Conversational implicatures cannot save divine command theory from the counterpossible terrible commands objection

Abstract

Critics of Divine Command Theory (DCT) have advanced the counterpossible terrible commands objection. They argue that DCT implies the counterpossible ‘If a necessarily morally perfect God commanded us to perform a terrible act, then the terrible act would be morally obligatory.’ However, this counterpossible is false. Hence, DCT is false. Philipp Kremers has proposed that the intuition that the counterpossible above is false is due to conversational implicatures. By providing a pragmatic explanation for the intuition, he thinks that DCT proponents can then maintain that the counterpossible is actually true. In this article, I argue that Kremers's conversational implicature response fails because (a) there is good reason to think that no conversational implicature arises given what critics of DCT have expressed, (b) a competent reader would not understand the critics' utterance that TCC is false as implicating that TCC* is false, and (c) the counterpossible terrible commands objection can be easily modified to be immune to the conversational implicature response by cancelling any potential implicature. Thus, an appeal to conversational implicatures cannot save DCT from the counterpossible terrible commands objection.

Introduction

According to Divine Command Theory (DCT), whatever God commands is morally obligatory. A problem with DCT is that it seems too implausibly to allow terrible acts, such as gratuitously sacrificing children, to possibly be morally right. DCT proponents think that this objection fails because it is impossible for God, a being that is necessarily morally perfect, to command us to perform terrible acts.

However, Wes Morriston (2009; 2012), Erik Wielenberg (2005, 41–49), and Walter Sinnott-Armstrong (2009a; 2009b, 102–106) each argue that appealing to God's necessary moral perfection does not fully address the problem because the objection can be reformulated by using a counterpossible (i.e. a counterfactual with an impossible antecedent). These critics of DCT argue that DCT implies the following counterpossible.

Terrible-Command-Counterpossible (TCC): If a necessarily morally perfect God commanded us to perform a terrible act, then the terrible act would be morally obligatory.

However, our intuitions tell us that TCC is false. Therefore, DCT is false. Call this the counterpossible terrible commands objection.

Most DCT proponents respond by noting that the orthodox account of counterfactuals entails that all counterpossibles are vacuously true (Wierenga 1983, 393–396; Quinn 2006, 81–82; Craig 2009, 172; Baggett and Walls 2011, 133; Flannagan 2012, 23; Evans 2013, 92–93; Flannagan 2022, 401–402). The view that all counterpossibles are vacuously true is known as vacuism. Given vacuism, TCC would be vacuously true, and so the counterpossible terrible commands objection fails.

Although most DCT proponents appeal to vacuism, Philipp Kremers notes that an appeal to vacuism is controversial. There is an ongoing complex debate between proponents of vacuism and proponents of non-vacuism (the view that some counterpossibles can be false, and some can be non-vacuously true). Kremers (2021, 607) says, ‘For this reason, a lot more work would need to be done in order to prove that the underlying mistake of the [counterpossible terrible commands objection] is its tacit commitment to non-vacuist semantics.’ Hence, Kremers (ibid.) proposes a new reply that does not rely on vacuism.

Kremers argues that the intuition that TCC is false is due to conversational implicatures. By providing a pragmatic explanation for the intuition, he thinks that DCT proponents can then maintain that TCC is actually true.

In this article, I argue that Kremers's conversational implicature response fails for three reasons: (a) there is good reason to think that no conversational implicature arises given what critics of DCT have expressed, (b) a competent reader would not understand the critics' utterance that TCC is false as implicating that TCC* is false, and (c) the counterpossible terrible commands objection can be easily modified to be immune to the conversational implicature response by cancelling any potential implicature. Thus, an appeal to conversational implicatures cannot save DCT from the counterpossible terrible commands objection. I conclude with a short suggestion for DCT proponents.

Kremers's conversational implicature response

To understand Kremers's conversational implicature response, we need first to understand what a conversational implicature is. Famously, H. P. Grice (1989, 26) points out that in our conversations, we usually assume that others obey a cooperative principle like: ‘Make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged.’ In many cases, when a speaker asserts p, assuming that the speaker is following the cooperative principle gives competent listeners reason to understand the speaker as conversationally implicating another proposition q, even though p does not logically entail q. More formally, when a speaker utters a sentence that expresses a proposition p, a competent listener who understands the context of the conversation would think that the speaker intends to convey another proposition q, even though p does not logically entail q. In such a case, we would say that the utterance p conversationally implicates q (ibid., 30–31). The proposition that is conversationally implicated is called a conversational implicature.¹

Here is an example from Grice (ibid., 32) that illustrates this. Imagine a conversation between A and B.

• A: Please help me. Where can I get petrol for my car?

• B: There is a garage nearby at 5th Ave.

A competent listener who understands the context of the conversation would take B's utterance to imply <A can get petrol from the garage at 5th Ave>. However, notice that the proposition that B expresses says nothing about whether A could get petrol from the garage. So, we judge that by uttering ‘There is a garage nearby at 5th Ave’, B was conversationally implicating that <A can get petrol from the garage at 5th Ave>.

Kremers (2021, 608) provides his own example from the Bible where Esther says, ‘If I perish, I perish’ (Esther 4:12). Kremers notes that Esther's statement is a mere tautology when taken literally. Her statement takes the form: If X, then X. However, we should not think that Esther is trying to communicate an uninformative tautology. Instead, given the context in which Esther uttered the sentence, we should infer that ‘she meant to express something more informative like “I am willing to risk my life”’ (ibid., 608). So, by uttering the sentence ‘If I perish, I perish’, Esther was conversationally implicating another proposition such as: <I am willing to risk my life>.

One thing important to notice from these examples is that the truth value of the proposition that a person expresses may differ from the truth value of the proposition which is conversationally implicated. For example, suppose the garage at 5th Ave repaired cars but did not sell petrol. <There is a garage nearby at 5th Ave> would be true. However, <A can get petrol from the garage at 5th Ave> would be false. So, although the proposition that B expressed would be true, the proposition that B conversationally implicates would be false.

With this background, we can look at Kremers's (ibid., 609–610) conversational implicature response. Kremers claims that in the context of discussing TCC, speakers know that it is metaphysically impossible for a necessarily perfectly morally good God to command a terrible act. Kremers (ibid., 609) then says,

it is very hard to see on which grounds a reasonable speaker might ever be entitled to assert a counterpossible like [TCC]. Since the antecedent of [TCC] is known to be metaphysically impossible, it does not seem to make any difference what exactly the consequent of [TCC] is. A literal interpretation of [TCC] seems to convey more information than we find appropriate in such a context. For this reason, I believe that asserting [TCC] in this context would violate the maxim of quantity of Grice's ‘Cooperative Principle’.

Given this, a sensible hearer who assumes Grice's cooperative principle would lead us to think that one who asserts TCC is trying to conversationally implicate another proposition.

Now, let God* stand for a being who is similar to God, except that the being is not morally perfect. Kremers (ibid., 610) thinks that by uttering TCC, speakers are conversationally implicating:

TCC*: If a God* commands a terrible act, then the terrible act would be morally obligatory.

Since it is possible for a God* to command a terrible act, TCC* is not a counterpossible. Instead, TCC* is an ordinary counterfactual with a metaphysically possible antecedent. Intuitively, TCC* is false because terrible acts are necessarily wrong.

Given this, Kremers (ibid., 610) goes on to claim that the reason why we judge TCC as false is because we have confused TCC with what TCC conversationally implicates, namely TCC*. In other words, our intuitions are misguided because our intuitions are actually tracking TCC* instead of TCC.

To support his claim further, Kremers (ibid., 613) appeals to psychological evidence. Kremers appeals to Joshua Mugg's work, which argues that research on the phenomenon of ‘cognitive decoupling’ ‘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 to assessing “the closest (partial) state of affairs” in which the antecedent is true’ (Mugg (2016), 450). Mugg (ibid., 449) argues that we pretend that the antecedent is true while we ‘screen out those beliefs that (with the antecedent of the counterfactual) imply contradictions’. Therefore, Kremers thinks that the intuition that TCC is false is actually the intuition that TCC* is false. So, Kremers (2021, 608–610) thinks that DCT proponents can still maintain that TCC is actually true.

The appeal to conversational implicatures fail

While Kremers's response to the counterpossible terrible commands objection is novel, I think the conversational implicature response fails for three reasons.

First, I think there is good reason to think that no conversational implicature arises given what critics of DCT have expressed. In our context, critics of DCT are non-vacuists who have made it known that they think that TCC is non-vacuously false. They have made it known that they think the consequent of TCC makes an important difference to what truth value TCC bears. Here are some examples from the three critics of DCT who advance the counterpossible terrible commands objection.

Sinnott-Armstrong (2009a, 106) says that ‘[E]ven if God in fact never would or could command us to rape, [DCT] still implies the counterfactual that, if God did command us to rape, then we would have a moral obligation to rape. That is absurd.’ Sinnott-Armstrong (2009b, 104) notes that his objection ‘might seem too tricky because they ask about the moral implications if God were to do something that God could not do by His very nature’. However, Sinnott-Armstrong (ibid., 104) notes that ‘[t]here are technical ways to handle counterfactuals with necessarily false antecedents’. Furthermore, we can consider a similar counterpossible which has an opposite consequent: ‘If God commanded us to rape, then rape would still be morally wrong.’ This counterpossible seems ‘plausible to most people, regardless of any technical details about counterfactuals with impossible antecedents’ (ibid., 104). So, we should think that a counterpossible like TCC is false.

Similarly, Wielenberg (2005, 49) says:

implicit in [DCT] is the notion that God has the power to make any logically consistent ethical claim true; it is only His character that prevents Him from being able to exercise this power. This implies that if, per impossibile, God were not loving, He could make it the case that it is obligatory for someone to inflict a gratuitous pummeling on another human being.

However, Wielenberg claims that such a counterpossible is false.

Similarly, when Morriston discusses the counterpossible terrible commands objection, he notes that some DCT proponents appeal to vacuism, but Morriston (2009, 250) thinks that this strategy is wrongheaded because it is intuitive to think ‘that there are lots of non vacuously true “if per impossibile” counterfactuals’. For example, Morriston (ibid., 266) says:

Suppose that it is indeed a metaphysically necessary truth that God is good. Then it is impossible for God to be evil. But surely it is true that if (per impossibile) God were evil, He would not be good, and false that if (per impossibile) God were evil, He would be good.

So, counterpossibles can be non-vacuously true or false.

In another paper discussing the same objection, Morriston even explicitly says that he rejects vacuism. Morriston (2012, 20) thinks that ‘Many counterfactual conditionals with impossible antecedents – counterpossibles, as I shall call them – seem to me to be non-vacuously true or false, and the assignment of truth-values in such cases need not be arbitrary.’

All of these examples show that critics of DCT are aware that they are dealing with counterpossibles, but they think that counterpossibles can either be non-vacuously true or false, and they think that TCC is false. When a reasonable speaker holds to non-vacuism, and has made this known to others, they would be entitled to talk about whether they think a counterpossible is true or false. They would not be conveying more information than appropriate in such a context. So, no conversational implicature would arise. Therefore, we have reason to think that a conversational implicature does not arise when critics of DCT assert that TCC is false in such a context.

Second, if a conversational implicature arises, then a competent listener (or reader) would understand the critics' utterance that TCC is false as implicating that TCC* is false. However, I am sceptical that a competent reader would take critics of DCT to be trying to convey that they think TCC* is false. I think that it is clear from the examples above that the critics of DCT are not trying to talk about some God* issuing commands in some possible world, or about an ordinary counterfactual like TCC*. Rather, they are talking about a necessarily morally perfect God, about impossible worlds, and about counterpossibles.

It would be odd to interpret critics as implicating TCC* which is about God* because God* is not morally perfect, but critics of DCT have specified that they are talking about a necessarily morally perfect God. As Kremers (2021, 609) himself notes, the context is one in which God's necessary goodness is specified already. It would also be odd to interpret critics as implicating TCC*, an ordinary counterfactual, given that critics have specified that they are talking about a counterpossible. So, I think that competent listeners (or readers) would not understand critics of DCT as trying to convey that they think that TCC* is false. Instead, critics of DCT are trying to convey that TCC is false.

I believe I am not alone in interpreting critics of DCT this way. Those who cite the critics of DCT and discuss their arguments take these critics as talking about expressing intuitions about a counterpossible (see Pruss 2009; Craig 2009, 172; Baggett and Walls 2011, 133; Flannagan 2012, 23; Evans 2013, 92–93; Davis and Franks 2015; Flannagan 2022, 401). The only exception I am aware of is Kremers's paper. Given the plausible assumption that those who cite and discuss the critics' arguments are competent listeners (or readers), we have good reason to think that competent listeners (or readers) would not take the critics' assertion that TCC is false as conversationally implicating that TCC* is false. Therefore, we have good reason to think that no conversational implicature arises.

Third, even if Kremers is right that a conversational implicature does arise, this does not show much. Critics of DCT can easily modify their argument. Conversational implicatures are standardly thought to be cancellable. Grice (1989, 44; see also ibid., 39) says that:

a conversational implicature that p is explicitly cancellable if, to the form of words the utterance of which putatively implicates that p, it is admissible to add but not p, or I do not mean to imply that p, and it is contextually cancellable if one can find situations in which the utterance of the form of words would simply not carry the implicature.

Grice's idea is that a speaker can explicitly cancel out a conversational implicature by adding a clause to state or imply that the speaker is opting out of implying another proposition. Let me illustrate this with the earlier example of the conversation between A and B. Suppose instead of saying ‘There is a garage nearby at 5th Ave’, B instead says ‘There is a garage nearby at 5th Ave, but it is closed for the week, and I am not sure where else you can get petrol from.’ In this case, the additional clause makes it clear that A cannot get petrol from the garage at 5th Ave. So, the conversational implicature <A can get petrol from the garage at 5th Ave> does not arise because it is cancelled by the additional clause.

Given this, critics can modify the counterpossible terrible commands objection to ensure that saying that TCC is false does not conversationally implicate that TCC* is false. For example, they can explicitly say that ‘TCC is false’ and make additional clarificatory remarks like, ‘In asserting that TCC is false, I do not mean to imply anything about what would happen if God* commanded us to perform a terrible act’ or ‘In asserting that TCC is false, I am not trying to talk about the counterfactual TCC*.’ By adding such clarificatory remarks, critics can ensure that the only proposition conveyed is <TCC is false> and the conversational implicature that <TCC* is false> never arises because it is cancelled by the additional remarks.² With such additional remarks, it seems that nothing much about the critics' argument would change. The additional remarks merely act as a clarification of the premise used in the counterpossible terrible commands objection. It makes sure that our intuitions would be directed towards TCC since the conversational implicature TCC* is cancelled. Therefore, the counterpossible terrible commands objection can be easily modified to be immune to the conversational implicature response.

In summary, I think Kremers's conversational implicature response fails because (a) there is good reason to think that no conversational implicature arises given what critics of DCT have expressed, (b) a competent reader would not understand the critics' utterance that TCC is false as implicating that TCC* is false, and (c) the counterpossible terrible commands objection can be easily modified to be immune to the conversational implicature response by cancelling any potential implicature.

Even though I think Kremers's conversational implicature response fails, I want to end with a suggestion for DCT proponents. Based on Kremers's argument, there seems to me to be another response along similar lines that DCT proponents might take and develop. Even though I think that TCC* is not conversationally implicated, perhaps Kremers is still right that our intuitions are actually tracking TCC* instead of TCC. Perhaps when we evaluate TCC, we tend to imagine a possible world in which God* commands a terrible act, and so we actually have TCC* in mind. This is supported by the psychological evidence found in Mugg's work, which Kremers mentions. So, instead of appealing to pragmatics, perhaps DCT proponents can instead respond to the counterpossible terrible commands objection by offering a psychological explanation of why critics judge that TCC is true – namely because their intuitions are actually about TCC*.