Hume is read now because of one move, and it is not any of his famous conclusions. It is the decision about what philosophy is for.

Everyone before him asked how our beliefs could be justified. Hume asked where they come from, found that the justification was not there, and then — instead of concluding that we should stop believing — asked what it is about us that makes us believe anyway. The question shifts from what entitles us to what explains us, and almost all later work on knowledge, science and mind is downstream of that shift.

The problem

He said he wanted to do for the mind what Newton had done for matter: find the few principles that govern it, by observation rather than by system-building. The method is the one Locke started and Hume applies without flinching. Every legitimate idea has to be traceable back to some impression — some actual experience it is a copy of. Show me the impression, or I do not know what you are talking about.

Applied honestly, that demand is a machine for producing unwelcome results, and Hume ran it on the concepts philosophy relied on most: necessary connection, the continuing self, the justification of inference from past to future. It did not find them. The arguments are in the book — causation, the problem of induction and bundle theory each carry their own page, and A Treatise of Human Nature has them as the book makes them.

What belongs on his page is what he did with the result.

The naturalist turn

A skeptic, finding that induction has no rational ground, concludes that we should suspend judgment. Hume does not, and his refusal is the most characteristic thing about him.

Nature, he says, has not left it up to us. We go on expecting the future to resemble the past because we are built to, not because we have worked out that we should, and the propensity operates whatever philosophy concludes. He reports playing backgammon and dining with friends, and finding that after a few hours his own skeptical conclusions look ridiculous to him — and then observes that this does not refute them either.

So the honest position is double. The arguments stand: there is no rational foundation here. And the beliefs stand: they are not the sort of thing reason was ever producing. Philosophy's job is to describe the mechanism, not to supply a warrant it was never providing.

This is why he matters more than a list of skeptical results would suggest. The skeptical results are the setup. The naturalism is the position.

Reason, and where it actually sits

The sentence everyone quotes is the one to start from: reason is, and ought only to be the slave of the passions.

It reads as provocation and is a technical claim. Reason, on Hume's account, does two things: it works out relations between ideas, and it works out matters of fact. Neither of those can move anyone. Knowing that the building is on fire gets you out of it only because you already want to live; the belief supplies the route and the desire supplies the going.

From which the ethics follows directly. If reason cannot motivate, and moral judgments plainly do, then moral judgments are not deliverances of reason. They are sentiments — the approval or disapproval that arises in a spectator who contemplates a character from a certain distance. Not arbitrary, because human beings are similar enough for the sentiments to converge, and the standpoint is corrected for self-interest. But not discovered, either. See the is–ought gap, which is the methodological half of the same thought.

What to read

A Treatise of Human Nature (1739–40), written in his twenties in France. Everything is here, at full strength and full length. It fell dead-born from the press, in his own phrase.

An Enquiry Concerning Human Understanding (1748) rewrites Book I of the Treatise for readers. Shorter, better organized, and Hume's own preference — though it drops material, including the discussion of space and time and much of the treatment of the self, which is why scholarship goes back to the Treatise anyway.

Dialogues Concerning Natural Religion, written over twenty-five years and published in 1779, three years after he died, because he knew what it would cost him alive. Philo takes the design argument apart: the universe is a single case, so there is nothing to compare it to; the analogy to human contrivance is weak and gets weaker the harder you press it; and whatever it licenses, it does not license the God anyone is arguing for. It is the best-written philosophy in English and it has its own page.

He also wrote a six-volume History of England that made him rich and famous and that almost nobody now reads. He considered himself a historian.

The objection

In the essay "Of National Characters", Hume added a footnote asserting that he was apt to suspect the negroes to be naturally inferior to the whites, that no civilized nation of that complexion had ever existed, and that the uniformity of the case made a natural difference likely. He revised the footnote in a later edition. He did not remove it.

The damage is not only moral, and this is the point worth making on a page about his method. The footnote is precisely the thing his philosophy exists to forbid: a sweeping causal claim, asserted from prejudice, with no examination of the evidence, by a man who had built a career on refusing exactly that move when other people made it. Immanuel Kant repeated the sentiment approvingly; James Beattie answered it at the time, on Hume's own evidential grounds, and was right.

It is worth holding both facts. The method is good enough that it convicts its author. That is a real point in the method's favor and no defense at all of the man.

Where he leads

Kant said Hume woke him from dogmatic slumber, and the entire Critical philosophy is an attempt to answer him without giving up on necessity. Adam Smith was his closest friend and the Theory of Moral Sentiments is a Humean book. The logical positivists claimed him as a founder. And the naturalist program — explain the believing rather than justify the belief — is now most of how philosophy of mind and philosophy of science are actually done.