Friday, March 01, 2013

On Hart's Kantian (?) Argument

There has been some discussion of Hart's recent salvo against natural law theory. I was going to say something about it, but I confess I'm utterly baffled by the argument. Hart argues that natural law theory cannot provide either categorical or hypothetical imperatives adequate to morality, but the two branches of the argument don't seem to cohere. His argument on the categorical imperative side is purely Kantian. But his argument on the hypothetical side is a sensible knave problem put in such strong terms that it would, if problematic for natural law theory, be equally a problem for the Kantian, who also cannot persuade Nietzsche. Perhaps this is why he also sometimes uses language suggesting not Kantianism but fideism? But it seems a little harsh to insist that a theory of practical rationality, which is what natural law theory is, can only be adequate if it is a completely compelling Kantian refutation of Nietzsche. This is a very specific thing to demand, and I'm not sure why it's being demanded.

Any natural law theorist already rejects the Kantian assumptions of Hart's argument. While a natural law theorist could have a sharp divide between categorical and hypothetical moral judgments, such a divide only makes sense if you concede Kant's analysis of judgment into three irreducible kinds. (The three kinds are categorical, hypothetical, and disjunctive. Disjunctive judgments don't come in at all here because Kant insists that moral judgments must have necessity, and his account of disjunctive judgments makes them concerned only with possibility. Hypothetical judgment can have a kind of necessity, but Kant argues that it doesn't have the right kind.) Kant's theory of judgment, while essential for understanding Kant, is not something that is widely shared. Most people usually treat hypothetical judgments as reducible to categorical judgments. Logically this is quite easy to do, actually, since all conditional propositions can be reformulated as categorical propositions and vice versa; the predicate calculus depends on this fact, since without the conversion it would not be able to handle basic syllogisms. (Kant's theory of judgment is not purely logical; it is in a sense intended to be pre-logical or meta-logical, so this disparity wouldn't necessarily be an issue for Kant himself.) Hart's argument that you can't get a categorical 'ought' from a conditional is certainly Kantian. But the question immediately arises: "Why would a natural law theorist be committed to a Kantian theory of moral judgment?"

Likewise, as I keep pointing out to people, it has never been true, as a matter of logic, to say that you can't get an 'ought' from an 'is'. Logically it's trivially easy to create 'is' statements that yield 'ought' statements. The standard forms of deontic logic, for example, all have what is called O-necessitation, or deontic necessitation; and it is equivalent to saying that you can get an ought from any logical theorem in the system (even if it is an 'is'). This is a very controversial rule, mostly tolerated because it massively simplifies deontic logic, making it well-behaved, but the point is that you can always get an 'ought' from an 'is' whenever the relevant modal inference rules allow you to do so. And in any given case, it's not a purely formal or logical matter what the relevant modal inference rules are.

There are, though, obviously specific contexts and obviously very specific sets of assumptions on which the claim would make sense or even be true. What about those? It made sense for Hume to put forward the dictum, because Hume argued for it on the basis of his account of relations, and did so against a position that made 'ought' a relation between ideas (which is why he talks about 'ought' as if it were a copula rather than a modal operator). But there's no reason to accept the Humean version of the dictum unless you accept Hume's insistence that the Humean theory of relations covers all legitimate kinds of relations, which few people do, and even if you did accept it, it would arguably only work for accounts that require that 'ought' be a relation between ideas, although this is a bit more debatable. Kant, who takes up Hume but whose Hume is a generalized Hume, Hume as idealized empiricist skeptic, also can put forward something like this claim. But in Kant it's grounded by the distinction between the phenomena and the noumena, or (to put it in different terms) by his account of intuition/experience and its limits. And again the question arises: "Why would a natural law theorist be committed to a Kantian account of experience?"

A Kantian theory of reason rules out any classical natural law theory, to be sure; but it obviously does so because natural law theory is a rival theory of practical reason. On Kant's account of morality, no moral law can have any empirical component, even if the empirical component is universal to human nature. This is precisely what Kant says about the desire for happiness: it is universal, but incapable of forming any part of a moral law because it is entirely concerned with pleasure, which presupposes the actual existence of something, which can only be established by intuition giving the concept an object, which means it is not a priori, which means it is not necessary, which means that it cannot ground a moral imperative in the proper sense of the word. It is certainly true that if you accept this entire line of reasoning you are committed to rejecting any classical form of natural law theory, but every single step in this chain is as controvertible as anything in natural law theory itself, and not a single one need be (or should be) accepted by a natural law theorist. Natural law theory insists that all moral judgments must be based on what is good; Kant insists that all judgments about the good must be based on moral judgment established independently of any idea of what is good, which is why moral judgment for Kant is based on the purely formal capacity of autonomous will to be autonomous. Natural law theory insists on a moral role for specifically human aspects of practical reason; Kant explicitly insists that any such specifically human aspect cannot have any moral role at all (unless by 'human' we just mean 'pertaining to any kind of rational being'). Kant denies any moral role to prudence because it deals with contingent matters; natural law theorists have historically insisted on the moral importance of prudence because it deals with contingent matters. Hart's (apparent) argument that Kantian considerations rule out natural law theory is surely right, but this is just because what we usually think of as natural law theories are anti-Kantian theories of practical reason.

In other words, it is true that natural law theory is not Kantianism. But where does that actually leave the natural law theorists? Short of a proof that Kantianism is the way to go -- which Hart doesn't seem to hold, either -- it seems to leave them exactly where they were.

I have similar kinds of questions on the hypothetical side of the argument, as well as with Hart's apparent fideism. And there are problems with putting so much emphasis on persuasion when talking about theories of rationality, which are concerned with things that can be true or false regardless of whether people are persuaded. (The fundamental idea of natural law theory is just that there are practical rational principles based on the notion of good, which are the practical counterparts of the logical principles we use in theoretical contexts; we don't usually judge logic on the basis of whether people are persuaded by arguments using logical principles. If someone doesn't want to be logical, that's a problem for them, not logic. You might look into what you could do in addition to logical reasoning, but you don't say, "Well, I guess that's it for logic." So what would make the proposal of natural law in practical matters so different from the case of logic in theoretical matters?) But I'll leave it at this, because, as I said above, I don't actually understand what the argument is.