In Defense of the Incompatibility of Hume’s Is/Ought Dichotomy and Theistic Metaethical Realism: A Response to Richard Bushey

Hume’s is/ought dichotomy is, roughly, the thesis that no statement about what one ought to do can be entailed by a statement about what is the case. In other words, no moral facts can be entailed by non-moral facts. Hume used his dichotomy to argue for ethical non-cognitivism, but contemporary metaethicists often employ the dichotomy to argue that ethical facts constitute a domain of facts independent from any other collection of facts (e.g. robust metaethical non-naturalists argue that there are objective moral truths which are independent of any of the natural facts [1]). While the is/ought dichotomy is controversial, there remain a number of significant philosophers who view the dichotomy as eminently plausible.

In a recent blog post, I responded to the argument that morality so obviously comes from the Christian God that, insofar as atheists believe in objective moral facts, they must be stealing from Christianity (the obviousness thesis). I argued that Hume’s is/ought dichotomy is incompatible with theistic metaethical realism, so that, if the is/ought dichotomy holds, then morality cannot originate with God. This is because, on the is/ought dichotomy, no moral fact could be reduced to any non-moral fact about God. Because the is/ought dichotomy is controversial, the obviousness thesis must be false. Note that this does not entail that theistic metaethical realism is false; perhaps the is/ought dichotomy is false and morality does originate with God, but that morality originates with God is not so obvious that atheists must be stealing from Christians. Or perhaps the is/ought dichotomy is true and whatever moral facts there are are independent of God (though God might still play some important role; for example, perhaps God, though not ultimately responsible for the existence of ethical facts, is responsible for creating creatures who can appreciate ethical facts).

In response, Richard Bushey has argued that theistic metaethical realism is not incompatible with the is/ought dichotomy. (Although Richard does not state in his article that he was responding to me, he communicated to me that he had my post in mind when he wrote his.) Here, I will respond by showing that the arguments Richard presents are unsound, though they raise a number of interesting (and challenging) philosophical issues.


First, I want to clarify my argument from my previous post, since I am not positive Richard understood it. He describes my argument this way:

So the atheist will argue that the theist is drawing an ought from an is in their rendering of the moral argument. For we propose that morality is within God’s nature. God is the Good. As such, we saying [sic] that being God is this way, that therefore, we should behave in some way. So then, is the moral argument guilty of the ‘is/ought’ fallacy?

I don’t think this captures what I was saying. Richard characterizes my argument as pointing out that theists cannot derive moral facts from the fact that ‘God is the Good’. However, the fact that ‘God is the Good’ is a moral fact; as such, deriving moral facts from ‘God is the Good’ would not be a violation of the is/ought dichotomy. Instead, my claim was that if moral facts cannot be derived from non-moral facts then moral facts cannot be derived from non-moral facts about God. (Incidentally, I don’t know where Richard surmises this business about humans behaving the same way as God. Trivially, it’s impossible for humans to behave the same way as God, not just because God is infinitely more powerful than us but also because God, if She exists, would appreciate many more moral facts than we do. As such, God would possess reasons for Her actions that are cognitively closed off for us.)

The other misconception which Richard seems to hold is that the atheist must be the one to argue that the is/ought dichotomy and theistic metaethics are incompatible. Theists can argue that the is/ought dichotomy is false, but, if it had been true, would have been incompatible with theistic metaethical realism. (Maximus Confesses, who is a theist that blogs on this network, has told me that this describes his position.) Or theists can argue that the is/ought dichotomy is true, incompatible with theistic metaethical realism, and that ethical facts are independent of God facts. Although this would require revising the notion of God’s aseity in ways that most theists would find unpalatable, there are theologians who would endorse it.

Now that I’ve proceeded through several clarifications, I move on to each of Richard’s arguments. First, Richard describes the moral argument as it has appeared in the contemporary philosophical literature [2]:

  1. If God does not exist, then objective moral values and duties do not exist.
  2. Objective moral values and duties do exist.
  3. Therefore, God exists (from 1 & 2 by modus tollens).

Richard points out — correctly — that my argument does not reject either premise 1 or 2. Instead, my argument is designed to show that the is/ought dichotomy, if true, entails that moral facts cannot be grounded in or reduced to non-moral facts about God. Still, my argument creates a problem for the moral argument that is worth our time to explore. As it turns out, the problem with the moral argument, as Richard has described it, has much to do with some general problems in classical logic concerning the semantics of conditional statements. Consider the following argument which parallels the moral argument in structure:

1’. If there are no leprechauns, then the sky is never blue.
2’. It is false that the sky is never blue.
3’. Therefore, there are leprechauns.

Notice a few things about 1’-3’. Between leprechaun believers and disbelievers, premise 2’ is non-controversially true. Leprechaun disbelievers have strong reasons to reject 1’; after all, they think that there are no leprechauns, but the sky is sometimes blue. However, leprechaun disbelievers cannot simply assert that 1’ is false without begging the question against the leprechaun believers. And the leprechaun believers have fairly good reasons for thinking that 1’ is true, because if the antecedent of any conditional statement is false then the statement is true.

However, we can imagine a leprechaun believer who is uncomfortable with 1’. While the leprechaun believer will evaluate 1’ as true if they are using classical logic, they will rightly ask what the existence or non-existence of leprechauns has to do with the color of the sky. They may even be introduced to an argument for the independence of facts concerning sky and leprechauns. Somehow, 1’-3’ is a sound argument, on their worldview, even though they may have reason to believe the antecedent and the consequent of 1’ have nothing to do with each other. That’s an odd situation.

However, this odd situation may be resolved by realizing that the truth-semantics of conditional statements in classical logic was a choice made by Gottlob Frege. Contemporary logicians have questioned Frege’s semantics and developed alternative relevance logics which demand that for a conditional to evaluate to true, the antecedent and the consequent of the conditional must be appropriately related [3]. Returning to the leprechaun argument, both leprechaun believers and disbelievers would evaluate 1’ as false because the existence (or non-existence) of leprechauns and the color of the sky are not appropriately related. Thus, 1’-3’ would constitute an unsound argument.

Let’s turn back to the moral argument. Suppose that the is/ought dichotomy is both true and incompatible with theistic metaethics. There are two possibilities. First possibility: while moral facts are not grounded in or reducible to non-moral facts about God, moral facts can exist only if God exists. Second possibility: moral facts are not grounded in or reducible to facts about God and can exist regardless of whether or not God exists. The first of these two possibilities is compatible with 1-3 while the second is not. But surely the first possibility is not what theists mean when they use the moral argument, given that, on the first possibility, God could play no explanatory role for moral facts. Yet, strangely, this is the position Richard takes:

As such, even if we were to grant this objection [e.g. that the is/ought dichotomy is incompatible with theistic metaethical realism], the [moral] argument would still remain undefeated. […] It seems that even granting the objection, the atheist would still be left unable to ground morality within their worldview. If anyone is to posit any form of moral realism, this argument restricts them to a theistic model of it.

Somehow, despite playing no explanatory role for morality, moral realism could only be true if theism were true. That seems altogether implausible. It seems far more plausible that if theism plays no explanatory role for morality, the truth of theism would be independent of the truth of moral realism. Therefore, I suggest that, if Richard concedes the is-ought dichotomy is both true and incompatible with theistic metaethical realism, he would be in the position of the leprechaun believer; in that case, he should employ relevance logic and deny 1 because theism and moral realism would not be relevantly related.

Nonetheless, Richard does think theism explains moral realism and spends the last two sections of his article arguing for how this might be so. In his first line of defence, Richard argues that I have misunderstood the distinction between moral value and moral duty. The former is grounded in God’s nature while the latter is grounded in God’s commands. Because these are distinct, no fact about what we ought to do (i.e. what our duties are) is derivable from facts about what God is (i.e. God’s nature). It is worth recalling why theists drew the distinction between God’s commands and nature in the first place. One might have thought that one could ground or reduce all moral facts to God’s commands (Naive Divine Command Theory, herein: NDCT). This faces the Euthyphro Dilemma: does God command us to do x because x is good or is x good because God commands us to x? The latter is compatible with NDCT, but entails that there are no further moral reasons for God’s commands. If there were no further reasons, then God’s commands can be completely arbitrary. There would be nothing to bar God from ordering us to slaughter innocent babies; and, if NDCT is true, God’s issuing such a command would make it morally obligatory for us to slaughter innocent babies. If there were further moral reasons for God’s commands, then those would be the real reasons that x is good and not God’s commands. The latter — that x is good because God commands us to x — is more plausible, but incompatible with NDCT.

In response, theists, such as William Lane Craig [4] and William Alston [5], argue that the Euthyphro Dilemma is a false dichotomy. On their view, God’s nature provides an ultimate standard of goodness and, because God’s commands are explained by God’s nature, God’s commands cannot be arbitrary. Richard has argued that God’s commands are independent of God’s nature, so that the facts concerning God’s commands cannot be reduced to facts concerning God’s nature. I don’t understand why he would endorse this reply, given that it renders God’s commands arbitrary in precisely the way Craig and Alston were trying to avoid.

Perhaps the fact that Richard took up this line of reasoning can be explained by the fact that he misreads me as stating that moral facts — which he may understand as identical to God’s commands — cannot be reduced to what God is (i.e. God’s nature). Nonetheless, there is a further problem for Richard’s view: there are moral facts not just about creatures and our duties, but about God’s essential goodness. Assuming the is/ought dichotomy, God’s essential goodness cannot be explained by any further non-moral facts about God, but it is difficult to understand how essential goodness might be explained by any further moral facts. The theist may claim that God’s essential goodness is simply identical to God, that God’s necessary existence needs no explanation, and, therefore, God’s essential goodness needs no further explanation. Yet this entails that goodness needs no explanation. Given that goodness would require no further explanation, it becomes difficult to see what is so objectionable about views on which morality is independent of God and requires no explanation for its existence (such as robust metaethical non-naturalism [6]).

Richard worries that robust metaethical non-naturalism posits no ultimate authority from which moral commands are issued. Perhaps moral obligation can only be derived from the commands of a relevant authority; thus, without God, there would be no one of suitable authority to issue commands and we would have no moral duties. Granting this objection, it may still be the case that moral properties ultimately originate independently of God, for perhaps God can be said to possess a perfectly good nature because God maximally participates in the Form of the Good (in the Platonic sense). But, if so, moral properties need not be grounded in God. It is worth noting that, although I will not explore them at length here, there are multiple secular metaethical theories that include an account of moral obligation.

Richard’s final argument — that morality must originate with a worthy authority — concedes the entire game. He writes:

[W]hen God offers a moral command to us, we know that it has become our duty, because God is a worthy authority. This is the sort of determination that we make all of the time. We recognize and determine the merit of any given authority figure and determine if we are going to submit to them. […] The only option for the objector is to say that we are deriving what we ought to do from the reality of our moral commands. But this seems to be something that we all recognize as perfectly legitimate, since we know the difference between pulling over for a police officer and pulling over for a complete stranger.

I agree with Richard that when we deem an authority figure to be legitimate, it is rational for us to obey their commands. I also agree that there is a difference between the commands of a police officer and those of a complete stranger, who may hold no authority. However, this is because we recognize the truth of a principle like the following:

A. We should follow the commands of worthy authorities.

Where does A come from on Richard’s view? A cannot originate in God because that would misconstrue the order of explanation. What we wanted to explain was why we should follow God’s commands. A explains why we should follow God’s commands. It would be backwards (or circular) to explain A in terms of God’s commands. So, A cannot originate in God’s nature or essence either, since, on the view Richard endorses, our obligations originate in God’s commands and A is an obligation. In other words, Richard seems to have conceded that the is/ought dichotomy is both true and incompatible with theistic metaethical realism, given that his view requires an additional moral obligation that must be independent of God.

So, in sum, if the is/ought dichotomy is true, then theistic metaethical realism is not.


[1] For a defense of robust meta-ethical non-naturalism, see, for example, Enoch, D. (2011) Taking Morality Seriously. Oxford: Oxford University Press.

[2] See, for example, Craig, W. (2007) “Theistic Critiques of Atheism”. In Michael Martin (ed), The Cambridge Companion to Atheism. Cambridge, Cambridge University Press, 82.

[3] Mares, E. (2014) “Relevance Logic”, In Edward Zalta (ed), The Stanford Encyclopedia of Philosophy. URL = <>.

[4] Craig, W. (1997). “The Indispensability of Theologicaly Meta-ethical Foundations For Morality”. Foundations 5: 9-12.

[5] Alston, W. (2002). “What Euthyphro Should Have Said”. In William Lane Craig (ed), Philosophy of Religion: A Reader and Guide. Edinburgh: Edinburgh University Press, 283-289.

[6] This argument originates in Wielenberg, E. (2009). “In Defense of Non-Natural, Non-Theistic Moral Realism”. Faith and Philosophy, 26 (1), 23-41.