The Necessity of Nature

Alfred J. Freddoso
University of Notre Dame

This paper lays out the main contours of an objectivistic account of natural necessity that locates its source within natural substances themselves. The key claims are that what occurs by a necessity of nature constitutes the culmination of deterministic natural tendencies and that these tendencies are themselves rooted in the natures or essences of natural substances. The paper concludes by discussing the notion of a law of nature as it emerges on this account.

You struggle out of bed early one dreary winter morning, find your way into the kitchen somehow, and, after a mandatory thirty seconds or so of groping for utensils, faucets, and dials, finally succeed in situating the kettle of water squarely over the blue gas flame. When you return from the other room ten minutes later, the water is beginning to boil. A familiar enough scene, yet one which, like many another such scene, gives rise to philosophical puzzlement.

It is no accident, nearly all of us will grant and, indeed, insist, that this water should be boiling at this time (call it time 'T'). But just what sort of nonaccidentalness or necessity is involved here? Obviously, the water's boiling at T is not a matter of metaphysical necessity; after all, it is not even metaphysically necessary that the water should so much as exist at T, much less be boiling at T. The necessity in question seems clearly distinct from the necessity that attaches to the past simply by virtue of its being the past; one can easily conceive of a world that has the same history as our world just after you put the kettle on the flame but in which someone intervenes before T to prevent the water from boiling.

What we have lighted on here is the modality commonly dubbed the necessity of nature, or natural necessity. Accordingly, given the scene described above, we might reasonably affirm that the proposition This water is boiling is true at T by a necessity of nature; or, to pass from the de dicto to the de re, we might reasonably affirm that it is by a necessity of nature that at T, this water has the property of being such that it is boiling. Still, agreement about the name cannot long conceal the deep disagreements about the thing itself.

Since the time of Hume, or, more accurately, since the appearance of occasionalism as championed by the likes of al-Ghazali and Gabriel Biel in the Middle Ages and by Malebranche and Berkeley in a subsequent era, philosophers have been propounding what we might aptly call subjectivistic explications of natural necessity.1 The occasionalists located the source of this necessity in God's free decision to bring about natural events in accordance with certain arbitrary but (barring only a few exceptions /216/ strictly adhered to rules, rules reflected in the empirical generalizations that scientists strive to discover and systematize. Modern thinkers, as is their wont, shift the focus from the divine mind to the human mind. Hume claimed to have found the source of natural necessity in the psychological habit whereby we come to expect the occurrence of events similar to the effect on the occasion of our experience of events similar to the cause; whereas later, more chastened Humeans have tended to ground natural necessity in one or another epistemic property, e.g., in the explanatory potential or degree of confirmation possessed by lawlike generalizations.2 Disparate though they be, these theories share in common the conviction that natural necessity, while dependent in a more or less direct way on some knowing or acting subject or subjects, is entirely independent of and extrinsic to the "objective" natural substances that are involved in ordinary causal interactions, or, to use a more neutral locution, ordinary causal sequences. Not surprisingly, subjectivistic accounts of natural necessity typically go hand in hand with what, from a realist point of view, appear to be unduly thin and attenuated metaphysical characterizations of natural substances. By way of corroboration, one need only cite Malebranche, Berkeley, and Hume.

Nearly all the subjectivistic accounts I know of handsomely repay close scrutiny; some are downright ingenious. I have argued at length elsewhere, for instance, that from a theistic standpoint, occasionalism provides a far more impressive and satisfying philosophy of nature than most contemporary theistic intellectuals have cared (or dared) to admit.3 In this paper, however, I will dutifully set aside such perverse thoughts and accept without argument the now widely held opinion that natural necessity, whatever else might be said about it, is an objective feature of the world. My goal, accordingly, will be to formulate the main contours of an objectivistic account of natural necessity, one that locates the source of this necessity within natural substances themselves.

I will begin in section I by characterizing two other families of modal notions that will be featured prominently in my account of natural necessity, namely, the metaphysical modalities, and what I have elsewhere dubbed the accidental modalities. Then I will proceed in section II to lay out, roughly and informally at first and more precisely later, the core elements of my account of natural necessity. The key claims embodied by this account are (i) that what is true at a given time by a necessity of nature constitutes the culmination at that time of deterministic natural tendencies and inclinations, and (ii) that such tendencies and inclinations are themselves grounded in the essences or natures of natural substances, specifically in the active and passive causal dispositions had essentially by those substances. Finally, in section III I will entertain and respond to some objections.


I

I will commence by explicating the metaphysical modalities and marking them off clearly from much that goes by the name "logical" or "conceptual" or "transcendental" necessity, impossibility, and contingency. The formulas that follow presuppose the necessary existence of such abstract entities as properties, propositions, states of /217/ affairs, and possible worlds, though the position I am advancing in this paper in no way hinges on the claim that these formulas cannot be replaced without loss by nominalistic paraphrases.4 1 am taking it for granted that propositions are tensed, so that many of them are capable of being true at one time in a given possible world and false at some other time in that same world. What's more, I am assuming that there is an exact isomorphism between propositions and states of affairs, and hence that states of affairs as well as propositions are tensed. Accordingly, I take a possible world to be a temporally ordered sequence of maximal possible states of affairs.5

Let' s start with the following equivalences for the de dicto metaphysical modalities:

    Proposition p is metaphysically necessary if and only if p is true at every moment in every possible world.

    Proposition p is metaphysically impossible if and only if p is false at every moment in every possible world.

    Proposition p is metaphysically contingent if and only if p is neither metaphysically necessary nor metaphysically impossible.

Notice that de dicto metaphysical necessity as characterized here has no obvious connection with the logical form of sentences that express metaphysically necessary propositions, or with the question whether such propositions are knowable a priori or a posteriori or at all, or with the question whether the truth of some such propositions is presupposed by the very possibility of any human being's or of any rational being's having knowledge. All these topics, engaging though they be, are irrelevant to the matter of just which metaphysical modality a given proposition has. So, for instance, if a proposition is logically necessary only if it is knowable a priori, or only if it is expressible by a sentence that is a truth of first-order logic or whose negation is self-contradictory, then not every metaphysically necessary proposition is logically necessary. Since some philosophers seem to use the term "logically necessary" with one or another of these connotations in mind, we must carefully distinguish metaphysical necessity from logical necessity so understood. Again, if a proposition is logically or conceptually necessary only if a sentence expressing it is such that (the meaning of) its subject contains (the meaning of) its predicate or such that whoever understands it thereby sees that (the meaning of) its subject contains (the meaning of) its predicate, then not every metaphysically necessary proposition is logically or conceptually necessary. Since some philosophers seem to use the terms "logically necessary" and "conceptually necessary" with one or another of these connotations in mind, we must carefully distinguish metaphysical necessity from logical or conceptual necessity so understood. Again, if a proposition is transcendentally necessary only if its truth is a basic presupposition of any conceptual framework in terms of which the world can be understood, or only if its truth is required in order for a human being or for a rational being to have knowledge of the world, then not every metaphysically necessary proposition is transcendentally necessary. For there are many metaphysically necessary propositions that are so exotic or arcane as to have nothing at all to do with the possibility that a human being (or a rational being) should have knowledge of the /218/ world. Since some philosophers seem to use the term "transcendentally necessary" with one or another of these connotations in mind, we must carefully distinguish metaphysical necessity from transcendental necessity so understood.

Now that these cautionary notes have been issued, we can move on to the de re metaphysical modalities. To coordinate this part of the paper with what follows, I will characterize these modalities with respect to a given time in a given possible world:

    It is metaphysically necessary for entity x to have property P at moment t in possible world w if and only if  

      (a) x exists at t in w, and
      (b) for any moment t* and any possible world w* such that x exists at t* in w*, x has P at t* in w*.

    It is metaphysically impossible for entity x to have property P at moment t in possible world w if and only if

      (a) x exists at t in w, and
      (b) for any moment t* and any possible world w* such that x exists at t* in w*, x lacks P at t* in w*.

    It is metaphysically contingent for entity x to have property P at moment t in possible world w if and only if

      (a) x exists at t in w, and
      (b) it is neither metaphysically necessary for x to have P at t in w nor metaphysically impossible for x to have P at t in w.

As I understand them, these de re modalities exhibit a similar independence of all questions concerning logical form, type of knowability, conceptual frameworks, and relations among concepts. I state this point partly by way of assertion and partly in order to announce an intention. When, dissociating himself from an illustrative example he has just used, D. M. Armstrong comments parenthetically that he does not himself "believe in essential properties, save relative to some conceptual scheme," I construe this remark as implying that he does not believe in essential properties at all.6 For, as I see it, a property P is essential to a given entity x only if x has P with metaphysical necessity and thus has P regardless of how (or even whether) x is conceptualized or described by any human being or by any rational creature whatsoever. In fact, I find it difficult even to imagine how the conceptual framework within which an entity is conceived of or described can have any bearing at all, one way or the other, on which properties are essential to it. If this is too controversial a thesis to be stated baldly and without extended argumentative support, then I will be content here simply to make known my intention to use the term 'essential property' only in the way just adumbrated. Given this usage, which is by no means idiosyncratic, Armstrong's statement can be correctly interpreted as implying, or at least strongly suggesting, that he believes that no entity actually has any property with metaphysical necessity or, hence, essentially. Of course, the above formulas merely characterize metaphysical necessity; none of them requires that any entity in fact has any property with metaphysical necessity or that any proposition is in fact metaphysically necessary. /219/

The other family of modalities I will use in fashioning my account of natural necessity are what I call the accidental modalities. Intuitively, a proposition is accidentally necessary at a given time just in case it is a metaphysically contingent proposition that can no longer be false at or after that time. Suppose, for instance, that Michael went swimming yesterday. Then most of us would be inclined to say that the proposition Michael has gone swimming, though metaphysically contingent, is now, given the history of our world, no longer possibly such that it is or will be false. If so, this proposition is now accidentally necessary, and its negation, It is not the case that Michael has ever gone swimming, is now accidentally impossible. Moreover, all but unrepentant fatalists will cheerfully concede that there are many future-tense propositions that, along with their negations, are now contingent in the corresponding sense. Clearly, a proposition' s accidental modality is relative to a given time. What' s more, since the accidental modalities are a function of the world's history at a given moment, and since the history of a possible world w at a moment t may differ from the history of another world w* at that same moment t, a proposition's accidental modality is relative to a given time in a given possible world. So we have these de dicto accidental modalities:

    Proposition p is accidentally necessary at moment t in possible world w if and only if

      (a) p is not metaphysically necessary, and
      (b) p is true at t and at every moment after t in every possible world w * such that w * shares the same history with w at t.

    Proposition p is accidentally impossible at moment t in possible world w if and only if

      (a) p is not metaphysically impossible, and
      (b) p is false at t and at every moment after t in every possible world w * such that w * shares the same history with w at t.

    Proposition p is accidentally contingent at moment t in possible world w if and only if

      (a) p is metaphysically contingent, and
      (b) p is neither accidentally necessary at t in w nor accidentally impossible at t in w.

What is it for two worlds to have the same history at a given moment? This question can be given substantively diverse answers, each associated with a distinctive response to the challenge of logical determinism. My own, Ockhamistic answer has been set out in detail elsewhere,7 but it might be worthwhile to outline it here, since I will be presupposing it in the discussion of the natural modalities.

The guiding idea is that what is temporally (as opposed to causally) independent at any given moment is wholly a function of the present-tense (or, as I prefer to say, immediate ) propositions true at that moment. All nonimmediate, or temporally dependent, propositions true at a time t are true at t only by virtue of the fact that the appropriate immediate propositions were or will be true at moments other than t. So, /220/ for instance, the nonimmediate proposition Michael has gone swimming is true now by virtue of the fact that the immediate proposition Michael is swimming was true at some past time. Moreover, for any moment t in any possible world w there is a set k of immediate propositions that determines all that is true at t in w in a temporally independent way, i.e., true at tbut not by virtue of what occurs in w at times other than t. (Keep in mind that the sort of dependence in question here is temporal, and not causal. ) I call k the submoment of t in w and say that k obtains in w when and only when each of its members is true. Then two worlds share the same history at t if and only if they share all and only the same submoments, obtaining in exactly the same order, prior to t.

Given this overall picture, how are we to distinguish precisely between immediate and nonimmediate propositions? My own explication of this distinction, laid out in the place alluded to above, is too complicated to be presented in passing here. Still, the intuitive idea is relatively clear: what is "purely present" (immediate) at a given moment is distinct from what is strictly past or strictly future (nonimmediate) at that moment. The fact that it is no mean feat to find a general algorithm for separating immediate from nonimmediate propositions does not in any way impugn this guiding intuition.

In addition to the de dicto accidental modalities, there are also de re accidental modalities. For example, if Michael went swimming yesterday and still exists today, we might reasonably claim that it is now accidentally necessary for him to have the property of having gone swimming, and that it is now accidentally impossible for him to have the property of never having gone swimming.8 That is, given the present history of the world, it is no longer possible for him ever to lack the former property or ever to have the latter. So, it seems, with the passage of time certain properties that are metaphysically contingent for a given entity to have become accidentally necessary or impossible for it to have, just as, it seems, with the passage of time certain propositions that are metaphysically contingent become accidentally necessary or impossible. By the same token, there are presumably many properties that are now not only metaphysically contingent for Michael to have but also accidentally contingent for him to have. The following formulas seem to capture these de re modalities:

    It is accidentally necessary for entity x to have property P at moment t in possible world w if and only if

      (a) x exists at t in w, and
      (b) it is not metaphysically necessary for x to have P at t in w, and
      (c) for any possible world w* such that w* shares the same history with w at t, and for any moment t* such that t* is at or after t and x exists at t* in w*, x has P at t* in w*.

    It is accidentally impossible for entity x to have property P at moment t in possible world w if and only if

      (a) x exists at t in w, and (b) it is not metaphysically impossible for x to have P at t in w, and (c) for any possible world w* such that w* shares the same history with w at t, and for any moment t* such that t* is at or after t and x exists at t* in w*, x lacks P at t * in w*. /221/

    It is accidentally contingent for entity x to have property P at moment t in possible world w if and only if

      (a) x exists at t in w, and (b) it is not metaphysically contingent for x to have P at t in w, and (c) it is neither accidentally necessary for x to have P at t in w nor accidentally impossible for x to have P at t in w.

Remember that the above formulas merely characterize the accidental modalities and do not by themselves require that any proposition is, in fact, accidentally necessary or impossible, or that it is now, in fact, accidentally necessary or impossible for any given entity to have any given property. So, as with the discussion of the metaphysical modalities above, many substantive questions are left open. I have simply provided a framework for expressing various theses about the metaphysical and accidental modalities and for asking questions about them and their relation to the natural modalities. For instance, it seems intuitively clear that the necessity of nature is not just a species of the necessity of the past. Pace logical determinism, accidental necessity does not seem to be determinative of the future in the strong way that natural necessity is. But just how do these two types of necessity differ from one another? And just how is natural necessity related to metaphysical necessity? The answers to these questions will, I hope, soon begin to emerge.


II

Nowadays universal causal determinism, though still revered as a regulative ideal in certain influential philosophical circles, has generally speaking lost the well-nigh irresistible charm it had for intellectuals as recently as a century ago. In science, indeterminism with respect to individual entities and events has cropped up in various sectors of physics, chemistry, and biology. In philosophy, strong voices have recently emerged challenging the reigning compatibilist consensus and proclaiming as of old that what I will tendentiously call 'genuinely' free action--be it on the part of God, angels, human beings, or nonhuman animals (with respect to at least some of their movements)--requires the absence of causal necessitation or determinism and the presence in the agent of an ability to choose and to do otherwise in the very same circumstances as those in which the free action is in fact chosen and performed. Since I am hardly in a position to gainsay the scientific case for indeterminism, and since I enthusiastically subscribe to the belief that genuine freedom of action is commonplace, I reject the thesis that every cause brings about its effects deterministically, along with its implication that everything that is caused to be true in the spatiotemporal universe is true by a necessity of nature. However, this rejection is not a part of, nor is it presupposed by, my account of the necessity of nature. The only thing presupposed is the extremely modest methodological claim, able to be conceded in good conscience by even the most fanatical champion of determinism, that in constructing an account of the necessity of nature, we ought to allow for at least the epistemic possibility that genuine indeterminism and genuine freedom are metaphysically possible. Accordingly, I will proceed in my exposition as if there are indeed indeterministic /222/ causes, both natural (of the sort studied in, say, elementary physics) and free (among which are human agents in some of their operations).9 Still, as will become clear below, the parts of my theory that are meant to accommodate the possible truth of causal indeterminism do not in any way rule out the possible truth of universal causal determinism.

Now if there is genuine indeterminism in the world, then some effects are produced directly by indeterministic causes and, hence, do not occur by a necessity of nature. What's more, even events that do occur by a necessity of nature (assuming that some do) may nonetheless be contingent in the sense that they have indeterministic (whether natural or free) causes somewhere in their causal ancestry. Take, for instance, the scene with which this paper began. When you placed the kettle of water on the flame, you initiated a causal sequence involving the flame, the kettle, and the water. Assume that no indeterministic causes intervened between the time you put the kettle on the flame and the time (T) at which the water began to boil. Then, clearly, it was by a necessity of nature that the water began to boil at T. Yet if you acted (genuinely) freely in putting the kettle on the flame in the first place, it follows that the water's boiling at T, though necessary because of the causes that were operative once you played your role, was not necessitated by any causal sequences that began at times before you acted. That is to say, for any such time t*, there are worlds that have the same history as our world at t* but in which even though no indeterministic causes intervene to prevent the water from boiling at T, it is nonetheless not the case that any water is boiling on your stove at T. In short, whenever causal indeterminism is introduced into the world, a certain ultimate sort of contingency diffuses itself outward from that time on, even along causal chains that are thoroughly deterministic. So the fact that an event occurs by a necessity of nature does not entail that there are no indeterministic causes in its causal lineage; it is only a completely deterministic world that is such that none of the events occurring in it has any indeterministic causes at all among its causal progenitors.

But why insist that the water's boiling at T is a matter of natural necessity, given that the water would not have boiled at T had you not freely put the kettle on the flame at the appropriate time? Presumably because the water's boiling at T was inevitable or, so to speak, automatic once the relevant causal mechanisms were set into motion and left unimpeded.

This is not to say that those mechanisms were unable to be impeded. In fact, there are any number of ways in which they might have been. The gas jet you steadfastly neglected to clean might finally have become clogged; your spouse, deeply concerned about the inordinate amount of coffee you have been consuming lately, might have decided in an imperious moment to put an abrupt halt to your unhealthy ways; the local utility company might have cut off the flow of gas into your home just before T in order to fix a leak two blocks away; it is even conceivable that a vigilant omnipotent agent might for some hidden reason have freely intervened, whether by omission or commission, to keep the water from boiling at T10--the possible variations are seemingly endless. Indeed, even though the water did in fact boil at T, it may still be the case that its boiling at T did not result by a necessity of nature from the causal sequence you /223/ initiated when you put the kettle on the flame; perhaps some sort of interference took place but was immediately compensated for in a manner dependent upon a further instance of indeterministic causation. Suppose, for example, that when the gas company turned off the natural gas, you noticed it at once and immediately switched over to your emergency supply of propane gas, so that the water began to boil at the very same moment, T, at which it would otherwise have begun to boil. In that case, the water's boiling at T was naturally necessary (let us suppose) because of the causes operative from the time at which you activated the propane tank, but it was not necessitated by the causes operative at the time you put the kettle on the stove.

The upshot is this: what occurs by a necessity of nature at a given time is what is produced by virtue of deterministic natural propensities that are operative before that time and that, since they have been left unimpeded, issue forth automatically in the necessitated event. When, as in the original case, the water begins to boil at T because of causes that act deterministically once you have freely placed the kettle on the flame, we can say that it is naturally necessary that the water boil at T, and we can point to the natural propensities present at the earlier time as evidence for our claim. For in every world which shares the same history with our world just after your action and in which no indeterministic causes intervene before T to prevent the water from boiling, the water begins to boil at T. So what is true by a necessity of nature is what is caused to be true automatically, as it were, once the relevant deterministic causes have been set into motion and left to themselves.

At this juncture, one might be tempted to ask why it is that, left to themselves, these causes automatically produce just this effect. Isn't it conceivable that rather than boiling at T, the water should instead turn into, say, a block of ice or a shrub or even a personal computer? I will return to this question later. For now it is enough to note that a very natural, though apparently controversial, way to respond is simply to claim that it is of the essence or, as the term 'natural necessity' suggests, of the nature of these causes to act in this way and to bring about this effect when unimpeded. Their producing such-and-such an effect in such-and-such circumstances flows from their being the sorts of things and, hence, the sorts of causes they are. They are what they are; they can be no other. And their action follows from what they are. My account of natural necessity will have this sort of essentialism built into it. For according to this account, if a given effect occurs by a necessity of nature, then it occurs in every possible world in which the same unimpeded deterministic causes are operative at the same time and under the same circumstances. There are, to be sure, further subtleties to be spelled out here, and I will attend to some of them below. But the general idea is easy enough to grasp: the deterministic causal propensities that issue forth in natural necessities are themselves grounded in basic causal dispositions, both active and passive, that are had essentially (and so with metaphysical necessity) by natural substances and that, in part, constitute their essences or natures.

How, then, might we build upon the metaphysical and accidental modalities in constructing an account of natural necessity? Let's begin with the de dicto natural modalities. First, a proposition that is true by a necessity of nature is made or caused to be true. Since no metaphysically necessary proposition can, I take it, be caused to /224/ be true by anyone or anything, and since no metaphysically impossible proposition can be true in any way at all, only metaphysically contingent propositions may be naturally necessary.11 By the same token, a proposition that is accidentally necessary at a given time is not one that is made true at that time by causes then operative. Rather, accidentally necessary propositions constitute the ineradicable present and future vestige of what was caused to be true at some previous time. Again, what is accidentally necessary or accidentally impossible at a given. time is entirely a function of the history of the world at that time. By contrast, what is naturally necessary or naturally impossible at a given time is in part a function of what is going on causally at that very time. So no proposition that is accidentally necessary at a given time is true by a necessity of nature at that time, and no proposition that is accidentally impossible at a given time is false by a necessity of nature at that time.

I pointed out above that a natural necessity can plausibly be thought of as the culmination of an unimpeded deterministic natural propensity. (In what follows, I will assume that natural propensities are divided into natural tendencies, which pertain to de dicto natural modality, and natural inclinations, which pertain to de re natural modality. Though natural propensities need not be deterministic, the ones relevant to an account of natural necessity and impossibility are just the deterministic ones.) The distinction between a natural necessity, on the one hand, and a deterministic natural propensity on the other, well-motivated in its own right, has the additional virtue of enabling us to forestall a possible anomaly in the use of terms like 'naturally necessary' and 'naturally impossible'. Given that miracles are at least conceivable, we might be tempted to characterize a miraculous event as one that occurs despite its being naturally impossible that it occur or, alternatively, despite its being naturally necessary that it not occur. To cite a standard medieval example, revived recently by Rom Harre and Edward Madden,12 when the three young men emerge unharmed from the fiery furnace, we might be tempted to say that they have emerged unharmed despite its being naturally impossible that they do so, or despite its being naturally necessary that they not emerge unharmed. But if we permit such usage, then we must deny that a proposition's being naturally necessary at a given time entails its being true at that time. Worse yet, since a proposition' s being naturally necessary should, it seems, be equated with its being true by a necessity of nature, we will be forced to deny that a proposition's being true by a necessity of nature at a given time entails its being true at that time.

However, if we distinguish de dicto natural necessities from deterministic natural propensities--more specifically, from deterministic natural tendencies--we can use the term 'naturally necessary' in such a way that it entails truth, and the term 'naturally impossible' in such a way that it entails falsehood. Then we can go on to distinguish various strengths of natural tendencies and to characterize miraculous events as those resulting from causal interference with a natural tendency so strong that only an omnipotent or 'supernatural' agent can have the power to impede or hinder it.13 An analogous point will hold below for the relationship between de re natural necessities and deterministic natural inclinations. /225/

How exactly should we conceive of natural propensities and their relation to the de dicto natural modalities? There is bound to be at least a bit of arbitrariness here in the formulation of a semitechnical vocabulary. My own preference is to think of a natural tendency as a relation that a world bears to a proposition and to think of a deterministic natural tendency as a natural tendency that, if left unimpeded, issues forth in the truth of a given proposition at a given time. As I conceive of them, such tendencies are ultimately a function (i) of the basic causal dispositions had essentially by natural substances and (ii) of the ways in which those dispositions are integrated with one another at the time at which the world is said to have the deterministic natural tendency in question. So, on this picture, a deterministic natural tendency is, as it were, an all-things-considered deterministic natural propensity on the part of the world at a given moment. Accordingly, we can think of the world as having at a moment t a determinis-tic natural propensity toward a proposition p at a moment t*, where t* occurs at or later than t, a propensity such that it automatically issues forth in p's truth at t* as long as it is unimpeded; and when it does issue forth in p's truth at t*, we then say that p is true at t* by a necessity of nature.

Notice, if a natural tendency is deterministic, then it will not be impeded by any natural (as opposed to free ) causes. It will not be impeded by deterministic natural causes because, as just noted, deterministic natural tendencies are ultima facie, and not merely prima facie, deterministic propensities relating the world at one moment to a proposition at a second moment. That is to say, such tendencies are a function not only of the natures of the deterministic causes that will be operative between the two moments in question but also of the history of the world and, hence, the arrangement of causes in the universe at the earlier moment. Nor can any indeterministic natural cause impede a deterministic natural tendency, since any propensity that is defeated by an indeterministic natural cause is by that very fact not a propensity that might flower into a natural necessity.

Consequently, only free causes can prevent deterministic natural tendencies from blossoming into full-blown natural necessities. Moreover, as hinted above, natural tendencies in general, and deterministic natural tendencies in particular, can be ordered according to the amount of power it would take to impede or hinder them. In our mundane example of the water's beginning to boil at T, we saw that the obstruction of a natural tendency need not be miraculous; it need not require very much power at all. At the other end of the scale, miracles are conceivable because we are able to conceive of instances of causal interference that exceed the power of any natural or nonomnipo-tent agent, or that at least exceed the power of any of the natural agents present in the relevant circumstances. So we have:

    At moment t, possible world w has a deterministic natural tendency toward proposition p at moment t* (where t* is at or after t) if and only if

      (a) p is not metaphysically necessary, and
      (b) p is not accidentally necessary at t* in w, and
      (c) p is true at t* in every possible world w* such that (i) w* shares the same history with w at t and (ii) no freely acting agent brings it about at or after t in w* that p is false at t*, and /226/
      (d) in at least one possible world w* such that (i) w* shares the same history with w at t and (ii) p is true at t* in w*, p's being true at t* is not brought about at or after t in w* either by a freely acting nonomnipotent agent or by an omnipotent agent acting freely and alone.

Condition (c) captures the idea that a deterministic natural tendency must be impeded in order not to result in the truth of the relevant proposition. Condition (d) has a twofold purpose. First, it captures the idea that if the world has a deterministic natural tendency toward a proposition, then the truth of that proposition need not result from any free action of the sort described. Second, the stipulation that the truth of the proposition need not be caused by an omnipotent agent acting freely and alone serves to mark off my account clearly from occasionalism, according to which events in nature are brought about by God alone. Also, this stipulation is worded in such a way as not to rule out a priori the classical theistic doctrine according to which it is metaphysi-cally necessary that any proposition caused to be true is such that God causally contributes (at least via his permission) to its truth.14

We can characterize a strong deterministic natural tendency along the same lines by simply adding another condition to the effect that no nonomnipotent agent has the power to bring it about at or after t in w that p is false at t*. So a strong deterministic natural tendency is one that only a miraculous action can obstruct. Keep in mind, however, that these explications of a deterministic natural tendency and of a strong deterministic natural tendency do not by themselves state that miracles or genuinely free actions are metaphysically possible. Instead, they are meant merely to reflect the fact that if miracles and genuinely free actions are, indeed, metaphysically possible, then this is relevant to the question of which deterministic natural tendencies and which strong deterministic natural tendencies a given world has at a given moment. We can now characterize the contrary of a deterministic natural tendency, what we might call a deterministic natural aversion, as follows:

    At time t, possible world w has a deterministic natural aversion to proposition p at time t* if and only if at t, w has a deterministic natural tendency toward not-p at t*.

We can characterize a strong deterministic natural aversion along the same lines by adding the condition that no nonomnipotent agent has the power to bring it about at or after t in w that p is true at t*. So only a miracle can thwart a strong deterministic natural aversion.

From here it is an easy step to the characterization of de dicto natural necessity, since a natural necessity is nothing more than a deterministic natural tendency come of age. So we have:

    Proposition p is naturally necessary (or: true by a necessity of nature ) at moment t in possible world w if and only if

      (a) p is true at t in w, and
      (b) there is a moment t* such that (i) at every moment from t* to t, w has a deterministic natural tendency toward p at t, and (ii) p's truth at t is not /227/ brought about at or after t* in w either by a freely acting nonomnipotent agent or by an omnipotent agent acting freely and alone.

To put it succinctly, a natural necessity results from an unimpeded and uninterrputed deterministic natural tendency. Also, we can explicate the idea of a proposition's being strongly naturally necessary by simply replacing 'deterministic natural tendency' in clause (b) (i) by 'strong deterministic natural tendency'. So a proposition is strongly naturally necessary at a given moment only if it would have taken a miracle to prevent it from being true at that moment.

Notice, by the way, that clause (b) (ii) is needed in order to exclude a peculiar but, to my mind, perfectly conceivable case. Suppose that just before T, but as close as you wish to T, some free agent (perhaps it would have to be an omnipotent agent) intervenes and impedes the world's deterministic natural tendency toward the proposition This water is boiling at T. But suppose further that, for whatever reason, this agent decides after all to bring it about directly that the water should begin boiling at T. In such a case, the proposition This water is boiling would obviously be naturally contingent at T, since brought about directly by a free cause. However, without the addition of clause (b) (ii), we would be forced to say that This water is boiling is naturally necessary at T. By contrast, once we add clause (b) (ii), our problem is solved. For no matter how close to T it is when our agent impedes the natural tendency in question, the truth at T of This water is boiling will have resulted directly from the agent's free action, and not from any deterministic natural tendency.15

Notice, too, that on the above account of natural necessity--and an analogous point holds as well for natural impossibility--future-tense propositions may be naturally necessary at given times. This is so, at least, if we grant a thesis that I have argued for at length in another place, namely, that if a cause C brings it about that a proposition p is true (false) at a time t, then C thereby brings it about that it has always been the case that p would be true (false) at t.16 Let T* be the moment at which you put the kettle on the flame, and assume that at T* the world has a deterministic natural tendency toward the proposition This water is boiling at T. It then turns out, on the above account of natural necessity, that if the water in fact boils at T because of the causes put into operation at T*, then the future-tense proposition This water will be boiling at T, which (as I would argue) was always true before T, becomes naturally necessary as well at every moment from T* up to, but not including, T.17 So some future-tense propositions that are accidentally contingent at a given time might nonetheless be naturally necessary at that time. This distinction between the natural and accidental modalities has often been blurred in discussions of logical determinism.

De dicto natural impossibility can be explicated as follows:

    Proposition p is naturally impossible (or: false by a necessity of nature ) at moment t in possible world w if and only if

      (a) p is false at t in w, and
      (b) there is a moment t* such that (i) at every moment from t* to t, w has a deterministic natural aversion to p at t and (ii) p's falsity at t is not brought about at or after t* in w either by a freely acting nonomnipotent agent or by an omnipotent agent acting freely and alone. /228/

We can characterize a proposition' s being strongly naturally impossible, or false by a strong necessity of nature, simply by replacing 'deterministic natural aversion' in clause (b) (i) by 'strong deterministic natural aversion'. So a proposition is strongly naturally impossible at a given moment only if it would have taken a miracle to render it true at that moment. And, parallel to what was said above in the case of natural necessity, clause (b) (ii) is meant to exclude the peculiar case in which a free agent first blocks the world's deterministic natural aversion to p at t and then nonetheless directly brings it about that p is false at t.

We can now say that a proposition p is naturally contingent at a moment t in a possible world w just in case p is metaphysically contingent, and accidentally contingent at t in w, and neither naturally necessary nor naturally impossible at t in w. So a proposition true at a given moment t is naturally contingent at t only if it is rendered true directly (i.e., without the mediation of other causes) by an indeterministic cause at t. Note that although logical determinism entails the very strong thesis that no true future-tense propositions are ever accidentally contingent, universal causal determinism entails only the much weaker thesis that no true future-tense propositions are ever naturally contingent. These two theses are often conflated in discussions of logical determinism.

However, we have seen that there is a further sort of contingency that a theory of natural modality should allow for, namely, the sort of contingency had by a proposition that, even though it might be true by a necessity of-nature at a given time t, is nonetheless such that at some past time, the world was not tending deterministically toward its truth at t. This sort of contingency can be explicated as follows:

    At moment t in possible world w, proposition p is contingent with respect to the causes operative at moment t* (where t* is at or before t) if and only if

      (a) p is metaphysically contingent, and
      (b) p is accidentally contingent at t in w, and
      (c) p is true at t in w, and
      (d) it is not the case that at t*, w has a deterministic natural tendency toward p at t.

So, in keeping with what was said above, even though the proposition This water is boiling is true at T by a necessity of nature, it is nonetheless contingent with respect to any deterministic causal sequence set in motion before you freely put the kettle on the flame.

Now for the de re natural modalities. On the picture I have been developing here, the natural tendencies in terms of which we explicate de dicto natural necessity reflect natural inclinations had by the various natural substances. Natural inclinations in general are propensities on the part of a substance toward the possession of a given property at a given time, and deterministic natural inclinations in particular are all-things-considered deterministic propensities of this sort. Parallel to the relationship between a world and a deterministic natural tendency, a substance has a deterministic natural inclination toward a property P at a time t only if it can be prevented from having P at t by the action of some free agent, perhaps a very powerful one. Consequently, /229/ no entity can have a deterministic natural propensity toward any property that it has essentially, and so with metaphysical necessity; nor can an entity have a deterministic natural propensity toward any property that it has by accidental necessity. What's more, deterministic natural inclinations are ultima facie, and not merely prima facie, deterministic propensities relating a substance at one moment to a property at a second moment. That is to say, such inclinations are a function not only of the natures of the deterministic causes that will be operative between the two moments in question, but also of the 'history of the world and, hence, the arrangement of causes in the universe at the earlier moment. More formally, we have:18

    At moment t in possible world w, entity x has a deterministic natural inclination toward property P at moment t* (where t* is at or after t) if and only if

      (a) x exists at t in w, and
      (b) it is not metaphysically necessary for x to have P at t* in w, and
      (c) it is not accidentally necessary for x to have P at t* in w, and
      (d) x has P at t* in every possible world w* such that (i) w* shares the same history with w at t, and (ii) no freely acting agent brings it about at or after t in w* that x does not have P at t*, and
      (e) in at least one possible world w* such that (i) w* shares the same history with w at t and (ii) x has P at t* in w*, x's having P at t* is not brought about at or after t in w* either by a freely acting nonomnipotent agent or by an omnipotent agent acting freely and alone.

Conditions (d) and (e) parallel conditions (c) and (d) in the formula for a deterministic natural tendency. To explicate what it is for an entity to have a strong deterministic inclination toward a given property, we need add only that no nonomnipotent agent has the power to bring it about at or after t in w that x does not have P at t*. So if a substance has a strong deterministic natural inclination toward a property at a given time, then only a miracle can prevent it from having that property at that time.

The contrary of a deterministic natural inclination, what I will call a determinis-tic natural repugnance, can be characterized in this way:

    At moment t in possible world w, entity x has a deterministic natural repugnance to property P at moment t* (where t* is at or after t) if and only if at t in w, x has a deterministic natural inclination toward the complement of P at t*.

Again, by adding the condition that no nonomnipotent agent has the power to bring it about at or after t in w that x has P at t*, we arrive at an explication of what it is for a substance to have a strong deterministic natural repugnance to a property. So if a substance has a strong deterministic natural repugnance to a property at a given time, then it takes a miracle for the substance to have the property at that time.

From here we simply follow a route analogous to the one already traveled with the de dicto modalities in order to characterize de re natural necessity and impossibility. To wit:

    It is naturally necessary for entity x to have property P at moment t in possible world w (or: x has P by a necessity of nature at t in w) if and only if /230/

      (a) x has P at T in w, and
      (b) there is a moment t* such that (i) at every moment from t* to t in w, x has a deterministic natural inclination toward P at t and (ii) x's having P at t is not brought about at or after t* in w either by a freely acting nonomnipotent agent or by an omnipotent agent acting freely and alone.

    It is naturally impossible for entity x to have property P at moment t in possible world w (or: x lacks P by a necessity of nature at t in w) if and only if

      (a) x lacks P at t in w, and
      (b) there is a moment t* such that (i) at every moment from t* to t in w, x has a deterministic natural repugnance to P at t and (ii) x's lacking P at t is not brought about at or after t* in w either by a freely acting nonomnipotent agent or by an omnipotent agent acting freely and alone.

Again, parallel to the explication of the de dicto modalities, we can give an account of what it is for a substance to have a property by a strong necessity of nature, or what it is for it to lack a property by a strong necessity of nature. We need merely preface the occurrences of 'deterministic natural inclination' and 'deterministic natural repugnance' in the above formulas with the word 'strong'. So if a substance has a given property by a strong necessity of nature at a given time, it could have lacked that property at that time only by virtue of a miraculous action; and if it lacks a given property by a strong necessity of nature at a given time, then it could have had that property at that time only by virtue of a miraculous action.

In addition, natural substances are said to act or to be acted upon by a necessity of nature just when it is naturally necessary for them to have the property of acting in such-and-such a way or the property of being acted upon in such-and-such a way. So, in our example, the flaming gas acts by a necessity of nature during the time when it is heating the water, and the water is acted upon by a necessity of nature during the time when it is being heated by the flame.19

We can now go on to say that it is naturally contingent for entity x to have property P at moment t in possible world w if and only if it is metaphysically contingent for x to have P at t in w, and accidentally contingent for x to have P at t in w, and neither naturally necessary for x to have P at t in w nor naturally impossible for x to have P at t in w. So if a substance x, in fact, has a given property P at a time t, it is naturally contingent for x to have P at t only if x has P at t by the direct action of some indeterministic cause, be it natural or free. And, as with the de dicto modalities, we must distinguish sharply between natural contingency and accidental contingency.

Finally, we can formulate the de re counterpart of the de dicto notion of a proposition's being contingent relative to the causes operative at a certain time:

    At moment t in possible world w, it is contingent for entity x to have property P relative to the causes operative at moment t* (where t* is at or before t) if and only if

      (a) it is metaphysically contingent for x to have P at t in w, and
      (b) it is accidentally contingent for x to have P at t in w, and
      (c) x has P at t in w, and
      (d) it is not the case that at t* in w, x has a deterministic natural inclination toward P at t. /231/

Thus, in our example, it is contingent for the water to have the property of boiling at T relative to the causes operative at times before you freely put the kettle on the flame, and it is contingent for the gas now being consumed in the flame to be so consumed relative to the causes that were operative before you ignited it by turning the appropriate dial.

There are other satellite notions we could go on to explicate, but these should be sufficient to convey clearly the main contours of my account of the natural modalities.20 To recapitulate, the key concept is that of a deterministic natural propensity, with such propensities being subdivided into deterministic natural tendencies (which are had by whole worlds) and deterministic natural inclinations (which are had by individual natural substances). I have previously claimed that deterministic natural tendencies merely reflect deterministic natural inclinations. Now I wish to emphasize the further point, adumbrated above, that such natural inclinations are in turn grounded in those basic causal dispositions that natural substances have essentially, and thus with metaphysical necessity.

The idea is this: by virtue of being the kind of thing it is, a natural substance is disposed toward making characteristic causal contributions, both active and passive, to the actualization of various states of affairs. These dispositions are triggered in some circumstances, frustrated in others, and, in still others, integrated with the dispositions of other substances to produce some intermediate effect. The all-things-considered natural tendencies and inclinations we have been speaking about, indeterministic as well as deterministic, are in effect the result of the integration of the causal dispositions of existing natural substances with the arrangement, both spatial and temporal, of these substances in the universe. So the fact that, say, this sample of phosphorus or this flower or this insect or this lepton has just these causal dispositions, i.e., active and passive causal powers, is a necessary fact about it, one that is true of it at any moment it exists in any possible world. Consequently, this sample of phosphorus could not have been a flower or an animal or a sample of hydrogen instead. It is of its essence or nature to be phosphorus and to have the causal dispositions, both active and passive, that are characteristic of phosphorus. And this is why natural necessity can be characterized in terms of what occurs in every possible world with a given history.21

Of course, none of the above requires that any proposition is, in fact, naturally necessary, or that any substance has, in fact, any property with natural necessity. Nor does the above account require the truth of essentialism. So it does not by itself generate the controversial theses hovering in the neighborhood. Still, it does require that if any proposition is ever naturally necessary, or if it is ever naturally necessary for any substance to have a given property, then it has to be the case that natural substances have natures and, thus, that they have essentially a full array of causal dispositions.


III

I will now briefly consider some objections:

(a) "It' s astonishing that you should have come this far in a paper on the necessity of nature without so much as alluding to natural laws. After all, contemporary discussions /232/ of natural necessity often focus from the beginning on such laws and treat little else. It is just the necessity of laws of nature, some would argue, that serves as the foundation for natural modalities of the sort you have explicated, since these modalities are best thought of as functions of (i) the history of the world at a given time and (ii) the laws of nature obtaining at that time. Yet your theory makes no mention of laws of nature."

The account of natural necessity proposed above can easily accommodate laws of nature as long as these laws are conceived of in the way I will now specify.22 First, notice that the foundational role ascribed in the objection to these laws corresponds exactly to the foundational role played in my account by the various active and passive causal dispositions that natural substances have as essential properties. Next, grant that a law of nature is of the form

    It is a law of nature that F' s are G,

and assume with such philosophers as D. M. Armstrong, Fred Dretske, and Michael Tooley that a statement of this form, far from being equivalent to the corresponding universal generalization, is instead correctly taken to express a modal relation between universal properties, a relation most perspicuously represented by the formula

    Being an F necessitates being a G.23

Assume further that the first property named in such a formula is in the paradigmatic instance a natural kind, that the second property named is an active or passive causal disposition, and that the sort of necessity linking them together is metaphysical necessity. Finally, suppose it to be a metaphysically necessary truth that any substance that instantiates a natural kind essentially instantiates that kind. From all this, it follows that, on my account, natural modalities are indeed functions of (i) the history of the world at a given time and (ii) the laws of nature obtaining at that time. For laws of nature so understood simply specify those metaphysically necessary connections between natural kinds and causal dispositions that, as I claimed at the end of section II, serve as the foundation of all deterministic natural tendencies, aversions, inclinations, and repugnances.

(b) "But few will agree that natural laws are metaphysically necessary propositions linking causal dispositions to natural kinds. Indeed, even those who argue that natural laws are relations among universals shy away from the thesis that such laws are metaphysically necessary. The most straightforward reason for their reluctance is that many laws of nature are manifestly not metaphysically necessary. How can you ignore this fact?"

Whoever sets out to analyze the notion of a law of nature has to contend right from the start with the evident fact that the term 'law of nature' is used in a variety of substantively distinct ways. So no resultant analysis will be wholly untainted by arbitrariness, though this is not to say that some terminological decisions might not be more illuminating than others. I myself am inclined to draw a sharp distinction between laws of nature, strictly speaking, which are metaphysically necessary propositions, and corresponding statements of regularity, which are metaphysically /233/ contingent propositions depicting what happens "always or for the most part," to use Aristotle' s phrase. For instance, the proposition Salt has a disposition to dissolve in water is a paradigmatic law of nature stating a metaphysically necessary connection between the natural kind salt and the causal property being disposed to dissolve in water. By contrast, the proposition Salt dissolves in water describes only a metaphysically contingent regularity, since in our world salt's disposition to dissolve in water will not normally be thwarted--as it is, say, when the water in question has already exhausted its capacity for dissolving salt. Further, the more stable and significant the relevant normal conditions, the more 'lawlike' the regularity, with the result that some regularities attain the status of what are sometimes called 'derived laws'. An example would be the laws of planetary motion operative in our solar system. Still, on the view I am defending, all such regularities are firmly grounded in natural laws, strictly speaking, since they merely reflect the continuous integration of these metaphysically necessary laws with metaphysically contingent antecedent conditions.

(c) "But isn't it obviously conceivable that the laws of nature should be suspended or violated, or at least that there should have been different such laws ? This is the basis for the Humean (or, to be fair, Ghazalian) dictum that there are no logically necessary connections between causes and effects, i.e., that there are no conceptual limitations on what sorts of causal transformations are possible. Yet, on your view, it is metaphysically necessary for, say, water to have a causal disposition to be heated when brought into suitable proximity to fire, and metaphysically impossible for it to have a causal disposition to be transformed into a personal computer under the same circumstances. So you are committed to claiming, pace Hume and al-Ghazali, that it is logically (and not just naturally) necessary that in those circumstances, the water should be heated, and that it is logically (and not just naturally) impossible that in those circumstances, the water should turn into a computer. Isn't this clearly wrong-headed?"

This objection provides me with a welcome opportunity to show how I am able to accommodate the truth contained in the Ghazali-Humean dictum without at the same time falling into the causal anarchism characteristic of occasionalism and posi-tivism. Recall that on my position, metaphysical necessity enters into natural modality in two distinct ways. First, the relation between the universals involved in a natural law is metaphysically necessary, with the result that laws like the one cited above have de dicto metaphysical necessity. (This does not mean, however, that the natural kinds they involve are necessarily instantiated.) Second, every possible natural substance essentially instantiates the natural kind that it instantiates, and so in every possible world in which it exists, each such substance has essentially, and hence with de re metaphysical necessity, the causal dispositions endemic to its natural kind.

So it is, indeed, impossible that there should have been different laws of nature if by this we mean either (i) that the substances that, in fact, exist could have had different natures, i.e., could have instantiated different natural kinds, or (ii) that the natural kinds instantiated by these substances could have been necessarily tied to basic causal dispositions different from those they are, in fact, necessarily tied to. Nonetheless, there could have been different laws of nature in the sense that it is metaphysically /234/ possible (i) that there should exist natural substances of kinds that have, in fact, never been instantiated and/or (ii) that the natural kinds that are, in fact, instantiated should not have been.24 So, for instance, my position does not rule out the possibility that there should have been elementary physical particles of types that are different from those that are, in fact, instantiated and that are such that if they had been instantiated instead, then the strengths of the fundamental forces in nature would be slightly different or even greatly different from what they are in our universe. Under those circumstances, the scientific theories that hold for our universe would not have held. Nor does my position rule out the possibility that natural kinds themselves are necessarily interrelated in such a way that certain of them can be instantiated only if certain others are; perhaps no natural kind instantiated in our universe is instantiated in worlds in which the elementary physical particles differ in kind from those that exist in our world. Further, my position does not rule out the possibility that a universe should be instantaneously transformed from one in which a certain set of natural kinds is instantiated into one in which some disjoint set of natural kinds is instantiated. A fortiori, my position does not rule out the more modest Ghazali-Humean possibility that the water on your stove should "turn into," or at least be replaced by, a personal computer when brought into proximity to fire. Whether this causal transformation is metaphysically possible depends only on whether it is metaphysically possible for there to be a causal agent powerful enough to effect it. Nothing I have said rules this out. Given that our current understanding of the natures of water, fire, and personal computers is substantially correct, all I am forced to deny is that the water's transformation into a personal computer on the occasion of its being brought into proximity to fire is a transformation that might result from the natural tendencies (whether deterministic or indeterministic) operative at the time in question. If this denial is sufficient to entail that transformations of this sort are 'logically' or 'conceptually' impossible in one or another of the senses alluded to in section I, then so be it. As I see it, the truth contained in the Ghazali-Humean dictum is adequately safeguarded as long as we do not antecedently rule out such transformations as metaphysically impossible.

So, correctly understood, the conception of natural laws most congenial to my account of natural necessity allows for ample flexibility regarding which natural laws hold at any given time or in any given possible world, and also regarding which causal transformations are metaphysically possible.

(d) "Despite what you say, this response does not preserve the whole truth contained in the Ghazali-Humean dictum. To see this clearly, call to mind the miracle of the fiery furnace. As told in chapter three of Daniel, the story seems to entail that real human flesh was exposed without protection to real fire and yet was not incinerated. This is clearly conceivable. Yet Harre and Madden, whose views you have been closely adhering to throughout this discussion of natural laws, rule it out as impossible.25 They do not, to be sure, rule out every kind of miraculous intervention in this situation. They allow, for instance, that God might conceivably have replaced the human flesh of the young men with stone for the duration of the fire. (This is analogous to the water's being replaced by a personal computer in your example. ) But clearly this is not enough. The miracle as told is conceivable, and your position cannot accommodate this fact." /235/

Anyone familiar with their pioneering work Causal Powers (now, inexplicably, out of print in America) will recognize how deeply indebted I am to Harre and Madden for my understanding of natural necessity and natural laws. As I see it, however, their discussion of the miracle of the fiery furnace is flawed in a way that misrepresents the real difference between their (and my) general understanding of natural modality, on the one hand, and the occasionalist/positivist position on the other hand.

As noted in the objection, Harre and Madden are perfectly willing to concede that human flesh exposed to fire might be instantaneously transformed into stone; yet they adamantly refuse to concede that unprotected human flesh might not be incinerated when exposed to fire. But why the asymmetry here? Neither of these events could be explained by an appeal to the natures of fire, human flesh, and stone--or, hence, by an appeal to any natural tendencies operative at the time in question--at least not if we suppose that our current scientific understanding of these natures is substantially correct. Both events would seem to involve the thwarting of the same deterministic natural tendency, namely, the world's tendency toward the proposition This human flesh is incinerated at the time in question. So why should just one of them be counted as metaphysically possible?

The only plausible response open to Harre and Madden is this: In the miracle story as told in Daniel, chapter three, all the conditions obtain that are necessary for the exercise of the relevant causal dispositions on the part of the fire and the flesh, and it is metaphysically impossible that all those conditions should obtain and yet that the flesh not be incinerated; but in the alternative story involving the transformation of the flesh into stone, one of those conditions fails to obtain, since an omnipotent being steps in at the last moment to interfere with the operative natural tendency by turning the flesh into stone.

This response, however, is far from adequate. I willingly grant that one of the conditions necessary for the fire's incinerating the flesh is that no agent, omnipotent or otherwise, should interfere with the natural tendency in question. Such a condition is explicitly built into the account of natural modality propounded in section II. But just which ways of interfering with this tendency are metaphysically possible? What are the metaphysically possible ways in which an agent might bring it about that the fire does not incinerate the flesh? Suppose for a moment that no natural substance can exercise an active causal disposition unless God acts along with it to bring about its natural effect--an idea that lay at the heart of late medieval and early modern theories of God's 'general concourse' or 'general concurrence'.26 In that case, God can bring it about that the (real) fire does not incinerate the (real) human flesh merely by withholding his causal concurrence, even if all the other conditions obtain that are necessary for the exercise of the fire' s active disposition to incinerate human flesh brought into proximity to it. On what grounds do Harre and Madden rule out such a possibility? What reasons do they have for thinking that the doctrine of divine general concurrence is metaphysically impossible? After all, to the naked philosophical eye, this doctrine does not seem incoherent; and it is certainly no more wondrous, shall we say, than the thesis, wholeheartedly endorsed by Harre and Madden, that God might conceivably transform human flesh into stone. /236/

Now it is not my purpose in this paper to defend or even to articulate the doctrine of divine general concurrence. The point I wish to make is simply that Harre and Madden have not provided a satisfactory argument---or, for that matter, any argument at all--against its possibility; hence, they have not furnished us with any reason for thinking that the ostensibly coherent mode of causal interference associated with this doctrine is metaphysically impossible. Perhaps it is, indeed, impossible; perhaps it is not. But the issue should certainly not be decided by fiat.

What' s worse, by putting such undue emphasis on the miracle of the fiery furnace, Harre and Madden give the impression that the distinctiveness of their (and my) general account of natural modality and natural laws depends crucially on the claim that the miracle as described in the book of Daniel is metaphysically impossible. Nothing could be more misleading. In order to distinguish our general account of natural modality and natural law from that of the occasionalists and positivists, we need only point out that even if this miracle is metaphysically possible (whether conceived of as the transformation of human flesh into stone or as the simple nonincineration of real human flesh), it cannot be accounted for in terms of the natural tendencies and inclinations operative at the time in question. That is why, pace al-Ghazali and Hume, this event and others like it must be treated as having a special causal ancestry and as not being of a piece with events that do constitute the culmination of natural tendencies and inclinations.

(e) "Despite what you have suggested, your position is not the only alternative to occasionalism and positivism. After all, those philosophers who reject the positivist conception of natural necessity and hold that natural laws are relations among universals reject as well the sort of view you are proposing. Their arguments are surely more persuasive than those you have cited so far."

Where, exactly, are these persuasive arguments? A few years ago, someone might have charged that if the laws of nature are metaphysically necessary, then they must be analytic or, at least, knowable a priori, and not empirically. Perhaps, on one of the meanings of 'logically necessary' mentioned in section I, it is true that every logically necessary proposition is analytic or knowable a priori. But my account entails only that the laws of nature are rnetaphysically necessary, not that they are logically necessary. And nowadays few philosophers will be inclined to insist that no metaphysically necessary proposition can be known empirically or a posteriori.

The fact is, I think, that there are no compelling arguments against the view of laws I am taking here. This judgment is borne out by a brief survey of some of the best contemporary work on laws of nature. Dretske and Tooley, for example, simply assume without argument that laws of nature, despite being relations among universals, are not metaphysically necessary but, instead, have a special modal status falling somewhere between metaphysical necessity, on the one hand, and unqualified metaphysical contingency on the other.27

Armstrong, to be sure, does treat the issue at some length, but as he is honest enough to admit, his arguments against the metaphysical necessity of natural laws all rest on the highly controvertible assumptions that form the basis of his distinctive /237/ metaphysical system. For instance, his main objection to the doctrine of strong necessity, i.e., the doctrine that laws of nature are metaphysically necessary, is that it requires that universals are necessary rather than contingent beings and, thus, violates the principle that there cannot be any uninstantiated universals. Elsewhere he ties his acceptance of this principle to his antecedent commitment to naturalism, the view that "nothing else exists except the single, spatio-temporal world, the world studied by physics, chemistry, cosmology and so on.''28 The objection will thus not have much power to sway those of us who do not fancy naturalism, or even those naturalists who feel that uninstantiated or necessarily existent universals pose no more threat to naturalism than instantiated and contingently existing ones do.

In rejecting the fallback to the doctrine of weak necessity, according to which two universals connected nomologically in any possible world are so connected in every possible world in which they both exist, Armstrong explains that this doctrine has a difficulty with certain kinds of complex laws. That difficulty need not detain us here, since it is a genuine difficulty only on the assumption that universals are contingent beings, an assumption I for one am not inclined to accept. However, it turns out that one way to avoid the difficulty is to adopt a view of laws that, leaving aside questions about whether universals are necessary or contingent beings, is much like the one I have proffered here. Armstrong retorts:

    The old tradition which links powers with necessity is here upheld. The Weak Necessity view requires irreducible powers. But if we wish to uphold a purely Actualist metaphysics, then it seems that we should reject the Weak Necessity view of laws of nature and declare them to be contingent.29

An actualist metaphysics, in turn, is one that

    debars us from postulating such properties as dispositions and powers where these are conceived of as properties over and above the categorical properties of objects. It is not denied that statements attributing dispositions and/or powers to objects, or sorts of objects, are often true. But the truth-makers or ontological ground for such true statements must always be found in the actual, or categorical, properties of the objects involved.30

Actualism, Armstrong tells us, is the "most difficult and uncertain" of the basic assumptions he makes in his treatment of laws of nature.31 Indeed it is. Why, after all, should we believe that every dispositional property must have a 'categorical basis'? What is a categorical basis, anyway? How exactly is it supposed to 'ground' a property, especially if, as Armstrong must apparently hold, it is only contingently related to the properties it grounds? How, for example, is salt' s disposition to dissolve in water grounded in its chemical structure? What exactly is wrong with the claim that when scientists explain a thing's causal dispositions and powers by appeal to its microstructure, they are explaining those dispositions and powers by appeal to the dispositions and powers of the relevant microentities? In some instances, the first set of dispositions and powers might be strictly reducible to the second; in others, the first set might necessarily emerge from the second. In any case, the fact that structural explanation /238/ has turned out to be a significant scientific tool does not, it would seem, militate against the philosophical understanding of natural modality and natural law proposed here.

Then, too, look at the consequences that follow from Armstrong's thesis that the laws of nature are metaphysically contingent and that none of the ties connecting individual natural substances with their causal dispositions are metaphysically necessary. First, on this thesis, it is a primitive, unexplained, and rather mysterious fact that the laws of nature that hold in our world are true. Second, even though, as in the passage cited above, Armstrong often insists on the importance of 'ontological grounds', there are on his own view no such grounds for the alleged possibility that propositions that are in fact false might have been laws of nature. At this point, the occasionalists could at least appeal to God' s power to promulgate laws of nature different from those he has in fact promulgated. (Does Dretske have this precedent in mind when he likens the modal status of natural laws to that of the duly promulgated, though contingent, constitutional delineation of the powers of the president of the United States?)32 But what will a self-avowed naturalist appeal to? Third, is it plausible, or even coherent, to suppose, as Armstrong must, that the kind salt could have been linked to the causal dispositions and powers associated in our world with, say, the kind pin oak? Finally, what of his rejection of the thesis that natural substances have various causal dispositions essentially or with metaphysical necessity? Is it at all reasonable to suppose that this grain of salt before me could have had all and only the causal dispositions characteristic of, say, a piece of phosphorus or a rhododendron or an armadillo, or that it could have belonged to some uninstantiated and, perhaps, unimaginable natural kind?

I do not mean to suggest by the tendentious tone of these questions that Armstrong and the others have no resources to draw upon here. I intend to explore the matter in more detail elsewhere. Nor do I mean to suggest that the view I have been defending is anything but 'old' or even old-fashioned. The account of natural modality proposed in this paper is, in fact, little more than a dusted-off and updated version of what was standard fare among the Aristotelian scholastics from Aquinas to Suarez. What I do mean to suggest, however, is that, at least on the surface, this old-fashioned account has virtues that we can now begin to appreciate anew in the current postpositivistic philosophical milieu.33


Notes

1. Berkeley himself frowns upon at least certain uses of the term 'occasion' because they presuppose a belief in matter. See Three Dialogues between Hylas and Philonous, edited by Robert M. Adams (Indianapolis, Ind. 1979), 54-55, and A Treatise concerning the Principles of Human Knowledge, edited by Kenneth Winkler (Indianapolis, Ind., 1982), 52-57. I myself prefer the name 'no-nature view' to the traditional 'occasionalism' and see Berkeley's account of God's causal activity in nature as, metaphysically speaking, the purest instance of such a doctrine. This subject is treated at length in my "Medieval Aristotelianism and the Case against Secondary Causation in Nature," as yet unpublished. Pertinent references to the work of the other philosophers mentioned here include problem XVII, "Refutation of Their Belief in the Impossibility of a Departure from the Natural Course of Events," 185-96 in al-Ghazali' s Tahafut al-falasi-fah: The Incoherence of the Philosophers, translated by Sabih Ahmad Kamali (Lahore, 1963); Gabriel Biel, Collectorium circa quattor libros Sententiarum IV, pt. 1, edited by Wilfridus Urbeck and Udo Hoffman /239/ (Tuebingen, 1975), question 1, "Utrum sacramenta legis novae sint causae effectivae gratiae," 1-36, especially 14-18 and 27-36 (I have made a translation of the relevant sections for anyone who might be interested); and Nicolas Malebranche: The Search After Truth and Elucidations of the Search After Truth, translated by Thomas M. Lennon and Paul J. Olscamp (Columbus, Ohio, 1980), bk. 6, pt. two, chaps. 2 and 3, 440-52, and elucidation 15, 657-83.

2. For a list of some of these epistemic properties, see pages 251-52 in Fred I. Dretske, "Laws of Nature," Philosophy of Science 44 (1977): 248-68.

3. See my "Medieval Aristotelianism and the Case against Secondary Causation in Nature," to appear in Divine and Human Action: Essays in the Metaphysics of Theism, edited by Thomas V. Morris (Ithaca, N.Y., forthcoming).

4. In fact, I find myself in deep sympathy with at least the main contours, if not all the details, of the theistic conceptualism proposed in Michael Loux's contribution to this volume.

5. Following Alvin Plantinga, I take a state of affairs S to be maximal just in case for any state of affairs S*, S either includes S* or precludes S*

6. D. M. Armstrong, What Is a Law of Nature? (Cambridge, 1983), 166.

7. See my "Accidental Necessity and Logical Determinism," The Journal of Philosophy 80 (1983):257-78.

8. If there are no tensed properties, but only tensed exemplifications instead, we can say that it is now accidentally necessary for Michael to have exemplified the property of swimming. The formulas below can all be restated accordingly. Incidentally, besides the de re temporal modalities spelled out here, there are several others that are philosophically significant. See Thomas V. Morris, "Properties, Modalities, and God," The Philosophical Review 93 (1984): 35-55.

9. I here endorse Elizabeth Anscombe's sundering of causation from determinism and deterministic explanation, since there seems to be nothing incoherent in the notion of an indeterministic cause. See her now classic paper "Causality and Determinism," reprinted in Causation and Conditionals, edited by Ernest Sosa (Oxford, 1975), 63-81. The root notion involved in causation is that of producing or bringing about or effecting some state of affairs, a notion that seems obviously independent of and prior to questions regarding determinism and indeterminism. The distinction between free and natural causes is a hallowed one, and although I have no precise definitions to offer, I assume that free causes are beings equipped with rather substantial cognitive and volitional capacities, whereas natural causes lack such capacities. Also, as should be obvious, I am assuming that causes are paradigmatically substances rather than events. However, events may still be causes in a derived sense as long as they involve causal contributions on the part of substances. Here I follow Rom Harre and Edward Madden, Causal Powers (Totowa, N.J., 1975), 5. This is an ontological issue that requires more extended discussion than I am able to provide here, so I must beg the indulgence of those who take so-called event causation to be basic. However, anyone unwilling to countenance even provisionally the idea that agent causation (on the part of natural as well as free agents) is primitive should be able to recast the following discussion of natural necessity into a more congenial idiom.

10. I allow for the possibility of interference by omission because this sort of interference figured prominently in late medieval and early modern theories of God's general concurrence in the ordinary course of nature. See the response to objection (d) in section III below.

11. I do not mean to rule out antecedently every sort of quasi-causal dependence on the part of necessary truths. For instance, even though propositions are not, strictly speaking, caused to exist, it still seems conceivable that they should depend for their existence on the divine intellect in a way analogous to that in which thoughts depend for their existence on those who think them. If this thesis is true, then the necessary truth There are propositions would depend upon God for its truth. Again, in the traditional explication of the Christian doctrine of the Trinity, even though the divine persons are uncaused and necessary beings, the Son is said to be begotten by the Father, so that the existence of the Son depends on that of the Father, but not vice versa. If this thesis is true, then the necessary truth The Son exists depends for its truth on the activity of the Father. The conceivability and coherence of such theses is closely tied to the claim that at least some subjunctive conditionals with metaphysically impossible antecedents are false.

12. Harre and Madden, Causal Powers, 46-47. The biblical story in question is found in the third chapter of the book of Daniel. /240/

13. Here and below I am building upon previous work by presupposing the correctness of the account of omnipotence found in Thomas P. Flint and Alfred J. Freddoso, "Maximal Power," in The Existence and Nature of God, edited by Alfred J. Freddoso (Notre Dame, Ind., 1983), 81-113.

14. Condition (d) has the desirable side effect of excluding the world's having a deterministic natural tendency toward propositions, e.g., Adam freely chooses to eat the apple, that cannot be brought about without the free action of a nonomnipotent agent. Also, condition (d) guarantees that if a proposition p is true at t in w but could not, given the history of the world, have come to be true at t in w had it not been brought about by the free action of some nonomnipotent agent or of some omnipotent agent acting alone, then w did not have before t a deterministic natural tendency toward it. So, for instance, if Adam eats the apple was freely brought about by some nonomnipotent agent at t in our world, then condition (d) ensures that the world was not deterministically tending toward its truth at any previous time. For in the choice-situation obtaining at the time Adam freely chose to eat the apple, this proposition could be rendered true only by a free action of one of the sorts just specified.

15. I leave open the question of whether an agent can simultaneously obstruct the natural tendency in question and bring about the relevant state of affairs. Such an agent, it seems, would be preventing and bringing about the same state of affairs at the same time.

16. See my "Accidental Necessity and Power over the Past," Pacific Philosophical Quarterly 63 (1982): 54-68. So anyone who brings it about that the water does not boil at T thereby brings it about that the proposition This water will boil at T has always been false and, more to the point, that it has been false from the time you acted up until T.

17. To be sure, characteristic indicators of the future tense in English, both 'will' and 'shall', on the one hand, and 'going to' on the other, are sometimes used to express natural tendencies (both deterlninistic and indeterministic) rather than unadorned future truth. In "Omniscience and the Future," which constitutes chapter three of Providence and Evil (Cambridge, 1977), Peter Geach emphasizes the former use of these locutions to the complete exclusion of the latter use. This seems to me to be a mistake, though it is true that the use of future-tense locutions to express natural tendencies (as well as intentions) is often overlooked.

18. The consequent in condition (d), namely, 'x does not have P at t* in w*', is being read in such a way as to be true if either (i) x does not exist at t* in w*, or (ii) x exists but lacks P at t* in w*.

19. The suggestion here and elsewhere that there are inanimate causal agents is likely to be offensive to pious philosophical ears nowadays. However, I not only do not apologize for this suggestion, but I even make bold to claim that many of the puzzles surrounding human agency can be dealt with adequately only if we begin by looking at agency in nature.

20. One such notion, obediential potency, occupies a prominent place in medieval discussions of the Christian doctrine of the Incarnation. It can be characterized as follows:

    At moment t in possible world w, entity x is in obediential potency toward property P at moment t* (where t* is at or after t) if and only if

      (a) at t in w, x has a strong natural repugnance to P at t*, and
      (b) God has the power to bring it about at or after t in w that x has P at t* .

See my "Human Nature, Potency and the Incarnation," Faith and Philosophy 3 (1986): 27-53, for a discussion of this and other natural modalities involved in the explication of the Incarnation. The present treatment of these modalities supersedes and, in some instances, corrects that found in the paper just alluded to.

21. Of course, things are somewhat more complicated than indicated so far, since essential dispositions must be carefully distinguished from various nonessential dispositions, capacities, and abilities that natural substances might come to have at various stages of their development (as in the case of living organisms) or in combinations with other substances (as in the case of chemical compounds). Still, all of these developmental and combinatorial properties, which are, strictly speaking, nonessential because their exemplification depends in part on contingent antecedent conditions, must on my account be firmly rooted in higher order dispositions that are essential to the things in question. See Harre and Madden, Causal Powers, especially chapters five and six, for a treatment of some of the complexities involved here.

22. I do not wish to give the impression that I have invented the account that follows. As I indicate below, much of what I have to say was at least implicit in late medieval and early modem philosophy. In /241/ addition, my views on natural laws have been deeply influenced by a close reading of the work of several contemporary philosophers, even if I do not find myself in complete agreement with them on all the relevant issues. The list includes Peter Geach, pt. II, "Aquinas," in Three Philosophers (Ithaca, 1961 ), especially 101-9; Milton Fisk, "Are There Necessary Connections in Nature?," Philosophy of Science 37 (1970): 385-404; Harre and Madden, Causal Powers, and also "Natural Powers and Powerful Natures," Philosophy 48 ( 1973): 209-30; Sydney Shoemaker, "Causality and Properties," in Time and Cause, edited by Peter van Inwagen (Dordrecht, 1980), 109-35; and Baruch Brody, Identity and Essence (Princeton, N.J., 1980), especially chapter six.

23. See Armstrong, What Is a Law of Nature?, chap. 6; Dretske, "Laws of Nature"; and Tooley, "The Nature of Laws," Canadian Journal of Philosophy 7 (1977): 667-98. To be accurate, Dretske denies that there is, strictly speaking, a modal relation between the universals, call them 'F' and 'G' , and affirms instead that such a relation holds only between an individual substance's being F and its being G. However, this point is not relevant to the discussion that follows.

24. Here we are faced with a choice. On the one hand, we might claim that all true propositions of the form 'It is metaphysically necessary that F's are G' (where 'F' stands for a natural kind and 'G' for an active or passive causal disposition) are laws of nature but that only those laws that involve instantiated natural kinds are operative laws of nature. On the other hand, we might claim instead that true propositions of the sort in question are laws of nature only if they involve instantiated natural kinds, whereas they are merely potential laws of nature if they involve uninstantiated natural kinds. I have no strong preference one way or the other, though I suspect that the second alternative is more in keeping with the manner in which we normally speak of laws of nature.

25. See Harre and Madden, Causal Powers, 46-47. AI-Ghazali appeals to a parallel story from the Quran (21: 69-70 and 37: 98) in which Abraham is thrown into a fire and emerges unharmed. The present objection, by the way, is very much in the spirit of J. L. Mackie's argument against irreducibly (or what Mackie calls 'distinctively') dispositional properties in Truth, Probability and Paradox (Oxford, 1973), 137-43.

26. The relevant medieval literature includes Aquinas, Summa Contra Gentiles III, chaps. 66-70, and De Potentia, question 3, art. 7; Luis de Molina, Liberi Arbitrii cure Gratiae Donis, Divina Praescientia, Providentia, Praedestinatione et Reprobatione Concordia, pt. II, "De concursu Dei generali"; and Francisco Suarez, Disputationes Metaphysicae, disputation 22, "De Prima Causa et alia eius actione, quae est cooperatio, seu concursus, cum causis secundis." (The selections from Molina and Suarez are not yet available in English translation, though I have made a translation of some of the relevant sections from Molina for anyone who might be interested.)

27. In "Laws of Nature," 263-64, Dretske has this to say:

    Although true statements having the form [of laws of nature] are not themselves necessary truths, nor do they describe a modal relationship between the respective qualities, the contingent relationship between properties that is described imposes a modal quality on the particular events falling within its scope. This F must be G. Why? Because F-ness is linked to G-ness; the one property yields or generates the other in much the way a change in the thermal conductivity of a metal yields a change in its electrical conductivity.

In "The Nature of Laws," 672-73, Tooley remarks:

    The idea of a statement about particulars being entailed by a statement about a relation among universals is familiar enough in another context, since some philosophers have maintained that analytical statements are true in virtue of relations among universals. In this latter case, the relations must be necessary ones, in order for the statement about particulars which is entailed to be itself logically necessary. Nomological statements, on the other hand, are not logically necessary, and because of this the relations among universals involved here must be contingent ones.

It may be that Dretske and Tooley take the contingency of natural laws for granted because they are thinking of these laws as linking natural kinds to the manifestation by instances of those kinds of the relevant dispositional /242/ properties. What I have said also entails that such links are contingent. On my view, however, natural laws link the kinds in question directly with the dispositions, and not with the manifestations of those dispositions. So perhaps Dretske and Tooley are calling 'laws' what I would call 'statements of regularity' instead.

28. Armstrong, What Is a Law of Nature?, 82.

29. Ibid., 168.

30. Ibid., 9.

31. Ibid.

32. Dretske, "Laws of Nature," 264-66.

33. I wish to thank Thomas Flint and Philip Quinn for their helpful comments.