If I say that I ought to do something, what am I actually saying? Let's take a fairly straightforward example. I want to be downtown by a certain time, so I say, "I ought to be at the bus stop to catch the 1M before 7:15." It's clear enough that what I'm really talking about here are ends and means: I have an end view (being downtown by a certain time) and I have a means for it (taking the 1M from the stop before 7:15). By saying that I ought to catch the 1M before 7:15, I'm saying that the means is the way to get to the end. It's clear enough that this is a matter about which I could be right or wrong. You might reply: "Oh, no; there is no possible way that taking the 1M will get you downtown by the time you need to be there. You ought to take the 174 instead; that's the only route that will get you there in time. It arrives at that stop about 15 minutes before the 1M." To which I might reply: "You're right, now that I think of it. I ought to be at the stop to catch the 174 by 7:00." Given that this is so we can see immediately that it can be trivially easy to derive an 'ought' claim from an 'is' claim: given the end, claims about the means to the end (which are 'is' claims) can yield 'ought' claims. It's important to see, too, that the "given the end" restriction is equally unmysterious. In the sort of context in which we talk about 'ought' and 'should', the end is merely establishing a problem to be solved. From 'is' claims about possible solutions to this problem (e.g., the 1M does not arrive by the required time, only the 174 is early enough and fast enough to arrive by the required time) we derive claims about the optimal solution to the problem (e.g., I ought not to take the 1M but instead ought to take the 174). That is all there is to it. In scenarios like our bus schedule scenario, 'ought' is a very ordinary thing, and 'ought' can easily be inferred from 'is'.
Making the 'ought' moral doesn't change anything about the 'ought' itself. Only the kind of end changes. Some ends are temporary and limited (like the one in our bus scenario) while some are general and standing (like living a great life). And there are no doubt other differences. The difference between a moral 'ought' and any other kind of 'ought' would be due entirely to the difference in the ends in question. Since ends are pegged as ends by being considered good in some way, this means that the differences between different kinds of 'ought' claims reduce down entirely to the differences between different ways in which things can be classified as good.
So I would claim, anyway, and at an abstract level this is the structure we find in Hume. Hume's account of obligation depends on his account of how we recognize good and bad. The differences between different kinds of 'ought' claims reduces to differences between different ways in which we can recognize something as good in some way. Hume makes a distinction between 'natural obligations' and 'moral obligations'. What Hume calls 'natural obligations' are those that follow from recognizing something as good for us, which is determined by our interests. What he calls 'moral obligations' are those that follow from recognizing something as good even without regard for our interests: "'Tis only when a character is considered in general, without reference to our particular interest, that it causes such a feeling or sentiment, as denominates morally good or evil" (T 3.2.4, SBN 472). (Hume uses an interesting analogy: recognizing something as morally good even though not in our interests is like recognizing that a man has a beautiful voice even though we can't stand the man. It can be done; indeed, much of our moral life is based on the fact that, no matter how often we let our interests bias our moral judgments, we can and at least sometimes do recognize that the two are irreconcilable.)
Hume is also right that is far more reasonable than the rationalist idea that obligations are relations or rapports. A problem with the rationalist view is that if I recognize (to use an example from Malebranche) that the soul is superior to the body, that in itself tells me nothing about my obligations. Does it mean (as Malebranche takes it to mean) that we should everywhere and under every circumstance prefer the things of the soul to the things of the body? Or does it simply propose a useful ideal, like saying that an A paper is superior to a B paper? The relation is not enough, even if it exists; something else must convert (so to speak) the relation into an end or good that will establish the right kind of obligation. In fairness, there are rationalists who recognize this. Malebranche himself, for instance, recognizes it: while the content of our obligations is determined by relations of ideas perceived in divine Order, what makes these relations of ideas obligatory is divine Love, which is, a love of Order. I won't get into the account here. But what it means is that the 'obligatoriness', so to speak, is not deriving from the relation at all.
Where Hume goes wrong, however, is in thinking of this in terms of a kind of pleasure. Hume would not have been alone in the period in thinking of ends entirely in terms of pleasure. It was a widespread view, and is found even among Hume's rationalist opponents -- Malebranche, for instance, had said that "pleasure is good and pain evil, and...the pleasure and pain that the Author of nature has attached to the use of certain things makes us judge whether they are good or evil" (LO 359). But there are serious problems to any such view. For instance, many cases of pleasure and pain can't be explained unless we allow them to be explained in terms of our perception of things as good or bad. The same gesture can induce laughter or fear depending on whether we see it as a good thing (like good-natured play) or a bad thing (like an attempt to murder us). Also, whatever link there may be between them, the like between pain and our recognition of evil, or pleasure and our recognition of good is not straightforward: pain and pleasure can pass over into each other, pains like the ache of exercise can be regarded in some cases as good (as the saying goes, pain is proof you are alive), and some pleasures can be regarded as bad. Indeed, this follows from Hume's own view; you can be pleased at a pain, pained at a pleasure. But then the relation between the two pairs, pain and pleasure, good and bad, is not going to be a straightforward one. It's much simpler to take pleasure and pain as either oblique to the question of good and bad, or as (at least sometimes) following on our recognition of something as good or bad.
There are a few more things to be said, but this is enough for a post.
By the way, I realize that I've not been explicit about references, using standard abbreviations in a number of posts. Here's a key:
T David Hume, A Treatise of Human Nature. The Treatise is divided into three books, each of which is divided into parts. Each part is divided into sections, and one can number the paragraphs in each section. Thus if you see a reference like T 18.104.22.168, that means Book I, Part I, Section I, paragraph 1.
SBN Despite Hume's popularity there is no scholarly, critical text Collected Works of Hume, although there has been one in the works for some time now. Since it's useful to have a standard edition for scholarly reference, Hume scholars have for a very long time used Nidditch's revised version of Selby-Bigge's editions of A Treatise of Human Nature and of Enquiries Concerning Human Understanding and Concerning the Principles of Morals. Hence the SBN. If we're talking about the Treatise, SBN 213 would mean page 213 of the Selby-Bigge/Nidditch edition. If we're talking about the Enquiries, SBN 213 would usually mean marginal number 213 in the Selby-Bigge/Nidditch edition. Sometimes, to distinguish the two volumes, E is used instead of SBN for the Enquiries.
NN Norton and Norton edited the Oxford edition of A Treatise of Human Nature. Exasperatingly, despite the fact that it would be nice to take it as a new standard edition, they didn't bother to put the SBN numbers in their text, so you can't use the NN edition to look up almost anything published by Hume scholars prior to the publication of the NN edition itself: it's all in SBN numbers. So instead of having a new standard edition, Hume scholars now have two standard editions. Since NN (unlike SBN) numbers the paragraphs, some Hume scholars don't bother to give the NN page numbers, but simply give the paragraph number. Were I writing for publication or even presentation, I would provide all three (the paragraph reference, the SBN page, and the NN page), but for blog posts I'm not going to worry about it.
LO This refers to the translation of Malebranche's A Search after Truth, by Lennon and Olscamp.