This review appeared in The Journal of Philosophy (1990), pp. 432-5.

The Concept of Physical Law. NORMAN SWARTZ. New York: Cambridge University Press, 1985. 220 + xi p.  Cloth $29.95.

This volume is devoted to a systematic defense of the "regularity" view of physical (i.e., natural) laws against the claims of those who would distinguish in some "ontological" way between generalizations which are merely "accidentally" true and those having genuine "nomic necessity." All contingent generalizations can, at best, truly state the uniformities present in the actual world. Attributions of some kind of modal force to some, alleging that they hold true not simply in the actual world but in all "physically possible" worlds as well, rest on faulty intuitions which cannot be backed up by any adequate theory. Much of the argument is framed in terms of the debate over existential assertions, rather than that over generalizations. Here the proponent of a distinction between the genuinely lawlike and the merely accidental claims a distinction between existentials that are "doomed" to falsity (since they conflict with genuine laws) and those merely false, and, hence, merely contradicting true accidental generalizations. Norman Swartz repeatedly insists that, insofar as the issue of possibility arises, it arises for all the existential assertions equally. They are, if false, "impossible" relative to some general assertion's being true. But only that kind of relative possibility is intelligible. Some further distinction in the (physical) modality of the generalizations whose truth guarantees their falsehood is merely an illusion.

What is wrong with the notion of nomic necessity? It is, Swartz says, impossible to explain what it means without invoking notions, such as "physically accessible possible worlds," that can themselves only be understood by utilizing the dubious notion of nomic necessity with which we started. The proponent of nomic necessity fails, Swartz claims, to give us an adequate epistemological account of how we could gain epistemic access to what goes on in the alleged other physically possible worlds and what does not happen in them. More specifically interesting in the attack on this particular modality is the absence, according to Swartz, of an adequate criterion to distinguish the genuinely lawlike from the merely accidental. Logical (and, perhaps, mathematical?) assertions are characterized by a distinctive formal structure that marks them off from all other assertions, so that we can, if we want, attribute a special modal status to them. But the genuinely lawlike generalizations are not so a priori distinguishable from the mere accidental generalizations. Nor is there some clear empirical means of determining which of the generalizations true of the world have nomic status.

In the usual Humean tradition, one would expect the author to spend a fair amount of time trying to offer us at least a "psychological" account of the source of our familiar intuitions about the distinctions between generalizations that support counterfactuals and those which do not, to note one traditional alleged distinction between the laws and the mere generalizations. We do not get too much help on this from Swartz, though. He does say that the counterfactual supporting generalizations are more "pervasive" than the others, having narrower ceteris paribus clauses implicitly attached to them. But by itself that is not much to go on. He says that just because a piggy bank has all copper coins in it we do not infer, "If this coin were in the piggy bank it would be copper." Our ground for denying the counterfactual force, he says, is the existence of lots of piggy banks with noncopper coins in them. But that, as Swartz acknowledges, will not, by itself do. We do infer, "If this were an electron it would have a unit charge," even though electrons are all particles and lots of particles do not have unit charges.

One might expect, at this point, the usual resort of the Humean to factors not directly related to the truth conditions of the assertion to make the distinction for us. Indeed, Swartz has some extended discussion of Nicholas Rescher's1 particular version of that sort of thesis where he advocates a version of the "role in the scientific hierarchy" model of "imputed lawlikeness." Swartz devotes his attention, though, to arguing that such a version of the grounds for the distinction between the lawlike and the accidental would not be what the proponent of nomic necessity as an "ontological" feature of the world intended. No, it would not. But, of course, the Humean line has always been that a reconstruction of the distinction based on epistemic, pragmatic, or "psychological" features is part and parcel of an adequate account that will do justice to the real grounds of the intuitions on which the "metaphysician" relied in trying to convince us of the existence "in the world" of the structure we are now, in a sense, denying. Swartz's line of boldly saying of all false existentials that they are equally impossible of truth (since all equally incompatible with some true general assertion or other, and all such generalizations are, modally, on a par) is, perhaps, less satisfying than the Humean program that "reconstructs" the very notion of possibility, explaining (or, if you want, explaining away) our systematic intuitions instead of just denying them.

The book has three sections, "Theory," "Applications," and "The Theory Extended." The issue of free will is taken up in the second part. Here Swartz begins with the plausible regularist line that free will is compatible with lawfulness as the regularist construes it. Locutions telling us that the law "makes" the choice go as it does are misleading. It is our choosing as we do, rather, that determines what laws hold, the laws being nothing but general summaries of the particular real happenings in the world. But he then goes on and argues for unpredictability as the sina qua non of a free choice. Even on a surface level this is a problematic thesis, of course. The old examples of the predictable free choice of the rational agent versus the unpredictable random activity of the unfree madman come to mind.

There are problems at deeper levels, too. Swartz uses a coarse-grained notion of event (or action) so that an action could be redescribed and placed under quite different laws than the kind of generalizations connecting unique antecedent to unique action which, he says, could not have been inferred from past experience. Swartz does take this up, suggesting that, even at the lower levels, the antecedent adequate for the action will be so complex that the proper regularity could not be inferred. But, of course, the usual line would be that the antecedent is just a congeries of simple facts connected to the action, also now a collection of simple facts, by the standardly inferred physical regularities of the world. Is it just then the complexity of initial conditions (rather than the noninferability of the appropriate regularity, as Swartz seems to suggest) that makes all actions appropriately called "free" genuinely unpredictable? Swartz also thinks that another "positive" condition on action's being free (over and above the compatibilists usual "negative" conditions, like the action being noncoerced, not the result of drugs or electrodes in the brain, etc.) along with unpredictability is that the actor not be "uniform" in his behavior. This is a tricky notion (being relative to the description of the action chosen). Nor is it so clear that even action under a guise that takes it as uniform should be, a fortiori, classified automatically as unfree.

At one point (144) Swartz seems to suggest that, if all events were rationally forecastable, this would provide support for the "autonomy theory," i.e., for the idea of a distinct kind of "ontological" nomic necessity. Since, he thinks, events are not universally so rationally predictable, no such support exists. But why would a world that satisfied the "Laplacean paradigm" support the "autonomy theory?" Just what is supposed to be incompatible between predictability (in principle or otherwise) of all events from inductively inferable generalizations and knowable singular initial conditions, and the view that the regularities in question are, like all regularities, nothing but true generalizations holding in the actual world? There seems to be some implicit assumption here that, if all the regularities were so inferable from less than an inspection of all their instances, this would indicate the truth of the necessitarian's claim that the laws "made" events go the way they go and did not just summarize how, in fact, the events do go. But why should Swartz's regularism be so threatened by such a possibility of inductive inferability of laws and predictability of singular events?

One place where Swartz's regularism may be going too far is where (137) he suggests that, even for logical necessity, talking of the events as "complying" with the law is misleading. Instead, he says, with the use of possible-world semantics we can see that even logical truths just "take their truth from the way the world is" (138). But summarizing what goes on in "all possible worlds" is not like summarizing what goes on in the actual world. What the possible worlds are is constrained by the class of logical truths, unless, I suppose you really hold to a "realistic" possible-worlds way of looking at things where the actual world is just our neighborhood of reality. At this point, I think Swartz's extension of his idea that the singular facts are primary and all generalization supervenient on them gets misleading.

Many other topics are taken up: miracles and marvels from the regularity perspective, the place of statistical generalization in the regularist view, dispositions as a putative ground for natural necessity, and others. Overall the book is very well structured. The positions taken and the arguments for them are laid out with clarity. Many very trenchant replies to misguided "refutations" of regularist views are forcefully and convincingly presented. The book is a welcome contribution to the recurrent debate on the nature of lawlike generality.


University of Michigan

1 Scientific Explanation (New York: Free Press. 1970)