Thursday, July 30, 2009

Aut Deus Aut Homo Malus and Division-Based Arguments

At AskPhilosophers.org, Alexander George and Peter Smith both answer a question about aut deus aut homo malus arguments, and in doing so miss what I think is a serious opportunity to discuss division-based arguments. The result is that the answers they provide are entirely unhelpful, and appear to be based on unsubstantiated assumptions about the argument in question.

Aut deus aut malus, like every other dilemmatic or trilemmatic argument, works by establishing a space of possibilities and dividing it up. We'll set aside inductive versions of division-based arguments for now, and assume that we've decided to do an eliminative one. There are several ways you can go with the argument from here, depending on what you are actually trying to do. In giving such an argument, you could be arguing ad hominem, in Locke's sense of the term: that is, the conclusion you reach is not about the true conclusion but the conclusion the other person should reach in order to be consistent. When the division has this purpose, you argue not on the basis of what you think is established but on the basis of what the other person has conceded or is assumed to concede. On the other hand, you could be arguing ad iudicium, in which case you are not arguing from the principles of the other persons but from things that are deemed to be facts or certain principles. In either case, you can run your own eliminative argument, rejecting possibilities until you have reached one, or you can simply leave the division at that, and let people sort themselves out on their own and, in effect, do their own eliminations.

What this shows us is that it is impossible to evaluate division-based arguments without regard for their context. We see this at every stage:

(1) Field of Possibilities: Are we considering all logical possibilities? Or are we considering only 'live options', actual claims people have made? Or are the possibilities something else? This can only be determined from the context of the argument.

(2) Principles of division: Likewise, when we get to the question of why we are dividing up the field of possibilities in the way we are, we have to know what we are trying to do. If I am trying to determine whether any penguins can fly, there are a number of ways I can divide up the penguin population; some of these ways of dividing the population up will make my argument easier to make and some will make it harder. If I am considering not just actual penguins but also any possible penguin, the principles according to which I'm dividing up the possibilities become even more crucial. In order to determine whether the principles of division are appropriate, and often even to determine what they are, we need to look at the context.

(3) Type of Argument: Arguments that are simply arguments that one's opponent should on their own principles concede some point are very different from arguments that everyone should, on principles that are true and established, concede a point. They have different standards of success -- indeed, different kinds of success entirely. And which is the relevant one will, again, depend entirely on what someone is doing in context.

(4) Grounds for elimination (where relevant): In giving reasons for eliminating options, you may be doing several things. You may, for instance, be trying to present a rigorous stand-alone argument. You may be trying to summarize a complex set of evidential arguments that are dealt with in more detail elsewhere. And you may be giving defeasible presumptions that would have to be overcome by the opposing side in order to make their case. Each one of these yields a different kind of conclusion (although each one might be stated in exactly the same words); each one has to be evaluated differently; and we can only determine which one is in play by looking at the context.

A corollary of all this is that we cannot assume that any two verbally similar cases of aut deus aut homo malus are actually logically similar until we have determined that their contexts force them into the same shape, so to speak. This type of argument affords some fairly easy examples of this point, in fact. The usual source for these kinds of arguments is C.S. Lewis, who uses them in several places. But Lewis always presents such arguments ad hominem (again, in Locke's sense of the term) and is often very careful to specify the sort of person he has in mind. He employs it as an argument to show that a certain type of position is implausible and apparently not consistent with the way we usually reason about people. But just as people who take up Pascal's Wager have a bad habit of dropping all the qualifications and limitations put on it by its original context without first checking to see whether they can really do so, so people picking up the aut deus aut malus argument have a bad habit of dropping its context, and all the constraints context places on it, without regard for how this changes the argument. For it does change the argument; they will be different arguments, however verbally similar. Some criticisms of one may apply to the other, but some may not, so it is essential in evaluating the argument that we not be fooled by verbal similarity into treating them as if they were the same argument. We cannot assume that what applies to Murray and Rea's version automatically applies to Kreeft's version, as Smith does, or that what is said of either of these automatically applies to any other version. The context for each will have to be examined to see what the arguer is trying to do with the argument. Aut deus aut homo malus is a very diverse family of arguments, not a single argument; and this is a feature it shares with many kinds of eliminative arguments.

It is this, not George's and Smith's approach (which consists of little more than making a series of vague assertions denying various parts of the argument, and in Smith's case, of making the silly assumption of logical similarity on the basis of mere verbal similarity), that gives you a reasonable assessment of the argument. The questioner isn't very precise about the original context of the argument, dropping most of the relevant details. I'd imagine that it's from Kreeft's Between Heaven and Hell, but it could be from another work. And that difference could be significant; in You Can Understand the Bible, for instance, the point of the argument is simply to show that in the Gospel of John the interpretation that Jewish leaders make of Jesus' words as presented in that book is unsurprising and reasonable; in Between Heaven and Hell it is presented as part of a discussion between two interlocutors and thus (1) is not a stand-alone argument; and (2) would have to be interpreted in light of the way in which those interlocutors are represented in the dialogue as a whole; and in Ecumenical Jihad he uses it simply as a parallel or analogue for a different argument. I forget how it's used in the Handbook, and can't find a relevant selection anywhere online; but if that's the source, its context would be crucial as well. The arguments all belong to the same family; but they are discernibly different arguments.

The fact that context is so key to pinning down even the full logical features of the argument is a reason why dilemmas, trilemmas, and other eliminative arguments can be very difficult to handle: since we have to go digging in order to come up with the version of the argument with which we are actually dealing, it's very easy for prejudices, both rational and irrational, to lead us to read things into the argument that are not there, or, at times, even read things out of the argument that are there. It's not a time to be sloppy and cavalier, but to recognize that, however simple such may seem (and they often do seem simple), they have many working parts and draw at every stage on the actual context in which the argument is given. Evaluating both the parts and the whole is not something that can be done with sweeping generalizations; trying to do it that way will leave us with the result we have here: vague, limping, weakly answers instead of robust and precise ones. And since eliminative arguments are actually quite common, we need to have the tools in hand to give the robust answers.