Suppose we make an ontological argument with the following general form:
- D (for divinity) is a consistent concept
- Every consistent concept is possibly instantiated
- D entails necessary existence
- D is actually instantiated
- A being who instantiated D would be God
- God exists
Or something like this. (Note that this formulation of the argument uses the modal principle that possibly necessarily p entails necessarily p.)
This sort of argument has a problem: If (3) is trivial, given how D is defined (for instance, if D is defined as the concept of necessary existence), then the argument is question-begging, for no one who didn’t already believe that D is instantiated would ever accept premise (1). However, if (3) is a surprising result, for which a sophisticated argument is required, then we might worry that D has other surprising consequences, some of them contradictions, and so reject (1).
Furthermore, it seems to me that for most theists the existence of God is more certain than any of the premises in the argument, so it doesn’t seem that the argument can be used to increase the confidence of someone who is already a tentative theist.
So can the ontological argument do anything? I say it can. Suppose we grant that the argument is question-begging, and that, for the theist, the conclusion is already more certain than the premises. Now, all question-begging arguments are valid, and some are sound, the problem is just that they can’t be used to convince people of the conclusion. A parallel case to this can be found in work on the foundations of arithmetic. When axiomatic set theory was being developed, people were coming up with axioms, which they weren’t sure were right, and using them to prove that 2+2=4. Now, since the premises of these arguments were quite questionable, and the conclusion was already known certainly to be true, this procedure can’t possibly have been intended to increase anyone’s confidence in 2+2=4. So what was the point? Well, they were trying to construct an axiomatic foundation for arithmetic, and 2+2=4 is a truth of arithmetic, so if the axioms don’t entail it they are obviously not the axioms we are looking for, and if they entail its contrary they are obviously false. However, if we can come up with axioms that have all the right entailments, we’ll have a reason for thinking our axioms are true, and we’ll be able to use them to answer the questions we weren’t sure about to begin with, and we also will have a much more systematic understanding of arithmetic. (Note that the method of ‘reflective equilibrium’ employed by many philosophers, especially in value theory, has a similar structure: you start by making up some proposed rules and seeing if they get the right answers in the cases where we know what the right answer is, then, if they do, apply them to the cases we aren’t sure about.)
I propose that the ontological argument might play a similar role in a theory of metaphysical theology. (Indeed, I think it does play a role somewhat similar to this in James Ross’s Philosophical Theology.) On this view, the ontological argument is not used directly to convince people of the existence of God. Rather, its premise (1) is proposed as a sort of axiom, on which a metaphysical system is to be built. This system is to be judged by its overall coherence and the plausibility of its consequences. The ontological argument might still have an indirect role to play in convincing people of the existence of God: if it turns out that assuming that the concept of, e.g., a being than which none greater can be conceived, or an infinitely perfect being, or whatever, is consistent leads to a highly attractive and systematic theory of metaphysics, wouldn’t that be a reason for accepting it? Next question: what are the consequences (apart from the existence of God) of supposing the candidate concepts to be consistent?
[cross-posted at blog.kennypearce.net]