Noa Latham

Noa Latham (University of Calgary), "Fundamental Laws," with commentary by Cei Maslen (Victoria University). Both the paper and commentary can be found here.

Comments

[Noa Latham]

Reply to Maslen

I’m grateful to Maslen for her careful working through of some issues in my paper. In an earlier draft I included a further nonHumean view in which laws are much as Maslen construes them on her NonHumean Method 1. Such laws do at least support intuitions of the kind illustrated in the example of the mile wide gold sphere. Perhaps they are the laws that “watch over” the facts without generating them or filling them out. But I couldn’t get any intuitive grasp of them or see how they could be defined. So I left them out, and used what Maslen calls the factmaking feature of laws as definitive of the two nonHumean conceptions of laws.

It seems to me that I am defining nonHumean laws on both generational and atemporal conceptions as strongly rather than weakly forwards (possibly probabilistically) deterministic, because the next slice of the universe is guaranteed from a single slice rather than a history of slices. Perhaps I could construct corresponding weak versions, but I’m not sure how. At any rate I do want to claim that a forwards deterministic generative law is not also backwards deterministic. It is not the case that in all possible worlds with laws L and a slice S2, an earlier slice S1 also exists. For there is a possible world with L that begins with S2 and advances forward in time from there. So I am inclined to stick to my view that an initial slice would be ultimately contingent if a universe is generated from it, while the generated slices would not be ultimately contingent. This simple picture of a generated universe in which all particular facts are ultimately explained in terms of fundamental laws and initial conditions I took for granted until I wrote this paper.

If a law on the atemporal conception is time-reversal invariant, then it is both forwards and backwards (possibly probabilistically) deterministic. But if it is not time-reversal invariant then I don’t want to say that that very law must be both forwards and backwards (possibly probabilistically) deterministic. Rather, I want to say that the law and a time-reverse of it must both be forwards (possibly probabilistically) deterministic. (We could have a deterministic law in one direction and a probabilistically deterministic law in the other.) I don’t think there is any serious disagreement between us here.

However, I am inclined to think Maslen is right that I was mistaken in claiming that none of the slices of a universe governed by nonHumean laws on the atemporal conception are ultimately contingent. It follows from my definition that all slices are ultimately contingent. This leaves the atemporal conception with no ontological advantage over the Humean conception—a conclusion I reached by the more long-winded route of claiming that a disjunction of the slices is ultimately contingent in such a universe.

I am happy to defer to Maslen’s theistic intuitions about the difference between Humean and nonHumean universes. But in case anyone should doubt that the distinction makes sense on the orthodox theistic view, I think it worth pointing out that it can clearly be maintained if one considers universes created by a limited deity.

There is unclarity in what I said about the generational conception of laws with regard to the role of an initial slice. I agree with Maslen that a generated universe need not be generated from a single slice. The only alternative, as far as I can see, is that it be generated from a conjunction of slices. I discussed this view in the section marked “A Proposed Modification of the Generational Conception”—a misleading heading as it is not the conception of generative laws that is modified. In that section I did mean to be starting a balance sheet for this view of a generated universe. The paper gets increasingly ragged at this point, so I’ll try to summarise that balance sheet here: On the positive side, this view captures what I take to be the intuitively plausible idea that there are slices of the universe that are not ultimately contingent but can be explained as generated by fundamental laws and an earlier particular fact about the universe. The rest is all negative, however. Any choice of a specific conjunction of slices from which the universe is generated would not be objective but would just be a systematisation of the facts. On any such choice, the slices that are part of the conjunction from which the rest of the universe is generated would not be explained as generated, even though they would seem to be generated. The only way I can see for a specific conjunction to have an objective special status would be for it to be selected by a deity. But even then it appears that such a selection would be completely arbitrary. A conjunction of slices from which the universe is generated would not be ultimately contingent. However, a universe generated from a conjunction of slices would have its fundamental laws and an infinite disjunction of conjunctions of slices as ultimately contingent facts—a greater stock of ultimately contingent facts than a pure Humean universe would have.

[Brad Weslake]

Thanks to the authors for two interesting papers. I have two comments on Latham's paper.

Firstly, Latham (p. 7) makes it seem as if the dispute between Armstrong and Lewis is over which facts are to be taken as primitive. But I thought the Lewis challenge to Armstrong does not simply concern the introduction of more primitive facts, but also concerns their alleged explanatory power. In particular, the challenge is to make sense of the idea of natural necessity -- the idea that the laws constrain how the facts can go. The problem isn't that the laws themselves are primitive and thereby can't be explained, but that their operation can't be explained. I take it that the Lewis argument is that since they can do no explaining (that their operation is at least as mysterious as the regularities they are supposed to explain), they are metaphysically otiose.

Secondly, given the distinction Latham (p. 9) makes between the "extrinsic" direction of time, defined as the direction of generation, and the "intrinsic" direction of time, roughly understood as the temporal asymmetries in particular matters of fact, I found Latham's constraint that extrinsic time can only be ordered in one intrinsic direction in a universe an unusual one. Why can't there be a time-slice somewhere in the middle of a universe which serves as the origin of extrinsic temporal direction, with further time-slices generated successively both ways (intrinsic past and intrinsic future) out to the temporal edges? This possibility should be no more surprising than the possibility Latham does consider, of the extrinsic direction being the reverse of the intrinsic direction. But then once we have countenanced these possibilities, it also seems natural to suppose that the extrinsic order of generation could go any which way with respect to the intrinsic temporal orientation -- extrinsic-first timeslice intrinsic-five, then extrinsic-second timeslice intrinsic-seven-hundred-and-two, and then so on arbitrarily until the universe is wholly filled in. (Surely God has all of these options at her disposal, to put it in terms of the theological metaphor?). This line of thought I think lends some further weight to the Humean view, since it shows just how underdetermined the generational laws would be by the particular matters of fact in such a non-Humean universe. If on the other hand Latham would like to keep the uni-directional constraint, I'd be interested in hearing more about how the claim that "time flows only one way within a single universe" can be justified, keeping in mind that it is extrinsic, generational time to which the claim applies.

[Brendan Leier]

In the big book, Hume introduces the notion of 'hidden causes'. Can we interpret Hume to be a causal skeptic if he believes in hidden causes? Aren't these two positions mutually exclusive?

B

[Noa Latham]

Hi Brad,

In response to your first question, I agree that Humeans claim that the operation of nonHumean laws is mysterious. But assuming that the nonHumean conception of laws cannot be shown to be incoherent, I want to argue that if there are such strange (to Humeans) but not incoherent entities as nonHumean laws, then much about the universe admits of a kind of ultimate explanation that is unavailable on a Humean view. So on the nonHumean generational view, not only are the fundamental laws themselves primitive in the sense of being ultimately contingent facts, but the conceptual fact that nonHumean laws generate slices has to be taken as a primitive conceptual fact.

Second question, first part: I think the disagreement between us is only terminological. I wanted to develop what I think is the intuitive nonHumean view that laws generate successive states of the universe in just one temporal direction—in the direction we describe as progressing from past to future. I also wanted to explore the possibility you raise of a universe generated in both temporal directions from an intermediary slice, but I preferred to call this “filling out” rather than “generating” the universe, and to call the corresponding conception of laws “atemporal” to distinguish it from the one-directional conception.

Second question, second part: I think one can conceive of a nonHumean universe in which the description of the “law” is a long instruction specifying in full detail what each slice of the universe is to be. But this would be no simpler than a Humean universe and thus not an attractive nonHumean option.