According to Leibniz, any answer to the question ‘why is there something rather than nothing?’ must bottom out in “a necessary being, which carries the reason for its existence within itself, otherwise we still would not have a sufficient reason at which we can stop” (Principles of Nature and Grace, sect. 8, tr. Woolhouse and Francks). The coherence of such a being has, however, been questioned. What would it be for a being to ‘carry the reason for its existence within itself?’ What kind of impossibility could there be in the supposition that some particular being does not exist? Earl Conee’s contribution to The Puzzle of Existence is devoted to arguing that no broadly Anselmian argument for the impossibility of the non-existence of God can succeed. Its relevance to the theme of the volume is not spelled out, but I take it that the above issues are in the background: Anselm’s argument purports to derive a contradiction from the supposition that there is no God. If the argument succeeded, it would thus amount to a defense of the existence of a necessary being, just the sort of regress-stopping being wanted for certain answers to the puzzle of existence.
Recall that Anselm’s general strategy is to argue that the Greatest Conceivable Being (GCB) must exist because existence is greater than non-existence. If the GCB did not exist, then it would be possible to conceive of a being, GCB+, who was just like GCB except that GCB+ exists. This would make GCB+ greater than GCB, but of course it is by definition impossible to conceive a being greater than GCB, so the supposition that GCB does not exist yields a contradiction.
According to Conee, the mistake in the argument is a confusion between the level of greatness a being must have in order to satisfy a certain conception and the level of greatness a being satisfying a particular concept actually has. Thus the concept unicorn requires more greatness than the concept horse, but the things satisfying the concept horse are greater than the things satisfying the concept unicorn because the latter are merely imaginary. When we conceive of a GCB, this conception requires more greatness than any other possible conception, but it does not follow from this that some other conception is not satisfied by greater things, if the latter conception (e.g., horse) has real instances and the GCB is merely imaginary.
Conee’s objection is reminiscent of two memorable remarks of Kant’s on this topic:
To posit a triangle and cancel its three angles is contradictory; but to cancel the triangle together with its three angles is not a contradiction (A594/B622).
A hundred actual dollars do not contain the least bit more than a hundred [merely] possible ones (A599/B627).
The general idea here is sometimes called the ‘conditionalizing strategy.’ The idea is that the concept or definition of a GCB tells us what has to be true of something in order for it to be a GCB. Even if we build existence into the concept or definition, the only result we get is that in order for anything to be a GCB, that thing must exist, but this is totally uninteresting, since it is also true that in order for anything to be a triangle, that thing must exist.
What Conee wants to show is that ‘an optimal version of Anselm’s argument’ falls to this sort of objection. In order to count as a ‘version of Anselm’s argument’ Conee says, an argument must proceed from the conception of a GCB to the absurdity of denying the GCB’s existing via the assumption that “existence mak[es] a positive difference toward … greatness” (115-116). Thus, although Conee talks in the notes about the prospects for an argument that talks about necessary existence, he does not address modal ontological arguments in detail.
Can an argument which is Anselmian in this sense escape the conditionalizing strategy? With the help of some controversial assumptions, I think it can. Here is an argument that the Fool cannot coherently say (affirm) in his heart that there is no God:
- If one cannot coherently conceive of x as F, then one cannot coherently affirm that x is F.
- Beings conceived of as real are conceived of as being greater than beings conceived of as merely imaginary/fictional.
- It is possible to conceive of a GCB as real.
- One cannot coherently conceive of a GCB as merely imaginary/fictional. (If one did, then either one would conceive the GCB as both real and merely imaginary/fictional, which is a contradiction, or else it would be possible to conceive of a being greater than the GCB, namely, a real being that is just like the GCB.)
- One cannot coherently affirm that a GCB is merely imaginary/fictional.
Conee discusses Meinongian and anti-Meinongian versions of the argument, but I think this version, which appeals to imaginary/fictional objects, but not non-existent objects, is more faithful to Anselm, since Anselm talks about ‘existing in the understanding.’ Presumably objects that exist in the understanding exist.
What the Fool denies, on this reading, is that God is real. He thinks that God is a mere fiction, an imaginary being. (Atheists cannot very well deny that there is a character called ‘God’ in a great many stories.) This helps the argument to escape Hume’s objection that whenever we conceive of anything we always conceive of it as existing, for there seems to be a significant difference between how I conceive of Abraham Lincoln and how I conceive of Sherlock Holmes: I conceive of Lincoln as a real historical person, and Holmes as a fictional character. It is plausible to suppose that this is really part of the content of my conception.
I see two main weaknesses for this argument. First, one could question whether, by conceiving of something as real, we actually conceive of it as being greater than if we conceive of it as merely fictional/imaginary. Perhaps unicorns are still conceived as greater than horses, even when I explicitly include the fictionality of unicorns in my conception. Second, there are tricky issues here about the very nature of fictions. For instance, according to the fiction about Holmes, Holmes is a real (i.e. non-fictional) detective. Now, perhaps the right thing to say about this is that, when engaging imaginatively with the fiction, the reader conceives of Holmes as real, but the reader (who knows she is reading fiction) does not affirm this conception. The conception she affirms is the conception of Holmes as fictional.
These are tricky issues. In any event, the argument I have given is, I submit, superior to the one Conee calls the ‘Optimal Anselmian Argument,’ at least in the sense that it is harder to see what’s wrong with mine.
(Cross-posted at blog.kennypearce.net.)
Consider a valid (given S5) ontological argument, for instance the Plantinga one:
1. Possibly, there is a maximally great being (or: God).
2. Necessarily, if x is a maximally great being (or: God), then x is maximally excellent in all worlds.
3. Therefore, there actually is a being that is maximally excellent in all worlds.
A standard approach to arguing against ontological arguments is to come up with a parody like “necessarily existing chimera” or a “maximally great island”. However, I think that a lot of such parodies achieve little because the parodic concept they operate with (a) seems much less natural than that of a maximally great being (or a positive property, in the Goedelian arguments), and (b) is not in common use.
Both of these points are relevant. The more natural a concept seems, the more likely it is that the concept is of something possible. It is easy to stipulate unnatural impossible concepts. And the more a concept is in common use, the more revisionary it is to claim that the concept is of something impossible. So when it comes to a competition between the possibility of a necessarily existing devil and the possibility of a maximally great being, the latter wins out. I think this point is applicable to all three major forms of the ontological argument: Plantinga, Goedel and Anselm.
Specifically in regard to the Plantinga argument, it is sometimes forgotten that decent ontological arguments do not start with stipulative premises. The Plantinga argument does not stipulate a maximally great being as maximally excellent in all worlds. Rather, it starts with the intuitive, fairly natural and commonly used (e.g., in monotheistic devotion) concept of a maximally great being, and offers a substantive philosophical claim about that concept, namely that anything that falls under that concept has maximal excellence in all worlds. This substantive claim may be justified directly by intuition or by an argument, but it is not stipulative.
This point also shows why it is mistaken to say that the Plantinga argument begs the question in its possibility premise (1), as it is often alleged to. The possibility premise in the Plantinga argument does not by itself justify the conclusion. It justifies the conclusion in conjunction with the substantive philosophical analysis yielding (2). Granted, premise (2) is less controversial than (1), but it is a substantive premise as can be seen from the fact that even some theistic philosophers (e.g., Swinburne) deny it.
Anyway, all of this shows that a good parody of an ontological argument would start with a concept that is natural and in common use and that has the property that its instantiation is incompatible with the existence of God. The best version of a parody argument I know is the argument given of Gale and others that possibly there is a gratuitous evil, and hence there is no God, since if God exists, there is no gratuitous evil in any world. The concept of a gratuitous evil does have a certain naturalness to it. It is a substantive point that gratuitous evils are incompatible with the existence of God–van Inwagen, for instance, denies the point–and hence the argument involves substantive non-logical analysis. But I think this argument still loses out to the ontological argument because the concept of a gratuitous evil is to a lesser extent in common use than the concept of God or a maximally great being. Working theists as Gale notes sometimes do worry that a particular evil might be gratuitous, but I think the compound concept–a gratuitous evil–is not in as common use as the concept of God.
In a piece (based on a post I made on Prosblogion almost two years ago) that has just come out in Religious Studies (with a response by Graham Oppy), I prove a certain theorem. Say that a property A is strongly positive iff, necessarily, having A essentially is a positive property. Assume the following three axioms:
- F1: If A is positive, ~A is not positive.
- F2: If A is positive and A entails B, then B is positive.
- N1: Necessary existence is positive.
Theorem T1: Given F1, F2 and N1, if A is a strongly positive property, then there exists a necessarily existing being that essentially has A.
- N2: Essential omniscience, essential omnipotence and essential perfect goodness are positive properties.
Then we get the following result.
Corollary C1: Given F1, F2, N1 and N2, there exists a necessary being that is essentially omniscient, and a necessary being that is essentially omnipotent, and a necessary being that is perfectly good.
But I was unable to prove, without assuming further controversial axioms, that there is one being that is omniscient and omnipotent and perfectly good. I can now do so as long as one grants the following axiom:
- N3: There is at least one strongly positive property that, necessarily, is uniqualizing.
A property is said to be uniqualizing provided that it is impossible for there to exist in one world two distinct things that have the property. For instance, being the tallest woman is uniqualizing. Note that it is prima facie possible Janet to have a uniqualizing property in one world and for Patricia to have the same property–but in a different world.
Theorem T4: Given F1, F2, N1 and N3, there exists a unique necessary being that has all the strongly positive properties.
Corollary: Given F1, F2, N1, N2 and N3, there necessarily exists an essentially omniscient, omnipotent and perfectly good being.
Moreover, I think a good case can be made (see point 1 below) that N2 implies N3, so in fact, the controversial axioms are going to be F1, F2, N1 and N2, just as in T1.