Emanuel Rutten sent me the following interesting argument which I am posting with his permission. Please make sure to be clear that if you cite this post, everything except the title, the preceding sentence and this sentence, is taken verbatim from Rutten. He has some other interesting arguments on his blog, some of which alas are in Dutch.
Take the following metaphysical principle, connecting possible worlds, knowledge and truth: 'If it is impossible to know that p, then p is necessarily false'. This principle seems cogent. For, if a given proposition p could be true, then, plausibly, there is some possible world in which some subject knows that p is true. In other words, if in *all* possible worlds *all* subjects do not know that some proposition is true, then, plausibly, that is because that very proposition cannot in fact be true.
Well, on a Cartesian view of knowledge, that is, to know p is to be certain that p is true, the above principle has an interesting consequence. For, take for p the proposition 'God does not exist'. It seems reasonable to hold that it is impossible to know that God does not exist. For, whatever the arguments against God, there will always be some (perhaps an extremely remote) possibility that God does exist after all, so that we can never truly say, on the Cartesian view, that we know that God does not exist. But then it follows that it is necessarily false that God does not exist. Hence, it is necessarily true that God exists. The principle thus entails theism. Is this new argument for theism convincing?
The atheist might object that it is also impossible to know that God exists. And thus, by similar reasoning, it would follow as well that it is necessarily true that God does not exist. However, I would argue that there is a possible world in which some subject can truly say that he or she knows that God exists. Take a possible word in which God exists and in which there is an afterlife, such that all who enter the afterlife in that world will encounter the divine. In that case, those subjects who enter the afterlife will in fact know that God exists. So, it is not impossible to know that God exists. Note that a similar move to reject the argument for theism is not open to the atheist. For, if God does not exist, then, plausibly, there is no afterlife. And besides, even if there would be an afterlife, then entering it would not bring a subject in the epistemic condition of knowing that God does not exist.
Now, the atheist might want to offer three further objections to the proposed argument, which shall I present and respond to in what follows. The second objection (#2) was suggested to me by Alexander Pruss.
The principle on which the argument for theism is based can be formulated as: 'If p is possibly true, then p is knowable'. This principle entails that every truth is knowable. But from that, as Fitch has shown in his 1963 paper 'A Logical Analysis of Some Value Concepts', it can be logically deduced that every truth is in fact known! An atheist might reason that this is a very problematic, if not absurd, consequence. Thus, as the atheist would have it, the proposed argument is not convincing and should be rejected.
I would respond as follows. Now, it is indeed the case, following Fitch, that the principle entails that every truth is in fact known (call this consequence T). But why hold that T is false? After all, for all we know, there might be an omniscient being in the actual world knowing all truths. (I recently read an excellent refutation of Dennis Whitcomb's argument that omniscience is impossible.) So, even though T does seem problematic for atheism, it does not follow that T is false. It would be begging the question for the atheist to deny T solely because T does not fit nicely with atheism and favors theism (since the theist can hold that God knows all truths). Indeed, the fact that my principle entails T is not sufficient to reject it. For, it would be unreasonable for the atheist to initially accept the principle as plausible (which I would contend it certainly is), but then, when it becomes clear (after a non-trivial complex deduction) that it has a consequence unpleasant for atheism, to reject the principle.
The atheist might refute my response to the first objection. After all, someone could, even encountering God in the afterlife, believe that he or she is dreaming, or hallucinating, or being deceived. Therefore, on the Cartesian view of knowledge, it is impossible to know that God exists after all. But then, by parallel reasoning, it also follows that, necessarily, God does not exist. And thus the new argument fails. My response would be that even if someone could always think that he or she is dreaming, hallucinating or being deceived, it still does not follow that it is impossible to know that God exists. For, take a possible world in which God exists. In this possible world there is a subject that knows that God exists, namely God. Indeed, in that world God knows that God exists. So, it is not impossible to know that God exists.
Another objection would be to argue that there might be some true mathematical Gödel sentence G that cannot be proven by any proper mathematical system. Hence, G is unknowable. But then not all truths are knowable, and therefore my principle (which entails that all truths are knowable) fails. My response would be that G is in fact knowable. For, there is a possible world in which G is known. Take again a possible world in which God exists. In that world God can be taken to know (at least) all mathematical truths by direct immediate intuition, and therefore God knows G as well.