Euthyphro and Omnipotence Part 2

This is the second part of my response to Brian Zamulinski’s criticism of my post from last October on the Euthyphro dilemma.

First, I want to point out that Zamulinski rejects the Divine Command Theory; his point in the post I am responding to is that my defense of the Euthyphro argument is unconvincing. However, he also criticizes the DCT for other reasons. (He also offers an interesting objection to Adams’ version of the DCT in a comment to Matthew Flannagan’s most recent article on the subject. And here on his own blog.)

Regardless, Zamulinski does believe that my arguments in support of the Euthyphro objection are unsuccessful. Here is the relevant quote again:

To repeat, Adam’s reformulated divine command theory is that morality is constituted by the commands of an essentially loving God.  Now, if E (for “entity”) essentially possesses P (for “property”), then E possesses P in all possible worlds in which E exists.  So, by hypothesis, God is loving in every world in which He exists.  God is not just contingently loving, that is, loving in at least one possible world in which He exists

On the basis of an analogy, Thibodeau claims that “we may know with certainty that an all-loving being will not issue a command to torture children, but, given that he is omnipotent, it remains the case that he can issue such a command.”  The analogy is to someone who will not but who could jump from the Empire State Building.  The analogy is irrelevant because the possible jumper is only a contingent jumper.  For the parallel to hold, it must be possible for an essentially non-jumping person to jump.  Thus, Thibodeau equivocates between an essentially loving being and a contingently loving being.

Zamulinski’s point, in essence, is that I can either claim that God can command torture or that he is essentially loving, but not both. In claiming that God is essentially loving and that he can command torture I have contradicted myself. Since there is no world in which an essentially loving being commands torture, it is not possible that he commands torture; to assume otherwise is incoherent. Now I actually did not say that an essentially loving being can command torture (though perhaps I implied it; and I’ll argue below that perhaps it really is true). What I said is that an all-loving being that is also omnipotent can command torture.

There are actually two distinct points to be made here. One point concerns the meaning of ‘can’ and the relationship between motive and ability/capacity; the other concerns how we ought to conceive of the motivational attributes of an omnipotent being. Let’s take the issue of ‘can’ first.  The original example that I chose was meant to show that the fact that a person may be unmotivated to do something does not imply that he cannot do that thing. That I am unmotivated to do something might imply that I won’t do it but it does not imply that I can’t. And even if my lack of motivation is due to some essential characteristic, while this does imply that there is no possible world in which I do that thing, it still does not follow that I can’t do it.

When we talk about God being all-loving, we are talking about the range of motives he will act from (or, perhaps more aptly put, the range of motives he can have). The point of claiming that God is essentially loving is that in every possible world his motivational set is such that he would never form the necessary motive to command something horrible like the gratuitous torture of a child. Given the plausible assumption that, for a free and rational being, performing an action requires having an appropriate motive (or pro-attitude) toward the action, it follows that if, in every possible world, God lacks pro-attitudes toward gratuitous violence (which is supposed to be a consequence of his all-loving nature), then, in every possible world, God does not command the gratuitous torture of children.

Does this imply that God cannot command the gratuitous torture of children? It seems to me that it does only if it is true that any statement of the form “P can a” (where ‘P’ is a person and ‘a’ is an action) implies that there is some possible world in which P does a. But I am not convinced that this implication holds.

It seems to me that it is possible for there to be a person who is constitutionally incapable of killing his own mother (in fact, it wouldn’t surprise me if there are many such people). To be more specific, such a person is such that he cannot form a pro-attitude toward the killing of his mother and this is true of him essentially. Call this person Smith. Even though it is impossible for Smith to form the requisite motive, and even though I just said that he is constitutionally incapable of killing his mother, I think that strictly speaking it remains the case that he can kill his mother. That is, he has the ability to acquire a gun and take it to his mother’s house, point it at her, and shoot it. This is something that Smith can do and that, for example, his dog cannot do.  It is true that we might describe Smith as being incapable of killing his mother but when we compare this alleged inability to other things that Smith cannot do (such as fly unassisted, dunk on a 15 foot rim, swim across the Pacific Ocean, survive in a vacuum, etc.), I think that it is appropriate to make a distinction between the things that Smith can’t do because of physical (non-motivational) limitations, and things that he can’t do because of his motivational nature. About the latter group, it is more accurate to say that Smith won’t do those things than that he can’t, full stop.  To say that Smith won’t is to say that he will never form the requisite pro-attitude; and it follows from this being an essential characteristic of Smith that, in every possible world, he won’t kill his mother. However, it remains true that, in many possible worlds (all those in which Smith is not severely incapacitated), he can kill his mother.

If I am right about this, then the fact that there is no possible world in which God commands torture has nothing to do with whether God can command torture. However, there is no immediate upshot as far as the Euthyphro dilemma goes. After all, even if we adopt my understanding of ‘can’ and the distinction between ‘can’t’ and ‘won’t’, it still remains the case that an essentially loving God won’t command torture, in every possible world. Zamulinski is certainly right about this much. However, there are some loose ends here that need to be cleared up, and in attempting to do so, I hope to show that this distinction between ‘won’t and ‘can’t’ really does reinforce the Euthyphro objection.

I’m arguing that ‘P can a’ does not imply ‘There is some possible world in which P does a.’ This is problematic because ‘It is possible that P does a’ is plausibly equivalent to (or at least implied by) ‘P can a.’ How can we maintain the intuitive link between ‘can’ and ‘possible’ given what I’ve said? One possibility is to lean on the notion of close counterparts. Let’s return again to Smith. Presumably, even though there are no worlds in which Smith kills his mother, there are worlds in which some person who is very much like Smith (is a male child of Mrs. Smith, born on Smith’s birthday, with many similar memories to Smith, the same of similar physical characteristics, etc.) who does form the motivation to kill his mother. Again, on the assumption that Smith is essentially incapable of having such a motive appear in his mind, any counterpart of Smith who does form such a motive would not be Smith. But this does not exclude the possibility that, in some world, there is a very close counterpart to Smith who does want to kill his mother.

So, my suggestion is that  ‘P can a’ implies either ‘There is some world in which P does a’ or ‘There is some world in which a close counterpart of P does a.’*  Where a-ing involves doing something that would entail the existence of a motive that is excluded from P’s motivational set because of P’s essential motivational traits,  to say that P can a is to say that there is some possible world in which a close counterpart to P (who has the appropriate motivational set) does a.

Now obviously the notion of close counterpart needs some fine-tuning. How close does a counterpart need to be in order to be a close counterpart? But I don’t think that I need to worry about that here in order to make my point. On my analysis, to say that God can command the gratuitous torture of children is to say that there is some world in which a close counterpart to God (who is not God, given the requisite changes to His motivational characteristics) does command the gratuitous torture of children.

Now I think that if there is a possible world in which a close counterpart to God commands the gratuitous torture of a child, then the divine command theory implies that, in this world, torture is obligatory. I am certain that Matthew Flannagan will disagree with this claim. But rather than trying to predict the nature of his objection, I will let him respond (if he wants to).

I said that there are two points to be made in response to Zamulinski, and so far I’ve only made one. The second concerns the question of whether an omnipotent being can have essential motivational characteristics. I am going to put off this discussion for another post.

*We can probably clean this up with a suitable definition of ‘close counterpart.’ If ‘counterpart’ is defined so as to not excluded identity (but not require it either) then we can get rid of the first disjunct. I ought to point out also that by using the term ‘counterpart’ I am not intending to suggest a commitment to Lewis’ Counterpart Theory.

Euthyphro and Omnipotence part 1

I recently came across a response to my October post on the Euthyphro Dilemma, which I thought it worth responding to. The response, written by philosopher Brian Zamulinski, was itself written in October, but I missed it until just a few days ago. In it Zamulinski says that my arguments defending the strength of the Euthyphro objection to the divine command theory are unsuccessful:

To repeat, Adam’s reformulated divine command theory is that morality is constituted by the commands of an essentially loving God.  Now, if E (for “entity”) essentially possesses P (for “property”), then E possesses P in all possible worlds in which E exists.  So, by hypothesis, God is loving in every world in which He exists.  God is not just contingently loving, that is, loving in at least one possible world in which He exists

On the basis of an analogy, Thibodeau claims that “we may know with certainty that an all-loving being will not issue a command to torture children, but, given that he is omnipotent, it remains the case that he can issue such a command.”  The analogy is to someone who will not but who could jump from the Empire State Building.  The analogy is irrelevant because the possible jumper is only a contingent jumper.  For the parallel to hold, it must be possible for an essentially non-jumping person to jump.  Thus, Thibodeau equivocates between an essentially loving being and a contingently loving being.  Thibodeau has a second objection that also fails:  it begs the question in that it presupposes that a non-loving being can create morality.

In my original post, I gave two arguments for the conclusion that the Euthyphro Dilemma defeats to the divine command theory (DCT) because it shows that the DCT implies that morality is arbitrary and contingent. My first argument was supposed to show that even an all-loving being is able to make cruel commands, and thus that the DCT implies that it is metaphysically possible that a cruel act such as torturing a child is morally obligatory. The second says, essentially, that even if we thought that an all-loving being cannot command torture, this does not help the DCT since it is possible that there exists a divine commander who enjoys (and commands) cruelty for its own sake. If it is possible that such a being (whom I called Asura) exists, then it is possible that torturing children is morally obligatory. Zamulinski says that both of these arguments fail; the first because I have misunderstood or misinterpreted Adams version of the divine command theory, the second because I have begged the question. In this post I will respond to the second of Zamulinski’s criticisms and I will follow up with a later post that will address his first criticism.

For now what I am particularly interested in is the following statement:

Thibodeau has a second objection that also fails:  it begs the question in that it presupposes that a non-loving being can create morality.

This claim strikes me as incredibly odd. Why, if Adams is permitted to assume that an all-loving being can create morality, am I not permitted to assume that a non-loving being can do the same thing? Now, it is true that I actually don’t believe that a non-loving being can create morality, I am assuming it only for the sake of creating a reductio of the divine command theory. But I cannot see that there is any problem with this assumption.

Adams version of the divine command theory assumes something that I believe is false: that an all-loving being can create morality. I believe this is false because I believe that no person, loving or otherwise, could have control over moral facts (that is, have the capacity to change moral facts or to bring them into existence). But I am not and was not trying to make this point. My point was only to draw out the absurd consequences of the divine command theory. My tack was to say that if we are permitted to assume that an all-loving being can create morality, then surely we are permitted to assume that a non-loving being can create morality. Zamulinski has not shown that this is an unreasonable argumentative move.

If we knew how God is able to create morality, then maybe we would be in a position to say that a non-loving being cannot do it (or at least that we are not warranted in supposing that he can). Maybe being all-loving endows God with some special capabilities that a non-all-loving being would not have. But has Adams actually shown how God creates morality? Well, the divine command theory says that God does it by issuing commands. But a non-loving omnipotent being can certainly issue commands. Then is there some reason to believe that the commands of a non-loving being would be ineffective, that they wouldn’t actually create morality even though God’s commands can? If this is what Adams or Zamulinski or anyone else believes, then we need an argument for it. We need to know why it is that being all-loving endows God with the capacity to create morality; we need to know how it works. There is no such argument that I am aware of. As it stands, given that Asura (the evil Creator from my example) is at least as powerful as God, it is reasonable to think that if there is something that God can do, then Asura can do it as well.

The structure of my argument, to which Zamulinski objects, is as follows:

(1)    If God can create morality, then so can Asura.

(2)    There is some possible world in which Asura commands the torture of children

Thus, (3) In that world, the torture of children is morally obligatory.

Thus, (4) There is some world in which the torture of children is morally obligatory.

Therefore, (5) It is metaphysically possible that the torture of children is morally obligatory.

I cannot see how premise (1) begs the question. The question is not, “Can a non-loving deity create morality?” but “Does the divine command theory have the consequence that morality is arbitrary and contingent?” Remember, Adams modification of the divine command theory was motivated by a need for a reply to the Euthyphro Dilemma, not because it is somehow difficult to believe that a being who is not all loving can create moral properties.

Again, my working assumption is that if there is some feat that God (assuming he exists) can accomplish, then, absent any obvious reason to think otherwise, we are justified in believing that any being that is omnipotent will be able to accomplish the same task. It is worth pointing out, however, that Adams’ God is limited in the things that he can do. According to Zamulinski, God cannot command the torture of children, for example. Thus, a being who is not essentially limited, in the way that Zamulinski, Adams, and Matthew Flannagan all agree that God is, can do more than God can do.

With this in mind, I will now reformulate my argument to explicitly refer to a being who is essentially unlimited rather than to the non-loving being Asura:

Conisder the supernatural being who we’ll call Yod: Yod is the omniscient, omnipotent, all-loving creator. In fact the only way in which Yod differs from God is that Yod is not essentially all-loving. There are worlds in which Yod is all-loving, but there are also worlds in which he is not. Yod is not essentially all-loving because he is omnipotent. Being omnipotent, it is possible for him to do anything, including issue cruel commands, such as that children be tortured. Issuing such a command may entail changing his character traits, but being omnipotent means being unlimited, which in turn entails not being limited by one’s own character traits. Since Yod is omnipotent, he can change his own character. Thus there is no problem in supposing that Yod, even though he is actually all-loving (and thus has not actually commanded the torture of children), can command the torture of children. Since Yod can command torture, there is some possible world in which he does command torture and thus, if the divine command theory is true, there is some possible world in which torturing children is morally obligatory.

Notice that this version of the argument does not assume that a non-loving being can create morality, it assumes that an all-loving being who is not essentially loving can create morality. This is not so far from Adams’ presupposition that an essentially loving being can create morality. And until we have some argument that shows why only an essentially loving being can create morality, if Adams’ presupposition is allowable, mine must be as well.

Notice also that I have claimed that Yod’s being omnipotent requires that he not be essentially loving. This observation, which is the basis of my claim that an omnipotent being is able to command torture, will be expanded and defended in my next post.

The Euthyphro Objection is Robust

I came across this article by Matt Flannagan criticizing a recent article by Jerry Coyne about secular morality (thanks to Jeffery Jay Lowder at the Secular Outpost). Coyne seems to want to make two distinct points: One, that atheists have a well-developed moral sense and thus you don’t need God to be a good person; and two, that morality cannot come from God. Flannagan makes some good observations about the relevance of the distinction between having a moral sense and being under a genuine moral obligation and shows that Coyne doesn’t always acknowledge this distinction. However, Flannagan himself is guilty of misunderstanding Coyne’s argument about the Euthyphro dilemma and he wants to downplay some of the serious problems that the dilemma creates for God-based moral theories. I’ll quote the relevant portion of Flannagan’s article:

The only time Coyne is remotely on point is when he argues that if moral obligations are constituted by God’s commands then morality becomes arbitrary; anything at all could be deemed ‘right’ as long as God has commanded it – even stealing or infanticide. Coyne suggests this argument is devastating and has known to be so by philosophers for hundreds of years.

In fact, since Adams’ publication, this argument has been subject to extensive criticism in the philosophical literature. So much so that today even Adams’ leading critics grant that it fails. Adams contended that moral obligations are, in fact, the commands of a loving and just God; therefore, it is possible for infanticide or theft to be right only if a fully informed, loving and just person could command things like infanticide and stealing. The assumption that this is possible seems dubious. The very reason Coyne cites examples such as infanticide and theft is because he considers them to be paradigms of conduct that no morally good person could ever knowingly entertain or endorse.

Coyne seems vaguely aware of the response, stating “Of course, you can argue that God would never sanction something like that because he’s a completely moral being, but then you’re still using some idea of morality that is independent of God.” Here he again falls into confusion. What his response shows is that people can have ideas about and recognise what counts as loving and just independently of their beliefs about God and his commands. Now this is true but this does not show that moral obligations can exist independently of the commands of a loving and just God. Coyne again fails to grasp the basic distinctions involved in discussions of God and morality.

I think Flannagan is wrong in his interpretation of Coyne’s argument. And I know that he is wrong about what Adam’s leading critics say about the validity of the arbitrariness objection that stems from the Euthyprho Dilemma. (One need only consult the work of Michael Martin, Erik Wielenberg, Mark C. Murphy, or even Richard Swinburne, to see that this is so). Regardless of what the academic consensus is, it is fairly easy to show that the arbitrariness objection is very powerful. But first, I want to address Flannagan’s misinterpretation of Coyne’s argument.

Coyne does not make the mistake that Flannagan accuses him of; he is not just saying that in order to judge God’s commands as moral or immoral we would have to have a moral sense that is independent of God. Rather, he is saying that we would need a standard of moral obligation that is independent of God. What Coyne has done is condense a bit of argumentative interaction between the purveyor of the Euthyphro objection and the defender of the divine command theory (DCT). One aspect of the Euthyphro objection is that, if the DCT is true, then morality is arbitrary. If the DCT is true, God can make any action (even something universally regarded as horrendous such as torturing small children) morally right just by commanding that we do it. But this conflicts strongly with our moral intuitions: it seems natural to believe that something as awful as torturing children could not possibly be morally right. But the DCT implies that this action, along with any act that causes unwarranted and horrendous suffering, could possibly be right (Note: the notion of possibility at use here is metaphysical possibility, not epistemic; more on this below.) One divine command theorist response to this is to say that a loving and moral God would never issue commands the require us to needlessly cause people to suffer (this is the response that Coyne mentions).

There are a few problems with this response. The most important (and the one that I think that Coyne had in mind) is that if we are to understand the reply to mean that a moral God would not issue immoral commands, then this in essence capitulates to the Euthyphro objection. That is to say, the response implies that there is a standard of morality that is independent of God against which he and his commands can be judged. But if morality is independent of God, then the DCT is false.

Consider: If God’s commands are the standard of right and wrong, then it makes no sense to say that one of his commands is immoral. Say he commands that every person kills at least one dog in their lifetime just for fun. If his commands establish the moral facts, that, e.g., an action is morally right (or wrong, as the case may be), then his command that we kill a dog establishes that killing dogs is obligatory. And it makes no sense to say that this command is immoral because killing dogs is morally wrong. On the DCT, under this scenario, killing dogs would be morally obligatory, full stop, just because God commanded that we do it. Thus, if the DCT is true, it is logically impossible for God to command us to do something that it would be morally wrong for us to do. The fact that God commanded us to do it establishes that it is morally right. The very important upshot of this for the purposes of the current discussion is that, on the DCT it is a logically necessary fact that every action that God commands us to do is a morally right action.

So now, if we say that God is a morally good being and that therefore he won’t issue immoral commands, we are assuming that there is a standard of morality that is independent of God. For according to what standard are God’s commands to be judged? We just saw that on the DCT, it is logically impossible for God to command us to perform an action that is immoral; but that is just because an action is morally right just in virtue of God’s commanding it. And this means that no matter what commands God issues, including that we kill dogs or torture children, those things would be morally right. So, if we want to say that God won’t issue those kinds of commands because he is moral, then we have to assume some standard, independent of God, according to which an act can be judged as moral or immoral. And this means that we would have to reject the DCT.

This is the point that Coyne was making when he said, “you can argue that God would never sanction something like that because he’s a completely moral being, but then you’re still using some idea of morality that is independent of God.”

Notice that this has nothing to do with appealing to a moral sense that is independent of God. The point is a logical one and does not depend on us having a moral sense or on there actually being genuine moral value. So I think Flannagan just misinterprets the gist of Coyne’s objection in the above quoted passage.

In any event, as Flannagan indicated, the debate does not end here because the divine command theorist may concede the point but still insist that all he needs is that God is all-loving, and he will get the same consequence (or at least one that is close enough); namely that God will not issue commands that require us to cause horrible pain and suffering (or do anything that we all agree would be horrendous). If developed in the appropriate direction, this reply can lead to a fully developed response to the arbitrariness objection. That response goes something like this: “God is necessarily an all-loving being. The commands that he issues flow naturally from his essential nature. Thus it would be impossible for an all-loving being to issue commands to kill, maim, or unjustly harm. So, in fact, it is not possible, on the DCT, that torturing children is morally right because, on the DCT, it is not possible for God to issue a command that we torture children.”

There are two problems with this response. The first problem is that when we are talking about what is metaphysically possible, we are talking about what can happen, not what will happen. So, if I want to know whether it is possible (in the relevant sense) for my friend to jump off of the Empire State building, I need to know only whether he can do it. It is irrelevant to this question whether or not my friend will do it. He may be an unusually content, satisfied, and happy person by nature who has absolutely no inclination toward suicide. I may conclude therefore, that he will not jump from the Empire State building. But it remains the case that he can do it. Similarly, we may know with certainty that an all-loving being will not issue a command to torture children, but, given that he is omnipotent, it remains the case that he can issue such a command. And if he can do it, then it is possible for him to do it. So, it is possible for an all-loving God to command that we torture kids and thus, on the DCT, it is possible that torturing kids is right.

But even if we could somehow respond to this concern, there is still a second problem. This problem stems not from a concern about what it is possible for God to do, but what is possible period. Consider:

The following is possible:

        (A) There exists an all-powerful creator that enjoys watching sentient beings suffer.

As I’ve done in the past, let’s call this horrible deity, ‘Asura.’

Given that (A) is possible, the following is also possible:

                (T) Asura commands that parents torture their babies.

To translate this into possible world semantics, we’ll say that there is a possible world (call it W_A) in which (A) and (T) are true. If the DCT is true, it follows that in W_A the following is true:

                (O) Torturing babies is morally obligatory for parents.

What all of this means is that it is possible that it is obligatory to torture babies.  And it’s important to note that I am not saying that it is epistemically possible, that for all we know torturing babies is obligatory (on the contrary, I think we know that torturing babies is wrong). Rather, I am saying that, if the DCT is true, then it follows that it is metaphysically possible that torturing babies is the right thing to do.

There are two relevant conclusions to draw from this: First, it shows that the arbitrariness objection cannot be answered via the claim that God is necessarily a loving being. Second, it demonstrates once again that the DCT has consequences that are fundamentally contrary to our moral intuitions. We cannot imagine that torturing babies could be right.  Torturing babies is wrong everywhere, every time, in all possible worlds. That is to say, torturing babies is necessarily wrong. Since it implies that it is possible for torturing babies to be obligatory, the DCT conflicts strongly with our moral intuitions.

So, contrary to Flannagan’s dismissal of it, the arbitrariness objection to the divine command theory is very much alive.