus, I do not believe that one can simply undermine my pragmatics-based defence of DCM* by claiming that (3)^– is so fundamentally different from (3) that it is implausible to assume that we tend to confuse them.