1. Human discourse is comprehensible because thoughts expressed by it are related in ways we expect. What are these relations between ideas?
2. Hume claims there are three basic principles of association: resemblance, contiguity and cause-and-effect.
3. All propositions of thought are claimed to fall into one of two classes. The first, relations of ideas, are logical and mathematical principles. These correspond roughly to Descartes’s contents of the understanding. The second class, matters of fact, are statements about the experienced world.
4. The truth of a relation of ideas is established merely by thinking it. Not so with matters of fact.
5. We can tell if something is a relation of ideas by trying to deny it and seeing if we get a contradiction. If we do, we have a relation of ideas. By contrast, the denial of a matter of fact, even if false, doesn’t produce a contradiction. It is always conceivable that a proposition stating a matter of fact might be wrong.
6. All knowledge of matters of fact is claimed to depend on inferring effects from causes.
7. Relations of cause and effect can come to be known only by experience of their `constant conjunction.’
8. Note, therefore, how in contrast to Descartes Hume starts out by drawing the strongest possible sort of distinction between knowledge of logical dependence and knowledge of ontological dependence.
9. We tend to forget that cause-and-effect relations that are very regularly observed ever had to be learned.
10. It is claimed that no effect can be inferred just from examining its cause.
11. Thus we cannot do science, as Descartes had imagined, by pure reason.
12. On pp. 21-22, Hume argues that there is no principle of reason that justifies induction, that is, that justifies our inferring from observed constant conjunction that the next similar case will resemble the observed ones. He doesn’t, of course, deny that we do some reason, or even that we must so reason. He’s denying that such reasoning involves any element of logic. It is, he claims, just something we do.
13. This is shown, he argues, by the fact that denying the conclusion of an induction produces no contradiction. No matter how constant a conjunction has been, it’s always conceivable that tomorrow it could cease to hold.
14. You can’t argue as follows: “I can infer, from the fact that the sun has risen every day, that it will rise tomorrow, because the future in general resembles the past.” What is supposed to justify your claim that the future in general remembers the past? Well, just the fact that it always has. But this tries to justify induction by appeal to induction, and so is circular. “It is impossible … that any argumenys from experience can prove the resemblance of the past to the future; since all these arguments are founded on the supposition of that resemblance.”
15. Hume argues that since children and animals perform inductions, induction can’t be based on rational argument. It’s simply a natural disposition.
16. This conclusion may not strike a modern person as greatly shocking. Note, however, that it denies the entire basis of Plato’s and Descartes’s philosophies, which rested on the idea that everything important we think we know must be demonstrable by some argument. Here is Hume claiming that one of the most basic things we think we know, which is the foundation for much of our other knowledge, can’t itself be established by reason, but is just a brute natural fact.
17. This is the first clear appearance in Western philosophy of naturalism, the view that the universe is not guided by an underlying prevailing reason, but just is.