There is a standard objection to the divine command theory (DCT) that runs as follows:

1. Either God’s commands are arbitrary or they are grounded in reasons.
2. Arbitrary commands cannot ground moral obligations.
3. If God’s commands are grounded in reasons, then it is those reasons, rather than God’s commands, that ground moral obligations.
4. Either way, God’s commands are superfluous; they do not ground moral obligations.

I’ll call this the Arbitrariness Argument (AA).  You can find versions of this argument in Walter Sinnot-Armstrong’s book Morality Without God?  and in his article “Why Traditional Theism Cannot Provide an Adequate Foundation for Morality.” Here is a quote from the latter:

Let’s assume that God commanded us not to rape. Did God have any reason to command this? If not, his command was arbitrary, and then it can’t make anything morally wrong. On the other hand, if God did have a reason to command us not to rape, then that reason is what makes rape morally wrong. The command itself is superfluous. Either way, morality cannot depend on God’s commands. [1]

Matthew Flannagan has recently criticized this argument [2]. Flannagan is a proponent of a version of the divine command theory; on his view God’s commands constitute moral obligations. To be more specific, Flannagan, following Robert Adams, maintains that moral obligations are constituted by the commands of a perfectly loving God. This qualification (that the commands be commands of a perfectly loving deity) is necessary to deal with an objection, stemming from the Euthyphro dilemma, concerning the possibility that God might command something horrendous (such as the torture of children). The objection runs as follows: an omnipotent being can command anything whatsoever. So, DCT (at least an unqualified version) entails that, since it is possible that God commands child torture, it is possible that child torture is morally obligatory. However, Flannagan’s version of DCT does not entail this since a perfectly loving being does not command horrible actions. God would not command child torture and so Flannagan’s view does not have the unfortunate consequence that child torture might be morally obligatory.

Now, on such a view such as Flannagan’s, in the absence of divine commands, there are no moral obligations. Thus Flannagan must maintain that, contrary to the AA, God’s reasons for his commands cannot, on their own, constitute moral obligations. But this appears to be problematic because, first, it is not too difficult to imagine what might motivate a perfectly loving deity to issue commands. God commands that we not rape, presumably, because rape causes severe, undue emotional and physical harm. Second, and more to the point, if this reason is enough to motivate a loving God to command that we not rape, surely the reason by itself is sufficient to make rape wrong. It thus appears that, once we admit (a) that God must have reasons for his commands (else, morality would be arbitrary) and (b) that God is perfectly loving and hence the reasons for his commands must involve concern for human beings (or sentient creature more generally), we should conclude that the reasons would be sufficient, on their own (that is, even in the absence of God’s commands), to constitute moral obligations.

So, Flannagan needs to show that a divine command theorist can maintain that God’s commands (and not merely his reasons) constitute our moral obligations, that Gods’ commands are grounded in reasons, and that those reasons are themselves grounded in love. Though I think that his argument is insightful I don’t think that it is successful, as I will now proceed to show.

Flannagan helpfully formalizes Sinnot-Armstrong’s  argument as follows:

(1) Either: (i) there is a reason, r, why God prohibits rape; or, (ii) there is no reason, r, why God prohibits rape.

(2) If there is no reason, r, why God prohibits rape, then God’s commands are arbitrary.

(3) If there is a reason, r, why God prohibits rape then, r, is what makes rape morally wrong.

(4) If r is what makes rape morally wrong then God’s commands are superfluous. [3]

Flannagan correctly points out that this argument is problematic since ‘makes’ is ambiguous. It can refer both to constitutive explanations and to motivational explanations. The divine command theorist can easily accept that God has motivational reasons to command that we not rape. In a sense, then, this motivational reason, whatever it is, makes it the case that rape is morally wrong since it explains why it is that God commanded that we not rape. But the divine command theorist need not be forced to conclude that God’s reason constitutes the wrongness of rape. That is, it is perfectly consistent to claim that God has reasons for his commands and that these reasons motivationally explain why, e.g. rape is wrong, and to also insist that these reasons do not constitute the wrongness of rape.

To be successful, AA must acknowledge this very real and important distinction. Flannagan therefore provides two updated reconstructions of Sinnot-Armstrong’s argument in which the distinction is explicitly acknowledged. Version 1:

(1) Either: (i) there is a reason, r, why God prohibits rape; or, (ii) there is no reason, r, why God prohibits rape.

(2) If there is no reason, r, why God prohibits rape, then God’s commands are arbitrary.

(3′) If there is a reason, r, why God prohibits rape then r motivationally explains why rape is wrong.

(4′) If r motivationally explains the wrongness of rape then God’s commands are superfluous. [4]

(3′), says Flannagan, is true. God’s reasons for prohibiting rape explain why God makes rape wrong; and so, in that sense, God’s reasons explain the wrongness of rape. The problem with this version is that (4′) is false. That a reason might motivationally explain something does not entail that it constitutes that thing. (more on this below). Flannagan’s view is that God’s commands constitute moral obligations. The fact that God’s reasons motivationally explain his commands does not entail that God’s commands are superfluous if, as Flannagan maintains, those commands constitute moral obligations. This version of the argument, then, is unsuccessful. We must instead look at a different version, one that explicitly deals with constitutive explanation. Version 2:

(1) Either: (i) there is a reason, r, why God prohibits rape; or, (ii) there is no reason, r, why God prohibits rape.

(2) If there is no reason, r, why God prohibits rape, then God’s commands are arbitrary.

(3’’) If there is a reason, r, why God prohibits rape then r is what constitutes the wrongness of rape.

(4’’) If r is what constitutes the wrongness of rape then God’s commands are explanatorily superfluous. [5]

I think that this version accurately captures the thought behind Sinnot-Armstrong’s argument. And I am inclined to think that it is a pretty good argument. However, Flannagan says that it does not work. The problem, he says, is that “(3”) is a non- sequitur. Armstrong contends if DCT is true, and Gods prohibitions constitute moral wrongness, then any reason God has for prohibiting rape must constitute the moral wrongness of rape.”[6]. Flannagan claims that this contention is wrong since it depends upon the following principle:

PI: If A is constituted by B, and someone has reasons r for bringing about B, then A is constituted by r.

But PI is false as was shown by Stephen Sullivan [7] who provides the following counter-example:  Consider a bachelor, Giorgio, who chooses to remain unmarried because he prefers to live alone. This reason (that he prefers to live alone) provides a motivational explanation for Giorgio’s choosing to remain unmarried, but it does not constitute his bachelorhood. Giorgio has a reason for being unmarried, his being unmarried constitutes his being a bachelor, but his reason for being unmarried does not constitute his being a bachelor. Thus, PI is false and Sinnot-Armstrong’s argument, since it relies on it, is unsound.

Is Flannagan correct that the arbitrariness argument relies on PI? I don’t see that it does. PI is extremely implausible on its face and does not capture the insight involved in the argument. In any event, the argument does not need it. A defender of AA need not be committed to PI but to something much more narrow in scope. A defender of AA can accept that PI is not generally true but assert that it is true of phenomena that involve reasons in the way that moral obligations do.

Morality involves reasons in two different respects; if I am morally obligated to do something, then I have a reason (or reasons) to do it. But, more than this, if I am obligated to do something, then there are reasons that I am under that obligation. For example, a parent is morally obligated to care for her children. But there are also reasons for this obligation. A parent is so-obligated because she is partly responsible for bringing the child, who is helpless on his own, into existence. A stranger living 500 miles away from the child has no reason to provide for that child and, again, there are reasons for this lack of obligation (namely, that the stranger does not know the child (or even that the child exists) and so does not have a special connection to the child and can do very little in any event).

The preceding is merely an attempt to sketch an example which illustrates the dual way in which morality is connected to rationality. Nothing depends on whether the analysis of the parent-child moral nexus that I provided is accurate. The point is only that moral obligations are reasons and that there are reasons for the presence (or absence) of moral obligations.

Given this, it is more plausible to read the arbitrariness argument as depending on a principle that is more specific than PI, one that concerns reasons in particular:

PR: If A is constituted by B, A is grounded in reasons and is itself a reason, and someone has reasons r for bringing about B, then A is constituted by r.

PR is a much more relevant to AA since moral obligations are reasons and are grounded in reasons. Nor is PR falsified by the bachelor example. Being a bachelor is not a reason, nor is it necessarily grounded in reasons. (I mean that many bachelors just happen to be bachelors, not out of choice; not for any particular reason.) So, on the plausible assumption that obligations are both reasons and are grounded in reasons, there is no need for AA to rely on PI. Thus, in showing that PI is false, Flannagan has not defeated AA. None of this proves that PR is true and I will not attempt to do so here. Nonetheless, Flannagan’s criticism of AA does not go through since it involves criticizing a principle that a defender of AA need not be committed to.

If you are interested, you can read Flannagan’s response to Richard Carrier’s criticism of his arguments here.

---

[1] Sinnot-Armstrong, W., “Why Traditional Theism Cannot Provide an Adequate Foundation for Morality” in R. Garcia & N. Kind eds. Is Goodness Without God Good Enough? A Debate on Faith, Secularism, and Ethics (Lanham, Md.: Rowan and Littlefield, 2009), p. 108.

[2] Flannagan, “Is Ethical Naturalism more Plausible than Supernaturalism? A Reply to Walter Sinnot-Armstrong” Philo 15, no. 1 (2012): 19-37.

[3] Ibid, 20.

[4] Ibid, 21.

[5] Ibid, 21.

[6] Ibid, 22.

[7] Sullivan, “Arbitrariness, Divine Commands and Morality” International Journal for the Philosophy of Religion 33, no.1 (1993): 33-45.