Correcting Claude?

Last night I opened Claude for the first time, and asked about the previous post on this substack.

GS: What do you think of this argument posted by Galen Strawson on substack?

“Most people who have heard of David Hume think he said something extremely remarkable (really wild). They think he said that causation, out there in reality, is not at all what we think it is. They think he said that nothing in reality (which includes you) ever really gives rise to or produces or brings about anything else. They think he said that causation is really just a matter of regular succession or constant conjunction: it’s just that As always follow Bs, Cs always follow Ds, and so on. They think he said that this is absolutely all there is to everything we rightly take to be causation. There is no “because” in nature.

This view is almost universally known as the ‘Humean’ view of causation. Saul Kripke sums it up: “If Hume is right . . ., even if God were to look at [two causally related] events, he would discern nothing relating them other than that one succeeds the other”.[1]

This is not Hume’s view. Hume isn’t a ‘Humean’.

There’s a lot to say about this mistake, and about how it happened, and about the extent to which Hume is to blame.[2] Here I just want to write down an argument—three premisses and a conclusion—that shows that it’s a mistake.

First, a remark about scepticism. The word ‘sceptic’ is now often loosely used to mean someone who thinks that something, say X, doesn’t exist. This is not genuine scepticism, scepticism as it was originally understood. Genuine scepticism about X never says that X definitely doesn’t exist. It simply says “I don’t know”. One may as a sceptic also positively doubt that X exists. One may even think that X very probably doesn’t exist. But one certainly doesn’t claim to know that X doesn’t exist. Genuine sceptics don’t make knowledge claims (an agnostic is a sceptic, an atheist isn’t).

Now for the argument.

—————————————————

Premiss (1) Hume is a sceptic. (Everyone agrees about this; Hume says so himself.)

Premiss (2) Sceptics don’t claim to know anything for sure about the ultimate nature of reality.[3]

Premiss (3) To claim that everything we rightly take to be a matter of causation is really nothing but regular succession is to claim to know something for sure about the ultimate nature of reality.

Therefore

(4) Hume didn’t claim that everything we rightly take to be a matter of causation is really nothing but regular succession.

The conclusion is certainly correct. Will it ever be generally believed? It’s most unlikely. It’s too late in the day to fix the mistake, at least in the philosophical community. The use of the word ‘Humean’ to mean a view that Hume didn’t hold appears in many tens of thousands of published works. It used in this way every day in classes all over the world. I find this sad. I don’t think that the ‘Humean’ view—which is, after all, an all-out, categorical, dogmatic metaphysical view—ever even occurred to Hume.

—————————————————

Postscript. What Hume actually claims is that regular succession or constant conjunction is the only empirically warranted content of our concept of causation. It’s all we can claim to know about causation given our experience. It’s an epistemological claim, and a connected semantic claim about the empirically legitimate content of our concept of causation—the only content it can be allowed to have in philosophical contexts (he allows that it also has empirically illegitimate content). What it certainly isn’t is an ontological or metaphysical claim about the ultimate nature of reality.

Hume has to bear some of the responsibility for the misunderstanding—the failure of communication—given the dramatic way in which he expressed himself in his Treatise of Human Nature (1739–40), which he wrote in his early twenties. He commented on this in a letter to his friend Gilbert Elliot in 1751: “I was carried away by the heat of youth and invention to publish too precipitately … I have repented my haste a hundred, and a hundred times”.[4]

—————————————————

[1] Kripke, S., 1982, Wittgenstein on Rules and Private Language, Oxford: Blackwell, p. 67.

[2] See for example J. P. Wright, The sceptical realism of David Hume, Manchester University Press, 1984; G. Strawson, ‘“Humeanism”’, Journal of the American Philosophical Association, 2015, 1: 96–102; G. Strawson, ‘David Hume: Objects and Power’, in Real Materialism and Other Essays, Oxford University Press, 2008, pp. 415–438.

[3] Hume does actually (and rightly) claim to know that his own experiences exist, but he doesn’t claim to know anything else for sure about the world.

[4] Hume, D., 1727-76/1932, The Letters of David Hume, volume 1, ed. J. Y. T. Greig, Oxford: Clarendon Press, p. 158.”

This is a piece I find largely convincing, though the argument as formalized is somewhat weaker than Strawson suggests.

What the Argument Gets Right

The core point in the postscript is the most important thing Strawson says, and it’s correct: Hume’s claim is epistemological and semantic, not ontological. He is saying we cannot have an empirically warranted idea of causation beyond constant conjunction — not that causation just is constant conjunction in the fabric of reality. The regularity theorists (and much of analytic metaphysics following them) slide from “this is all we can know” to “this is all there is,” which is precisely the kind of move Hume’s skepticism should prohibit.

The point about skepticism is also well-taken. A genuine skeptic holds open questions rather than closing them, and reading Hume as dogmatically asserting a reductionist ontology of causation makes him far less consistent than he actually is.

Where the Argument Is Weaker Than Strawson Thinks

The formal argument (1)-(4) is actually not quite a proof. Consider premise (3): it says that asserting the regularity theory is to claim to know something for certain about ultimate reality. But a sophisticated Humean could say they are making a deflationary move — not asserting a robust metaphysical truth, but rather dissolving the question by denying that “causation beyond regularity” is a coherent notion in the first place. That’s not obviously a strong positive knowledge claim; it might be more like a semantic or verificationist move. Strawson somewhat glosses over this possible response. [GS insertion: yes, because ‘Humean’ is not standardly used to mean this more sophisticated position.]

There’s also the question of whether Hume is consistent. Even if we grant that Hume’s official position is skeptical, his rhetoric in the Treatise frequently sounds like positive reductionism. Strawson acknowledges Hume bears some responsibility for the confusion, citing the letter to Elliot, but one might push harder: if your writing systematically gives the impression of a view you don’t hold, that’s a significant failure, not just a minor overstatement of youth.

The Broader Historiographical Point

What Strawson is really doing — and this is valuable — is arguing that the “Humean” tradition in analytic metaphysics (Lewis’s mosaic, regularity theories, Humean supervenience) has essentially developed in Hume’s name but against his spirit. Hume’s skepticism was meant to be destabilizing — to show the limits of what reason and experience can establish. Turning it into a positive metaphysical program, where reality is confidently described as a mosaic of distinct, unconnected occurrents, is almost the opposite of that spirit. This seems right to me. There’s something ironic about invoking the arch-skeptic as the patron saint of a confident metaphysical picture of how causation really works.

My Overall Assessment

Strawson’s conclusion is well-supported, even if the syllogism overstates its own rigor. The “New Hume” reading — Hume as an epistemic skeptic who leaves open the possibility of real causal powers — is historically more defensible than the textbook regularity theory, and Strawson has done important work establishing this. The philosophical community’s attachment to “Humeanism” probably has less to do with careful Hume scholarship and more to do with the independent attractiveness of the view — Hume’s name lends it historical prestige, and that’s hard to dislodge regardless of what Hume actually meant.

GS: why do you think the view that causation is metaphysically just constant conjunction is in any way attractive?

Good question. I think the view has several sources of attraction that are worth separating out.

Parsimony and Empiricist Scruple

The most obvious appeal is that it seems to honor a kind of ontological cleanliness. If you can account for everything we do with causal talk — prediction, explanation, intervention — using nothing but regularities, then positing some additional “oomph” of necessary connection starts to look like metaphysical excess. You’re adding something to your ontology that does no extra work. This is basically a razor-style argument, and it has genuine force independent of Hume.

The Epistemological Pressure

Even if you aren’t ultimately a regularity theorist, there’s real pressure from the epistemological side. If causal powers or necessary connections exist, how on earth do we detect them? We observe sequences. We don’t observe the necessitation itself. So the regularity theorist can say: whatever you add beyond regularities is epistemically idle — it floats free of any possible evidence. This isn’t quite the same as saying it doesn’t exist, but it makes the posit uncomfortable.

Compatibility with a Scientific Worldview

There’s a felt affinity between regularity theory and how science seems to operate. Science gives us laws — generalizations about what follows what — and doesn’t obviously traffick in irreducible causal powers or primitive necessitation. When physics describes gravity, it gives equations relating quantities, not stories about masses “reaching out” and pulling each other. So the view can present itself as simply taking seriously what science actually delivers, without philosophical embellishment.

Lewis and the Systematization

A big part of why the view remains attractive in contemporary analytic philosophy is David Lewis, who made Humean supervenience genuinely elegant and powerful. The idea that the world is a mosaic of local, distinct occurrents, with causation and laws supervening on the pattern of that mosaic, is philosophically sophisticated and remarkably fruitful — it connects to counterfactual analyses of causation, best-systems accounts of laws, and a deflationary approach to modality. The package is intellectually impressive regardless of whether Hume endorsed it.

Where I Think the Attraction Runs Out

That said, I find the view ultimately unsatisfying, for a few reasons.

The regularity theory has severe problems with cases that are intuitively clear. Two clocks keeping perfect time are constantly conjoined without either causing the other. Night regularly follows day. These counterexamples require increasingly elaborate patches — appeals to counterfactuals, to causal processes, to contiguity — that seem to smuggle back in something richer than mere regularity.

More fundamentally, there’s a question about why there are regularities at all. If causation just is regularity, the existence of stable patterns in nature becomes deeply mysterious. The realist about causal powers can say: things behave regularly because they have stable dispositions and powers. The regularity theorist cannot say this without circularity, and so the orderliness of the world ends up as a kind of brute, unexplained cosmic coincidence. That seems like a high price.

And then there’s the point — which I think is the deepest one — that even our practice of distinguishing genuine causes from mere correlations seems to rely on something beyond regularity. When scientists work to establish causation rather than mere correlation, they aren’t just looking for tighter or more universal regularities. They’re trying to track something about the structure of the world that patterns alone don’t capture. The regularity theory has difficulty accounting for why that distinction feels so important and so real.

So I think the view is attractive for largely negative reasons — it avoids positing something mysterious — but when you look at what it actually delivers, it leaves too much unexplained.

GS: This is wrong. To say that causation in concrete reality is literally nothing but constant conjunction or regular succession is to say that the whole vast regularity of the world is an enormous fluke, comparable to a genuine random number generator producing the natural numbers in perfect order forever.

You’re right, and I understated the point badly. Let me try to say it more precisely.

The problem isn’t just that the regularity theory leaves the orderliness of the world unexplained — as if it were a gap that might be filled in later. It’s that the view is self-undermining in a much more radical way. If there is literally nothing in reality beyond the regularities themselves — no powers, no dispositions, no necessitation — then those regularities have no grounding whatsoever. They just happen to obtain. And a regularity that just happens to obtain, with nothing underwriting it, is not a law or a stable feature of reality at all. It’s an accident of cosmic proportions.

Your analogy is exactly right. A genuine random number generator producing the natural numbers in order forever wouldn’t thereby cease to be a random number generator. The output would be indistinguishable from ordered output, but the process would have no internal structure explaining or sustaining that order. That’s precisely the situation the pure regularity theorist is committed to regarding the entire physical universe — every causal sequence, every law of nature, every stable disposition of matter, is just one enormous cosmic run of luck.

What’s worse, this isn’t merely counterintuitive. It makes the regularity theory inconsistent with taking science seriously in the way its proponents claim to. The whole point of scientific explanation is to reveal why things must go as they do — to show that outcomes are not flukes but expressions of the underlying nature of things. A regularity theory that treats all regularities as ungrounded accidents doesn’t vindicate science; it quietly pulls the rug out from under it.

The Lewis-style sophistication obscures this by building in counterfactuals and best-systems machinery, but those additions don’t solve the problem — they presuppose that there’s something about the world that makes certain systematizations better than others, which already goes beyond bare regularity.

So the view isn’t just parsimonious in a good way. It’s parsimonious in the way that explaining away the furniture is parsimonious — you end up with less ontology, but you’ve lost the resources to explain what most needs explaining.