Supervenience, expressivism and theistic ethics

Abstract

Expressivism is supposed to have an advantage over moral realism, in that it can explain why it is a conceptual truth that the moral supervenes on the natural, even though the natural does not entail the moral. I develop an analogy between expressivism and a version of theistic moral realism, and argue that this version of theistic moral realism shares any advantage that expressivism might have. It may be that the alleged advantage that expressivism has over moral realism is overstated. Nevertheless, possible analogies between expressivism and theistic moral realism have received very little attention in the past, and are worth considering for their own sake.

1 Introduction

Expressivism, it is frequently thought, has an advantage over moral realism, in that it can easily account for the fact that it is a conceptual truth that the moral supervenes on the natural, even though the natural does not entail the moral.

But if, as I shall argue in this paper, a relevant analogy can be made between expressivism and a particular version of theistic moral realism, which I shall call Divine Expressivism, then this version of theistic moral realism shares any advantage that expressivism might have over other versions of moral realism.

Here is my argument:

• Premise 1 Expressivism has an advantage over (non-theistic) moral realism in terms of the ability to explain why it is a conceptual truth that the moral supervenes on the natural.

• Premise 2 Divine Expressivism is analogous to expressivism in relevant ways.

• Conclusion Divine Expressivism also has an advantage over (non-theistic) moral realism in terms of the ability to explain why it is a conceptual truth that the moral supervenes on the natural.

In Sect. 2 I shall outline the argument for Premise 1. In Sect. 3 I shall discuss a version of theistic moral realism which (I argue) is analogous to expressivism in a way which allows it to share any advantage that expressivism might have. Finally, in Sect. 4 I shall address an objection to my argument.

My conclusion will be fairly modest. My main purpose is to highlight the analogy between expressivism and theistic moral realism. It may be that the alleged advantage that expressivism has over moral realism is overstated, in which case there may be no significant advantage for theistic moral realism to share. Nevertheless, possible analogies between expressivism and theistic moral realism have received very little attention in the past, and are worth considering for their own sake.

2 The supervenience challenge

Most philosophers agree that it is a conceptual truth that the moral status of a thing cannot change over time without some change in the thing’s natural properties, and that two things which are identical in terms of their natural properties must also be identical in terms of their moral status. Anybody who denies this would, according to Blackburn (1973: 115–6), be guilty of ‘a logical and not merely a moral mistake’.

Furthermore, many philosophers are committed to a non-entailment thesis, according to which the truth of at least some moral propositions is not entailed by any natural fact (ibid.: 120). It is important to note that entailment is understood here as a logical or conceptual relation, rather than, for example, a metaphysical relation. Suppose you and I agree about all the natural features of an action, and yet I conclude that this action is wrong and you conclude that it is not wrong. One of us is clearly mistaken, but in this case it is a moral and not a logical mistake that is being made. If the moral features of an action are entailed by its natural features, then any disagreement of this sort would mean that one of us was conceptually confused. As Blackburn points out, his argument does not require the claim that the truth of no moral proposition is logically entailed by a natural fact, but only the weaker claim that there are some moral propositions whose truth is not entailed by any natural fact.

Now, Blackburn thinks that combining the supervenience thesis and the non-entailment thesis is problematic under moral realism.¹ He poses the problem in terms of a ban on ‘mixed worlds’. Suppose that φ and ψ are two actions which are identical to one another in all their natural features. According to the non-entailment thesis, it is not logically necessary that something with these natural properties has any particular moral status, and so there is a logically possible world w₁ where φ is morally good and a logically possible world w₂ where φ is morally bad. According to the supervenience thesis, it is logically necessary that ψ has the same moral status as φ in each world, and so it is logically necessary that, given the moral status of φ in these two worlds, ψ is morally good in w₁ and is morally bad in w₂. There is no logically possible world w₃, where φ is morally good and ψ is morally bad, or vice versa; such a world would be a ‘mixed world’, where two things identical in natural properties differ morally. What Blackburn finds puzzling is this ban on ‘mixed worlds’ (1985: 135).

Similarly, although there is a logically possible world w₁ where φ has particular natural properties and is morally good, and there is a logically possible world w₂ where φ has the same natural properties and yet is morally bad, there is, according to the supervenience thesis, no logically possible world w₃, where φ has particular natural properties at time t₁ and is morally good and has the same natural properties at time t₂ but is morally bad. As Blackburn says:

we are asked to make intelligible the notion of a state of affairs subject to the constraint that its existence does not follow from the naturalistic facts being as they are but its continued existence follows from the natural facts staying as they are. (1973: 119, his emphasis)

The challenge for moral realists is to explain this ban on ‘mixed worlds’.

A simple way out of this problem would be to deny one of Blackburn’s two theses. One might deny the supervenience thesis by arguing that it is only metaphysically rather than conceptually necessary that the moral supervenes on the natural, and since Blackburn does not deny that the natural metaphysically entails the moral the problem would disappear. Klagge (1984: 374–5) at one point proposed this solution although he later recanted (1988: 465, including his footnote 8), apparently agreeing with Blackburn that obeying the supervenience constraint is constitutive of making genuine moral judgments. Hills (2009: 174) suggests that it may not be even metaphysically necessary for the moral to supervene on the natural—there may be metaphysically possible worlds where utilitarianism is true at one time and Kantian deontology becomes true subsequently.

Alternatively, one might deny the non-entailment thesis by insisting that the moral is conceptually entailed by the natural, so that it is not just true, but a conceptual truth, that things have the moral status which they do in fact have. Cuneo & Shafer-Landau (2014: 404−15) argue that it is a conceptual truth that at least some things are wrong, which leads them to conclude that moral error theorists, who deny that anything is objectively morally right or wrong, must be ‘conceptually deficient’, though not necessarily ‘conceptually confused’. However, they do not claim that all moral truths are conceptually necessary, and Blackburn’s non-entailment thesis claims only that there are some moral propositions which are not entailed by natural ones.

Many moral realists accept both of Blackburn’s theses, and these people might seek to avoid his challenge by arguing that the ban on ‘mixed worlds’ is no more of a problem for realism than it is for expressivism. But Blackburn denies that it is a problem for expressivism. According to expressivism, a statement like ‘φ-ing is wrong’ is an expression of an attitude—perhaps an attitude of disapproval towards people φ-ing. As Blackburn points out:

There can be no question that we often choose, admire, commend, or desire, objects because of their naturalistic properties. Now it is not possible to hold an attitude to a thing because of its possessing certain properties and, at the same time, not hold that attitude to another thing that is believed to have the same properties. The nonexistence of the attitude in the second case shows that it is not because of the shared properties that I hold it in the first case. Now, moral attitudes are to be held towards things because of their naturalistic properties. Therefore it is not possible to hold a moral attitude to one thing, believe a second to be exactly alike, yet at the same time not hold the same attitude to the second thing. Anybody who appears to do this is convicted of mis-identifying a caprice as a moral opinion. (1973: 122)

Hence, if we believe two things are identical in terms of their natural properties, our moral attitudes towards them will be identical.² There is a consistency constraint on our moral attitudes. Even though a particular moral attitude towards something is not entailed by that thing’s natural properties, it is logically necessary that I hold a particular moral attitude towards one thing given that I hold this attitude towards something else which is identical in terms of natural features and given that a moral attitude must be based on something’s natural properties in order to qualify as a moral attitude.

Blackburn acknowledges that there is a similar consistency constraint on moral beliefs; if a moral realist believes something to be wrong on the basis of its natural features, then the realist must also believe something else to be wrong, so long as the two things are identical in terms of their natural features. Unfortunately, however, as Blackburn points out, this

merely shows the realist putting conditions upon what can be believed to be the truth, not upon what is the truth. Our belief, he is saying, has to be consistent across naturalistic similarities—but this is no explanation of why, on his theory, the truth has to be. (1973: 122, his emphasis)

It looks, then, as if expressivism has an advantage over moral realism in terms of explaining moral supervenience. Expressivism has the comparatively straightforward task of explaining a consistency constraint on attitudes, whereas moral realism has the much more difficult task of explaining a consistency constraint on facts and properties. Blackburn doesn’t claim that this refutes moral realism, but he does think it leaves moral realism with a problem, and is therefore a point in favour of expressivism.

3 Expressivism and theistic ethics

Now, if Blackburn is right to think that expressivism has an advantage over (non-theistic) moral realism in this respect, then it looks as if theistic moral realism can share this advantage. For if moral properties relate in some way to God’s attitudes then it is possible to explain supervenience, as Blackburn does, on the basis of a consistency constraint on attitudes–not human attitudes this time, but God’s.³

How might moral properties relate to God’s attitudes? The precise details are unimportant for my purposes here, but I think a plausible relationship would be one of property-identity: the property of wrongness is identical to the property of being disapproved of by God. Other moral properties would then relate to God’s attitudes in analogous ways. The important thing to note, though, is that I am concerned with God’s attitudes of approval and disapproval, rather than, for example, his commands. If Blackburn is correct to think that there is a consistency constraint on our attitudes, then there must presumably also be a consistency constraint on God’s attitudes, but it need not follow that there is a consistency constraint on God’s commands.

God’s attitudes seem less vulnerable to the charge of arbitrariness than his commands might be, and can therefore better accommodate the supervenience constraint. It can sometimes be quite rational to give an arbitrary command, as is illustrated by Mark Murphy. He gives the example of Jane and Tom, each of whom could jump in the river to save a drowning child; however, if they both jump in they would get in each other’s way and fail to save the child:

So I give a command: ‘Jane, jump in and save the child! Tom, stay on the bank!’ This is a reasonable act of commanding: I had reason to command one and only one of them to save the child; but there was no relevant general property to distinguish Tom and Jane; so I gave a command that was not determined by their general properties. It was random; I just picked. (Murphy 2002: 90)

I assume that Murphy’s point is not so much that the issuing of arbitrary commands cannot give rise to obligations; after all, if Jane and Tom are both junior lifeguards, and the command is issued by the senior lifeguard whom they have a duty to obey, then presumably Jane does acquire an obligation which Tom does not have. Rather, his point is surely that the basis or foundation of morality cannot be arbitrary in this way. If it is rational for the lifeguard to give arbitrary commands, then it seems plausible to suppose that it could sometimes be rational for God to give arbitrary commands. However, I don’t think it makes sense to think of God’s attitudes of approval and disapproval as being arbitrary in this sense. God might disapprove of Jane and Tom failing to agree amongst themselves which of them should save the child, but unless there is some relevant difference in their situations I don’t think it makes sense for him to disapprove of one of them failing to save the child unless he also disapproves of the other for failing to save the child. This strikes me as an important reason for preferring a theory which identifies moral wrongness with God’s attitudes rather than with God’s commands.

There are additional reasons why I think it makes more sense to identify wrongness with God’s attitudes rather than his commands, most of which do not concern me here. Certainly, if we judge something to be wrong it seems much more plausible to suppose that we will disapprove of it than that we will issue commands regarding it, and the reverse is also true: we are more likely to think that somebody judges something to be wrong if they disapprove of it than if they merely command us not to do it. This seems to give at least some reason to believe that what really is wrong is more closely associated with God’s attitudes than with God’s commands. Furthermore, an action is either commanded or not commanded, whereas there are subtle differences between different types of attitudes, which allows for different moral properties to be identified with different attitudes. If something is obligatory in virtue of being commanded by God then it is difficult to see how non-obligatory goodness relates to God’s commands; and in fact many Divine Command Theorists relate only deontic moral properties with God’s commands (cf. Adams, 1979; 1999). Even here, though, it seems difficult to explain how, of two wrong actions, both of which are forbidden by God, one might nevertheless be morally worse than the other. A theory whereby moral properties are related to God’s attitudes rather than his commands allows for nuances which seem problematic for Divine Command Theory.

This does not mean, of course, that God’s commands are irrelevant to morality, but merely that they are not the ultimate basis or foundation of morality. God’s commands could still create new obligations; just as Jane acquires a new obligation when the senior lifeguard commands her to jump in the river, so she would presumably acquire a new obligation if God commanded her to jump in the river. This is the case assuming we have an obligation to obey God—in other words, if God disapproves of disobedience to his commands. According to the property-identity thesis I am proposing, however, to be morally wrong just is to be disapproved of by God, and so to ask why we have an obligation to avoid those acts of which God disapproves would be like asking why we have an obligation to avoid wrong actions.

Briefly, we might identify moral properties with God’s attitudes by beginning with the following claim:

• (1) For all agents X and all actions φ, for it to be morally wrong for X to φ just is for God to disapprove of X φ-ing.

By taking into account not just the fact of God’s disapproval, but also the degree, or intensity, of his disapproval, we are able to make comparative claims about the moral status of different actions:

• (2) For all agents X, all actions φ and all actions ψ, for it to be morally worse for X to φ than for X to ψ just is for God to disapprove more strongly of X φ-ing than of X ψ-ing.

An act is morally permissible if it is not wrong, and an act is morally obligatory if it is morally wrong not to do it. Hence, the following two claims follow from (1):

• (3) For all agents X and all actions φ, for it to be morally permissible for X to φ just is for God not to disapprove of X φ-ing.

• (4) For all agents X and all actions φ, for it to be morally obligatory for X to φ just is for God to disapprove of X not φ-ing.

As just mentioned, it is not just deontic moral properties that could be related to God’s attitudes; if moral rightness and wrongness relate to God’s attitudes of disapproval, then non-obligatory moral goodness might relate to his attitude of approval (by which I mean positive approval as opposed to an absence of disapproval):

• (5) For all agents X and all actions φ, for it to be morally good for X to φ just is for God to approve of X φ-ing.

If God disapproves of X not φ-ing then he presumably approves of X φ-ing, which means that obligatory acts are also good acts. However, the reverse is not true: God might approve of X φ-ing without disapproving of X not φ-ing, and so not all good acts are obligatory. Supererogatory acts would perhaps be those good, but non-obligatory, acts which involve great sacrifice on the part of the agent. Acts of which God neither approves nor disapproves would presumably be morally neutral.

It is necessary to say something about what I mean by God’s attitudes. I mean something stronger than a mere preference, although no doubt the attitudes will involve preferences. To say that God disapproves of something says more about his values than his preferences. Blackburn (1998: 67) says that someone’s values are ‘those concerns that he is also concerned to preserve: the ones by which he stands. He would contemplate losing them only with dismay’. This is not true of mere preferences; on a particular occasion, I may prefer raspberry ripple ice cream to neapolitan, but this needn’t be something that is important to me, and it needn’t bother me to think that on another occasion I may prefer neopolitan. Preferences can be arbitrary in a way that values are not. Sinclair (2014: 429) suggests that moral attitudes, but not preferences, ‘are sometimes very similar to (if not identical with) serious negotiating concerns’, and he explains what he means by a ‘serious negotiating concern’ as follows:

It is, first, a concern, that is, it is a motivationally-infused non-cognitive attitude that disposes an agent to favour certain courses of action over others.. .. Second a serious negotiating concern is serious. A serious concern is (i) reflectively endorsed, (ii) one about which the agent is resistant to change and (iii) one the satisfaction and preservation of which is considered by the agent to be very important (measuring importance psychologically, in terms of motivational strength and pervasiveness within the agent’s motivational profile).. .. Finally, a serious negotiating concern is a negotiating stance, that is, one which one hopes will be shared by others, and which is partly constituted by a disposition to engage with others in ways which (one hopes) will lead to them sharing it. (428)

This certainly seems to capture something of the distinction between moral attitudes and preferences. Like human attitudes, God’s attitudes of approval and disapproval should be understood as serious negotiating concerns. They are concerns about which God is resistant to change, the satisfaction and preservation of which is considered by God to be very important (in the sense discussed by Sinclair). Furthermore, they are partly constituted by a disposition on God’s part to engage with humans, in whatever way is appropriate for divine/human engagement, with the aim that we might come to share his concerns.⁴

Although this is surely part of what it means to have this sort of attitude, there is an additional feature, which is the disposition to feel hostility towards those who do not act in accordance with one’s concerns. Of course, having this disposition doesn’t mean that hostility will always be felt (Bjornsson & McPherson, 2014: 16). Perhaps if, as many theists believe, God is gracious, he will not actually always be hostile towards those who act in ways in which he disapproves, but this need not mean that his disapproval doesn’t involve the disposition to feel hostility towards people who act in these ways.

In addition, perhaps the type of attitude in question might also, at least sometimes, be accompanied by a disposition to feel hostility towards those who do not share the attitude. Blackburn pictures a ‘staircase of emotional ascent’, with simple preferences at the bottom, and our sentiments becoming more like moral attitudes as we ascend. Higher up, we may feel hostility towards those who do not share the attitude in question, and higher still, we may feel hostility towards those who do share the attitude but fail to feel hostility towards others who do not (1998: 9). Blackburn thinks that moral attitudes may be found at different places on the staircase; disapproving of torturing children for fun, for example, he thinks is towards the top, where dissent is beyond the pale, but some moral attitudes might be a bit lower down. Perhaps the same is true with God; perhaps there are some things where God’s disapproval is such that he is also disposed to disapprove of those who fail to share his disapproval, and perhaps there are other things where his disapproval does not include this feature.

Although in talking of God’s attitudes of approval and disapproval I have in mind something along these lines, I do not claim that this perfectly captures what the relevant attitude is. Contemporary expressivists have done a lot of work in developing proposals as to how we might understand the exact nature of the non-cognitive attitude in question (cf. Blackburn, 1998: 320; Gibbard 2003: 7; Sinclair 2009: 137), and possibly some of these proposals could be developed for my account too.⁵ The precise details are not all that important for my purposes here; what I want to emphasise is the interesting analogy between God’s attitudes and the sort of attitudes that contemporary expressivists have in mind when they talk about moral judgments, and how such an analogy enables the type of theistic account I have just outlined to share any advantage that expressivism might have in terms of explaining supervenience.

It is not difficult to see the analogy between expressivism and the account I have just outlined. For both expressivism and the above account, there is an important connection between certain types of attitudes and certain moral statements. According to expressivism, when I say ‘stealing is wrong’, this statement is an expression of my attitude towards stealing. According to the above account, when I say ‘stealing is wrong’, this statement is an assertion of a belief, but its truth depends on God’s attitude towards stealing. The analogy could be expressed in the following way: according to expressivism, the sincerity of the statement ‘φ is wrong’ depends on the speaker having an attitude of disapproval towards φ-ing, whereas according to the above account, the truth of the statement ‘φ is wrong’ depends on God having an attitude of disapproval towards φ-ing.

Because of the analogy with expressivism, I call the above account ‘Divine Expressivism’. As far as I am aware, Unwin (2008) is the only other philosopher to have noticed possible parallels between expressivism and theistic ethics. Unwin and I both hold that ‘what is fundamentally significant here is that both expressivism and [theistic ethics] tend to be concerned with similar types of attitudes’, where the difference ‘concerns just who has the attitudes in question’ (Unwin, 2008: 671).

However, it is also important to emphasise the disanalogy between expressivism and Divine Expressivism. Firstly, expressivism is a version of moral non-cognitivism, whereas Divine Expressivism is a form of moral cognitivism, specifically moral realism.⁶ According to expressivism there are no moral properties, whereas according to Divine Expressivism there are moral properties; however according to Divine Expressivism, there are no moral properties conceptually prior to God’s attitudes of approval and disapproval. Secondly, whereas Divine Expressivism relates moral properties to God’s attitudes, expressivism relates moral judgments to certain human attitudes.

Given that the theory I am proposing is a cognitivist theory, it might be thought that the use of the term ‘Divine Expressivism’ to describe it is misleading or inappropriate. Unwin, in discussing a view similar to the view I am describing here, says that such a view

does not merely resemble expressivism in a few curious and unexpected ways. Rather, it is a species of expressivism! Human affective and conative attitudes (or, at least, some of them) are characterized here as the internalizations of God’s own attitudes. .. ; but moral judgments are still deemed to be nothing over and above their ‘expressions’ in the sense understood by expressivists. (2008: 675, emphasis his).

Now, I would agree that many of our moral attitudes are (imperfect) internalizations of God’s own attitudes. In Biblical terms, God has written his laws on human hearts, even the hearts of those who don’t believe in him.⁷ How he does this, or even whether he does it primarily on us as individuals, as communities or as a species, is beyond my purposes for this paper. Nevertheless, if, as theists believe, we are created by God, then it seems plausible that we are created in such a way that our attitudes will reflect God’s own attitudes of approval and disapproval; this may indeed be part of what it means to be made in God’s image, as is believed by theists in the Judeo-Christian tradition.

But even if our moral attitudes are (imperfect) internalizations of God’s own attitudes, is Unwin right to think that our moral judgments are nothing over and above their ‘expressions’? It seems to me that this is wrong. Even if our moral judgments are expressions of our attitudes, they are more than this: they are also expressions of beliefs. There is nothing new in the idea that a moral judgement can be both an expression of a belief and an expression of an attitude—such a hybrid view has been defended by Copp (2001), among others, and is a view that I find quite convincing. It is a view which is open equally to theists and non-theists.

But perhaps Unwin does not mean that our judgments are expressions of our own attitudes, but expressions of God’s attitudes, which we have internalised. Clearly, this is a view which is open only to theists. It would still be a mistake to say that a moral judgement is nothing ‘over and above’ this, since it is clearly an expression of a belief too—for example, the belief that some action is wrong, not necessarily the belief that God disapproves of the action in question.⁸ Nevertheless, I think that it is plausible that the claim that some action is wrong (whether spoken by a theist or a non-theist) is both an expression of the speaker’s belief that the action in question is wrong and also (so long as the claim is true) an expression of God’s attitude of disapproval towards the action (as well as, typically, an expression of the speaker’s own attitude of disapproval towards the action).

Suppose, then, that the claim ‘stealing is wrong’ expresses God’s attitude of disapproval towards stealing; who is it that is expressing God’s attitude? Is it the speaker, or is it God expressing his attitude through the speaker? It seems to me that the answer should be both. Clearly, in making a moral judgment, it is the speaker who is stating that something has a particular moral status, and so it makes sense to say that it is the speaker who is expressing God’s attitude. But the speaker’s judgments need not just happen to coincide with God’s attitudes; rather God expresses his attitudes to us so that we internalise them. So I think it also makes sense to say that God is expressing his attitudes when we make correct moral judgments.

Presumably, due to our own fallibility, our attitudes are only an imperfect expression of God’s attitudes. We frequently make false moral judgments. In what sense, if any, can these false moral judgments be said to be expressions of God’s attitudes? Perhaps the simplest answer is to say that it is only true moral judgments which are expressions of God’s attitudes. But we might also want to say that, to the extent that false moral judgments at least aim at truth, they also aim at expressing God’s attitudes, some more successfully than others.

Our moral judgments, whether we believe in God or not, are therefore imperfect expressions of God’s attitudes, with our correct moral judgments being clearer or more perfect expressions of his attitudes. They are not assertions regarding God’s existence or expressions of any belief about God, but are nevertheless expressions of God’s attitudes. As such, I think the label ‘Divine Expressivism’ apt. But whether the label is apt or not, my main point is that, by highlighting the analogy between expressivism and Divine Expressivism, I can show that any advantage that expressivism has over (non-theistic) moral realism in terms of explaining the supervenience of the moral on the natural is shared by Divine Expressivism.

This is how Divine Expressivism can explain why it is a conceptual truth that the moral supervenes on the natural, without denying the non-entailment thesis: since God’s attitudes towards things are based on their natural properties⁹, it is conceptually impossible for him to have different attitudes towards two things which are identical in terms of their relevant natural properties, and so the supervenience thesis holds.¹⁰ Nevertheless, it need not be conceptually necessary that he has the attitude towards something which he does in fact have, in which case the non-entailment thesis also holds.¹¹

Given Blackburn’s argument that it is conceptually impossible for somebody to hold different attitudes towards two things which are identical in terms of their relevant natural features, it follows that God’s attitudes must be consistent in this sense. However, there may be a potential problem lurking for Divine Expressivism. Expressivism doesn’t rule out the possibility of us humans changing our minds about what we approve and disapprove of. I may come to disapprove of something which I previously approved of, and in doing so I may acknowledge that in some sense it is preferable to disapprove of it than to approve of it—perhaps I regret my previous attitude, and in some cases (depending on how far it is up the staircase of emotional ascent) I may even disapprove of myself for having previously approved of it. This doesn’t raise a problem for supervenience, since I presumably initially approved not just of the thing itself but also of everything else which was identical in terms of (relevant) natural features, and I presumably now disapprove of all these additional things too. However, in the case of God, if moral properties are related to God’s attitudes in the way in which I have proposed, then if God were to change his attitude towards a thing then its moral status would change, in which case supervenience would be undermined. In order to preserve supervenience, then, I need to be able to say that, unlike with humans, God’s attitudes cannot change over time. Can I say this?

Although many theists believe that God is immutable and therefore incapable of change at all, I do not want to commit myself one way or the other on this issue. Fortunately I don’t have to, since all that is needed to ensure that the moral supervenes on the natural is that God must be incapable of change in terms of his moral attitudes of approval and disapproval. It is not difficult to show why this must be the case. In cases where God’s attitude is quite high up the staircase of emotional ascent, presumably God could not begin to disapprove of something which he had previously approved of without disapproving of his former self, and likewise he could not begin to approve of something which he previously disapproved of without his former self disapproving of his latter self. Now, it is surely incoherent to suppose that God at time t₂ could disapprove of God at time t₁, since if this were possible then God at time t₂ would clearly not consider God at time t₁ to be that than which none greater can be conceived—which is true by definition. So, where God’s attitudes are towards the top of the staircase of emotional ascent, it is incoherent to suppose that these attitudes could change.

It may be that all God’s attitudes are such that he cannot begin to disapprove of what he formerly approved of without disapproving of his former self, in which case the above argument would apply to all God’s attitudes. But suppose God has some attitudes which, although moral attitudes, are nevertheless a bit lower on the staircase of emotional ascent, and such that God need not disapprove of somebody who failed to share these attitudes. Perhaps he wouldn’t find any fault with his former self for failing to share these attitudes. However, even here it seems incoherent to suppose that God could change his attitudes of approval and disapproval. Remember that the type of attitude in question is akin to a serious concern about which one is resistant to change and which one cannot contemplate losing without dismay. Now, it seems incoherent to me to suppose that God could give up an attitude under resistance—it would seem to suggest that he has been overpowered by somebody greater than himself, which is clearly inconsistent with the concept of God as that than which none greater can be conceived. Similarly, it seems incoherent to suppose that God could willingly relinquish a concern which he cannot contemplate losing without dismay. We sometimes change in our attitudes because we come to see things from a different perspective, one which perhaps we have not considered before—but presumably there is no perspective which God, being omniscient, has not considered.¹² I therefore conclude that it is conceptually impossible for God’s attitudes of approval and disapproval to undergo change.

In view of the above argument, I think that it is a conceptual truth that God’s attitudes of approval and disapproval are stable and constant. He cannot disapprove of what he once approved of, or vice versa. Hence Divine Expressivism can explain why it is a conceptual truth that there can be no change in a thing’s moral status unless there is some change in its natural properties. Explaining supervenience does not therefore appear to be a problem for Divine Expressivism. If Blackburn is correct in thinking it is a problem for non-theistic moral realism then it looks as if Divine Expressivism has an advantage over non-theistic moral realism in this respect.

This conclusion is ironic, since accounts which derive morality from God are frequently considered to be especially vulnerable to the problem of explaining how the moral supervenes on the non-moral. I have shown that this need not be the case.

4 An objection

There is, unfortunately, an objection to my argument. Even if it is true that the property of moral wrongness is identical to the property of being disapproved of by God, and even if it is a conceptual truth that God’s attitudes of approval and disapproval are consistent, it does not follow that it is a conceptual truth that the property of moral wrongness is identical to the property of being disapproved of by God. Property-identity claims, if true, are metaphysically necessary but need not be conceptually necessary. For example, it is metaphysically necessary, but not conceptually necessary, that water is identical to H₂O. Unless it is conceptually necessary that the property of wrongness is identical to the property of being disapproved of by God, it may seem that I have failed to show why it is conceptually necessary that the moral supervenes on the natural. I have failed to show why supervenience holds in those conceptually possible worlds where moral properties have nothing to do with God.

This is the type of objection which is raised by Mitchell (2017) against moral realism generally, but specifically targeting Shafer-Landau and Jackson. Whilst Shafer-Landau denies that moral properties are identical to natural properties, he does think that the moral is ‘entirely and exhaustively constituted by the descriptive’, which enables him to argue that ‘a moral fact supervenes on a particular concatenation of descriptive facts just because these facts realize the moral property in question’ (Shafer-Landau, 2003: 76−7).

But, objects Mitchell (2017: 2915, emphasis in original),

the constitution thesis itself does not seem to be a conceptual truth, for it is apparently open to, say, Moorean non-naturalists and expressivists to deny that the moral is constituted by the natural without misusing moral concepts. But then it is hard to see how an appeal to constitution can help with supervenience as a conceptual truth. It certainly cannot give us a further explanation: even if constitution explains supervenience itself, it won’t even begin to explain why supervenience is a conceptual truth.

Jackson, on the other hand, thinks that there is an analytical equivalence between moral claims and natural claims, arguing that there is no ethical claim ‘that cannot in principle be made in purely descriptive vocabulary’ so that ‘for every ethical sentence, there is some equivalent purely descriptive sentence’. Ethical language may be indispensable in practice due to the complexity of the infinite disjunction of descriptive sentences involved, but, he insists, ‘we could in principle say it all in descriptive language’. Nevertheless, Jackson acknowledges that it is only ‘an implicit part. . . of our understanding of ethical terms and sentences that they serve to mark distinctions among the descriptive ways things are’ (Jackson, 1998: 123–5).

Mitchell’s objection to Jackson is twofold; firstly, he thinks that even if Jackson’s theory does solve the supervenience problem it is only able to do so at the huge cost of insisting that ‘all those who deny his account, e.g. Moorean non-naturalists, are not just wrong, but guilty of (albeit unobvious) conceptual incompetence’ (2017: 2917, emphasis his). But, more importantly, he argues that there is a misfit between Jackson’s claims relating our merely implicit understandings and the explicit, obvious, supervenience claims:

Supervenience is not a merely implicit part of our understanding of ethical vocabulary. It is something quite obvious, quite explicit: flouting the constraint is not seen as a respectable move within a murky philosophical controversy, but a flagrant show of confusion. So even if Jackson is right that our implicit understanding of ethical vocabulary gives it the purpose of marking descriptive distinctions, this would only explain the supervenience constraint inasmuch as it is likewise implicit, leaving its further status as something obvious and explicit as mysterious as ever. (ibid, emphasis his)

How does Mitchell’s point relate to my account? It seems I am faced with a dilemma. On the first horn, I might argue that it is merely a metaphysical, but not conceptual, truth that the property of wrongness is identical to the property of being disapproved of by God, in which case Mitchell’s argument against Shafer-Landau appears to apply to me too. On the second horn, I might argue that it is a conceptual truth, but an unobvious and implicit one, that the property of wrongness is identical to the property of being disapproved of by God, in which case it looks as if I must say, implausibly, that those who deny my account, such as all non-theistic moral realists, are guilty of (albeit unobvious) conceptually incompetence, and furthermore I would have failed to show why it is an explicit and obvious conceptual necessity that the moral supervenes on the natural.¹³

Although Adams (1973), in developing his modified Divine Command Theory, originally proposed that ‘morally wrong’ in the mouths of theists (but not non-theists) means ‘forbidden by God’ he subsequently abandoned this idea, claiming instead that the property of wrongness is identical to the property of being forbidden by God even though there is no conceptual claim being made here (1979; 1999). Adams thinks that if it were to turn out that God does not exist then the property of wrongness would turn out to be identical to some other property after all—just as, if it were to turn out that hydrogen and oxygen are unable to combine in the way needed to form H₂O, water would turn out to be some other compound after all.¹⁴ Things are different when it comes to analytic or conceptual claims; it is conceptually true that a bachelor is an unmarried man, and this means that, should it turn out that all men are married, rather than concluding that the term ‘bachelor’ refers to some other property, we would conclude instead that there are no bachelors.

I differ from Adams in an important respect. I am more inclined than he is to say that, were it to turn out (in this world) that God does not exist, then it would turn out that moral properties do not exist, in which case moral error theory would be true. This is because moral properties, if they exist, make demands on us, and I find it difficult to see how a mindless universe can make this sort of demand on us. If moral properties exist, then it looks as if there is an objectively correct way for us to live; we are somehow meant to live our lives one way rather than another, and this seems ‘queer’ in a world without God. Elsewhere (Taylor, 2020), I have argued that this is how we should understand Mackie’s queerness argument, which explains why Mackie himself thought that the existence of God might resolve the queerness problem (cf. Mackie, 1977: 48, 230–1; 1982: 115–7). I therefore think that, unlike Adams, I would be tempted towards moral error theory if I were to become convinced of the non-existence of God.

Nevertheless, however implausible it may seem to me that moral properties could exist in the absence of God, this certainly does not prove that it is a conceptual truth that they would not exist unless God exists, nor that it is a conceptual truth that the property of wrongness is identical to the property of being disapproved of by God. Is there anything else I can say then that might help to make a conceptual connection between moral properties and God?

I think there is something I can say, but first I think it worth pointing out that, if this is a problem for my view, then it is not clear that expressivism fares any better. If expressivism is true, then it is either true as a matter of conceptual necessity or else it is not. Suppose first of all that it is not; suppose that there are conceptually possible worlds where moral realism is true, worlds where robust, non-minimalist moral properties of the type believed in by moral realists exist. Then expressivists face the same challenge that realists face, the challenge of explaining why some of these conceptually possible worlds cannot be morally mixed worlds. Blackburn’s supervenience argument appears to confront not just those who believe that moral properties actually exist, but also those who believe that the existence of moral properties is conceptually possible. In order to explain why there are no conceptually possible mixed worlds, it looks as if expressivism has to insist not just that there are no moral properties in this world, but that there are no conceptually possible worlds containing moral properties (in the robust non-minimalist sense understood by moral realists).

Perhaps expressivists would be happy to go down this route. Since expressivism has traditionally been understood as a thesis in semantics, presumably it must be true as a matter of conceptual necessity if it is true at all.¹⁵ But then we are faced again with the apparently implausible consequence that most people must be conceptually deficient in their use of moral terms—it may be that even to wonder whether moral properties exist ‘is to betray conceptual confusion’ (Joyce, 2016: Section 1). If Jackson’s view is implausible because it has the consequence that Moorean non-naturalists are guilty of (albeit unobvious) conceptual incompetence, then expressivism, if understood as a conceptual truth, must be just as implausible since Moorean non-naturalists would turn out to be just as guilty. Furthermore, expressivism would face the other problem which Mitchell poses for Jackson’s view: if expressivism is a conceptual truth then it would seem to be an unobvious one. But then the question arises as to why supervenience should be an obvious conceptual truth.

Expressivists typically appeal to the function of moral judgments to support expressivism. The distinctive function of moral judgments, according to Sinclair (2009: 136) is ‘the expression of affective attitudes for the purpose of mutual co-ordination of action’. It is not clear exactly how the function of moral judgments should be determined, but one option would be to adopt a teleosemantic account of functions, according to which ‘the function of a type is that effect for which it has been selected’, which is ‘what tokens of that type have done in the past that accounts for the continued survival and proliferation of tokens of that type’ (Sinclair, 2012: 642). Sinclair suggests that the evolutionary function of moral judgments is interpersonal coordination. This is because it is frequently thought by those who believe that morality evolved that it did so because it enabled interpersonal coordination, and that those individuals or groups which were better at interpersonal coordination were better equipped to leave descendents. Moral judgments, then, are ‘products of a mechanism that allows groups of interacting individuals to co-ordinate their actions and emotions for mutual benefit. The function of the moral habit is therefore to produce mutually beneficial co-operative patterns of action and emotion’ (ibid: 649). Although Sinclair does not think that the teleosemantic account he discusses provides conclusive support for expressivism, he suggests it ‘should make us lean in its favour’ (661).

However, even if it is true, as some believe, that the evolutionary success of our moral concepts had nothing to do with tracking truth (cf. Street, 2006), and even if it is true that the purpose of morality is ‘broadly, to augment social cohesion’ (Joyce, 2014:119), it is surely important to consider how this is achieved; Joyce (ibid.: 119−20) thinks it does this ‘by providing people with beliefs concerning actions having moral qualities’. He is not denying that these beliefs arose as a result of our emotions or attitudes (or rather, those of our ancestors) being projected onto the world; his point is simply that for moral judgments to have an evolutionary advantage, it is necessary, or at least beneficial, that our ancestors actually believed (albeit falsely, according to Joyce) that these moral judgments were objectively true, that they tracked externally binding rules (ibid.: 119−20). I am inclined to believe then that even if we accept the teleosemantic claims, we should favour cognitivism over non-cognitivism. Hence, thinking about why or how moral concepts evolved does not seem to show that expressivism is a conceptual truth, or that it is a conceptual truth that robust moral properties do not exist.

In fact, a case might even be made for saying that moral judgments on the one hand and judgments about what God commands or approves of on the other hand could have the same function as one another. Suppose that our concept of God evolved; it is not difficult to see how the judgment that God approved of those things which promoted social cohesion might have had exactly the same evolutionary advantage as the judgment that those things were morally right or good. Assuming there is an evolutionary advantage in behaving in ways that promote social cohesion, a community which is convinced that acting in such ways is morally right may well be more likely to act in such ways, but so surely is a community which is convinced that God approves of such actions, perhaps especially if this belief is accompanied by the belief that he will reward such actions or punish failure to act in these ways (Joyce, 2001: 729; De Waal 2013: 18). All this is consistent, of course, with both moral judgments and judgments about God’s attitudes of approval and disapproval being in fact a projection of our own (or our ancestors’) attitudes and emotions—but it is also consistent with these attitudes and emotions themselves being an expression of God’s attitudes and emotions.

Indeed, it has been suggested by Anscombe (1958) that at least some moral concepts (specifically obligation) arose in the context of theism and make little sense outside of this context. Nevertheless, even though the concept makes little sense outside the context of theism, we have, to a large extent, abandoned theism but kept the moral concepts. She argues that our concepts of moral obligation require a concept of divine law, but that once the concept of divine law is abandoned, ‘it is a natural result that the concepts of “obligation,” of being bound or required as by a law, should remain though they had lost their root’ (ibid: 6).

My reading of Anscombe is that she thinks it is at least close to being a conceptual truth that moral obligation requires the existence of God, which is why she thinks the concept of obligation, although persisting after theism has been abandoned, now makes little or no sense and should have been abandoned along with theism. The idea that it is at least close to being a conceptual truth that moral obligation requires God is indicated by her analogy with criminality; she suggests that the retention of the concepts of obligation once the concept of divine law was lost would be similar to the retention of the notion of ‘criminal’ after abandoning criminal law and criminal courts (ibid.). What she seems to be saying is that, just as being a criminal conceptually entails the existence of a criminal law and criminal courts, so being obligatory conceptually entails the existence of God and divine law. Retaining the concept of ‘criminal’ without the existence of criminal courts and criminal law and retaining the concept of ‘moral obligation’ without the existence of God and divine law would both result in ‘the interesting [situation] of the survival of a concept outside the framework or thought that made it a really intelligible one’ (ibid.).

Here, then, is the required conceptual connection between moral obligation and God. If it is true that the concept of obligation arose in the context of theism in something like the way Anscombe believed, then we have an explanation as to why it is a conceptual truth that the moral supervenes on the natural. The concept of obligation includes within it the concept of supervenience because of the context in which the concept of obligation arose, together with the fact that it is conceptually impossible for God’s attitudes of approval and disapproval to change, for the reasons argued for in the previous section.

The explanation as to why it is a conceptual truth that the moral supervenes on the natural, according to this line of thought, would be historical: the concept of supervenience was initially included in the concept of moral obligation because of the links with theism. When theism was abandoned, the concept of moral obligation survived, including within it the concept of moral supervenience.

Even a non-theistic moral realist could in principle accept this historical explanation as to why it is that we consider it a conceptual truth that the moral supervenes on the natural. However, in denying the existence of God, such a person would be left unable to explain why it should be that the moral in fact supervenes on the natural. This need not mean that such a person is conceptually confused; they may simply be left with a mystery.

Although I think that there is some merit in the argument just outlined, there are problems with it. Perhaps the most serious problem is that it assumes that Anscombe is correct to think that the concept of obligation arose in the context of theism. Anscombe mentions Jews, Christians and Stoics as people whom she believes had a concept of obligation (which she believes is lacking in Aristotle), and whom she believes held a divine law concept of obligation; but an examination of many different ancient cultures would be necessary in order to substantiate her claims. De Waal argues that morality is much older than religion, although it is not clear that he thinks that morality as understood by moral realists predates religion, since he acknowledges that ‘the view of morality as a set of immutable principles, or laws, that are ours to discover ultimately comes from religion’ (2013: 23)—a view which seems consistent with Anscombe’s view.¹⁶ A second problem is that Anscombe is talking mainly about moral obligation, and moral supervenience does not apply just to obligation. However, her comment (1958: 2) that the way we talk about moral goodness is also lacking in Aristotle suggests that she may have more than moral obligation in mind.

In spite of these problems, I think there may well have initially been some sort of conceptual connection between morality (or at least moral obligation) and God, and I think this would help to explain why supervenience is included in the concept of morality (or at least obligation). It is true that many people who appear to be competent in their use of moral concepts would deny the sort of theory I have proposed, and because of this I am reluctant to insist that it is a conceptual truth that the property of moral wrongness is identical to the property of being disapproved of by God; but it is also true that many people who appear to be competent in their use of moral concepts would deny expressivism, and so there is just as much reason to doubt that expressivism is a conceptual truth.¹⁷

If a metaethical theory can only explain why it is a conceptual truth that the moral supervenes on the natural if that metaethical theory is itself a conceptual truth, then it may be that neither expressivism nor Divine Expressivism can explain why moral supervenience is a conceptual truth. But this is no reason to reject Premise 2 of the argument which I outlined at the beginning of this paper; if the conclusion of that argument is rejected, it should be because of problems with Premise 1, not because of problems with Premise 2.

5 Conclusion

I have argued in this paper that any advantage that expressivism may have over (non-theistic) moral realism in terms of being able to explain why it is a conceptual truth that the moral supervenes on the natural is shared by a theistic version of moral realism which I have called Divine Expressivism. This doesn’t necessarily mean that non-theistic moral realism is incoherent, and any advantage may be outweighed by other advantages enjoyed by (non-theistic) moral realism. Many moral realists will no doubt think that introducing the concept of God is a greater cost than being unable to explain supervenience. Nevertheless, the possibility of an analogy between expressivism and certain forms of theistic ethics is of interest in itself, and could benefit from more attention in the future, whether or not one believes that there is a God.