Friday, March 11, 2011

On Ontological Arguments

I haven't been keeping up recently on what has been going on in the atheist blogosphere, but apparently there's been some discussion of the ontological argument; Sean Carroll gives links and his own discussion.

One of the things I think is interesting is that not a single one of the posts, and this includes Sean's, which however at least makes an effort, shows a real sign that the authors know what they are talking about. One of the tell-tale signs of this is the repeated talk of "the ontological argument" as if there were one, which leads them repeatedly to conflate elements of completely different arguments. Sean, for instance, outright says, "Anselm assumes that perfection is possible, and that to exist necessarily is more perfect than to exist contingently," which is quite the feat given that Anselm's argument doesn't involve any premises concerned with perfection or contingency or necessity. To be fair, he's misled here by how he's reading Suber, who he is following: Suber is in fact giving a slight restructuring of Hartshorne's version of the ontological argument, which is itself an attempt to rework Anselm's argument in Proslogion III. It's a worthwhile version of the argument to look at, since Hartshorne's work is one of the causes for re-awakened interest in the ontological argument and for the realization that the Kantian criticism makes assumptions about ontological arguments that do not fit all ontological arguments. And Hartshorne's version can be considered broadly Anselmian in inspiration (hence Suber's mentions of Anselm in the course of laying out the argument, which should, however, be more qualified than they are). But the fit between Hartshorne and Anselm is pretty loose; as Hartshorne himself would put it, it's at least the difference between a 'classical' and a 'neoclassical' view. We are, by the time we get to Sean's post, at several removes from anything purely Anselmian. And, indeed, the argument no longer has anything more than very broad analogies to Anselm's original; the terminology and course of the argument has become (broadly) Cartesian rather than Anselmian, but without any of the Cartesian background that makes Cartesian arguments non-arbitrary.

Likewise, Sean asks, "What exactly is this 'perfection' whose existence and necessity we are debating?" but never seems to consider that when faced with such a question the best thing to do might be to go to the evidence and see what, exactly, this 'perfection' is that is being debated. That he actually took the trouble to rely on Suber is a step up from most of the discussion; but while he comes closest, he still seems not fully to understand the idea that serious examination of arguments requires looking at evidence -- the evidence about what is motivating their use, the evidence about what their logical structure is, and the evidence about what undergirds their premises and assumptions.

But the further sign of trouble is that none of the criticisms bear much similarity to respectable criticisms of ontological arguments. Sean's, for instance, which, again, at least makes an effort, begins with perplexity about what 'perfection' might be, makes up some things about what it might be (again without any evidence that any of it is relevant in the context of this particular version of the argument, and the fact that he ultimately reduces it to 'good-sounding quality' is pretty much evidence in itself that he doesn't have any evidence that he just doesn't happen to mention), and dismisses it all with a vague amateur psychological diagnosis. It is, to be fair, an intelligent attempt to appeal to ignorance, simply make up some possible interpretations of the terms used, and explain the appeal of the argument without any appeal to evidence, but that will only get one so far. For a first impression, fine; for conclusive dismissal, not so much.

There are serious problems with taking ontological arguments as proofs; they mostly vary quite a bit among the different varieties and subvarieties because we call 'ontological arguments' arguments using very different concepts, assumptions, and logical principles, on the basis of a broad sense of similarity.* But they do exist. What strikes me, though, is how confident these atheists are in their dismissal, which they back up by very weak-tea kinds of refutations (unrigorous, uninformed, and largely floating free of actual evidence). Reasonable people can be misled by ontological arguments; from the fact that they are misled, however, it doesn't follow that those who are not misled are particularly reasonable. Jerry Coyne tries to analyze and evaluate the argument with a cartoon, for goodness' sake, and while this may be a weak joke, it's clear that not all the commenters treat it as such; it's like watching kindergartners try to explain the issues with Hilbert's second problem. But such is what one gets when one rejects arguments wholly because one thinks they have to go wrong somewhere and not because of any clear understanding.

________
*Cosmological arguments and design arguments, especially the latter, despite considerable variation all share several broad positive features with each other; ontological arguments are all a priori arguments for existence in some way, but that seems the only common factor. Part of this is seen in the fact that none of the characterizations standardly given of ontological arguments fit all the varieties of them. Suber, just to give an example, says "The gist of the argument is to prove that God exists from the mere possibility, or mere idea, of God's existence. It asserts that God's essence implies God's existence." But none of these three starting points, possibility, idea, and essence, are the same thing, and their actual relations to each other are highly controversial topics. So all Suber has really done is give a disjunctive characterization of the group -- ontological arguments start with possibility or idea or essence. And this is not uncommon. The problem here, of course, is that the label for the group comes through Kant, but Kant only had close knowledge of one subvariety of the large group of arguments that are now called 'ontological arguments', Wolffian standardizations of the Leibnizian modification of the Cartesian version, with perhaps some knowledge of a few other closely related variations. Since then other arguments have been placed under the label due to some resemblance or other to this or that variety of the argument, with the result that it no longer labels a coherent group. (Something similar has been happening with 'cosmological arguments' and 'teleological arguments' and 'moral arguments', as well, but not so swiftly. Of the three, the 'cosmological argument' label seems to me to have reached the point where one must get very general indeed to characterize the whole family, while 'moral argument' has stayed, so far, more tightly focused on a small group of arguments, and 'teleological or design argument' seems to me to be in an intermediary stage between the two. If this is in fact so, it raises the interesting history-of-philosophy question of why the labels have deteriorated at different rates. Some initial guesses, again, if we really do have this differential deterioation: part of this may just be due to what philosophers have taken the most interest in, and part may be due to the degree to which discussion of the argument has detached itself from historical anchors in order to treat arguments as freestanding. The latter would encourage the swift multiplication of variations without constraint to a particular form or type.)