I gave this talk as part of a course requirement in Philosophy of Science at Utrecht University. I hope you like it!
Much ink has been spilled on justifying induction. I am more interested in how Hume describes inductive reasoning.
Transcript
Today I want to present Hume’s account of inductive reasoning, contrast it with Descartes, and raise a couple of questions I think it leaves unanswered.
The starting point is what Peter Lipton and others call the problem of description — not to be confused with the problem of justification. The problem of description asks: how do we arrive at matters of fact about the world, when those facts are underdetermined by the evidence available to us?
Rationalism: causes contain their effects
To see what Hume is arguing against, we first need Descartes’ picture. I’ve depicted it here as a set of Russian nested dolls. The dolls represent ideas — some function as causes, some as effects. One billiard ball moving toward another is a cause; the second ball bouncing off it is an effect. Like the dolls, causes contain their effects: each effect can in turn be a cause containing further effects, nested inside one another.
The key claim is that we can derive facts about nature through reason alone. Take billiard balls: why does the second ball bounce off the first? Because the first ball is made of matter, and matter, for Descartes, just is extension. By the definition of extension, two bodies cannot occupy the same position — that would be absurd. So the second ball must, by logical necessity, bounce off the first. We’ve derived a fact about the world through pure reason.
Empiricism: no necessary connection
Hume rejected this. For him, all reasoning about nature rests on experience. A better image than nested dolls is a chain of dominoes: each domino is an idea, and every idea ultimately traces back to a sense impression or a memory of one. Each domino has a cause side and an effect side — any idea can function as either. Unlike the nested dolls, there’s no necessary connection between adjacent dominoes. In principle, any arrangement is possible; the arrangement we actually observe is a matter of custom, not logic.
Science, on this picture, is the project of reducing many effects to a few shared causes — building long chains of dominoes from a small number of starting points. A single domino representing gravity, for instance, can be linked into chains explaining both projectile motion and elliptical orbits.
Belief as feeling
Back to the billiard balls: it’s a matter of fact that they communicate motion by impulse — a fact underdetermined by the evidence, since we can coherently imagine the next collision going differently. So why do we believe it?
Hume’s answer, from Section 5, §40 of the Enquiry: belief is “nothing but a conception more intense and steady than what attends the mere fictions of the imagination” — a conception that is, in his words, more “vivid, firm, and forcible”. That’s why we favor this explanation over any alternative (say, invoking the Flying Spaghetti Monster, to borrow Dawkins’ phrase) — not because it’s better justified by logic, but because we have a stronger conception of it.
Why does it feel more solid? Because for Hume, belief is a feeling — a conception accompanied by a more vivid emotional charge than mere fiction. And that vividness is simply the product of repetition: every time we use an explanation, and every time we experience sensory input matching any of the causal links composing this explanation, our conception of those links grows more vivid. That’s the effect of custom. Hume doesn’t ground custom in anything deeper — he treats it as an instinct granted by a benevolent God, and he’s content with that, since he takes pure logical reason to be far more fickle than experience.
Two open questions
Nelson Goodman called this style of account “passably effective” 1and proposes his own new riddle of induction, which I won’t get into here. Instead, I want to push on Hume’s account directly with two questions.
First: we tend to prefer theories that explain surprising evidence. I don’t see how Hume’s account explains this preference. My guess is he’d appeal to vividness again — surprising things stick in memory longer than unsurprising ones — but the text doesn’t make this move explicitly, and I’m not fully convinced it would work.
Second, and less obviously: Hume divides the operations of the mind into three types — contiguity, resemblance, and causation. But he also claims that all reasoning about matters of fact reduces to causation. If that’s right, what role is left for contiguity and resemblance? It isn’t clear to me from the text.
- With apologies to the reader: those words are actually attributed to Hume’s own judgment of his solution, rather than Goodman’s. The full quote is: “All this seems to me quite wrong. I think Hume grasped the central question and considered his answer to be passably effective. And I think his answer is reasonable and relevant, even if it is not entirely satisfactory.” ↩︎
Leave a Reply