Whitcomb's argument

Whitcomb's argument for atheism relies on two key claims:

1. (1) Omniscience is impossible; that is, it is not possible that anyone be omniscient.

2. (2) If God exists, then God is omniscient.

For my purposes here, it is strictly speaking irrelevant how Whitcomb argues for the truth of (1). That is, I want simply to grant to Whitcomb that he has shown that omniscience is impossible. The question of this article is simply what would follow from this fact. For the curious, however, Whitcomb's basic idea is that the possibility of omniscience conflicts with various plausible theses concerning the nature of grounding. Whitcomb gives what he calls a ‘rough sketch’ of his argument as follows, and, I think, nicely sums up the core argument of his paper:

So much for Whitcomb's argument for the impossibility of omniscience.Footnote ³ But note: here Whitcomb simply says that it is false that someone is omniscient. Of course, strictly speaking, Whitcomb ultimately wants something stronger; presumably, what Whitcomb wants is that the given reductio assumption is necessarily false. As he says, he's going to argue not simply that it is false that someone is omniscient, but that omniscience is impossible. So we should understand Whitcomb as follows: the reductio shows not only that no one is omniscient, but that omniscience is impossible.Footnote ⁴

For my purposes, the crucial part of Whitcomb's reasoning is this. After (allegedly) reducing to absurdity the notion that someone is (or could be) omniscient, Whitcomb says: ‘But if God exists, then he is omniscient. Therefore, God does not exist.’ Now, how does Whitcomb argue for the truth of this conditional, viz. that if God exists, God is omniscient? Well, strictly speaking, he doesn't. He just assumes it. As he says:

But now we find ourselves in quite the peculiar position. Surely Whitcomb is right that there is a long, venerable philosophical and theological tradition that tells us that if God exists then God is omniscient. But surely members of this tradition would elect to rethink this commitment, were it to become evident that omniscience is in fact impossible. That is, clearly, this tradition has claimed that God must be omniscient against the background of supposing that omniscience is not in fact impossible, and surely this tradition never would have wanted to ascribe omniscience to God in the first place were it to have been known all along that Whitcomb is right and that omniscience is impossible. After all, members of this tradition who take God to exist are presumably not so philosophically incompetent as knowingly to ascribe to God essential properties they regard to be impossible, thus implying that the existence of the God they believe in is impossible as well. Thus, far from showing that the ‘God of this tradition’ does not exist, Whitcomb's arguments show – again, if they show anything – only that members of this tradition were mistaken from the start to think that God would have to be omniscient. This is, of course, still a highly interesting, important result. But the result is not atheism. Whitcomb is simply overreaching.

Of course, the tradition at issue here is the Anselmian tradition of perfect-being theology. The core commitment of this tradition is the claim that God is the greatest possible being – or, if you like, the claim that, if God exists, then God is the greatest possible being. Better yet, the core idea is that the concept of God is the concept of the greatest possible being; God=df. the greatest possible being. Thus, the question whether God exists is just the question whether the greatest possible being exists. Now, as Whitcomb notes, this tradition has tended to make certain claims about what a perfect (that is, greatest possible) being would in fact be like. For instance, most in this tradition have wanted to say that God is (or would have to be) omnipotent and perfectly morally good. Here it is crucial to see that these claims in fact function simply as (so to speak) guesses (even if good guesses) concerning what a perfect being would be like, but they are not essential components of the tradition per se. Someone could reject the claim that God is (or would have to be) omnipotent, and still be a member of this tradition. Such a person would simply have to have some story about why a perfect being would not be omnipotent after all; perhaps, as many have in fact argued, omnipotence is incompatible with perfect moral goodness, and it is a greater perfection to be morally perfect than to be omnipotent. (See, for instance, Geach (Reference Geach1977) ch. 1.) So there has in fact been and continues to be deep dispute amongst those who agree that God is (or must be) a perfect being about just which further attributes such a being would indeed possess. This is hardly surprising; after all, we might both agree that God is (or would have to be) a perfect being, but differ radically in our conceptions of what perfection would require.

In this light, it seems that Whitcomb is unfairly holding an ultimately secondary commitment concerning what perfection would require against those whose more fundamental commitment is only to perfection itself. In other words, it is true that most theists have wanted to say that God is omniscient if God exists. But then Whitcomb comes along and shows – let us suppose – that omniscience is impossible. He then says, ‘Now, you said that God is omniscient if God exists, but omniscience is impossible. So there is no God!’ Clearly, I think, the theist should simply say, ‘Well, hold on. I did say that God is omniscient if God exists, perhaps even essentially so. But, thanks to your argument, I now see that that was a mistake, and I'd like to change my mind.’ In order to get an argument for atheism off the ground, Whitcomb would have to reply: ‘But you can't change your mind. You still have to say that if God exists, then God is omniscient.’ But why can't the theist change her mind in these circumstances, and why must she continue to insist, in the face of this new data, that if God exists, God is omniscient? Perhaps someone will say: changing your mind in this way is ad hoc. Well, whatever that may come to, it seems false: the move is not ad hoc at all, for we now have excellent reason to suppose that omniscience is in fact impossible, and so excellent reason to suppose that a perfect being would not have to be omniscient. (I return to this point below.) So it seems patently unsatisfactory in the context of providing an argument for atheism simply to assume – once having argued that omniscience is impossible – that God nevertheless must be omniscient. This is a premise that no one – especially members of the relevant tradition – would (or anyway should) want to continue to endorse, given Whitcomb's conclusions. If Whitcomb wants to argue for atheism from the impossibility of omniscience, then he owes us an argument that if God exists, then God is omniscient. He can't simply hold our apparently ill-advised former commitment to God's omniscience against us.

But what argument could Whitcomb give for the premise that if God exists, then God is omniscient – after successfully arguing that omniscience is impossible? This is the trouble. Of course, he could say, quite rightly, that it does seem initially plausible to suppose that a perfect being would have to be omniscient, however omniscience is ultimately to be understood – a question with which we need not be concerned. But it turns out that omniscience is impossible – incompatible, as it were, with the laws of metaphysics – and so a mere chimera, a seeming perfection that turned out to be no perfection at all, but instead an illusory ideal no one or nothing could ever attain. And so, however much it seemed to us initially that a perfect being would have to meet this (alleged) ideal, this is ultimately something we should deny, if omniscience is impossible. In other words, you can't cogently argue that omniscience is a perfection, and so its possession would be required by a perfect being, but then go on to argue that omniscience is impossible – for then it is no perfection at all. So it seems hopeless (in this context) for Whitcomb to argue on broadly ‘intuitive’ grounds that God, since perfect, would have to be omniscient.

More likely is that someone would argue as follows: the meaning of ‘God’ includes omniscience; that is, it is analytic that God is omniscient if God exists. In other words, if we found out that some being was not omniscient, then whatever other properties that being may have, we could straightaway conclude, just by knowing the meaning of the word ‘God’, that this being is not God. But this is false. What is in fact analytic (if anything is) is only that God is a perfect being if God exists; omniscience may or may not be entailed by perfection, but in denying that God is omniscient, you wouldn't have to be conceptually confused. You would just have to have some story concerning why perfection doesn't after all require omniscience. Of course, it is perfectly open to Whitcomb (or anyone else) simply to stipulate the meaning of the word ‘God’, so that, on this stipulation, it is indeed analytic that God is omniscient. And then, it would be perfectly open to Whitcomb to argue that, in showing that omniscience is impossible, he has thereby shown that God – so defined – does not exist.

Perhaps, in other words, Whitcomb might want to retreat to the following position. As he says, he is simply taking the tradition at its word; he is simply taking it that if God exists, then God is omniscient. And so he might reply as follows:

There is nothing stopping Whitcomb or anyone else from saying something like this, even if it is an odd, tendentious sort of thing to say. All we should note is that it is a far cry from showing that God, conceived of (or defined to be) a certain way, does not exist, to showing that God does not exist. And that's what Whitcomb consistently represents himself as doing: arguing that there is no God, not just that there is no God thought of thus-wise, and so forth. If all Whitcomb really wanted to say is that there is no such being as ‘an omniscient God’ (or some such), but that for all that God (that is, a perfect being) may still exist, despite not being omniscient, then that, I think, is what he should have said.

I'll close this section with an analogy. Consider once more the (alleged) incompatibility of omnipotence and perfect moral goodness. The thought is this. Omnipotence requires the ability to do anything, including the ability to perform morally wrong actions. Perfect goodness, however, requires that one be unable to perform morally wrong actions. So omnipotence and perfect goodness are mutually incompatible.Footnote ⁶ Such an argument may or not be successful. But it is folly to suppose that the success of such an argument could provide any reason whatever for supposing that God does not exist, where God=df. the greatest possible being. This is despite the fact that a great many theists (and others) have supposed that if God (=the greatest possible being) exists, then God is both omnipotent and perfectly good. What such an incompatibility would show, clearly, is not that the greatest possible being does not exist, but that the greatest possible being need not be both omnipotent and perfectly good, and that a being's not being both is no mark against this being's perfection. After all, if having both of these properties is impossible, then the greatest possible being doesn't have both of these properties, since no possible being has them. Similarly, if having the property of omniscience is impossible, then the greatest possible being does not have this property, since no possible being has it. So it is not the case that if God (=df. the greatest possible being) exists, then God is omniscient – anyway, not if omniscience is impossible. So, again, even if Whitcomb is right about omniscience, his argument for atheism fails.

An objection

Earlier I claimed the following: if we have excellent reason to suppose that omniscience is in fact impossible, we have excellent reason to suppose that the greatest possible being would not have to be omniscient. More generally, the point is this: if we have excellent reason to suppose that the having of a given property is impossible, then we have excellent reason to suppose that the greatest possible being would not have to have that property. In other words, I seem to be claiming that being perfect (being the greatest possible being) does not require the having of any impossible properties. If the having of a property is impossible, then it is not required of the greatest possible being that she has that property. Being the greatest possible being – being perfect – does not entail being anything impossible. It's as simple as that.

But wait. If being the greatest possible being doesn't entail being anything impossible, this seems plainly to entail that being the greatest possible being is indeed possible – that is, that a perfect being possibly exists. But if a perfect being possibly exists, then, plausibly, a perfect being actually exists. The idea here is simple. First, it is plausible that a perfect being exists necessarily if at all. Given an S5 modal logic, it follows that if a perfect being is possible (that is, exists in some world), then a perfect being is necessary (that is, exists in every world), and hence actual. Thus, by claiming that being the greatest possible being does not entail being anything impossible, I seem to be in effect simply claiming that being the greatest possible being is possible, and thus that God exists. (The claim that being the greatest possible being is possible is the ‘possibility premise’ of the so-called ‘modal ontological argument’.Footnote ⁷ ) Thus, no one who is not already committed to theism will accept the operative assumption underlying my reply to the argumentative strategy in question, viz., that if the having of a certain property is impossible, then the greatest possible being need not have it. This is, so the thought goes, simply to beg the question against the atheist, who is already committed (in virtue of her atheism) to the claim that being the greatest possible being does entail being something impossible. What do I have to say for myself?Footnote ⁸

We can respond to such an objection as follows. The operative claim here is only that perfection does not require the having of any further properties, besides perfection, if those further properties are impossible. That is, being the greatest possible being at least doesn't entail being something else that is impossible. In other words, being perfect doesn't entail being anything impossible – except, perhaps, being perfect itself. I freely admit that being the greatest possible being itself may be an impossible property. My claim is only that we could have no reason to suppose that being the greatest possible being should require some further property that is impossible, such as omniscience, if the having of that further, distinct, property is impossible. Now, I have no precise way to individuate properties. However, I can at least say the following. If two properties are not necessarily coinstantiated, if they can be instantiated at all, then they are not the same property. From this alone it would seem to follow that being the greatest possible being (being perfect) and omniscience are not the same properties. Seemingly, that is, if these properties can be instantiated at all, then omniscience could be instantiated without perfection; if any being could be omniscient, then a being could be omniscient without being (in all respects) perfect.Footnote ⁹

So it doesn't follow from what I want to say to Whitcomb that I in effect assume the possibility of a perfect being, and hence the existence of God. Again, all I want to say is that if the having of a certain property distinct from perfection is impossible, then being perfect doesn't require the having of that property. I am not here assuming that being perfect is in fact possible; it may not be. And I think my position here makes intuitive sense. Consider what seem to be the most promising ways of attacking the possibility premise of the modal ontological argument, viz. that, possibly, a perfect being exists. Arguably, the most promising such ways attack the notion of perfection directly. For instance, I might employ the familiar thought that for any being there is, there could always have been one better.Footnote ¹⁰ Thus, the very notion of a ‘greatest possible being’ or ‘perfect being’ is simply incoherent – or anyway no more coherent than the notion of the greatest possible number. This seems (in principle) like a perfectly fair way of responding to the modal ontological argument, and, indeed, objecting to the existence of God. But what I am committed to here, and what I think is right, is this. If you want to object to the possibility premise of the modal ontological argument, then you had better not do so by first arguing that being the greatest possible being would require the having of some further, distinct properties, and then arguing that the having of those properties is impossible. As I tried to explain above, such a strategy is hopeless.

A comparison

In his 2008 paper, ‘A new defence of Anselmian theism’, Yujin Nagasawa pursues a strategy of responding to (certain) arguments for atheism that is deeply similar to the one I have developed above. Nagasawa distinguishes between three different types of arguments for atheism, the first two of which are as follows:

• Type A: Arguments which purport to show the incoherence of the divine attributes.

• Type B: Arguments which purport to show inconsistency between the divine attributes.

A standard example of a Type A argument, Nagasawa says, employs the paradox of the stone. The paradox purports to show the incoherence of the concept of omnipotence. But if God exists, God is supposed to be omnipotent. So God doesn't exist. Nagasawa also notes arguments from Patrick Grim to the effect that omniscience is impossible, since (for Cantorian reasons) there is no set of all truths.Footnote ¹¹ We thus get an argument exactly parallel to the one developed by Whitcomb: omniscience is impossible, but if God exists, God is supposed to be omniscient, so God doesn't exist. Further, we have already considered a prominent example of an argument of Type B: omnipotence and perfect moral goodness are incompatible, but if God exists, God is meant to be both, so God doesn't exist.

Nagasawa shows, conclusively in my opinion, that arguments of Type A and Type B are destined to fail, if they are meant to show that there is no greatest possible being. Of course, they may show that there is no ‘omniGod’, where omniGod=df. a being that is omnipotent, omniscient, and perfectly morally good. However, Nagasawa notes, the Anselmian theist (someone who believes that God is the greatest possible being) needn't believe that God=omniGod. In response to any argument of Type A or Type B, the Anselmian theist can simply happily admit that God is not (say) omnipotent, omniscient, or both omnipotent and perfectly morally good. She can instead maintain that God has the maximal degree of knowledge, power, and goodness (or whatever other (putatively) great-making properties are in question) it is jointly possible to have. This is, I believe, precisely the right result. As I see it, arguments of Type A and Type B should be cast in terms of what God would be like, were God to exist, not as arguments concerning whether God exists in the first place. Arguments of Type A and Type B, if cast as arguments for atheism, are, I think, entirely on the wrong track.

There is one further sort of argument for atheism that Nagasawa considers, however, and here I believe the strategy under consideration does not show that it is similarly ‘entirely on the wrong track’. The third type of argument is as follows:

• Type C: Arguments which purport to show the inconsistency between the set of the divine attributes and a contingent fact.

As Nagasawa notes, the most prominent argument of Type C – and, in fact, the most prominent argument for atheism – is the argument from evil. Now, Nagasawa contends that moving from the ‘omniGod’ thesis to the ‘MaximalGod’ thesis (the thesis that God has the maximal degree of the relevant great-making properties it is jointly possible to have) similarly undercuts arguments of Type C, such as the problem of evil – at any rate, as standardly presented. For the problem of evil, as standardly presented, purports to show that the existence of an omnipotent, omniscient, and perfectly good being is impossible, given that evil exists. However, Nagasawa notes, the Anselmian theist could again admit that the existence of evil shows that there is no omniGod. But the Anselmian theist could then maintain that this does not, without further argumentation, imply that there is no MaximalGod. And that's what matters.

There is, I think, something right about Nagasawa's position here. If the argument under consideration simply purports to show that there is no omniGod, we do not yet have an argument for atheism, appropriately conceived, for, so far, we have not been told that being the greatest possible being implies being omniGod. This is a perfectly fair point, as far as it goes. Nevertheless, there are, I think, salient differences between arguments of types A and B and those of type C, not adequately captured in Nagasawa's discussion. For consider the following. Suppose that, prior to assessing arguments such as those of Whitcomb's for the impossibility of omniscience, a particular theist takes it that God is omniscient. After all, omniscience would seem to be a great-making property. Now the theist confronts Whitcomb's argument, or perhaps arguments from Patrick Grim that there is no set of all truths. As I have argued above, such a theist can simply ‘give up’ God's omniscience, and do so in an entirely principled way that is perfectly consistent with her belief that God is the greatest possible being. She can ‘give up’ God's omniscience, given the relevant new results, precisely because the new results, in themselves, show (or would show) that omniscience is not after all a great-making property.

But consider someone who might similarly wish to ‘give up’ God's omniscience (or omnipotence or perfect goodness or …) in response to the problem of evil. Suppose, again, that prior to considering the problem of evil, a particular theist takes it that God is omniscient, since omniscience would seem to be a great-making property. She then confronts the argument from evil, and accepts on its basis that the existence of evil is incompatible with the existence of a perfectly good, omnipotent, omniscient being. She then says: ‘I suppose then that God must not be omniscient: God must not know about [or have been able to prevent, or have been good enough to want to prevent] the evils we see in the world.’ Now, this sort of ‘giving up on’ omniscience is not parallel to someone's ‘giving up on’ omniscience in response to arguments such as those of Whitcomb and Grim. For someone giving up on omniscience in this way does not, just in virtue of her consideration of the problem of evil, have reason to deny that omniscience is a perfection. In short, she has no story, so far, concerning why God is still the greatest possible being, despite not having the relevant property. As I see it, there is something problematic and implausible about giving up a traditional divine attribute (which one would otherwise think belongs to God) in response to the problem of evil, precisely because such a denial (unless supplemented by further considerations) is in direct tension with the claim that God is the greatest possible being. One needs a story concerning why God is still the greatest possible being, despite not having the relevant property. Arguments of types A and B have built into them the resources for providing such a story. Arguments of type C – in particular, the problem of evil – do not.

More important, however, is the following. For the reasons we have considered, it seems that arguments of type A and type B are, in a certain sense, entirely on the wrong track. The sorts of issues considered under the heading of such arguments – issues, say, concerning the coherence of omniscience and omnipotence – provide no reason at all, in themselves, for concluding that God does not exist. The theist could simply elect not to engage such issues at all. She could simply let the relevant cards fall where they may: if it turns out that omniscience is impossible, then so be it – we've discovered that God needn't be (and isn't) omniscient. If omnipotence and perfect goodness are incompatible, then so be it – God needn't be (and isn't) both. In response to Whitcomb's or Grim's arguments concerning omniscience, for instance, the theist does not have to become an expert on the nature of grounding or set theory in order to defend her theism. She will only have to do so if she wants to defend a particular thesis about what a perfect being would in fact be like. However, it does not seem similarly plausible that a theist can (in the same way) simply sidestep the traditional sorts of concerns that have gone under the heading of the problem of evil. The sorts of issues there discussed – issues having to do, for instance, with whether this is the best of all possible worlds, and what would justify permission of a certain kind of evil – cannot simply be ‘sidestepped’, given the theist's commitment to the existence of the greatest possible being. For the basic issue still remains: is it at all plausible that there exists the greatest possible being, given the facts of evil that we see in the world? Nothing that I have said in this article blunts the force of this basic problem. Moving from the ‘omniGod’ thesis to the ‘maximalGod’ thesis certainly does not blunt it – or even address it. It simply invites the (sophisticated) proponent of the argument from evil to reframe his objection accordingly: given the facts of evil, it is implausible that there exists the greatest possible being.Footnote ¹² Nothing about the maximalGod thesis renders it automatically immune to this sort of challenge. But the maximalGod thesis is automatically immune to challenges arising from arguments of type A or type B.