An Improved Ontological Argument for Atheism

Alvin Plantinga (in)famously proved the existence of God using the axiom S5 in modal logic. Axiom S5 states that if it is possibly necessarily the case that p, then it is necessarily the case that p. In other words, if there is any possible world in which it is a necessary truth that p, then p is a necessary truth in all worlds. (In case the reader has some phobia to talk of possible worlds, we can state the axiom another way: if it can be shown that there is a possible maximally consistent set of sentences in which ‘necessarily, p’ occurs, then that statement must appear in all maximally consistent sets of sentences.) Plantinga defines God in such a way that if God did exist, then God would be necessarily existent. Using axiom S5, the modal ontological argument follows. A simplified version of the modal ontological argument – which skips over much of the detail and explication Plantinga is careful to provide – might proceed as follows:

  1. If it is possibly necessarily the case that God exists then it is necessarily the case that God exists.
  2. It is possibly necessarily the case that God exists.
  3. Therefore, it is necessarily the case that God exists.
  4. Therefore, it is actually the case that God exists.

There have been those who were skeptical of modal logic generally or of those modal logics which use S5 in particular (e.g. Quine). Put modal skeptics aside; most philosophers agree that modal logic captures something important about our talk of possibility and necessity. The trouble for the theist, as Plantinga is careful to point out, is that a parallel argument can be constructed for the atheist:

  1. If it is possibly necessarily the case that God does not exist then it is necessarily the case that God does not exist.
  2. It is possibly necessarily the case that God does not exist.
  3. Therefore, it is necessarily the case that God does not exist.
  4. Therefore, it is actually the case that God does not exist.

The two arguments can be further simplified to two statements:

T. Possibly, God exists → God exists.

A. Possibly, God does not exist → God does not exist.

The difficulty rests in stating which of the antecedent in T or A are more plausible. The question remains: is it possible that God exists? Or is it possible that God does not exist?

On the one hand, if the atheist claims that God does not exist, they have to show that there is no possible world at which God exists (or, for the metaphysically phobic, that there is no maximally consistent set of sentences in which ‘God exists’ appears). That’s a rather tall order and the theist would be right to respond with incredulity: it’s not even possible that God exists?

Furthermore, the theist’s work seems to be cut out for them: all they have to do is to point out how reasonable it is to say that God is at least metaphysically possible. Certainly, the concept of God does not appear (at least prima facie) to contain any contradictions. So the theist might declare victory (and Plantinga does call his ontological argument victorious).

On the other hand, the atheist should caution the theist that they declared victory too soon. It seems fairly plausible that there is at least one possible world at which God does not exist (or that there is at least one consistent set of sentences in which ‘God exists’ does not appear). For the theist to declare that there is no such world is a rather tall order; the atheist would be right to respond with incredulity: it’s not even possible that there is no God? And, again, the atheist’s work seems to be cut out for them: all they have to do is to say that God’s non-existence is at least metaphysically possible. Certainly, atheism does not appear (at least prima facie) to contain any contradictions. So the atheist, as the theist, might declare victory (and we might call the resulting atheistic ontological argument victorious).

So the stalemate rests. However, I’ve noticed that my Cosmological Euthyphro Dilemma (CED) may contain a possible solution in favor of the atheist. In what follows, I will briefly summarize my CED argument and then show why I think it has the potential to resolve the impasse between T and A. I note that one of the resident theists on this blog network, MaximusConfesses, has provided me with a possible Anselmian response to CED. I will assume, for this article, that the Anselmian reply does not work, but I will postpone my response to the Anselmian reply to a future blog post.

What is CED? Traditionally, theists have posited that both that God created the universe and that God is absolutely ultimate (e.g. aseity). CED poses a dilemma to the theist: did God’s reasons for creating the universe originate in God (perhaps in Her nature or essence) or did they originate independently of God? If God’s reasons originated in God, there are two problems for the theist. First, because God’s nature or essence are both necessarily existent, God’s reasons would be necessarily existent. But if God’s reasons are necessarily existent, then God could not have chosen to do other than what God did; thus, God does not have free-will (in either the libertarian or compatibalist senses, which both require true counterfactuals about one’s actions). This is incompatible with God’s existence because God is defined as an agent; but if God does not have free-will, then God is not an agent. So God does not exist. Second, given that God would create the same universe in all possible worlds, the universe that we are in would be the universe that necessarily exists. Ditto for anything else that exists, for, on a traditional conception of God’s ultimacy, everything that exists is either God or created by God. So, metaphysical possibility is destroyed; everything is necessarily existent. But this also leads to God’s non-existence because most philosophers are not willing to give up the notion of metaphysical possibility or contingency. Thus, if we posit that some other states of affairs are possible, then God does not exist.

The other possibility is that God’s reasons for creating originate (somehow) independently of God. But if so, there is at least one thing – namely, the source of God’s reasons – that is either more ultimate than God or just as ultimate as God. Neither of these possibilities square with the conception of God as most ultimate. Thus, if God possesses reasons and these reasons must originate independently of God, then we are led into contradiction; so God does not exist.

There is a third possibility. Many of the traditional defenses of libertarian free-will rely on the notion that some sort of mysterious action without reason may be possible. For example, consider Buridan’s Ass. We imagine a hungry donkey equidistant between two equally valuable sources of food. If the donkey does not choose one food source over the other, then the donkey will starve. But if the donkey does choose one food source over the other, this cannot be with any sort of reason, for the donkey has just as much reason to choose one food source as the other. Intuitively, the donkey will choose one food source over the other; so, if our intuitions are correct, we must be capable of acting without reason (somehow). Perhaps this will save the theist from the CED argument.

Unfortunately, this third possibility cannot save theism from the CED argument either. If God acts without reason, this is no better than God acting randomly. It is difficult to see why this possibility would be desirable to the theist, though, historically, some theists did choose this option. (See Lovejoy’s The Great Chain of Being.) At the very least, if God acts randomly, God acts capriciously. It is difficult to see how this possibility would be consistent with the notion that God is perfectly good or all loving.

CED destroys a number of traditional arguments for God’s existence. For example, the design argument posits that our universe, or various structures in our universe, appear to have been designed for various purposes and that they were not merely created at random. But if the only way to save theism, given the CED, is for God to have created randomly, then design is not what we would expect to see in our universe. God could have no purpose for Her creation, for she evidently created without reasons.

Or consider the argument from contingency. The argument from contingency maintains that our universe is metaphysically contingent (i.e. our universe is not metaphysically necessary) and requires the existence of a metaphysically necessary being to explain its existence. The argument from contingency is not consistent with the notion that God’s reasons originate within God, for, in that possibility, our universe would not be contingent. One might think the argument from contingency is consistent with the possibility that God’s reasons originate independently of God, provided that wherever God’s reasons originate is contingent. But this cannot be so, for if the source of God’s reasons is contingent then either that source was created by God or contingent things do not require necessary creators after all. If the source was created by God, then we arrive back at CED: where did God’s reasons for creating that source originate? Lastly, perhaps one could think that the argument from contingency was consistent with God creating the universe without reason. But then it becomes difficult to explain why a theistic explanation would be better than positing that the universe originated at random without God, for both explanations posit that the universe was created without reason.

As promised, I now proceed by explicating how the CED argument might break the stalemate between T and A. One way of stating the stalemate is that we need some reason for thinking that it is more plausible for there to be at least one possible world without God than at least one possible world with God (or vice versa). The CED argument can provide that reason.

Recall that, on the possibility that God’s reasons originate within God, if God exists, there can be no other metaphysically possible worlds. It’s possible to show, using familiar arguments in modal logic, that there being no other way that the could be is entailed by possibly, God exists. So T and A become:

T’. Possibly, God exists → There is no other way that the world could be.

A. Possibly, God does not exist → God does not exist.

Using T, together with the statement that there is some other way that the world can be, we arrive at the antecedent to A via modus tollens. And it’s far more plausible that there is some other way that the world could be than that there is no other way that the world could be. So the stalemate is settled in terms of A. Thus: God does not exist and God’s existence is, probably, metaphysically impossible.

But what about the other possibilities, that either God’s reasons for creating originate independently of God or that God creates without reason? First, consider the possibility in which there is some God independent source for God’s reasons. In that case, there are two possibilities: either the source is necessarily existent or the source is contingent. Suppose that the source is necessarily existent. In that case, God possesses the same reasons in all possible worlds, so we return to T’ and A is again more plausible. Suppose that the source is contingent. In that case, God may possess different reasons in different possible worlds, but which reasons God might have for any of God’s actions would depend upon some contingent object. Strangely, this turns God into a slave of some God-independent object; this contradicts the ontological argument’s supposition that God is the greatest of all possible beings (this also contradicts the notion of God’s impassability). The only possibility that remains is that God has no reason for God’s actions. But, in that case, God’s actions are arbitrary and capricious. At best, God would act randomly. Again, it’s difficult to understand how a capricious God that acts arbitrarily is the best of all possible beings. And, again, this possibility is inconsistent with the theistic ontological argument. So the atheistic ontological argument remains.

2 Comments

  • Daniel Malcor
    May 18, 2015 - 1:14 am | Permalink

    Which “God” did we prove/disprove? Since this works for all definitions of gods it is invalid before it starts.

    • Dan Linford
      May 18, 2015 - 9:31 am | Permalink

      This article responds to classical theism, in which God is defined as a necessarily existent being, who created the universe, who is omniscient, omnipotent, omnibenevolent, and personal.

      Nonetheless, you didn’t respond to any of my arguments so it is difficult to know what to make of your claim that my article “works for all definitions of gods” but is “invalid before it starts”. Perhaps what you mean is that my article does _not_ work for all definitions of ‘God’, but then I never claimed that it did.

  • Leave a Reply