Introductory: A Popular Argument
Very often, especially in elementary discussions in the philosophy of science, one encounters an argument like the following, to the effect that a certain view about laws of nature is dead wrong:
1. Suppose that the (naïve) Humean view is correct – that is to say, suppose that every true general statement is a statement of a law of nature.
2. Suppose that, as a matter of fact, (*)On 8 May 1945, every coin in Henry’s right pocket was silver.
3. Hence (*), being a true general statement, is a law of nature.
4. Every law of nature supports its relevant counterfactuals.
5. However, (*) does not support all of its relevant counterfactuals; e.g., if this non-silver penny were in Henry’s pocket on VE day, it would not have been silver.
6. Hence (*) is not a law of nature.
7. Therefore, the Humean view is incorrect.Footnote 1
The purpose of this article is to cast some doubts on the legitimacy of step 5 of the above argument. Roughly speaking, I will argue that without already having assumed that (*) is not a law of nature, one cannot say of (*) that it does not support its relevant counterfactuals. Hence the displayed argument does not work. For all I say here, the Humean view may be incorrect, and there might be other reasons to think that (*) is not a law of nature—it is not my aim to dispute the conclusion of the argument or the truth of 6; my task is to show that this particular argument is not a good one. At the end of the article, I will briefly comment on the thesis of counterfactual support (premise 4) and point to an empiricist-friendly candidate for explaining why the thesis is thought to be true.
For the sake of discussion, I will not challenge a presupposition of the argument under consideration, namely that the notion of a law of nature is neither incoherent nor useless.
Counterfactual Support: Definition
“Law-statements support counterfactuals,” writes David Armstrong right at the beginning of a section on counterfactuals in his What Is a Law of Nature? (46).Footnote 2 Defining what exactly is meant by “counterfactual support” is not a minuscule project.Footnote 3 To simplify, I will forget about the relevancy requirement in our discussion of the thesis that laws of nature support their relevant counterfactuals.Footnote 4 Plus, I will focus on those laws whose statements are of the form “Every F is G.” I also will be less than ideally careful about the distinction between laws and their statements. Now for the definition.
Suppose that, as a matter of fact, every F is a G. We say that this fact supports its counterfactuals if for every non-F object a, if a were an F it would also be a G. Footnote 5
It is widely held that real laws of nature – whatever they are – support their counterfactuals. Thus in an old paper, Chisholm asks us to imagine that every man in a certain park at a certain time is Irish. Now, presumably, Ivan the non-Irish would not have fallen in the extension of “is Irish” if at the specified time he were at that park.Footnote 6 In this respect, “Every man in that park at that time is Irish” is different from “Copper always expands upon heating,” to use an example from Nagel (51). Take a piece c of copper that has never been heated, and destroy it so that it will never be heated in future either. We do assent to “c would have been expanded, had it been heated.”
Examples are abundant; to make my main point, I will focus on a well-known case by Michael Tooley.
Laws and Counterfactuals: The Problem of Epistemic Priority
Smith plants only apples in his garden, and successfully takes all kinds of precautions to prevent the growth of any other fruit there. Tooley writes, “This regularity [viz., All the fruit in Smith’s garden are apples] would not be a law, since laws must support counterfactuals” (1987, 43). Tooley’s intuition, as well as many others’, is that although the generalization
(S) All the fruit in Smith’s garden are apples is true, it is not a law of nature. Several features of (S) – such as its reference to a particular garden – may contribute to its non-lawhood; but the point of the example is that (S) fails to support counterfactuals like
(O) If this orange were in Smith’s garden, it would be an apple.
As I represented it, Tooley’s choice of a counterfactual relevant to (S) leaves something to be desired. For if one assumes, perhaps justly, that apples and oranges form two distinct natural kinds, then one would be inclined to think that no orange could possibly be an apple, so that the consequent of (O) turns out to be metaphysically impossible. Now, such degenerate cases of counterfactuals (as well as those with impossible antecedents or necessary consequents) are not of much interest – at least when we are dealing with the issue of counterfactual support of laws of nature. But I admit that the defect is not hard to remedy. Tooley could have presented a scenario where Smith plants only red apples in his garden, and a relevant counterfactual would be: If this green apple were in Smith’s garden, it would be red. Since my argument against the revised version would be the same, mutatis mutandis, as the one I will present below, in the remainder of my discussion I will put away this concern about (O) and deal with Tooley’s own example.
My qualm concerning Tooley’s intuition is that I doubt if we have a firm understanding of the truth-conditions of counterfactuals prior to our understanding of what a law of nature is. Or, what is more to the point here: I doubt if we have an understanding of the truth or falsehood of any given counterfactual prior to having an opinion about the lawhood of some statements related to it. It is not the case that we first have some opinion about the falsity of (O) and then we use it to see that (S) is not a law; rather, I submit, we say that (O) is not true because, inter alia, we do not think of (S) as a law – we do not think that, as a law and in normal circumstances, whether a piece of fruit is an apple depends on where it is. (Perhaps in this case something stronger holds: we tacitly consider it as a consequence of a law that the applehood of a piece of fruit is independent of its location.)Footnote 7
Or suppose that a force acts on an object o (mass: 3kg), and o’s velocity increases with the rate of 10ms-2. We assent to “Had the force been less than 30N, o’s acceleration would not be this much,” because we think of something like F = ma as a law of nature. Our judgment about the truth of this counterfactual is grounded in our positive judgment about the lawhood of Newton’s second law. In the absence of such a law, we would be clueless about the truth-value of the mentioned counterfactual statement.
That the judgment about the truth of a counterfactual comes after, and is grounded in, the judgment about the lawhood of certain relevant statements becomes clearer if we consider cases where a speaker does not know about the status of the relevant generalized claim as a law. Suppose that someone has contracted syphilis. She also had unprotected sex. Question: Would she still have gotten infected had she abstained? We do not consider ourselves justified in answering affirmatively unless we think of something like “Syphilis is an STD (modulo congenital cases)” as a law. No laws; no judgments about truth-values of counterfactuals.
A caveat might be in order here. I argued that our judgment about the truth-value of every counterfactual is grounded in our judgments about the lawhood of some relevant statements. This might be challenged by examples of the following kind. Suppose we know that, for some bizarre reason, Harry consistently observes the rule of not having non-silver coins in his pocket. Then we might be inclined to think that if a coin were in Harry’s pocket, it would have been silver. Similarly, we might be inclined to assent to “If I had attended the royal wedding, I would have dressed formally.” (Though the inclination to assent to these two counterfactuals is, I surmise, weaker than our inclination to assent to something like “If I had applied some force to this object, it would have accelerated,” which is backed by a natural law.) But no one would say that Harry’s idiosyncrasies or the regal dress code count as laws of nature. So it seems that we do have, after all, examples of judging the truth-value of a counterfactual independently of any judgment about the status of a relevant general claim as a law of nature.
I think that this is an objection to the letter, not the spirit, of my proposal. Although the role of the thesis of counterfactual support can be understood as demarcating accidental generalizations from laws of nature, it need not be described so narrowly. A better description, I believe, is that counterfactual support is supposed to distinguish accidental generalizations from non-accidental ones, be they laws of nature or not. Thus, suppose we have noticed that all people at the royal wedding were dressed formally, and that we have noticed the same about all people at a philosophy seminar. Of these two general truths, we judge that the latter, but not the former, is accidental, and a mark of the difference is that the former supports the counterfactual “If I had attended the royal wedding, I would have dressed formally,” but the latter does not support “If I had attended the philosophy seminar, I would have dressed formally.” And the gist of what I have been urging stands: our judgment about the truth-value of a counterfactual is grounded in our judgment about the non-accidentality of some relevant general statements. It seems to me, then, that I can adequately reply to the objection simply by replacing the law discourse by the non-accidental one. But let me resume talking about “laws of nature.”
Back to the discussion I digressed from two paragraphs ago: It seems reasonable to deny that the putative relationship between laws and counterfactuals makes counterfactuals significant, so far as judgments about the lawhood of particular statements are concerned. Whether or not laws support their counterfactuals, tracking the truth of counterfactuals cannot be used to settle the question about lawhood of a statement like (*). Thus the argument mentioned at the beginning of this article does not work.
It should be clear that the present worry about laws and counterfactuals is out-and-out epistemological. Here I do not deny that counterfactuals might have objective, judgment-independent truth-values autonomous from relevant laws; my thesis, rather, is that one cannot find out those truth-values, should they exist, without laws. If correct, this shows that counterfactuals cannot help the scientist to settle the lawhood of a particular statement like (*) – nor can they help philosophers in such a decision. So, from this specific epistemic respect, counterfactuals are scientifically unimportant.
Counterfactual Support: An Empiricist Conjecture
But do real laws of nature support their counterfactuals anyhow?Footnote 8 I will briefly suggest a modality-free analysis, agreeable to the palate of a radical (imaginary) empiricist who might hold the naïve Humean view, of why counterfactual support is considered a requirement for laws.
The piece c of copper Nagel talks about (section 2, above) is destroyed; we cannot heat it and see if it expands. Even if we are not as scrupulous as Nagel to make the non-testability explicit, it should be clear, with a moment’s reflection, that to the extent that we are talking about counterfactual conditionals, the antecedent of the conditionals in question are unalterably false (this may explain why giving truth-conditions for counterfactuals is considerably more difficult than the case of material conditionals); thus one cannot test them directly empirically.
Perhaps the claim that counterfactuals are not empirically testable is less obvious for statements like (O): can’t we empirically determine the truth-value of (O) by trying to bring an orange to Smith’s garden? This might be what Tooley has in mind in the passage partially quoted above:
This regularity would not be a law, since laws must support counterfactuals, and the regularity in question would not provide any ground for believing that if Smith had attempted to put an orange in the garden, he would have failed.
On the one hand, Tooley is talking about counterfactuals. On the other hand, as if he is justifying his claim about the falsehood of a counterfactual, he says that one has no grounds to believe that no orange could be brought to the garden without ceasing to be an orange.Footnote 9 However, one should guard against a lurking confusion here. Trying to bring an orange to the garden is not to test the truth of any counterfactual whatsoever. If we bring an orange to the garden and examine its applehood, we are examining a new actual case to see whether it confirms or disconfirms the general statement (S): It will either give us a counterexample to (S), or else will provide another instance of it. The counter-factuality of a counterfactual lies in the fact that its antecedent is forever false; you cannot make the antecedent true and see if the consequent obtains. To think otherwise is to confuse “If a piece of fruit were in Smith’s garden, it would have been an apple” with “If a piece of fruit is in Smith’s garden, it is an apple.” (What is a counterfactual situation is this: Had Smith, who never actually tried to bring an orange into his garden, tried to do so, the orange would have transformed into an apple. We have no empirical way to test this counterfactual.)
Without accusing Tooley of being confused about this, here is, then, a conjecture about the counterfactual-support assumption: I submit that occasionally, in thinking about counterfactual support, one mistakes the desirability of inductive support for the alleged requirement of counterfactual support.Footnote 10, Footnote 11
