1.

Many philosophers suppose that David Hume holds a ‘regularity’ theory of causation according to which—to put it in plain English—nothing ever really causally influences anything else in any way at all. The world is highly regular in its behavior, according to this theory; occurrences of one particular kind regularly and indeed invariably follow occurrences of another particular kind. But nothing that happens ever actually influences what happens next in any way. Nothing ever actually brings anything else about.

On this view Hume's claim about causation is not simply that we can never form any sort of empirically respectable—genuinely empirically contentful or meaningful—conception of causal power or causal influence; although he is certainly saying at least this. He's not allowing that there might possibly be such a thing as causal influence while stressing that we can't clearly conceive what such influence might involve. He's saying that there is as a matter of certain metaphysical fact no real causal influence of anything on anything—ever. He's endorsing what one might call the ‘pure’ or ‘outright ontological’ regularity view of causation.

Saul Kripke expresses this view of Hume well when he writes that ‘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’ (Reference Kripke1982: 67). A. J. Ayer writes that on the Humean view ‘in nature one thing just happens after another’ (Ayer Reference Ayer1973: 183). Woolhouse says that ‘Hume's conclusion [is] that so far as the external objects which are causes and effects are concerned there is only constant conjunction’; so far as the ‘operations of natural bodies’ are concerned, ‘regularity and constant conjunction are all that exist’ (Reference Woolhouse1988: 149–50). According to O’Hear, Humeans hold that there is not ‘any more to causality than “regularity of succession”’ (Reference O’Hear1985: 60). Putnam speaks of the ‘Humean definition of total cause—A is the (total) cause of B if and only if an A-type event is always followed in time by a B-type event’ (Reference Putnam and Putnam1981: 214). Jackson invites us to ‘consider . . . the possible world where every particular fact is as it is in our world, but there are no causes or effects at all. Every regular conjunction is an accidental one, not a causal one. Call it, for obvious reasons, the Hume world’ (Reference Jackson1977: 5).

Kripke's formulation of the once standard view of Hume is clear and accurate—wholly sufficient. I’ve added some more quotations because an anonymous reviewer for Mind judged my initial characterization of the once standard view of Hume to be ‘ridiculous’; ‘crude’, ‘unworthy’, ‘a parody’. These words fit the standard attribution of the pure regularity view to Hume well, but they don't fit my description of the attribution. The view once almost universally attributed to Hume is in everyday English, in which a spade is a spade, the view that there is as a matter of certain metaphysical fact no such thing as causal influence. It's not the thoroughly respectable (if still questionable) epistemological view that we can know nothing more of cause than regular succession. It's not the far more plausible view that regular succession is all we can legitimately mean by the word ‘cause’ in philosophy if we endorse a certain strict empiricist theory of meaning.

Fortunately, the once standard view of Hume has lost a lot of ground in the past thirty years.Footnote ¹ This is a good thing, if only because the standard view is flatly inconsistent with the fact that Hume—whatever else he is or is not— is a skeptic. He's a committed albeit ‘moderate’ (224/1.4.3.10) skeptic who rejects all claims to certain knowledge of the ultimate or intrinsic contingent nature of reality—except in the case of his own immediate experiences, which are, as he says, ‘perfectly known’ (366/2.2.6.2). It follows that it cannot be right to think that Hume claims to know that nothing ever actually influences anything else in any way at all.

This is already enough to show that the old view is false. We don't need to examine the details of Hume's discussion of causation. When we do, and proceed with care, we find that Hume as a ‘moderate’ skeptic never questions the idea that there is genuine causal influence. He takes it for granted and merely insists on the sense in which we can never hope to form any empirically respectable grasp of its intrinsic nature. And yet the adjective ‘Humean’ is still widely used as a shorthand term for the pure ontological regularity view.

2.

This use was powerfully reinforced by a well-known passage in David Lewis's introduction to the second volume of his Philosophical Papers:

In another place Lewis writes as follows:

Since quite a number of people trace their use of ‘Humean’ back to Lewis, it may be helpful to report an exchange that took place in June 1997.

Earlier that year I had sent Lewis a paper (Strawson Reference Strawson and Strawson2000) that argues that Hume is not a pure regularity theorist and is in effect a realist about causal influence—a suitably nondogmatic ‘sceptical realist’, in John Wright's terms (Wright Reference Wright1983). After dealing with some other business, I asked Lewis whether he agreed with the thesis of the paper. Lewis replied by fax the same day (June 23; quoted by kind permission of Stephanie Lewis):

The view in question hasn't been knocked down decisively. Don Garrett has recently suggested that it is as well supported by Hume's text as the old ‘pure’ regularity view (Garrett Reference Garrett, Beebee, Menzies and Hitchcock2009; for some older statements of various aspects of the skeptical realist view see Wright Reference Wright1983; Broughton Reference Broughton1987; Craig Reference Craig1987; and Strawson Reference Strawson1989). Many now agree that Hume's fundamental point about causation is epistemological-cum-semantical, not ontological-metaphysical. His claim is that we can't ever observe anything more than constant conjunction when it comes to whatever it is that is the reason why the world is regular in its operations, and so we can't form any clear and distinct or empirically legitimate or philosophically respectable concept of causation as something more than constant conjunction. The phenomenon of constant conjunction is ubiquitous. The trouble is precisely that, in Hume's words, ‘we cannot penetrate into the reason of the conjunction’ (93/1.3.6.15; my emphasis). ‘The scenes of the universe are continually shifting’, he says, ‘and one object follows another in an uninterrupted succession; but the power or force, which actuates the whole machine, is entirely concealed from us, and never discovers itself in any of the sensible qualities of body’ (63–4/7.8; my emphasis).

The terminological conclusion seems plain. Those who wish to use terms accurately should stop using the word ‘Humean’ in the old unqualified way. Will this happen? Many people hate changing their ways. They hate it even more when they’re told that they should. They usually say that it doesn't matter that a name is not appropriate or accurate.

Lewis, for one, disagrees. Those who wish to follow Lewis should stop using ‘Humean’ in the old way. The point holds independently of the fact that many now think that Hume is at bottom a realist about causal influence in a nondogmatic way that is entirely compatible with his moderate skepticism (for a good recent statement see, e.g., Kail Reference Kail2007 and also Wright Reference Wright2009). The point holds even if you agree with Lewis (as I hope you do not) that ‘the Hume of popular (mis?)understanding’ is ‘vastly more interesting’ than the skeptical realist Hume.

Perhaps we could use lowercase ‘humean’ rather than ‘Humean’—a proposal made by Gideon Rosen (in correspondence). That would be a start, so long as the first use of ‘humean’ in any piece of work was accompanied by an explanatory footnote of the kind proposed by Lewis. But I don't think that Lewis would think it enough, if his view were consistently and grossly misrepresented, to have the capital letter ‘L’ of his surname replaced by lowercase ‘l’; a point put to me by Edward Craig. How would you feel if it happened to you? You don't have to be overly self-concerned to think that it would be a pity.

Perhaps it's time to try something else. In the case of the supervenience thesis, we can replace ‘Humean’ by ‘Lewisian.’ The doctrine that ‘all there is to the world is a vast mosaic of local matters of particular fact’ takes on a deeply different and far more interesting shine when coupled with a ‘four-dimensionalist’ or ‘block universe’ conception of reality. But this conception of reality, although part and parcel of Lewisian supervenience, is no part of anything Humean.Footnote ²

Might we more generally replace ‘Humean’ everywhere by ‘Brownian’, in recognition of Thomas Brown? Brown does have a claim to be the first true regularity theorist (see, e.g., Kemp Smith Reference Kemp Smith1941: 91n; Strawson Reference Strawson1989: 7; Wright Reference Wright and Jones2005; Dixon Reference Dixon and Dixon2010: 11–16; Psillos Reference Psillos, Allen and Stoneham2011), but it's not clear that the claim is correct. Brown does write that

and that

At crucial points, however, he qualifies such claims with remarks to the effect that this is all causation can be to us—allowing that there may be something more. He's concerned with the notion of ‘a cause, in the fullest definition which it philosophically admits’ (Reference Brown1818: 17), and his awareness in 1818 of the limitations of definitions, when it comes to attempts to capture the nature of concretely existing phenomena, is the same as Hume's in 1739 and 1748. His criterion of philosophical admissibility is also the same as Hume's where concretely existing phenomena are concerned. It's a matter of clarity and distinctness of content of the sort that can in his view be conferred only by proper empirical grounding in experience. There's no reason to suppose that Brown thinks that reality doesn't and can't extend further than properly empirically grounded human conception. There's no more reason to think this of Brown than of Hume, who remarked in 1740, in a clarificatory addition to Book 1 of the Treatise, that

3.

It's not hard to see how the mistaken view of Hume arose. At one point in the Treatise Hume writes that

In the Enquiry he initially rephrases this as follows:

Surely, to offer a definition of something is to say what it is?

In fact, not—this is not so in the eighteenth century: not when it is a matter of defining a concretely existing phenomenon like causation, rather than squareness, say, or triangularity. As Edmund Burke notes, writing nine years after the publication of the first Enquiry,

Priestley concurs twenty years later:

This is what Hume was doing when he gave a definition of causation. It explains why he thinks his definitions are ‘exact’ and ‘precise’ (169/1.3.14.30) and ‘just’ (76/7.29)—they exactly capture our best human take on causation—even as he judges them to be irremediably ‘imperfect’ (76/7.29). A ‘more perfect definition’ would be one that ‘point[ed] out that circumstance in the cause, which gives it a connexion with its effect’ (77/7.29), but we can't do this. All we can detect is constant conjunction; ‘we cannot penetrate into the reason of the conjunction’ (93/1.3.6.15; my emphasis).

This understanding of definition isn't restricted to the eighteenth century. In 1927 Russell uses ‘define’ in the same sense as Hume, Burke, and Priestley when characterizing matter: ‘all that we ought to assume is series of groups of events, connected by discoverable laws. These series we may define as ‘matter’. Whether there is matter in any other sense, no one can tell’ (1927: 93). Russell makes it clear that to give a definition is not to make an ontological declaration.