Wednesday, December 1, 2010

Plantinga’s ontological argument

Alvin Plantinga famously defends a version of the ontological argument that makes use of the notion of possible worlds. As is typically done, we might think of a “possible world” as a complete way that things might have been. In the actual world I am writing up this blog post, but I could have decided instead to go pour myself a Scotch. (Since it’s still morning, I won’t – I can wait an hour.) So, we might say that there is a possible world more or less like the actual world – Obama is still president, I still teach and write philosophy, and so forth – except that instead of writing up this blog post at this particular moment, I am pouring myself a Scotch. (Naturally there will be some other differences that follow from this one.) We can imagine possible worlds that are even more different or less different in various ways – a possible world where the Allies lost World War II, a possible world in which human beings never existed, a possible world exactly like the actual one except that the book next to me sits a millimeter farther to the right than it actually does, and so forth. Not everything is a possible world, though. There is no possible world where 2 + 2 = 5 or in which squares are round.

Philosophers make use of the notion of possible worlds in all sorts of ways. For example, it is sometimes suggested that we can analyze the essence of a thing in terms of possible worlds: What is essential to X is what X has in every possible world, what is non-essential is what X has in some worlds but not others. It sometimes suggested that modality in general can be analyzed in terms of possible worlds: A necessary truth is one that is true in every possible world, a possible truth one that is true in at least one possible world, a contingent truth one that is true in some worlds but not others, an impossible proposition one that is true in no possible world. Plantinga, again, makes use of the notion in order to reformulate the ontological argument famously invented by Anselm. We might summarize his version (presented in The Nature of Necessity and elsewhere) as follows:

1. There is a possible world W in which there exists a being with maximal greatness.

2. Maximal greatness entails having maximal excellence in every possible world.

3. Maximal excellence entails omniscience, omnipotence, and moral perfection in every possible world.

4. So in W there exists a being which is omniscient, omnipotent, and morally perfect in every possible world.

5. So in W the proposition “There is no omniscient, omnipotent, and morally perfect being” is impossible.

6. But what is impossible in one possible world is impossible in every possible world.

7. So the proposition “There is no omniscient, omnipotent, and morally perfect being” is impossible in the actual world.

8. So there is in the actual world an omniscient, omnipotent, and morally perfect being.

Plantinga famously concedes that a rational person need not accept this argument, and claims only that a rational person could accept it. The reason is that while he thinks a rational person could accept its first and key premise, another rational person could doubt it. One reason it might be doubted, Plantinga tells us, is that a rational person could believe that there is a possible world in which the property of “no-maximality” – that is, the property of being such that there is no maximally great being – is exemplified. And if this is possible, then the first and key premise of Plantinga’s argument is false. In short, Plantinga allows that while a reasonable person could accept his ontological argument, another reasonable person could accept instead the following rival argument:

1. No-maximality is possibly exemplified.

2. If no-maximality is possibly exemplified, then maximal greatness is impossible.

3. So maximal greatness is impossible.

In The Miracle of Theism, atheist J. L. Mackie argues that even this concession of Plantinga’s overstates the value of his ontological argument. For it is not at all clear, Mackie says, that a rational person can treat the question of whether to accept either Plantinga’s argument or its “no-maximality” rival as a toss-up, as if we would be within our epistemic rights to choose whichever one strikes our fancy. Why wouldn’t suspense of judgment in the face of such a deadlock, a refusal to endorse either argument, be the more rational option? Indeed, if anything it is the “no-maximality” argument that would be the more rational choice, Mackie suggests, in light of Ockham’s razor.

But though I do not myself endorse Plantinga’s argument, I think these objections from Mackie have no force, and that even Plantinga sells himself short. For it is simply implausible to suppose that, other things being equal, the key premises of Plantinga’s argument and its “no-maximality” rival are on an epistemic par. To see why, consider the following parallel claims:

U: There is a possible world containing unicorns.

NU: “No-unicornality,” the property of there being no unicorns in any possible world, is possibly exemplified.

Are U and NU on an epistemic par? Surely not. NU is really nothing more than a denial of U. But U is extremely plausible, at least if we accept the whole “possible worlds” way of talking about these things in the first place. It essentially amounts to the uncontroversial claim that there is no contradiction entailed by our concept of a unicorn. And the burden of proof is surely on someone who denies this to show that there is a contradiction. It would be no good for him to say “Well, even after carefully analyzing the concept of a unicorn I can’t point to any contradiction, but for all we know there might be one anyway, so NU is just as plausible a claim as U.” It is obviously not just as plausible, for a failed attempt to discover a contradiction in some concept itself provides at least some actual evidence to think the concept describes a real possibility, while to make the mere assertion that there might nevertheless be a contradiction is not to provide evidence of anything. The mere suggestion that NU might be true thus in no way stalemates the defender of U. All other things being equal, we should accept U and reject NU, until such time as the defender of NU gives us actual reason to believe it.

But the “no-maximality” premise of the rival to Plantinga’s ontological argument seems in no relevant way different from NU. It is really just the assertion that a maximally great being is not possible, and thus merely an assertion to the effect that Plantinga’s first and key premise is false. And while Plantinga’s concept of a maximally great being is obviously more complicated and harder to evaluate with confidence than the concept of a unicorn, it seems no less true in this case that merely to suggest that a maximally great being is not possible in no way puts us in any kind of deadlock. Unless someone has actually given evidence to think that Plantinga’s concept of a maximally great being entails a contradiction or is otherwise incoherent, the rational position (again, at least if we buy the whole “possible worlds” framework in the first place) would be to accept his key premise rather than the key premise of the “no-maximality” argument, and rather than suspending judgment.

(Mackie’s assumption that Ockham’s razor is relevant here – he speaks of not multiplying entities beyond necessity – also seems very odd to me. Appealing to Ockham’s razor is clearly in order when you are dealing with alternative explanations each of which is already known to be at least in principle possible, and are trying to weigh probabilities in light of empirical evidence. But questions about semantics, logical relationships, conceptual and metaphysical possibilities, and the like – the sorts of issues we are considering when trying to decide whether Plantinga’s key premise or its rival is correct – are not like that. The whole idea of applying Ockham’s razor to such issues seems to be a category mistake. But I won’t pursue the thought further here.)

Other objections to Plantinga are also oversold. There is, for example, the tired “parody objection” that critics have been trotting out against ontological arguments since Gaunilo, and which I suggested in my previous post have no force, at least against the most plausible versions of such arguments. For example, John Hick suggests (in his An Interpretation of Religion) that Plantinga’s reasoning could equally well be used to argue for the existence of a maximally evil being, one that is omnipotent, omniscient, and morally depraved in every possible world. The problem with this objection is that it assumes that good and evil are on a metaphysical par, and as I have had reason to note before, that is by no means an uncontroversial (or in my view correct) assumption.

But defending the idea that evil is a privation would require a defense of the more general, classical metaphysics on which it rests. And there lies the rub. For Plantinga is not a classical (i.e. Platonic, Aristotelian, or Scholastic) metaphysician. That is reflected not only in the way he conceives of God’s omnipotence, omniscience, and “moral perfection” – we’ve noted before that Plantinga is a “theistic personalist” rather than a classical theist – but also in the more general metaphysical apparatus he deploys in presenting his ontological argument. From a classical metaphysical point of view, and certainly from an Aristotelian-Thomistic (A-T) point of view, the “possible worlds” approach is simply misguided from the start (for reasons we’ve also had occasion to discuss before). Many no doubt think that Plantinga’s argument is at least an improvement on Anselm’s. I think it is quite the opposite. In no way do I intend that as a slight against Plantinga; on the contrary, The Nature of Necessity is, as no one familiar with it needs me to point out, a testament to his brilliance. But it is also, like the best of the work of the moderns in general, a brilliant mistake. A sound natural theology must be grounded in a sound metaphysics, which means a classical (and preferably A-T) metaphysics. Within the context of a classical metaphysics, Anselm developed as deep and plausible an ontological argument as anyone ever has. But (so we A-T types think) even he couldn’t pull it off.

24 comments:

David said...

The peculiar thing to me is the claim about excellence "in every possible world". One possible problem with "possible worlds" (as referred to in your previous post from June) is that God isn't simply one more thing in the world; but we can probably get away with "possible worlds" if they are just a way of talking. However, as soon as you starting talking of something that spans possible worlds, you are implicitly positing some kind of meta-world which contains them all, and then God has to be something inside that world — or rather, outside it, but certainly outside all the little possible worlds. But then the argument doesn't prove what it's supposed to.

Anselm's version definitely has a lot of meat, though. I wonder how much of Aquinas's objection to it is in some sense practical: the argument does in a way seem to demand knowing God in order to prove Him, which of course is not terribly helpful in trying to convince someone, or to learn something new. If you already have to prove God's existence first, why not go on then to the next question and use that knowledge to demonstrate something new? But of course Anselm isn't trying to persuade himself that God does exist, he's trying to understand God better. Since God does exist and is necessary, it certainly seems as though an ontological proof ought to work.

Domini Canes said...

@David: "[A]s soon as you starting talking of something that spans possible worlds, you are implicitly positing some kind of meta-world which contains them all, and then God has to be something inside that world."

No. That Socrates did not teach Plato in a world W is, given his transworld identity, true of Socrates in every world. No need of postulating some "meta-world" arises.

Crude said...

A weird question. Since the theistic personalist and the classical theist are often cast as believing in beings that are noticeably distinct from each other, is it logically possible for the two deities to exist at once? (God as conceived by the theistic personalist, and God as conceived by the classical theist.)

A strange question - or a dumb one, I suppose - but still, I'm curious.

William said...

Great post. I happen to agree (reluctantly) that the ontological argument fails. I found it interesting though that it doesn't fail in the way that modern philosophers claim it fails. When I read Plantinga's argument I found the whole 'possible worlds' argument not really squaring up with a more traditional essentialist idea of the the world.

One thing that I must criticise though is the continual use of terms like 'If the modern used an A-T understanding', 'This is not what A-T would say', 'A-T says against modern concepts' and so on. Medieval/baroque (I shudder at the thought of Jesuit baroque) philosophers disagreed on a whole load of things. And simply appealing to the tradition without stating what precisely the tradition does say seems to slightly odd.

In any case, I know that this is a blog which is a confined media. I know also that you do explain some of the distinctions that a A-T philosopher makes as against the moderns. And I know that your books do so (brilliantly I might add) as well. So I can see why you do it, I can just see why some people might find it annoying. It may come across as 'I, Edward Feser, have some secret knowledge, it is the A-T tradition', and it might seem doctrinaire.

Anyway, great post none-the-less.

Holopupenko said...

Crude:

Even the lead up to the question you pose is problematic. IF God is understood as utter actualization (no potentiality whatsoever) and unbounded in all senses of perfection, completeness, oneness, etc., then even the "theistic personalist" vs. "classic theist" distinction you forward-loads the question.

The only way one would be able to distinguish between such "gods" is for one of the "gods" to have something in lesser degree (or not at all) than the other "god". But if this is the case, then the neither of those "gods" is God understood above.

This is why, perhaps too strongly and perhaps incorrectly, I reject Plantinga's "theistic personalist" notion of God on its face. It strays to close (for me) to suggesting we can positively know something about God in the univocal sense. No, because the only univocal knowledge we can have of God is through the via negativa (God is not this or that finitely-predicated thing)--which, at the end of the day, ensures He's not just another being among beings. The only positive knowledge we can have of God is through analogous language (God does not exist; God is Existence itself, etc.)--which ensures God is not a cause among causes but Universal Cause.

George R. said...

Here’s another flaw in the ontological argument:

The first thing we can know about God is that he exists. Now since we can only arrive at the knowledge of what we don’t know by means of what we do know, we cannot arrive at the knowledge of God’s existence by means of things we can’t yet know, such as that He is the “greatest being that can be conceived” or “a being which is omniscient, omnipotent, and morally perfect in every possible world.” And if the “greatest being that can be conceived” or “a being which is omniscient, omnipotent, and morally perfect in every possible world” must exist, this, too, can only be learned by means of things we already know. Therefore, the proof of God’s existence is necessarily a posteriori, there’s no way around it.

I think this is pretty much St. Thomas’s argument.

Another bad thing about the ontological argument is that atheists and agnostics can detect the flaw in it, and then claim that the “proofs of God’s existence” are sophistical.

Anonymous said...

This recent series of posts brings to mind something I read a while back on Russell Blackford's blog, an atheist, a philosopher, and the author of a recent, popular atheist manifesto entitled "50 Voices of Disbelief." Apparently the idea is, if ontological arguments for theism fail, all arguments for theism fail:

http://metamagician3000.blogspot.com/2010/04/marke-but-this-flea.html

"The Argument from Contingency gets nowhere, and it acquires whatever thin veneer of plausibility it might have only by relying covertly on archaic concepts of necessary being and the like. Alas ... you can have a concept, and attach to it the high falutin' formula "absolute actuality", but you do not thereby guarantee that the concept is instantiated in the real world. Thus, God would be just as contingent as anything else.

You'd think the numerous historical failures (including that of Goedel!) to get an ontological argument off the ground for more than a few confused seconds would provide enough warning as to why contingency arguments inevitably come crashing down in flames. If ontological arguments fail, you don't have a necessarily-actualised concept of God, but if you don't have that you don't a decent argument from contingency. The Argument from Contingency is parasitic on the Ontological Argument."


Snark aside, I think Kant said something similar in his 1st Critique. It sounds extremely flimsy to me, but I wondered what you more philosophically-educated posters would think of this supposedly wholesale objection to theism.

Crude said...

Holopupenko,

Even the lead up to the question you pose is problematic. IF God is understood as utter actualization (no potentiality whatsoever) and unbounded in all senses of perfection, completeness, oneness, etc., then even the "theistic personalist" vs. "classic theist" distinction you forward-loads the question.

Well, I had the feeling even using the word "God" there was going to cause trouble. Take out the word "God" if that helps anything and say 'The being envisioned by theistic personalists' versus 'The being envisioned by classical theists' - any better?

Eric said...

Professor Feser, what do you think about the claim that even if the stronger versions of the ontological argument fail to prove God's existence, they at least demonstrate that *if* God exists, he exists necessarily, and if God does not exist, he cannot possibly exist. If the argument does this much, it puts the atheist/skeptic in a difficult position: he can't merely argue that God happens not to exist, but that God's existence is impossible. (If I remember correctly, van Inwagen has said as much about the stronger versions of the Ontological argument.) Or, to put it another way, it shows that the atheist must grant that if God's existence is even possible, then God exists.

Domini Canes said...

@Crude: No, I don't believe that the God of classical theism and the "God" of theistic personalism can coexist, for both affirm to their respective Divinities attributes instantiable only once, for example, being the cause of all aside from oneself, existence a se, etc. For both to exist, therefore, would require that two things partake of an attribute which can be partaken of only once, which is absurd.

David said...

Domini Canes: That Socrates did not teach Plato in a world W is, given his transworld identity, true of Socrates in every world. No need of postulating some "meta-world" arises.

Well, there is of course the outside realm of thought from which we are surveying all these possible worlds, but even allowing for that, your example is different from saying it's possible that there is a maximally-excellent being, where "maximal excellence" already contains a notion of "possible". It's not a statement about a possible world w1 or w2, ... but a statement about the entire set of possible worlds, W. In one set of worlds, W1, there is a being with all the same properties in each wn, but there is another set, W2, where that being has certain properties in only some worlds. Then we're trying to figure out which set Wn is the set of, er, actual possible worlds.

Domini Canes said...

"Well, there is of course the outside realm of thought from which we are surveying all these possible worlds[.]"

I'm not quite certain what you mean. If you mean that we somehow are outside all possible worlds when we contemplate them, that that is simply false. Socrates considers a possible world W from α (the world that actually obtains), not some no man's land in between the two. Nor do we somehow get an inter-world perspective in considering other possible worlds: to say that "W is the best of all possible worlds" is merely to say (axiom S5) that "in α, W is the best of all possible worlds."

[Y]our example is different from saying it's possible that there is a maximally-excellent being, where "maximal excellence" already contains a notion of "possible". It's not a statement about a possible world w1 or w2, ... but a statement about the entire set of possible worlds, W.

Not in the way Plantinga is speaking of the matter. For Plantinga, to say that p is possible (♢p) is just to say that "in some possible world W, p." Now, if you object to this view of modality, you are perfectly within your rights and in excellent company. But then you are attacking possible worlds semantics as such and not Plantinga's application thereof.

Domini Canes said...

Whoops! Ignore that appeal to axiom S5.

Dianelos Georgoudis said...

I have a problem with premise (2). After all, a world where nothing exists is possible.

TheOFloinn said...

I have a problem with premise (2). After all, a world where nothing exists is possible.

But since "world" in this sense is the set of everything that exists, without a single existent there would be no world. Even if you allow the antique Newtonian notion of and empty absolute space - deSitter space - you would have space existing.

Dianelos Georgoudis said...

TheOFloinn:

You write: “ But since "world" in this sense is the set of everything that exists, without a single existent there would be no world.

Rather, every possible world corresponds to a set of existents. The world where nothing exists corresponds to the empty set.

More precisely, a world is defined by the set of all true propositions in this world. And a world is possible when this set of propositions does not entail a logical contradiction. The world Z where nothing exists is defined by a rather small set of true propositions, namely {“Z exists”, “For all X different than Z, X does not exist”}, and is thus provably possible.

Even if you allow the antique Newtonian notion of and empty absolute space - deSitter space - you would have space existing.

Space does not exist in Z. Neither time. Nor numbers. Nor God.

Do you see any reason why Z is not possible?

TheOFloinn said...

a world is defined by the set of all true propositions in this world.

Space does not exist in Z. Neither time. Nor numbers. Nor God.

Do you see any reason why Z is not possible?

TOF
Apparently, the propositions {"Z exists", "For all X different than Z, X does not exist"}exist in this otherwise empty world. Also the empty set itself is a thing (though granted its topology is rather coarse!) And since Ø≤Ø there may be other problems, too.

However, I was thinking "physically possible" and you are thinking "imaginable," so we are as ships passing in the night.

Dianelos Georgoudis said...

TOF,

You write: “Apparently, the propositions {"Z exists", "For all X different than Z, X does not exist"}exist in this otherwise empty world.

Not all. These propositions are about Z but exist in our world and not there. In Z there are no propositions, no beliefs, and of course no minds to hold them. There is nothing there.

Domini Canes,

You write: “What is more, you are committing the mistake of confusing the imaginable with the really possible ("see The Last Superstition for details").

I’ve read TLS but I don’t see its relevance for it’s not like A-T applies there. What is imaginable may not be really possible, on the other hand my argument is not “I can imagine Z therefore Z is a possible world”. Rather my argument is “I can define Z, and that definition implies that there are no contradictory true beliefs about Z, therefore Z is provably a possible world.”

My point is this: If “God exists necessarily” is true then it cannot mean “God exists in all possible worlds”, because the latter is false as there are possible worlds in which God does not exists.

John said...

great blog! i would love if someone would explain how it is that plantinga’s ontological argument is possibly an improvement upon anselm—or eric, if you or van inwagen have more to say about this “if possible god, then actual god” argument, since that seems to be where plantinga points.

everything feels familiar until the crucial step six, where he makes the leap that if something is impossible in one possible world it is impossible in every possible world. i don’t know why this follows. why not just say let’s imagine a world where it is impossible that there is not a god, and then skip to step six? the argument doesn’t really depend on the elaborate definition of god, does it? if we imagine a world where it is impossible for god to be any color other than red-orange, then because what is impossible in one world is impossible in all worlds, god must be red-orange. okay.

i like that he shifts emphasis away from anselm’s derivation of a necessary being by making his necessary existence an acceptable feature of a possible world, but he still requires anselm’s contingency when he makes the transfer back to our world. it is hiding in step four, where he defines the being a second time, this time with the important corollary that true greatness implies greatness everywhere. he outsources it to planet W, but how well does it travel?

shifting the location of the concept to a possible world might make it feel more domesticated, but if 2 + 2 must still = 4 in every possible world, then i’m not sure how a being formed from a concept slips beneath the radar. for me and my possible worlds, i’m actually fine with it, and also fine with 2 + 2 = 5, for that matter, but neither of these are capable of making the transfer back to our world, i don’t think, even if our world exists merely as one among possible worlds. but by the time we get to step six we are not concerned with familiar ontological leaps (that thing nothing greater than which can be imagined, or that perfect being whose essence must include existence), because we are busy trying to figure out if the notion of the possible is transferable to our world. and it appears that it is! and so it should be! after all, we exported it in the first place by playing the game of possible worlds. in fact we don’t need the imaginary world to say it at all: “it is possible that there is a being of maximal greatness.”

that’s actually the statement that we get, i think, from the proof, and we can make such statements about anything right here (no reason for mackie to get upset about the rationality of wagering for the impossibility of god in another possible world!). “it is possible that unicorns exist” is always a safer bet than “it is impossible that unicorns exist.” but this says nothing about the likelihood of our finding a unicorn. it only says very unhelpfully that it is possible; and it also says something about the categories possible and impossible, the former being open-ended and the latter being categorical. this is the strength i see in feser’s response to mackie (and i agree that okham’s razor is irrelevent here): that it is always more rational to say “maybe” than it is to say “no chance.” but still, imagining a necessary unicorn on planet W doesn’t make it necessary here or even probable!

Leo Carton Mollica said...

@John:

everything feels familiar until the crucial step six, where he makes the leap that if something is impossible in one possible world it is impossible in every possible world. i don’t know why this follows.

While it might seem strange, the rule that modal properties are necessary and thus common to all worlds (known as Becker's postulate) is pretty well accepted by contemporary modal logicians. I believe that Plantinga discusses it in The Nature of Necessity, which I heartily encourage you to read if you're interested in his ontological argument or modality more generally.

John said...

leo, thanks for your help, but i’m still confused. when you say “modal properties are necessary and thus common to all worlds” does that mean simply that having a modal property of one kind or another is necessary for anything in any world, or that actual existence as a specific modal property is necessary? surely not that latter, or else everything we can imagine would necessarily exist, no? this is very different from saying that necessity somewhere is necessity everywhere (2 + 2 = 4). my objection is less to becker’s postulate and more to its use here to shift focus from anselm’s unfounded premise, now the first half of plantinga's redux: god is necessary in world w.

it’s no easier or more difficult for me to imagine god being necessary in world w than to imagine 2 + 2 = 5 being necessary in world w, since it sounds like either claim would need to be necessary here first, or somewhere. outsourcing to w would be a lot more fun if possible worlds were a little less like our own! necessity as a modal property, before it can claim universality, must claim necessity, and probably by someone from this world (like anselm). in other words if necessity is a consistent modal property from one world to the next, then one good test for the necessity of something on w is to see if it is necessary here! please correct me if my logic is fishy.

GRACEMAN said...

Faith without reason is irrational; reason without faith is unbelievable.

Rick Taylor said...

For it is simply implausible to suppose that, other things being equal, the key premises of Plantinga’s argument and its “no-maximality” rival are on an epistemic par. To see why, consider the following parallel claims:

U: There is a possible world containing unicorns.

NU: “No-unicornality,” the property of there being no unicorns in any possible world, is possibly exemplified.
-------------
This is not at all a fair way a fair way to compare the two claims. More accurate would be

U: A maximally excellent being is present in every possible world.

NU; There is at least one world in which a maximally excellent being doesn't exist.

Clearly the doubter is making the weaker claim. Plantinga makes it appear otherwise by using a verbal sleight of hand, defining a Maximally great being to be a maximally excellent being that exists in all possible worlds, and then pointing out we only need to show it exists in one possible world to show it exists. Well yes, but you defined an MGB to be a being that exists in all possible worlds. That doesn't somehow make it's existence in all possible worlds more reasonable.

spandy said...

2+2=5 When there are extremely large values of 2 (2.49 , 2.51) laurance krauss was wearing a t-shirt which , he had made, at a debate