I'm a Humean, and I like necessary connections

In metaphysics, "Humeans" are people who believe that truths about laws of nature, counterfactuals, dispositions and the like (truths about what must or would be the case) are in some sense reducible to non-modal truths (about what is the case).

One way to be a Humean is to deny that there are any laws of natures, non-trivial counterfactuals, etc.: if there are no modal truths, then trivially all modal truths are reducible to non-modal truths. On this account, there are no "necessary connections between distinct existences": eating arsenic might in fact be followed by death, but it could just as well be followed by hiccups or anything else.

I don't know anybody who holds that view. At any rate, a lot of Humeans believe in laws of nature, counterfactuals, and all that stuff. But they also believe that these modal facts are made true by non-modal facts, perhaps about regularities in the distribution of categorical properties. On this view, there are necessary connections between distinct existences. It may well be (nomologically) necessary that consuming large amounts of arsenic leads to death.

Unfortunately, some people keep portraying Humeanism as the denial of necessary connections. Please don't do that. It's like saying that physicalism is the view that there are no mental properties. Eliminativisism about the mental is a particular kind of physicalism (a rather unpopular kind). Similarly, eliminativism about necessary connections is just one kind of Humeanism -- a kind that hardly anyone believes in.

The association of Humeanism with the rejection of necessary connections is especially popular among anti-Humeans. For example, Galen Strawson suggests in his book "The secret connexion" that according to the regularity theory of causation,

the regularity of the sequence of events in the world as a whole is just like the regularity of the sequence of natural numbers 0,1,2,3,4,5,..., in the extraordinary but obviously logically possible case in which the sequence is being generated by a true random number generator. (p.24)

(Obligatory Dilbert cartoon)

Analogies like this are a standard tool in anti-Humean rhetoric. But they have little force against Humeans who believe in laws of nature. If there is a law of nature according to which state 1 is followed by 2, 2 by 3, etc., then in what sense is the resulting sequence like the output of a random number generator? If the laws are deterministic, then state 1 must -- with nomological necessity -- be followed by state 2. No such necessary connection exists between the numbers generated by a random number generator. To accept a nomologically necessary connection between state 1 and state 2, all you need is to accept the relevant law. It doesn't matter if you also think the law is reducible to regularities.

Now some non-Humeans believe that certain successions of events in our world are not merely nomically necessary, but absolutely, metaphysically necessary. Humeans typically don't believe that: it is not metaphysically necessary, on their view, that consuming arsenic leads to death. So could we characterise Humeanism as the denial of metaphysically necessary connections between distinct existences?

This proposal at least has the advantage that some people actually hold the relevant view. However, many people who are commonly classified as non-Humeans also hold that view. To believe that truths concerning the laws of nature are irreducible to non-modal truths is not the same as to believe that the laws of nature hold with metaphysical necessity.

Another difficulty with this proposal is to clarify "distinct existences". Humeans like Lewis don't deny that e.g. people may have there origins essentially, and so that there are metaphysically necessary connections between distinct people. Similarly, Lewis would have happily accepted metaphysically necessary connections between distinct properties: perhaps the property of consuming arsenic in a world with the actual laws of nature is connected by metaphysical necessity to the property of dying, and I suppose these two properties are distinct (whatever that exactly means -- it better not mean that the two properties are not connected by metaphysical necessity: this would make the supposedly Humean doctrine analytic).

One might try to restrict the "distinct existences" to intrinsic properties, or express the denial of necessary connections in terms of intrinsic duplication. Unfortunately, "intrinsic" is used in many different ways, and we would have to make sure we're not talking past each other. For example, it seems to me that Lewis means something quite different by "intrinsic" than contemporary necessitarians about laws. For Lewis, an intrinisc property is essentially a property P such that a thing X having P entails nothing at all about the rest of the universe. Necessitarians generally do not think that properties like mass and charge are intrinsic in this sense, although they often call such properties "intrinsic". (I am not sure what they mean by that.)

This is not to say that there is no interesting distinction between various forms of necessitarianism and their alternatives. For example, necessitarians tend to regard causation as Lewis-intrinsic, and mass and charge as Lewis-extrinsic, which is opposite to Lewis's own account. But this somewhat delicate debate about the intrinsicality of various properties is only very indirectly related to the question of whether nomic truths are reducible to categorical truths.


# on 21 November 2010, 18:39

Hi Wo,

As many Humeans, you seem to want to have your cake and eat it too. To paraphrase Lewis, just because you call something 'nomological necessity' it doesn't mean it's a form of necessity. If all there is to, say, its being nomologically necessary that arsenic is poisonous is, say, that 'arsenic is poisonous' is a theorem in the best deductive system (or any other "Humean" story you might want to tell), then there really is no necessary connection between being made of arsenic and being poisonous, no matter how much you try to create the linguistic illusion that there is one to sweeten the pill.

# on 24 November 2010, 07:40

Hi Gabriele,

of course I can't expect you as a non-Humean to agree with me about what kinds of things qualify as laws of nature, or as necessary connections. My plea is rather for fairness in describing the Humean position. As a Humean, I believe that (1) there are no extra facts beyond what's made true by the Humean mosaic, and (2) laws of nature are certain kinds of regularities in the Humean mosaic. But when people discuss Humeanism, they often ignore part (2), which I regard as the core of the Humean position.

Compare functionalism in philosophy of mind. Functionalists typically believe that (1*) there are no extra facts beyond what's made true by the physical facts, and (2*) mental states are functionally characterised physical states. If you're not a functionalist, you will probably disagree with (2*). But this doesn't mean you can correctly present functionalism as the view that there are no mental states -- that people and animals are like simple robots, without any beliefs, desires, or experiences. This would mean to ignore part (2*) of the functionalist position. According to functionalism, there _are_ mental states.

Similarly, it is wrong to present Humeanism as the view that there are no necessary connections. This is not Humeanism. It is part (1) of Humeanism combined with a non-Humean account of necessary connections in place of the Humean's (2). This combination is a highly implausible form of eliminativism, and I think we can quickly agree that it is mistaken. Ignoring half of either position, it is just as far from Humeanism as it is from anti-Humeanism.

# on 25 November 2010, 12:13

Hi Wo,

I don't think the analogy is a good one. Humeans can claim to believe in laws of nature. However, they cannot claim to believe in necessary connections (under any reasonable interpretation of 'necessary connections'), which I thought was the point of your post. Humean laws of nature are not and do not giver raise to necessary connections. (Even an atheist can claim to believe that God exist provided that 'God' is understood to refer to the neighbour's dog but then they can't also reasonably claim to believe God to be omnipotent unless they also reintertpret the meaning of 'omnipotent'.). Most Humeans are quite candid about this (see Helen Beebee as an example) and I think that denying that would be to murky the waters even more.

# on 26 November 2010, 02:00

Oh, I see. By "necessary", I simply meant "nomologically necessary". That is, for there to be a necessary connection between consuming arsenic and death it is enough that there is a law of nature according to which consuming arsenic leads to death. On this usage (which doesn't strike me as unreasonable), there isn't any difference between believing in laws and believing in necessary connections. But you're right that one could mean other things by "necessary connection", e.g. "connection grounded in non-Humean states of affairs". In this sense, Humeans obviously don't believe in necessary connections. I wasn't aware that the notion is used in such a theoretically loaded way. The point of my post was that Humeans should not be portrayed as denying nomologically necessary connections, as they often are.

# on 26 November 2010, 10:17

The point I was trying to make is that there is nothing necessary in the Humean connection between arsenic ingestion and death or in what Humeans call 'nomological necessity'. If it is a law of nature that arsenic causes death, it, according to Humeans, supervenes on the pattern of arsenic ingestions and deaths in the Humean mosaic. So the pattern is more fundamental than the law and so the law plays no role whatsoever in the production of the pattern or in the going from a arsenic ingestion "tile" in the mosaic to a death "tile".

Again, just because you call something 'nomological necessity' it doesn't make it necessary. I can stipulate to call whatever I type on a Friday 'nomologically necessary' but this doesn't make it necessary.

# on 27 November 2010, 18:05

Gabriele is right of course that calling a proposition "nomologically necessary" doesn't make it necessary. However, if a proposition plays the role of laws in scientific inference e.g. in explanation, in accounts of counterfactuals and causation, in confirmation etc. then it may well deserve to be called "nomologically necessary". It is arguable (in fact has been argued) that Lewisian laws do play these roles. True that Lewis-laws are not fundamental and do not "produce" anything in the fundamental way-whatever that is- that proponents of metaphysically fundamental account of laws think laws do. But L-laws do produce in the more ordinary sense of playing their part in accounts e.g of how electron positron interactions produce gamma ray photons. If being metaphysically fundamental and "metaphysically producing" are required for nomological necessity then Lewis laws are not nomologically necessary. But saying that these are required doesn't make it so.

# on 28 November 2010, 11:23

Hi Barry,

Thanks for the reply. You were actually one of the people I was thinking of when I said that most Humeans are quite clear about their (I think unusual) use of notions such as 'law of nature'. However, I really don't see how its being used in a certain way in scientific inferences can confer a special *metaphysical* status to a proposition (or to the regularity it describes). Nor can I see how their playing a role in accounts of the occurrence of certain events occur make L-laws *produce* anything (other than explanations that is). It seems to put the cart before the horse in a way that reminds me of those old-school regularists that maintained that a law of nature is a universal generalization that is explanatory.

Moreover, I thought that Wo's point was that as a Humean he can believe in necessay connections. And I really don't see in what (non-idiosyncratic) sense of 'necessary connections' Humean laws (or L-laws in particular) can provide him with necessary connections between events. (Of course, he can believe that events are connected by "L-necessary connections").

# on 28 November 2010, 20:43

Hi Gabrielle,

Lewis does endorse logically (metaphysically) necessary connections between properties. e.g. between the property of having the mass of a proton and having mass since he construes properties as classes of possibilia and the former property is a subset of the latter one. So there is no conflict between Humeanism and necessary connections between properties. But the real issue between us is whether a connection between properties that is grounded in an L-law counts as "nomologically necessary." If "nomological necessity" is construed as "logical necessity" (as e.g. Shoemaker, Bird do) then L-laws are not "nomologically necessary." But the Humean is proposing an account of laws on which laws are contingent. So some other sense of "necessity" is at issue. If one accepts Lewis' account of laws (of course this is a big "if" and there are lots of arguments pro and con)then I see no reason why "nomologically necessary" cannot be defined in terms of the L-laws. On the other hand, if your view is that a relation counts as genuinely necessary only if it is logically necessary or is fundamental or only if it isn't reducible to the "Humean mosaic" then I see why you would think connections between properties entailed L-laws aren't necessary in your sense. Naturally, this is going to seem question begging from a Humean point of view. By the way, if you think of laws as contingently existing metaphysically fundamental items or features that in some way "produce" regularities then I don't see why that qualifies as making the regularities necessary (except in so far as they are "nomologically necessary" i.e. a consequence of laws). Those who hold this view should explain what these items are and how they manage to "produce" in this super metaphysical sense regularities.

# on 30 November 2010, 10:42

Hi Barry,

As I said, I see no reason for Humeans not to use the term 'nomologically necessary' as they please (although I do think they muddle things by using it as they do and I wish they were clearer about their use). Also, I do believe that the so-called laws of nature are metaphysically necessary (btw I take this to be the view of Shoemaker and Bird as well) but my problem with wo's "necessary" connections between events/properties/whathaveyou is neither that they are not metaphysically necessary nor that they are not fundamental (well at least that's not the best way to express my worry). The problem is that the so-called "necessary" connection plays no role in the generation of the pattern (so maybe it has to do with fundamentality after all). I think that that is a minimal condition for any meaningful talk of necessary connections (no matter how weak you want that necessity to be) when it coems to laws of nature.

Add a comment

Please leave these fields blank (spam trap):

No HTML please.
You can edit this comment until 30 minutes after posting.