Hume finished this in his twenties, at La Flèche in France — the Jesuit college where Descartes had been a student a century earlier — and published it anonymously in three installments. It sold almost nothing. His own verdict, written near the end of his life, is the most quoted sentence about it: it fell dead-born from the press.

He spent much of his career rewriting its contents in shorter forms and came to wish he were not judged by it. Posterity has ignored him on this. The Treatise is where the arguments are strongest and where several of them exist at all.

What reading it is like

Three books, about 270,000 words, in eighteenth-century prose that is elegant in stretches and exhausting across chapters. Hume was young and shows it: the book is ambitious, digressive, and organized by the architecture of a system rather than by what a reader needs.

Do not start here. Hume rewrote Book I as An Enquiry Concerning Human Understanding (1748) and Book III as the second Enquiry (1751), and both are better made and a fifth of the length. Read those first, then come back for what they left out.

What they left out is the reason to come back. The treatment of the self is in the Treatise and not the Enquiry. The account of the passions is only here. The is–ought paragraph is only here. And the conclusion of Book I — Hume describing what it is like to have argued himself into a position he cannot live in — is the best writing in the book and has no counterpart anywhere else in his work.

Two practical notes. The Abstract, which Hume wrote anonymously in 1740 to publicize the book, is eight pages and is the best short guide to what he thought the argument was. And the Appendix to Book III contains his own retraction of the account of personal identity, which is unusual enough to be worth reading on its own.

Book I: the understanding

The machinery first. Perceptions come in two kinds: impressions, which are what we feel and sense, forceful and immediate; and ideas, which are fainter copies of them in thought and memory. Every simple idea is a copy of some prior impression. That single rule is the whole of the method — to test a concept, hunt for the impression it came from.

Then he runs it on causation, in Part III, and the result is the most famous argument in modern philosophy. Watch one billiard ball strike another and you observe succession — this then that — and contiguity, and repetition. You do not observe the necessity. There is no impression of the making-happen. And since there is no impression, the idea did not come from the events; it came from us, from the habit repetition builds, and we project the felt expectation outward and call it a power in the world.

The corollary is worse. All reasoning from experience assumes that the future will resemble the past, and that assumption cannot be established by experience without using itself, nor by demonstration, since a changed future implies no contradiction. See causation and the problem of induction, which carry the standing treatments.

Part IV turns the method on the mind itself. Look inward for the self that has the perceptions and you find only more perceptions — a bundle, in the word everyone quotes, with no owner to be found. See bundle theory.

Book I then ends in the passage nobody expects from a systematic treatise: Hume alone in his room, having concluded that neither his senses nor his reason can be defended, describing the condition as a kind of forlorn solitude, and reporting that dinner and backgammon cure it and that the cure is not an argument.

Book II: the passions

The least read and the most underrated. Hume builds a mechanics of the emotions — pride and humility, love and hatred — each produced by a double relation between an idea and an impression, and the whole thing runs on sympathy, the capacity by which another person's feeling becomes, by communication, something like our own.

Two things here do real work elsewhere. Sympathy is the mechanism that makes his ethics possible, and it is what Adam Smith takes and rebuilds. And II.iii.3 contains the sentence about reason being the slave of the passions, which is not a flourish but the conclusion of an argument that reason alone can neither produce nor prevent any action.

Book III: morals

Morality is not a matter of fact discovered by reason but a matter of sentiment — the approval a spectator feels toward a character considered from a general point of view. The generality is what keeps it from being mere taste: we correct for our own position, because otherwise we could not talk to each other about it at all.

Then, three paragraphs from the end of Part I Section I, the observation that made the book famous in a way Hume did not anticipate. Every system of morality he has read proceeds for a while with is and is not, and then suddenly all the propositions are joined by ought and ought not — a change that is imperceptible, is of the last consequence, and is never explained. He asks that a reason be given. See the is–ought gap.

Part II, on justice, is the other place the book is still mined. Justice, for Hume, is an artificial virtue — not a natural sentiment like kindness but a convention that arises because it serves everyone, and that comes to command feeling only once it is established.

Key passages

Introduction — the announcement of the experimental method applied to moral subjects.

I.i.1 — impressions and ideas; the copy principle.

I.iii.2–8 — the analysis of causation.

I.iii.6 — the circularity in any attempt to justify induction.

I.iii.14 — the two definitions of cause, and the origin of the idea of necessary connection.

I.iv.6 — the bundle account of the self.

I.iv.7 — the conclusion of Book I; the forlorn solitude, the backgammon.

II.iii.3 — reason as the slave of the passions.

III.i.1 — the is–ought paragraph.

III.ii.1–2 — justice as an artificial virtue.

Appendix — Hume's dissatisfaction with his own account of personal identity.

What is contested

Whether Hume is a skeptic about causation or a realist of a chastened kind. The "New Hume" debate turns on whether he denies that there are real powers in nature or only denies that we can have an idea of them. Both readings have text on their side, which is the problem.

What the Appendix retraction is about. Hume says two of his principles are inconsistent and he cannot reconcile them, and he does not say clearly which two. The literature on this one paragraph is very large.

Whether the naturalism rescues the skepticism or dodges it. Explaining why we believe what we cannot justify is either the honest end of the inquiry or a change of subject, depending on who is writing.

Whether Book II belongs. Contemporary readers take Book I and Book III and leave the passions out, which is a decision Hume did not make and which the recent work on emotion has begun to reverse.