Alarming divine command conditionals and counterpossibles

If we assume with Davis and Franks that commanding an SS is incompatible with God's moral nature, then DCC is a counterpossible – a counterfactual with an impossible antecedent – and the proper handling of such counterpossibles becomes central to the debate. Hence the title of Davis and Franks's article: ‘Counterpossibles and the terrible divine command deity’.

On what might be called the Standard View, there are no false counterpossibles. Where p is an impossible proposition, ‘>’ is the counterfactual connective, and q is any proposition whatever, it is stipulated that p > q is true. So if DCC is a counterpossible, matters resolve themselves rather quickly. DCC is classified as a trivial and uninformative truth, and the Critic's objection to the DCT collapses.

Although Davis and Franks and I follow rather different paths and reach rather different conclusions, neither they nor I follow the Standard View. As ‘theistic activists’, Davis and Franks embrace ‘the null world hypothesis’, according to which there would be nothing at all if (per impossibile) God did not exist. This is obviously not meant to be true merely in virtue of the impossibility of God's non-existence. Rather it is intended to express the radical dependence of everything else – ‘the visible and the invisible’ – on God.

Given the null world hypothesis, Davis and Franks define what they take to be an important class of false counterpossibles – those whose antecedents involve God's non-existence, and whose consequents imply that something or other exists. The heart of their article consists in an argument for saying that alarming divine command conditionals like DCC belong to this class of falsehoods. I'll have a good deal to say about this argument presently.

In Morriston (Reference Morriston2009) and (Reference Morriston2012), I took a rather different path. I did not pretend to have a recipe for deciding, in every case, whether a given counterpossible is true or false, but I did at several points introduce counterpossibles that I take to be non-vacuously true. For example, I said that even if God could not bring Himself to command an SS (on account of his essential goodness) it would still be the case that if (per impossibile) God chose to give such a command, He would succeed in giving it. In another place, I offered the following illustration of a non-vacuously true counterpossible: if (per impossibile) a completely truthful and omniscient being said that two-plus-two equals five, then two-plus-two would equal five.

My thought was that in cases like these a counterpossible is true because of the way in which the content of its antecedent is related to the content of its consequent, and not because its antecedent is impossible.Footnote ² In the first of the above-mentioned cases, I thought it would be generally agreed that a maximally powerful being would succeed in accomplishing whatever it chose to accomplish. And in the second case, I thought it would be obvious to anyone that whatever a completely truthful and omniscient being said would have to be true.

Davis and Franks point out that if this approach to counterpossibles were adopted, it would be easy to ‘construct a parallel argument for the truth of DCC’. ‘For surely’, they say, ‘whatever God, a perfectly morally good being, commands must be morally obligatory’ (Davis & Franks (Reference Davis and Franks2015), 9). From this they conclude that my way of assigning truth-values to counterpossibles undermines my own argument against the DCT.

I must pause briefly to clear up a misunderstanding. Although the Critic's argument (including his claim that DCC is false) was featured in the introductory section of Morriston (Reference Morriston2009), I intended to use that argument only as a springboard for a more nuanced discussion. I did not (in propria persona) say that DCC is false – at least not when it is construed as a counterpossible.Footnote ³ In Morriston (Reference Morriston2012), my position was much clearer. Immediately following the ‘omniscient and completely truthful being’ example, which Davis and Franks cite, I said that whatever a perfect being commands must be morally obligatory – and I immediately drew the conclusion that the DCT escapes refutation ‘because the alarming-sounding counterpossibles implied by it turn out to true’ (Morriston (Reference Morriston2012), 21).Footnote ⁴

So I agree that my way of assigning truth-values to counterpossibles can be used against the Critic in something like the way suggested by Davis and Franks. For them, of course, this is only a dialectical manoeuvre. They think this is not the proper way to defend the DCT – indeed, they say that it is ‘desperately misguided’ (Davis & Franks (Reference Davis and Franks2015), 10–11). They complain that ‘Morriston's method’ places ‘undue emphasis on concept inclusion as the sole factor for sifting true from false counterpossibles’, and that it fails to tell us ‘what the extra-conceptual world would have been like had [the] antecedent [of the counterpossible] been true’. In particular, they say, it fails to tell us ‘what the world would have been like, if God had commanded one of Morriston's Sacrificial Scenarios’ (ibid., 10).

So, then, what do Davis and Franks think the world would be like if God commanded a thing like that? Their rather surprising answer appears to be that there would be nothing at all – no God, no divine commands, and certainly no obligation to perform an SS. From this they infer that DCC and its ilk are all false. But they insist that this is not a problem for the ‘savvy’ DCT-ist, since she is not committed to the truth of DCC.

Davis and Franks begin making their argument for the falsity of DCT by introducing the concept of a ‘state of affairs involving God's non-existence’.

The antecedent of DCC is a SIGN, because God's commanding an SS involves God's not being perfectly good.Footnote ⁵ And I presume that Davis and Franks would say the same of any proposition they deemed incompatible with one of God's essential properties.

Davis and Franks's next move is to declare their allegiance to theistic activism and the ‘null world hypothesis’. They quote Thomas Morris with approval:

With the null world hypothesis and the doctrine of SIGNs in place, Davis and Franks are ready to show that DCC is false. Here is the critical part of their argument.

However, (13) and DCC jointly entail

I wouldn't have thought we'd need to invoke the null world hypothesis to see that (14) is false, but I certainly agree that it is false. So if DCC, together with other correct principles, really does entail (14), it must be false as well.

Is this argument successful? Let's take a closer look. Davis and Franks derive (14) from DCC and (13) by invoking ‘the principle (p ⇔ q) & (p > r); hence (q > r)’. And they claim to have derived (13) from God's essential moral perfection, ‘together with the fact that commanding an SS is a mark of gross moral imperfection’. Both derivations raise questions.

To see that there is a problem here, notice first that (13) is equivalent to the conjunction of

and

Both claims are decidedly odd. One would have thought that God could not command anything without existing, and that non-existent beings could command nothing. It's hard to see why God's moral perfection (and the consequent impossibility of His commanding an SS) should make us think otherwise.

The explanation, I think, is that Davis and Franks are saying only that ‘God commands an SS’ and ‘God does not exist’ strictly imply one another. Given God's essential moral perfection and the fact that commanding an SS is (necessarily) ‘a mark of gross moral imperfection’, it follows that ‘God commands an SS’ is necessarily false. So if we assume (as Davis and Franks surely do) that ‘God does not exist’ is also necessarily false, then we have a pair of impossibilities on our hands. They strictly imply one another because it's impossible for either to be true whilst the other is false. Understood in this limited way, (13₁) and (13₂) are both true, and (13) is a strict equivalence.

With this understood, let's take a closer look at the way in which Davis and Franks attempt to derive (14) from (13) and DCC. Their argument consists in an application of the following principle of counterfactual reasoning.

Given that (13) is a strict equivalence, Davis and Franks's application of PC is formally correct. But all is not well. Here is another application of PC that should give them pause. From (13) and

it follows (in accordance with PC) that:

Davis and Franks must say that F is false, since its antecedent involves God's non-existence and its consequent entails the existence of something.Footnote ⁶ But I'd be surprised if they would be willing to say that this provides a reductio of T, since that would bring them into conflict with the principle that any proposition counterfactually implies itself. Yet the pattern of reasoning is precisely the same as the one Davis and Franks use to show that DCC must be false. Clearly, something has gone wrong.

At this point, Davis and Franks must choose between two alternatives. They can revert to the standard view of counterpossibles, and say that the counterpossibles under consideration (F, T, (14), and DCC) are vacuously true; or they can say that the standard rules for inferring counterfactual conditionals from strict implications do not apply to non-vacuous counterpossibles like these. Either way, their attempted reductio of DCC fails.

But I wonder: why do Davis and Franks bother with (13) and (14) and PC at all? After all, they think the antecedent of DCC is a SIGN. So why don't they dispose of DCC directly, in exactly the way they dispose of (14)? After all, DCC has the same consequent as (14). So if its antecedent involves God's non-existence, then (by Davis and Franks's standard) DCC must be false because its consequent ‘implies the existence of at least one thing’.

But does the antecedent of DCC involve God's non-existence? Lurking just beneath the surface of Davis and Franks's explanations, there is in fact a three-step argument for saying that it does – an argument that appeals only to premises and principles that Davis and Franks clearly accept.

1. 1. God's commanding an SS involves God's failing to be morally perfect.

2. 2. God's failing to be morally perfect involves God's non-existence.

3. 3. Therefore, God's commanding an SS involves God's non-existence.

I think this must be very close to what Davis and Franks have in mind – certainly all the ingredients are theirs. They say that ‘commanding that an SS be performed is a mark of gross moral imperfection’ (11), and they really do seem to mean that commanding an SS is incompatible in a very deep way with being morally perfect. Indeed, they say that the DCT's Critic is right in thinking that ‘only a “terrible” deity would command such “a terrible thing” ’ (17). They also hold that ‘being perfectly morally good is a property God could not have failed to possess without failing to exist’ (11, my italics), and from this they conclude that any proposition ‘involving’ God's not being perfectly morally good’ also ‘involves’ God's non-existence (11). So I think it's quite clear that Davis and Franks are committed both to the premises of the argument and to the claim that its conclusion follows from those premises.

What are we to make of the argument? One obvious problem is that it's hard to see how its conclusion could be true. One would have thought that God's commanding an SS (or, indeed, anything at all) would involve the existence, not the non-existence, of God. If that's right, then something must have gone wrong.

To investigate the matter properly, we need to know what logical relation is signified by the word ‘involves’. Davis and Franks don't provide a definition, but to judge from an earlier article by Davis (Reference Davis2006), I think what they have in mind must be a type of ‘non-strict, relevant implication’. The details of Davis's account need not detain us.Footnote ⁷ What matters is that on his account impossible propositions have non-trivial implications, and that the counterpossible conditionals corresponding to those implications are non-vacuously true. For example, Davis says that ‘Three is the sum of two even numbers’ relevantly (and counterfactually) implies ‘Three is evenly divisible by two’, but does not relevantly (or counterfactually) imply ‘There is a moon orbiting the earth’ (ibid., 383).Footnote ⁸

So let's suppose that each of the involvement claims in our three-step argument is to be construed as a ‘relevant implication’ claim. Then for each step there is a corresponding non-vacuous counterfactual, and our assessment of that step will be mirrored by our assessment of that counterfactual. It will be easiest to see where the argument goes astray if we begin by considering a counterfactual version of it.

1. 4. If God were to command an SS, God would fail to be morally perfect.

2. 5. If God failed to be morally perfect, God would not exist.

3. 6. Therefore, if God were to command an SS, God would not exist.

This bit of counterfactual reasoning is fallacious. The problem is that the evaluative context within which premise 4 is true is one in which neither 5 nor 6 can be true. Premise 4 invites us to consider a situation in which God exists and commands an SS, and in which commanding an SS is assumed to be a non-negotiable ‘mark of gross moral imperfection’. Given this context for assessing 4, God would fail to satisfy an appropriate standard of moral perfection if He commanded an SS – but He would most certainly not fail to exist. So given this context premise 5 is false.Footnote ⁹

To get premise 5 to come out true, we require an evaluative context in which possessing moral perfection is assumed to be a non-negotiable necessary condition of God's existence – a context in which ‘being perfectly morally good is a property God could not have failed to possess without failing to exist’ (Davis & Franks (Reference Davis and Franks2015), 11; my italics). But within this evaluative context – within, that is, a context in which God's essential moral perfection is held constant, it's just not true that God would be morally imperfect if (per impossibile) He commanded an SS. What's true instead is that if God commanded such a thing, His doing so would not be a ‘mark of gross moral imperfection’.

With this point about context-sensitivity in mind, let's take another look at the inference of 3 from 1 and 2. If 1 and 2 are construed as ‘relevant implication’ claims, then 1 entails 4 and 2 entails 5. Since there is no evaluative context in which 4 and 5 are both true, neither is there one in which 1 and 2 are both true. So the argument is impaled on one or the other horn of a familiar dilemma. Either we keep the evaluative context constant or we don't. If we do, then one of the premises (1 or 2) must be false. If we don't, then the conclusion (3) doesn't follow from the premises.

I myself have been unable think of any evaluative context in which either conclusion (3 or 6) is non-vacuously true. So I think it is probably fair to conclude that the antecedent of DCC does not (pace Davis and Franks) involve God's non-existence. Still, it's worth taking a moment to look at what happens when we combine Davis and Franks's doctrine of SIGNs with the null world hypothesis, and apply them to the antecedent of DCC.

Davis and Franks claim to have provided ‘a more discriminating theistic basis for tracing out the counterfactual implications of impossible propositions involving God's non-existence’ (ibid.). Their leading idea is that any counterpossible whose antecedent is a SIGN, and whose consequent implies that anything at all exists, must be false. So, then, if the antecedent of DCC (‘God commands an SS’) were a SIGN, the following counterpossibles would both be false.

• • If God were to command an SS, then something would have been commanded by someone.

• • If God were to command an SS, then someone would have been commanded to do something.

By my lights, these are straightforward and non-vacuous truths. In both cases the consequent is transparently related to the antecedent in such a way as to guarantee the truth of the counterpossible claim.Footnote ¹⁰ So (I say) something must be wrong with any way of ‘tracing out counterfactual implications’ on which they are false.

It will be recalled, however, that Davis and Franks complain that this way of assigning truth-values to counterpossibles fails to ask what ‘the extra-conceptual world’ would have been like if their impossible antecedents had been true (ibid., 10). This objection is confused. From the fact that the grounds for making a claim are conceptual, it does not follow that the claim is only about concepts. No one would deny, for example, that if there were bachelors on Mars, there would be men on Mars. That is one thing we can safely say about what an extra-conceptual world in which there were bachelors on Mars would be like. That the grounds on which this is reasonably asserted are purely conceptual is neither here nor there.

So, then, do you want to know something about what the world would be like if God commanded an SS? The first thing to do is to nail down the appropriate context for answering this question. Are we to assume that we're talking about a God who can't fail to be morally perfect? If so, then the answer has to start with the thought that it would be a world in which a morally perfect God commands an SS. In that case, of course, it would be a world in which there is a God who issues at least one command. Reflecting on the concept of a command, we can also say that there would be creatures capable of being commanded to do various things. In these and many other respects, things would be as they are in reality. But there is one important respect in which they would be different – there would be a circumstance in which sacrificing ten-year-olds in a gruesome and painful way is not wrong.

If this answer seems unacceptable, that may be because we are so appalled by the Sacrificial Scenario, and so certain that such a thing could not be morally permissible (much less obligatory), that we are tempted to carry this conviction over into the counterpossible situation in which a perfectly wise and morally good God commands it.Footnote ¹¹ But if we keep firmly in mind that the counterpossible situation in question is one in which a perfectly morally good God commands an SS, together with the thought that nothing such a God commands could be morally wrong, it will be clear that the wrongness of what's been commanded is not a feature of that (impossible) situation.

This implication will seem less damaging to the DCT if one bears in mind that, as Alexander Pruss has pointed out, every substantive metaethical theory of moral obligation has similar counterpossible implications (Pruss (Reference Pruss2009), 434–435). The Kantian, for example, must concede that if, per impossibile, torturing the innocent were categorically required by reason, then torturing the innocent would be morally obligatory. And the Rawlsian must concede that if, per impossibile, free agents negotiating under the veil of ignorance came up with a system that required torturing the innocent, then torturing the innocent would be morally obligatory. But no one thinks that either of these theories is refuted merely by the fact that it has such a counterpossible implication.

It is perhaps easier to imagine a godlike being commanding cruel sacrifices than it is to imagine the torture of the innocent being categorically required by reason, or being approved under the veil of ignorance, and this difference may give greater salience to the question, ‘What if God commanded that?’ But this difference largely disappears when we try to imagine a God who is perfectly morally good commanding cruel sacrifices. (See ibid., 439.)

For these reasons, I do think that a DCT-ist can provide an effective response to the Critic if she builds God's moral perfection into her account of moral obligation. She would also, I think, be well advised to include the requirement that God's moral perfection is one of His essential features. If she does not do this, she leaves open the possibility of circumstances in which God is not good and issues terrible commands. Her theory won't imply that it would be morally obligatory to obey such commands, but it will fail to imply that it would be wrong to obey them.

Needed: A theistic theory of moral goodness

There is still, in my judgement, at least one important missing ingredient. The DCT-ist needs to say more about God's moral goodness and the way in which it precludes the possibility of God's commanding anything like an SS. This is where I think the real trouble begins. (See Morriston (Reference Morriston2009), 252–254.) I'll briefly explain why I think so, and then consider some further points of disagreement with Davis and Franks.

Coming up with a satisfactory explanation of the fact that God's moral perfection rules out commanding such things is not nearly as straightforward as it might seem. The immediately obvious answer – that it would be wrong for anyone to command them, and that a morally perfect being would never do what is wrong – might be correct, but it implies that there are moral requirements to which God Himself is subject. This fits poorly with the DCT, according to which all moral obligations are fixed by God's commands. What the DCT-ist needs is an account of God's essential moral perfection that ‘gives it enough positive content to rule out the possibility of hideous divine commands, but does so without compromising the strong view of divine sovereignty that provides so much of the motivation for the DCT’ (ibid.).

At this point, the DCT-ist may invite us to consider various aspects of God's moral nature. God, she may say, is morally good in so far as He possesses properties like love and justice and mercy and compassion. Indeed, He is maximally good because he possesses these and other like characteristics to the maximum possible degree. It is obviously impossible for a being having these properties to command an SS. So if we can assume that they are among God's essential properties, it follows that there is no possible world in which He commands an SS.

Does this take care of business for the DCT-ist? I think not – at least not if she is anxious to preserve a strong view of divine sovereignty. The problem is that on the view under consideration, it looks very much as if the ultimate standard of moral goodness consists in a cluster of good-making properties that have no essential reference to God. What matters to any person's moral goodness will be the degree to which she is loving and just and the rest. If God exists and possesses all the relevant good-making properties to the maximum possible degree, then He is maximally good. But it is equally true that finite persons are good to the degree that they are loving and just and so on.

To drive this point home, I have sometimes pressed a Euthyphro-like query. ‘Is God good because He has these good-making properties? Or are they good-making because God has them?’ (ibid., 253). If we take the first alternative and say that God is good by virtue of having these properties, divine sovereignty is compromised. God and creatures are alike subject to an independent standard of goodness – a standard that has no essential reference to God. But the second alternative seems quite incredible to me. Why ‘should it make any difference to the good-makingness of compassion if there is (or isn't) a supremely compassionate God?’ (ibid.).

Davis and Franks respond by saying that my Euthyphro-like query poses a false dilemma. It is neither the case that God is good because He has these good-making properties, nor that they are good-making because God has them. There is ‘no “how” or “why” about it’, they say.

Without further elaboration, this proposal does not address the principal issue I had meant to raise. It will be recalled that my challenge was to provide an account of God's moral goodness ‘that gives it enough positive content to rule out the possibility of hideous divine commands’ without compromising divine sovereignty (Morriston (Reference Morriston2009), 253–254). Saying merely that goodness is one of God's ‘basic’ properties gives it no positive content. It leaves us with a number of unanswered questions: questions about the relationship between the property of being good and other morally significant properties like being loving or being just; about the way in which God's moral goodness precludes His commanding an SS; and about how, if at all, goodness in creatures depends on God's goodness.

Davis and Franks know that I was looking for more.

I am not sure how Davis and Franks mean to answer this question. To answer it in the affirmative would, after all, be to embrace something like the right side of my Euthyphro-like query, though with an important qualification. The idea would be that these properties are good-making because God has them, and because He has the basic property of being good.

Even with this qualification, the view seems quite implausible to me. As far as I can see, it makes no difference to the good-makingness of (say) compassion whether or not there is a supremely compassionate (and good) God. Davis and Franks are not impressed. They say that this commits me to affirming that:

But (they continue) the DCT-ist who is ‘armed with theistic activism’ can easily reject (16). From her point of view, it's just another false counterpossible, ‘for it implies that something would exist even if God did not’ (ibid.).

That's as may be, but my point here doesn't require the truth of (16). I invite the DCT-ist – ‘armed’, if she likes, with the null world hypothesis – to consider instead:

(16*) does not imply that ‘something would exist if God did not’. What it does imply is that the good-makingness of compassion does not depend on God's being compassionate.

I anticipate Davis and Franks's response. They will surely say that compassion is one of those properties that God could not have failed to possess without failing to exist, and that the antecedent of (16*) is therefore a state of affairs involving God's non-existence (a SIGN). In that case, they will say, (16*) is just as objectionable as (16), and for the same reason. It implies that ‘something would exist if God did not’.

I am sceptical of this manoeuvre. It seems to me that one can intelligibly ask what things would have been like if God's moral nature had been different. But what if I'm wrong about this? What if the antecedent of (16*) is a SIGN? And what if that does bring it into conflict with the null world hypothesis? Would this warrant us in concluding that compassion is good-making because, and only because, God is compassionate and good? It would not. By itself, the null world hypothesis gives us no specific information about the different ways in which different sorts of things depend on particular facts about God. Consequently, it gives us no reason to think that God's goodness and compassion play a special role in making it the case that compassion is good-making for creatures.