Sunday, January 02, 2022

Ought and Ought

 Sam Harris recently had a tweet on the claim that you can't get an 'ought' from an 'is':

There were quite a few really clumsy attempts to dunk on this. It is in fact a legitimate question. Unsurprisingly, none of the dunkers managed actually to get a coherent argument against it in tweet form. The closest, however, is interesting, because I've seen it before; it fails miserably, but it seems to have some purchase even in some philosophical quarters. The attempted response is something like this: The question is badly formed because it equivocates between the 'moral' reading of 'ought' and the 'epistemic' reading of 'ought'.  The claim that you can't get an 'ought' from an 'is' applies to moral oughts, not epistemic oughts.

It should be obvious that on its own this is utterly inadequate, since Harris explicitly asks a question about why two kinds of cases are being distinguished, and the totality of this answer so far is that it is because they ought to be distinguished. This simply takes the apparent problem (that in case A people use the is-ought prohibition and in case B they don't), describes it, and treats it as the solution to the problem. Obviously we need to know why this distinction is made in such a way that at least an analogous is-ought prohibition doesn't carry over, particularly given that we use analogous vocabulary to talk about them. And when you start pressing this, while you can distinguish 'moral' cases from 'epistemic' cases, it's very difficult to motivate a distinction that would be this sharp; the most natural distinctions between the two are purely practical, not fundamental. And, in the finest analytic tradition of explaining things that seem more clear by means of things that are less clear, there is not actually much in the way of a consensus view on how epistemic oughts work, particularly one that would really explain the disanalogy. For instance, sometimes you find people trying to build the distinction on the ground that epistemic oughts are "role ought"; but some moral oughts are role oughts, and, indeed, there are entire ethical systems in which most moral oughts are role oughts, and very few in which none are.

There are in addition reasons to worry about concluding too quickly that the two kinds of ought are so different. It's pretty clear that historically, we get 'epistemic' oughts on analogy with 'moral' oughts to begin with. Epistemic oughts don't look like every kind of moral ought, but they do look and act very much like abstract moral oughts -- i.e., moral oughts like 'People ought to get along' or 'It ought to be the case that people in need can get help'. You can model both fairly well with a D-type modal logical system, for instance. But such logical systems have deontic necessitation, in which case even at a purely formal level, before you even start talking about what 'ought' means, you get violations of the is-ought prohibition. There are plenty of ethical and also epistemological approaches in which at least some epistemic oughts are also moral oughts, and at least some moral oughts are also epistemic oughts. (This is true of classical utilitarianism, moral positivist deontologies, Aristotelianism, Stoicism, and Confucianism, to take just a few examples.) The list of reasons why the distinction, even if it can be made, is just not the right kind of distinction for answering the question could be made quite a bit longer.

It's an interesting question what social conditions lead to so many intelligent people treating as self-evident a principle that sits so poorly with logic, ethics, and perhaps universal human practice, that is not actually accepted by the philosopher to whom it is most attributed (Hume), that requires endless epicycles in order to deal with very common-sensical counterexamples, that makes it practically impossible to figure out what moral obligations are, and that no one can ever actually defend very well. People have tried to overthrow the principle of noncontradiction for less. But so it happened, and so it still to a great extent is.