A positive answer to Kant's famous question, ‘How is metaphysics possible?’, is given—one that represents metaphysics as being an autonomous and indispensable intellectual discipline whose task is to chart the domain of real possibilities. The notions of ‘real’ or ‘metaphysical’ possibility and necessity are defended and distinguished from those of various other species of modality, such as physical, logical, and conceptual possibility or necessity. The enemies of metaphysics, from relativists to those who seek to subordinate metaphysics to empirical science, epistemology, or the philosophy of language, are argued to hold positions that are either incoherent or self‐defeating.
In the Preface, I explained that the overall objective of this book is to help to restore metaphysics to a central position in philosophy as the most fundamental form of rational inquiry, with its own distinctive methods and criteria of validation. But if such a project is not to be stifled even before its inception, we need to have some assurance that its aim is a coherent and legitimate one. To provide such assurance is the purpose of this first chapter, which will also serve as an introduction to some of the book's main themes.
1. What Is Metaphysics?
It was Immanuel Kant who first asked ‘How is metaphysics possible?’ 1 But Kant supposed the subject‐matter of metaphysics to comprise a priori synthetic truths, and we have since come to doubt whether there is such a class of truths, not least because the analytic/synthetic distinction has itself become doubtful following W. V. Quine's attack on it. 2 The a priori/a posteriori distinction, on the other hand, is still taken seriously, and indeed has had new life breathed into it by the work of Saul Kripke. 3 But Kripke's work, too, undermines Kant's epistemological assumptions by implying that necessary truths need not be a priori nor contingent truths a posteriori. In the light of these reconceptions, Kant's question ‘How is metaphysics possible?’ needs to be asked anew, and even to be understood in a new way. But the question is quite as pressing for us as it was for Kant. Metaphysics is under assault from many sides, both from within the ranks of philosophers and from various external forces. Despite these attacks, metaphysics has been enjoying something of a revival amongst so‐called analytic philosophers, after a barren period during which first logical positivism and then ordinary language philosophy (p.2) prevailed, both of them hostile to metaphysical speculation. Analytic philosophers are no longer antipathetic to arguments concerning the nature of substance, the reality of universals or the existence of abstract entities. So we need to understand what can legitimate such inquiries and what epistemological status their conclusions can justifiably lay claim to.
As a preliminary, we need to fix upon some reasonably uncontentious characterization of what metaphysics should be understood to have as its main concern. Traditionally, metaphysics has been thought of as the systematic study of the most fundamental structure of reality—and, indeed, that is the view of it which I should like to support. Understanding the aim of metaphysics in this way makes defending the possibility of metaphysics a substantial and problematic task, and for that reason one well worth exploring. By contrast, to understand the aim of metaphysics as something less ambitious than this—for instance, as the attempt to analyse our currently accepted ways of talking about what we unreflectively take to be certain general features of the world we live in—would make its justification less difficult, but only at the expense of making metaphysics less interesting and less worth doing. Should metaphysics as traditionally conceived turn out to be impossible, we may still be able to pursue these less ambitious projects; but let us not pretend that in doing so we would be doing anything worth dignifying by the name ‘metaphysics’.
I confess that characterizing metaphysics as the systematic study of the most fundamental structure of reality hardly sounds very precise. But I am not offering a definition, imprecision in which would indeed be a defect. I do not think it would be at all profitable to pursue a definition of ‘metaphysics’, because the danger would be that it would be framed in a way which favoured one metaphysical position over others—for instance, in a way which presumed the reality of space, or of causation, when the reality of these things may be denied by some metaphysical systems. A similar defect would attend any attempt to specify the subject‐matter of metaphysics by means of a list of topics it supposedly embraces. One has only to look at the highly varied contents of modern textbooks of metaphysics to see how arbitrary such an approach would be. Of course, a consequence of not providing an absolutely clear‐cut delineation of the province of metaphysics is that metaphysics may not appear clearly distinct from certain other enterprises, such as those undertaken in the name of the empirical sciences. However, although I shall later argue that all empirical science presupposes metaphysics, I do not in fact believe that a clear‐cut distinction should be made between metaphysical concerns and some of the more theoretical concerns of science. Drawing sharp boundaries in such matters is unhelpful, and is not needed in order to maintain (p.3) that metaphysical concerns are sufficiently distinctive to form the core of a relatively independent discipline—one whose intellectual credentials are worthy of exploration. In any case, my hope is that when I come, in due course, to attempt to argue on my own account for the possibility of metaphysics, my conception of the subject‐matter and methods of metaphysics will emerge more clearly. In effect, I shall be filling out my preliminary characterization of metaphysics by characterizing metaphysics as the discipline made possible by the sort of defence that I shall offer.
2. How Is Metaphysics Possible?
Before I present that defence, I want to look briefly at some rival answers to the question ‘How is metaphysics possible?’ which are currently popular in various quarters. Some of these answers are frankly antimetaphysical whilst others, in my opinion, downgrade the status of metaphysics to that of something unworthy of the name and are thus covertly antimetaphysical too. An antimetaphysical answer to our question is, then, either one which simply denies that metaphysics, as traditionally conceived, is possible, or else one which defends the possibility of something else under the name of ‘metaphysics’, while implicitly abandoning the real thing. Other answers which I shall reject do attempt to defend the possibility of something recognizably akin to metaphysics as traditionally conceived, but do so in ways I consider to be unsatisfactory. The four positions which I shall consider—only to reject—I shall call (somewhat tendentiously) relativism, scientism, neo‐Kantianism, and semanticism. These are all ‘modern’ views, though some obviously have historical precedents. I shall not examine, here, certain more traditional positions—such as ‘rationalism’ and ‘empiricism’, as supposedly exemplified by such historical figures as Descartes and Locke—because these were positions developed before Kant raised his momentous question, ‘How is metaphysics possible?’
The answer of relativism—for instance, in its latest ‘deconstructionist’ guise—to this question is, quite simply, that metaphysics is not possible, because metaphysics is the misbegotten product of western intellectual hubris, the mistaken search for a non‐existent ‘objective’ and ‘total’ truth, guided by supposedly timeless and universal principles of logic. Truth and reason, according to this view, are culture‐bound concepts of strictly limited utility. The notion that there could be a ‘fundamental structure of reality’ for us to discern is deemed absurd and paradoxical, because what we call ‘reality’ is (supposedly) always just some human construction saturated (p.4) by interest‐driven interpretation. My reply to antimetaphysical claims of this sort is as follows. First, to the extent that they amount to mere assertions, unsupported by reasoned argument, they do not merit serious attention. The fact that their proponents are often contemptuous of the value of reasoned argument itself—this being one of the prime objects of their attack—by no means obliges the defenders of metaphysics to take them seriously. If the relativists wish to decry the notion of reasoned argument as a parochial cultural artefact, then they deny themselves any basis for their own assertions but ingrained prejudice, and we should be content to let them wallow in it if it gives them satisfaction. Secondly, if these claims are alleged to be supported by evidence—for instance, of a sociological or anthropological sort—then it has to be pointed out that in fact such evidence as is available supports nothing so extreme: it does not, and cannot, show that human beings are incapable of escaping from culture‐bound and interest‐driven conceptions of their world, but at most that they often fail to do so. Indeed, the very fact thatsomehuman beings have discovered that many human beings fail to do this shows, if it shows anything, that we are not incapable of making such an escape. Thirdly, it is characteristic of the relativist attack on metaphysics that it deliberately distorts what it seeks to denounce. It represents metaphysicians as laying claim to infallible insight into eternal and universal truths, uncoloured by any human perspective. But only the most naive or dogmatic metaphysician would make such bald claims. One of the primary aims of metaphysics is precisely to understand, in some measure, our own relation to the rest of reality, and of necessity it undertakes to do this from the position in which we find ourselves. The fact that we cannot stand outside ourselves to study that relation need not imply that it cannot be studied by us at all.
A second popular response to the question of how metaphysics is possible is to claim that it is so because, to the extent that there is a legitimate province of metaphysical inquiry, it is one entirely catered for by the empirical sciences. On this view, it is these sciences, if anything, that can tell us about the fundamental structure of reality. This being so, there can be no scope for a distinctly ‘philosophical’ approach to the questions of metaphysics, conceived as an approach different in its methods or objects from those already embraced by the empirical sciences. For instance, if there are fundamental questions to be raised about the existence and nature of space and time, it seems to follow that these can only be answered by such sciences as cosmology and quantum physics. There can be no room for ‘armchair’ philosophical speculation or ‘conceptual analysis’ as routes to addressing such questions. To the extent that metaphysical questions are genuinely answerable, it will be said that they are being (p.5) answered by people working in departments of physics, not by people working in departments of philosophy.
A view like this is often to be found expressed, either explicitly or in a barely disguised form, in modern books of popular science, designed to convey to a lay audience the arcane deliverances of the latest theories in physics—theories claiming, for instance, that space ‘really’ has many more than three dimensions or that the universe is the result of a quantum fluctuation in the vacuum and so came from ‘nothing’. In a subtler form, a devotion to scientism—as I call the doctrine that such legitimate metaphysical questions as there are belong to the province of the empirical sciences—is even to be found, ironically enough, in many departments of philosophy. A version of it has there become dignified under the title of ‘naturalized epistemology’. This is the view that all human knowledge—including any metaphysical knowledge that we may lay claim to—is a product of our biological nature as cognizing animals, and is hence to be studied by the methods of the life sciences, including psychology and evolutionary theory. Thus we see a bizarre situation arising in which popularizing scientists denounce the pretensions of philosophers while many of the people whom they are criticizing have in fact already abdicated any claim to know better than the scientists how to address the questions of metaphysics.
In my opinion, both types of devotee of scientism—those drawn from the ranks of scientists and those drawn from the ranks of soi‐disant philosophers—exhibit a blinkered dogmatism which is the very antithesis of genuine philosophy. Both fail to see that science presupposes metaphysics and that the role of philosophy is quite as much normative as descriptive—with everything, including science, coming within its critical purview. Scientists inevitably make metaphysical assumptions, whether explicitly or implicitly, in proposing and testing their theories—assumptions which go beyond anything that science itself can legitimate. These assumptions need to be examined critically, whether by scientists themselves or by philosophers—and either way, the critical philosophical thinking that must be done cannot look to the methods and objects of empirical science for its model. Empirical science at most tells us what is the case, not what must or may be (but happens not to be) the case. Metaphysics deals in possibilities. And only if we can delimit the scope of the possible can we hope to determine empirically what is actual. This is why empirical science is dependent upon metaphysics and cannot usurp the latter's proper role.
A third response to our question ‘How is metaphysics possible?’ is, unlike the first two, genuinely philosophical, drawing its inspiration from Kant—whence I call it neo‐Kantian. According to this view, metaphysics (p.6) does not and cannot tell us anything about objective reality ‘as it is in itself’, if indeed the notion of such a reality even makes sense. But it can tell us something about certain fundamentally necessary features of our thought about reality. For instance, it may be able to establish that we must think of the objects of perception as being situated in space and time and as being related causally to one another—perhaps on the grounds that, as Kant himself held, a recognition of ourselves as self‐conscious beings whose thoughts and experiences are ordered in time requires us to make reference to such a world of perceptible objects. By opting for a less ambitious aim, it is hoped that the possibility of a suitably modest metaphysics may be secured. However, such a position is fatally flawed, if its intention is to render ‘metaphysical’ claims legitimate by construing them as not venturing to speak of how things really are, as opposed to how we must think of them as being. Forwe, if we are anything, are part of reality ourselves, as are our thoughts, so that to purport to make claims about allegedly necessary features of our thoughts while simultaneously denying that anything is being claimed about the nature of ‘reality’ is to contradict oneself. Trying to make metaphysics safe by drawing in its horns in this way is an exercise which is doomed to failure.
Still more unsatisfactory is any attempt to legitimate metaphysics by enfeebling it yet further, by construing its claims as being merely descriptive of a conceptual scheme we find ourselves possessed of, without even purporting to establish the inevitability of that scheme. Metaphysical inquiry must at least be critical, so that to the extent that it deals in ‘concepts’ it cannot rest content to describe or analyse the concepts that we happen to have, but should, rather, seek to revise and refine these concepts where necessary. 4 But the point of such revision, if it has a point, can only be to render our concepts truer to reality—ensuring the mere internal consistency of a conceptual scheme is too modest an aim, because many mutually incompatible schemes could equally possess this feature. 5 If it is suggested that choice between such schemes can be rationally made by selecting what does least violence to our ‘natural’ beliefs or ‘intuitions’, leaving us in a condition of ‘reflective equilibrium’, then again it must be objected that an exercise of this sort does not deserve to be dignified by the name ‘metaphysics’, because we have no right to suppose that our natural beliefs reflect the fundamental structure of reality. It is one of the few (p.7) virtues of scientism that it recognizes this fact, because it recognizes that our natural beliefs are products of evolutionary processes which are geared to the practical demands of survival rather than to the theoretical demands of metaphysical truth.
The fourth and final response I wish to criticize is what I call semanticism. This is the view, notably subscribed to by Michael Dummett, that metaphysical questions can in principle be resolved by recourse to (and only by recourse to) the theory of meaning. 6 Thus, whether we are entitled to take a ‘realist’ view of some area of discourse, such as talk about set theory, or quantum physics, or the past, is to be decided by whether or not an adequate theory of meaning for that area of discourse will assign ‘realist’ truth‐conditions to its sentences, that is, truth‐conditions which reflect a commitment to the principle of bivalence as far as those sentences are concerned. In fact it seems that semanticism of this sort does not differ fundamentally from what I have been calling neo‐Kantianism, because its proponents such as Dummett regard the theory of meaning as providing the only legitimate basis for a theory of the structure and content of thought. 7 Thus semanticism only gives a linguistic dress to a kind of approach to metaphysics which we have already considered and rejected. If anything, semanticism threatens to reduce metaphysics to something even more parochial, by making answers to its questions turn on the linguistic practices of an arbitrarily chosen human community. No doubt the semanticist will claim that the features of the theory of meaning which are pertinent to metaphysical concerns are ‘deep’ ones, which transcend differences between different human speech communities. But what sort of basis would such a claim have? If merely anthropological, we are back to scientism or to some sort of relativism. Nor can the theory of meaning itself determine what counts as ‘deep’.
The basic problem with semanticism is that, to the extent that one can legitimately appeal to considerations of meaning in order to answer metaphysical questions, the considerations in question must not be merely ones of what we do mean—for there is no guarantee that we mean anything very precise or coherent by what we say—but, rather, they must be considerations of what we should mean. This is to reiterate the point that metaphysics must be critical and potentially revisionary of our currently accepted concepts and beliefs. However, questions of what we should mean cannot be answered wholly from within the theory of meaning, but require recourse to independent metaphysical argument. An illustration (p.8) of this is provided by the debate over what constitutes an object—a metaphysical debate if ever there was one, and one to which we shall return in detail in Chapter 2 . The semanticist will say, typically, that an object is to be understood as a possible referent of a singular term, and will contend that the notion of a singular term is one which can be explicated in a way which does not depend on a prior notion of what constitutes an object—for instance, by reference to the characteristic logical behaviour of singular terms as exhibited by the patterns of valid inference sustained by sentences containing them. 8 However, it is readily apparent that our existing language contains expressions which qualify as singular terms by any such criterion and yet which it would be extravagant to suppose make reference to objects of any sort. An example would be a definite description such as ‘the grin on John's face’. Ordinary language even gives the appearance ofquantifying over such spurious ‘objects’, as in a sentence like ‘John is wearing a broad grin’. Of course, such a sentence is satisfactorily paraphrasable by one avoiding a quantifier, namely, ‘John is grinning broadly’. But paraphrase is a symmetrical relation, so that there are no resources wholly within one's theory of meaning for the language wherewith to decide which of these two sentences is to be regarded as more accurately reflecting the ontology of its speakers. Much less does such a theory enable us to decide what an ‘object’ is, or what objects the world really contains. These questions can only be addressed by independent metaphysical argument, if they can be legitimately addressed at all. 9 The linguistic, or semanticist, approach to questions of metaphysics inevitably leads to a doctrine of extreme ontological relativity, as some of its proponents have realized. 10 In this guise, therefore, it collapses into a version of the first approach considered above, which I called relativism.
3. Metaphysical Possibility and the Possibility of Metaphysics
The time has now come for me to offer my own answer to the question of whether, and if so how, metaphysics is possible. My view is that it is indeed (p.9) possible: that is, I hold that it is possible to achieve reasonable answers to questions concerning the fundamental structure of reality—questions more fundamental than any that can be competently addressed by empirical science. But I do not claim that metaphysics on its own can, in general, tell us what there is. Rather—to a first approximation—I hold that metaphysics by itself only tells us what there could be. But given that metaphysics has told us this, experience can then tell us which of various alternative metaphysical possibilities is plausibly true in actuality. The point is that although what is actual must for that very reason be possible, experience alone cannot determine what is actual, in the absence of a metaphysical delimitation of the possible. In short, metaphysics itself is possible—indeed necessary—as a form of rational human inquiry because metaphysical possibility is an inescapable determinant of actuality. Stated in this highly abstract and condensed way, my answer may appear obscure and even gnomic, so my task in what remains of this chapter will be to unfold its implications.
So far I have begun to forge a link between the possibility of metaphysics and the notion of metaphysical possibility. The idea is that the realm of metaphysical possibility is a genuine one which needs to be explored, or at least assumed, before any claim to truth in actuality can be legitimated by experience. And this is a realm which cannot, of course, be explored solely by the methods of the empirical sciences, precisely because they merely purport to establish what is true in actuality on the basis of experience, and hence presuppose metaphysics. But it may be objected here that the only sort of possibility which the empirical sciences presuppose is logical possibility—and that this can be established without recourse to a distinct discipline of metaphysics, because logical possibility is simply a matter of compliance with the a priori laws of logic. In short, it may be urged that the only precondition which needs to be met by the theories of empirical science, before they are tested in the court of experience, is that they should not entail a logical contradiction. However, in the first place, the deliverances of experience itself can only be assessed in the light of metaphysical possibility and, in the second, such possibility is not simply tantamount to mere logical possibility as characterized a moment ago. I shall develop the latter point more fully in the next section when I discuss how metaphysical possibility may be defined, but some preliminary observations concerning it are appropriate here.
The logical possibility of a proposition or set of propositions, as characterized a moment ago, is simply a matter of its (or their) not entailing a logical contradiction. But metaphysical possibility is something quite distinct from this. In the first place, it is not—or, at least, is not merely—the possibility of aproposition (or set of propositions), but rather the (p.10) possibility of a state of affairs (one which isrepresentable, no doubt, by a proposition): and so in this sense it is a ‘real’, or de re, possibility. The notion of a state of affairs, of course, is itself a metaphysical notion, just one of a large family of such notions, some more basic than others. Other notions in this family are the notions of an object, aproperty, a relation, an individual, a kind, a part, a substance, existence, identity, instantiation, and indeed possibility, along with its correlative, necessity. Some of these notions are definable in terms of others, though precisely how they should be defined is itself a matter for metaphysical debate. Thus—as we shall see more fully in Chapter 6 —a substance might be defined to be an object which does notdepend for its existence upon any other object (where dependency is defined in terms of necessity). 11These metaphysical notions are not purely ‘logical’ notions: they are ontological. 12 They concernbeing and its modes, whereas logic, properly understood, does not concern being in general but, rather, the formal properties of and relations between propositions (which constitute only a small part of what there is). Moreover, these metaphysical notions are in a certain sense transcendental, in that they are not derivative from experience but are on the contrary to be invoked in construing what experience reveals of reality. Clearly, they are closely related to the categories of Aristotle and Kant, but my account of them crucially differs from Kant's (and is thereby closer to Aristotle's) in that I regard them as being genuinely applicable to reality and not merely to our thought about reality. They are not categories of thought, but categories of being. This is not, however, to say that the applicability of a given category to reality can, in general, be determined wholly a priori—only its possibleapplicability may be thus determinable. For instance, we may not be able to establish a priori that there actually are any substances, only that there could be. Only by recourse to experience, perhaps, can we have reason to believe that there are.
Of course, the ‘semanticist’ will claim that these ‘categories’ are just reflections of, and entirely derivative from, semantico‐syntactic features of the natural languages which we happen to employ—the notion of an ‘object’ corresponding to that of a singular term, the notion of a ‘property’ corresponding to that of a predicate, and so on. But I have already explained why I consider this view to invert the proper order of explanation. Such correspondences of this sort as may exist do so because any language which has evolved as a means to expressing truths about reality (p.11) must embody some recognition, however partial and distorted, of the metaphysical categories in terms of which the fundamental structure of reality is to be articulated. Since there is room for debate about that structure, it is unsurprising that different natural languages reflect some metaphysical categories more prominently than others. Such differences reflect, in all probability, differences in the tacitly held metaphysical beliefs of different human speech communities. But although linguistic structure can perhaps serve to reinforce and entrench those beliefs, the ‘Whorfian’ view that linguistic structure is the source of them is, I consider, quite unsustainable. 13
I began to talk about metaphysical categories in the course of introducing the notion of metaphysical possibility and urging that it differs from that of mere logical possibility. The metaphysical possibility of a state of affairs is not determined simply by the absence of contradiction in the propositions used to describe it—though, of course, such an absence of contradiction is a minimal requirement of metaphysical possibility. Consider, thus, an example par excellence of a question of metaphysical possibility: the question of whether there could be vague objects, that is, objects for which there could fail to be a fact of the matter as to their identity or diversity in certain circumstances. (We shall return to this and related questions at much greater length in Chapter 3 .) Now, it is true enough that a number of philosophers, notably Gareth Evans, have argued against this possibility by attempting to derive a contradiction from the supposition that a given identity statement is of indeterminate truth‐value, a supposition which may be expressed by a proposition of the form ‘∇(a = b)’. 14 As it happens, I consider that this argument is flawed, and consequently that a contradiction cannot be derived from the supposition in question. 15 It doesn't follow, however, that I must consider the existence of vague objects to be metaphysically possible. Indeed, I have grave doubts about this, because—as I shall explain in Chapter 2 —I consider that the only metaphysically defensible notion of an object is precisely that of an entity which possesses determinate identity‐conditions. 16 Thus, although I have, elsewhere, referred to the domain of subatomic particles as providing putative examples of ‘objects’ whose diachronic (p.12) identity may, in certain circumstances, be vague, 17 at a deeper level I am sympathetic to the view that what the relevant empirical evidence shows is that it is wrong to think of electrons and the like as really being objects at all. (Let it not be supposed, however, that this is just a verbal issue about the meaning of a word, ‘object’: the deeper point is that a satisfactory system of metaphysics needs to mark a fundamental division between those entities that do, and those that do not, possess determinate identity‐conditions—and appropriate use of the term ‘object’ serves precisely this purpose.) Thus we see that the validity of a claim that a certain state of affairs is metaphysically possible does not simply hinge on the question of whether or not the propositions used to describe it entail a contradiction, but rather on the question of whether acceptable metaphysical principles and categories permit the existence of that state of affairs. And this is a matter for distinctly metaphysical debate. A general comparison might be drawn here with questions of what is morally possible or permissible, which again cannot be settled purely by considerations of logic since at some stage distinctly moral notions must be brought in to play a substantive role in any argument for the moral permissibility of a given state of affairs.
Let us look at another group of examples in order to reinforce this conclusion. Metaphysicians have long debated about the possibility of change and the reality of time—and we too shall be exploring these questions in some depth in Chapters 4 and 5 . However, they are neither purely empirical nor purely logical questions. How we should conceive of time is itself a metaphysical question—a question of how, if at all, the notion of time is to be related to more fundamental metaphysical notions, including the categories. For instance, one view (the ‘Aristotelian’ view) is that time is the dimension in which alone a substance can receive contrary qualities, and this, if correct, would seem to imply at least three things: that substances must be able to persist identically through qualitative change, that there cannot be time without change, and that the unity of time rests upon the persistence of substances. Although I happen to agree with these claims, 18 I readily concede that they are controversial. But what I want to insist on just here is, first, that these issues are potentially capable of resolution through rational debate, and, secondly, that the kind of argument involved in such debate is distinctly metaphysical. Showing that time is metaphysically possible is not just a matter of demonstrating the logical (p.13)consistency of temporal discourse—by, for example, rebutting McTaggart's arguments to the contrary19 —nor is it simply a matter of formulating a consistent physical theory of time, along the lines of, say, Einstein's special theory of relativity. Einstein's theory makes certain fundamental claims about time—for instance, that simultaneity is relative and that the velocity of light cannot be surpassed—but that it does indeed concern time, and that in having such a concern it concerns something whose reality ispossible, are metaphysical questions which no merely scientific theory of this sort can settle.
While on the subject of time and change, here is one final, and quite specific, example which will serve to illustrate my general point. The example concerns the ‘possibility’ of one thing's becoming two, without thereby ceasing to exist—the intended implication being that in such a situation we would supposedly have, at a later time, two numerically distinct things which were formerly numerically identical. Although, very arguably, one can describe such a situation without logical contradiction, metaphysical reasoning appears to rule it out, on the grounds that no change in the objects concerned could occur at an appropriate time to effect their separation. (We are assuming, then, that the separation could not simply occur uncaused—itself a metaphysical assumption.) For if the putatively effective change were to occur prior to the supposed separation, it would be too early—since then it would have to affect both of the supposed objects in exactly the same way, they being, ex hypothesi, identical at that time. On the other hand, if the change were to occur subsequently to the supposed separation, it would obviously occur too late to have caused it, barring the possibility of backward causation. (This is not, of course, to deny that one thing can become two new things by ceasing to exist, nor that one thing can continue to exist while giving birth to another, as in ordinary cases of division or fission: it is only to deny that two things which, allegedly, were formerly one and the same can have been made to become separate and thereby non‐identical.)
4. The Nature of Metaphysical Necessity
The time has now come for me to present a more rigorous characterization of metaphysical possibility than is provided by the illustrative examples supplied above. Of course, notions of possibility and necessity are interdefinable. With that in mind, I shall approach the task in hand via (p.14) the notion of metaphysical necessity. Here, then, is the question that I should like to focus on for the time being: what exactly is logical necessity and how, if at all, does it differ from metaphysical necessity? For so far I have been implying that metaphysical necessity is quite distinct from logical necessity—and, indeed, in an important sense I think that this is true—but, at the same time, I must now acknowledge that there is one perfectly acceptable conception of ‘logical’ necessity which represents it as either coinciding with or embracing metaphysical necessity. It will be seen, however, that this acknowledgement does not in any way undermine the general thrust of my remarks in the previous section.
In the previous section I was content to characterize logical possibility somewhat vaguely in terms of compliance with the laws of logic, with logical necessity understood correspondingly. But in fact one can distinguish three different grades of logical necessity, as follows. First there is the strictly logically necessary—that which is true in virtue of the laws of logic alone. Secondly there is the narrowlylogically necessary—that which is true in virtue of the laws of logic together with definitions of non‐logical terms. And thirdly there is the broadly logically necessary—that which is true in every logically possible world, that is, in every possible world in which the laws of logic hold true. Now, one might very reasonably contend that this last grade of logical necessity is in fact coextensive withmetaphysical necessity—indeed, that these are just two different names for the same thing. In characterizing ‘broadly’ logical necessity in this way and associating it with the notion of metaphysical necessity, I am just following in an already well‐established tradition 20 —although I am sensible of the danger that this tradition may lead incautious philosophers to overlook the very division between logic and metaphysics which I was concerned to identify in the previous section. Some philosophers, of course, also speak of ‘conceptual’ necessity as being synonymous with some kind of logical necessity. I can agree with this, so long as the conceptually necessary is identified with what I have just called thenarrowly logically necessary. For I take it that the ‘conceptually’ necessary is that which is true solely in virtue of concepts together with the laws of logic.
Of course, what the laws of logic are is a matter for dispute, but we need not go into that here. There are some fairly uncontroversial candidates, such as the law of non‐contradiction: for any proposition P, it is not the case both that P and that not‐P. Thus, ‘It is not the case both that Ferdy (p.15) is a female fox and that Ferdy is not a female fox’ is strictly logically necessary, because it is an instance of that law. By contrast, ‘It is not the case both that Ferdy is a vixen and that Ferdy is not a female fox’ is onlynarrowly logically necessary, in my terms, because it can only be turned into an instance of that law by drawing on the definition of ‘vixen’, which is a non‐logical term. (What is a logical term is, of course, once again open to dispute, but no sensible account will treat ‘vixen’ as a logical term.)
What about an example of broadly logical necessity—more specifically, an example of such a necessity which is not also either a strictly or a narrowly logical necessity? A well‐known and plausible candidate would be ‘Water is H2O’. It may be objected that this proposition cannot be true in every logically possible world, because water does not exist in every such world. However, this difficulty can be circumvented fairly easily, in one way or another: for instance, one can distinguish between ‘strong’ and ‘weak’ necessity, saying that a proposition is weakly necessary (in the broadly logical sense) just in case it is true in every logically possible world in which its referring expressions are non‐empty. Then, assuming that ‘Water is H2O’ is to be analysed as an identity statement flanked by two referring expressions or names, it will turn out to be only weakly necessary (in the broadly logical sense). Notice, though, that if ‘Water is H2O’ is analysed instead as meaning ‘For any x, x is water if and only if xis H2O’, the difficulty goes away by itself, because ‘Water is H2O’ then turns out to be vacuously true in all worlds in which water doesn't exist (that is, in which nothing is water and in which, by the same token, nothing is H2O). What is crucial, however, is that it is not in virtue of the laws of logic plus definitions alone that ‘Water is H2O’ is true in all logically possible worlds (or, alternatively, in all such worlds in which its referring expressions are non‐empty)—and so it is neither strictly nor narrowly logically necessary, as I use these terms. Other possible candidates for the status of broadly logically necessary truths which are neither strictly nor narrowly logically necessary would be ‘Hesperus is Phosphorus’, ‘God exists’, ‘Nothing is both red and green all over at the same time’, and ‘This pain is mine’.
So, in virtue of what is ‘Water is H2O’ broadly logically necessary? Why, in virtue of the nature of water! Not, thus, in virtue of the laws of logic together with the concepts or definitions of water and H2O. It seems perfectly appropriate, then, to call this kind of necessity metaphysical necessity, since its ground is ontological rather than formal or conceptual. This might be a reason, indeed, for reserving the term ‘metaphysical necessity’ for those broadly logical necessities which are not also narrowly or strictly logically necessary. We could then also legitimately say, as many philosophers do, that it is a frequent (though not necessarily universal) feature of (p.16) metaphysical necessities that they are not knowable a priori—in contrast with strictly and narrowly logical necessities, which characteristically are knowable a priori. However, these epistemological and semantic matters are not our real focus of concern just here. Whether to define ‘metaphysical necessity’ as synonymous with ‘broadly logical necessity’ or as synonymous with ‘broadly logical necessity which is neither strictly nor narrowly logical necessity’ is ultimately just a matter for decision. Either definition will do, provided we stick to it. However, for the sake of simplicity and out of respect for the usage of other philosophers, I shall adopt the first option. But I should just emphasize that although I happily concede that the metaphysical necessity of a proposition such as ‘Water is H2O’ (or of the state of affairs which that proposition represents) is not knowable a priori—because it rests upon the nature of water, which is not thus knowable—I still want to insist that, as I put it in the previous section, experience alone cannot determine what is actual, in the absence of a metaphysical delimitation of the possible. Typically, such a delimitation will appeal to the metaphysical categories, which are indeed knowable purely a priori, unlike natural kinds such as water. (This is a point which I shall develop in much greater detail in Chapter 8 .)
5. Hale's Dilemma
It has to be acknowledged that not all philosophers are happy with the notion of metaphysical necessity. Thus, in a recent paper, Bob Hale presents what appears to be a serious dilemma for the friends of metaphysical necessity, based on an appeal to what he calls the generalized form of McFetridge's Thesis, that is, the thesis that if it is logically necessary that P, then there is no sense of ‘possible’ in which not‐P is possible. 21 Hale sees the dilemma as arising as follows:
However, setting aside the question of the soundness of Hale's argument for McFetridge's Thesis, I should like to query his reasons for thinking that each of the horns of the supposed dilemma should be uncongenial to the friends of metaphysical necessity. At the root of my objections is a dissatisfaction with Hale's conception of logical necessity.the argument for McFetridge's Thesis, if sound, does establish that if it is logically necessary thatp, then it is in no sense possible that not‐p, and so, in particular, that it is not metaphysically possible that not‐p, i.e. that it is metaphysically necessary that p. But then either the converse entailment holds quite generally, or it does not. If not, then it can be metaphysically necessary that p but logically possible that not‐p, so that metaphysical necessity is not, after all, absolute. If, on the other hand, whatever is metaphysically necessary is also logically necessary, then even if we have two notionally distinct kinds of necessity, both of them absolute, (p.17) they coincide in extension. Neither alternative is—or so it seems—congenial to the friends of metaphysical necessity. (p. 98)
Hale seems to be following orthodoxy in distinguishing between two kinds of logical necessity: ‘narrowly or strictly logical necessity’ and ‘broadly logical necessity’:
However, while this terminology is familiar, Hale's interpretation of it is idiosyncratic. For Hale, as we see, quite explicitly identifies ‘broadly logical necessity’ with ‘conceptual necessity’, whereas the orthodox tradition—exemplified by Plantinga and Forbes and by myself in the previous section—understands ‘broadly logical necessity’ to mean ‘truth in all logically possible worlds’: and the lattercannot be identified with ‘conceptual necessity’ because it is not, as a perfectly general rule, knowable a priori. Indeed, according to the tradition, as I understand it, ‘metaphysical necessity’ just is ‘broadly logical necessity’ as just defined, or at least is a sub‐category of the latter. Hale himself remarks, at one point, thatWe may distinguish between narrowly or strictly logical necessity and broadly logical necessity; I take the former to be a special case of the latter, and make no distinction between [the latter] and conceptual necessity. Hereafter, when I speak of logical necessity without further qualification, it is broadly logical necessity that I intend. (p. 94)
And this is more or less correct—on his understanding of the notion of broadly logical necessity. But it is not correct if, as the orthodox tradition has it, broadly logical necessity is identified with truth in all logically possible worlds.supposedly metaphysically necessary truths are typically held to be knowable only a posteriori, while [broadly] logical necessities are . . . knowable a priori—so there cannot, on pain of contradiction, be extensional coincidence [between them]. (pp. 98–9)
Now let us see how these considerations affect the cogency of Hale's attack on the notion of metaphysical necessity. I shall set aside any doubts about his argument for McFetridge's Thesis, because I accept that, on his understanding of what constitutes ‘broadly logical necessity’, the thesis is nearly enough correct. I shall focus, then, on the dilemma that this is supposed to create for the friends of metaphysical necessity. The first horn of the dilemma is supposed to arise if the metaphysician allows that it can be (p.18) metaphysically necessary that P and yet logically possible that not‐P—for then, Hale says, it must be conceded that metaphysical necessity is not ‘absolute’ but ‘relative’. Now, whether the metaphysician should indeed allow this depends on whether we understand the notion of ‘broadly logical possibility’ in Hale's sense or in the orthodox sense, that is, as being equivalent to conceptual possibility or as being equivalent to truth in some logically possible world. But, setting that matter aside for the moment, what is supposed to be the problem, for the friends of metaphysical necessity, about conceding that metaphysical necessity is not ‘absolute’ but ‘relative’, in Hale's terms?
At this point we need to understand exactly what Hale means by ‘relative’ necessity. This is what he says:
According to Hale, then, the metaphysician who concedes that metaphysical necessity is relative must concede that to claim that it is metaphysically necessary that P is to claim no more and no less than that, for some body of statements M, it is a logical consequence of M that P. But why should this not be congenial to the friends of metaphysical necessity? I see no harm whatever in conceding the point in question. No doubt there is some set of true statements or propositions M—roughly, the ones that characterize the ‘natures’ of all the entities that populate the world—such that every metaphysically necessary statement or proposition P is a logical consequence of M (that is, could be deduced from M). Thus ‘Water is H2O’ would be a consequence of M, in virtue of M's including a characterization of the nature of water. Of course, in this case, at least, the consequence is a trivial one, since the best way to characterize the nature of water is in terms of the very proposition in question, ‘Water is H2O’. However, we should not be tempted to think that any of this implies that metaphysical necessity has a status akin to, say, that of physical necessity. For—by my account, at least—every statement in M will itself express a broadly logical necessity, that is, will be true in every logically possible world. So it is not as though we are saying that ‘It is metaphysically necessary that P’ is analysable as meaning, ordefinitionally equivalent to, ‘P is a consequence of M’ (where M is defined as being something like the set of all truths about the natures of all entities). For we cannot define the notion of something's ‘nature’ independently of the notion of metaphysical necessity. (Moreover, we already have a definition of ‘metaphysical necessity’, in terms of truth in all logically possible worlds, and this definition doesn't mention ‘natures’. Of course, it may be debated how illuminating (p.19) our definition ultimately is—and that will depend on how illuminating one takes talk about ‘possible worlds’ to be. I shall not, however, pursue that issue just now, though we shall return to questions concerning possible worlds in later chapters.)By saying that a notion of necessity—φ‐necessity—is relative I mean that there is some body of statements Φ such that to claim that it is φ‐necessary that p is to claim no more and no less than that it is a logical consequence of Φ that p. (p. 93)
So what I want to say is that the friends of metaphysical necessity can quite happily concede that, technically speaking, metaphysical necessity qualifies as a species of ‘relative’ necessity, in Hale's terms. Here I observe that, concerning the ‘relativity’ of ‘relative’ necessity, Hale remarks:
But, whereas in the case of physical necessity it is clear that members of the relevant set Φ may be false in some logically possible worlds, this is not so, by my account, in the case of metaphysical necessity. Metaphysical necessity, by my account, is as ‘hard’ as any kind of necessity could be.The relativity of φ‐necessity consists in the fact that, whilst no φ‐necessary statement can be false, provided that all the members of Φ are true, it is not excluded that there are other senses of ‘possible’ in which the members of Φ may be false. (p. 93)
Why, however, does Hale think that conceding the relativity of metaphysical necessity should be uncongenial to the friends of metaphysical necessity? For the following reason:
What Hale is saying here is that if metaphysical necessity is not ‘absolute’, then it can be metaphysically necessary that P and yet still be the case that not‐P is true in some logically possible world. But Hale is not at all entitled to claim this, given that what he means by ‘broadly logical possibility’ is conceptual possibility, rather than (as the orthodox tradition has it) truth in some logically possible world. All that Hale is entitled to claim, on his own terms, is that if metaphysical necessity is not ‘absolute’, then it can be metaphysically necessary that P and yet still be the case that not‐P is conceptually possible. But to concede the latter should be by no means uncongenial to the friends of metaphysical necessity. For instance, they should quite happily allow that, while it is metaphysically necessary that water is H2O, it is none the less conceptually possible that water is not H2O—for the very point that the metaphysicians want to make is that the necessity of that identity is grounded in the nature of water rather than in ourconcepts of ‘water’ and ‘H2O’.To accept that metaphysical necessity is not absolute is to acknowledge that while it is, say, metaphysically necessary that heat is mean kinetic energy of molecules, there are possible worlds—logically possible worlds—in which this is not so [and yet] . . . what the metaphysicians wanted to maintain is that, given that heat is mean kinetic energy of molecules, there are no possible worlds in which heat is not so constituted (p. 98)
Having disarmed the first horn of Hale's supposed dilemma, let us turn (p.20) to the second, which is supposed to arise if the metaphysician contends that logical and metaphysical necessity are extensionally coincident. What is supposed to be the problem in this case? This is what Hale says:
Once again we have a ready answer to Hale's worries. Certainly, the metaphysician should not concede that metaphysical necessity coincides with what Hale calls ‘broadly logical necessity’, that is, withconceptual necessity, but he can happily allow that it coincides with what the orthodox tradition understands by ‘broadly logical necessity’, namely, truth in all logically possible worlds—for, as I pointed out earlier, the latter is not, as a perfectly general rule, knowable a priori.accepting that metaphysical and logical necessity are extensionally coincident is scarcely less unattractive. How could it be held that e.g. the identity of heat with mean kinetic energy of molecules is logically necessary? Furthermore, such supposedly metaphysically necessary truths are typically held to be knowable only a posteriori, while logical necessities are . . . knowable a priori—so there cannot, on pain of contradiction, be extensional coincidence. (pp. 98–9)
To conclude: I do not believe that Hale's arguments pose any real threat to the friends of metaphysical necessity. McFetridge's Thesis is correct (or nearly enough so), but innocuous, if construed as a thesis concerning conceptual necessity. The metaphysician should be prepared to agree that metaphysical necessity is not ‘absolute’, in Hale's terms, in as much as it can be metaphysically necessary that P and yet conceptually possible that not‐P. But this doesn't imply that the metaphysician must agree that it can be metaphysically necessary that P and yet not‐P be true in some logically possible world. For what is crucially overlooked by Hale is the fact that to say that not‐P is conceptually possible is not equivalent to saying that not‐P is true in some logically possible world. In a perfectly good sense, then, metaphysical necessity is ‘absolute’—namely, in the sense that if P is metaphysically necessary, there is no logically possible world (and so no world whatever) in which not‐P is true.
Our original question (first posed in section 4 ) was: what is logical necessity and how, if at all, does it differ from metaphysical necessity? My answer is: the term ‘logical necessity’ is multiply (three‐way) ambiguous and in one, but only one, of its senses can it be said to coincide in meaning with the term ‘metaphysical necessity’. A proposition is necessary in this sense just in case it is true in every logically possible world, that is, in every world in which the laws of logic hold true. This is, in a perfectly good sense, as ‘hard’ a kind of necessity as there can be—even though it is consistent with saying that a proposition P is necessary in this sense to say that not‐P is none the less possible in another sense, namely, in the sense that the truth of not‐P is not ruled out by the laws of logic together with any (p.21)non‐logical concepts which P involves. This kind of necessity is not, in many cases, knowable a priori, precisely because it is not, in many cases, grounded in logic and concepts but, rather, in the ‘natures’ of things.
6. Logic and Metaphysics
I say above that P is ‘strictly’ logically necessary just in case P is true ‘in virtue of the laws of logic alone’. However, this may seem to raise the spectre of a regress problem of the kind which Hale himself investigates towards the end of his paper. To say that P is true in virtue of the laws of logic alone is, seemingly, to say that P either is a law of logic itself, or else follows from—is a logical consequence of—such laws. But, in the latter case, what about the status of the proposition stating that P is a logical consequence of the laws? Should we regard that proposition, too, as being strictly logically necessary? If so, then we must say either that that proposition is itself one of the laws, or else that it is a consequence of those laws. And in the latter case, we shall then have to consider the yet further proposition that the proposition in question is a consequence of the laws—and so it goes on.
Is there really a problem lurking here? Let us consider a specific case of a strictly logically necessary proposition, such as ‘It is not the case both that Ferdy is a female fox and that Ferdy is not a female fox’. This is not itself a law of logic, because it contains non‐logical terms. But it is a logical consequence—because it is an instance—of the logical law of non‐contradiction: ‘For any proposition P, it is not the case both that P and that not‐P’. What we then have to consider is the status of the further proposition: ‘The proposition that it is not the case both that Ferdy is a female fox and that Ferdy is not a female fox is an instance of the logical law that for any proposition P, it is not the case both that P and that not‐P’. Call this proposition ‘X’. Ought we to say that X, like the two propositions to which it refers, is strictly logically necessary? Clearly, X is not itself a law of logic. So is it then a logical consequence of the laws of logic—and if so, what are the relevant laws? The answer seems to be that we would do well not to look for such laws, because even if we found some, a structurally similar problem would arise one level up.
But what makes X true, and necessarily so, if it is not strictly logically necessary? Well, what Xexpresses is a relationship between certain propositions, that is, between certain abstract entities which exist in every possible world. And that relationship obtains in every logically possible world, so that X is broadly logically necessary. Isn't that enough? Strictly logically necessary propositions are ones that are true in virtue of the laws (p.22) of logic alone—and sometimes this is a matter of the propositions in question being logical consequences of those laws: but true propositions stating the existence of logical consequence relations between other propositions are only broadly logically necessary—they express metaphysical necessities. The lesson would seem to be that logic, in the strict sense, is undergirded by metaphysics—as indeed is every other intellectual discipline. This serves to reinforce our earlier conclusion that logic by itself cannot be called upon to do the work of metaphysics.
7. Metaphysics, Actuality, and Experience
Having now explained why I hold that the realm of metaphysical possibility constitutes a distinct arena of rational inquiry, I want to return to the issue of what metaphysics can tell us about reality. I have suggested that metaphysics by itself can only tell us what is metaphysically possible, not which of various alternative metaphysical possibilities actually obtains. But this claim clearly needs some refinement. First, it would be unwise to rule out altogether the possibility of metaphysics by itself (or at most in conjunction with logic) establishing the existence of some metaphysically necessary states of affairs. Indeed, it would seem incoherent to attempt to rule this out, given that one can establish that some states of affairs are not metaphysically possible, by deriving a contradiction from the supposition that they are: for to establish, thus, that it is not metaphysically possible for S to exist is to establish that it is metaphysically necessary for S not to exist. However, what can be established by such purely logical means will not be a substantive metaphysical necessity, but rather just a logical impossibility (in the strict or narrow sense). A substantive metaphysical necessity, such as the necessary existence of God or of time, would be very much harder to establish. Second, the conclusions of metaphysical arguments will often have the form of conditional statements, which are themselves shown by such arguments to be unconditionally true. For instance, such a conclusion might be that if time is real, thensome persisting substance must exist. The fact that metaphysics often delivers conclusions of this form in no way conflicts with the claim that the primary object of metaphysics is to establish the metaphysical possibility of various states of affairs and that it does not, in general, tell us by itself that such‐and‐such a state of affairs actually obtains. Indeed, establishing the possibility of a state of affairs is often done precisely by showing that it necessarily would obtain if some other state of affairs, which has already been shown to be possible, were to obtain.
How, then, are we to form rational judgements as to which of various (p.23) metaphysically possible alternatives do actually obtain? In a word: by experience. Knowing how the world could be in respect of its fundamental structure, we must judge as best we can how it is by determining how well our experience can be accommodated with this or that alternative metaphysical possibility as regards that structure. 22 This may appear to give metaphysical theorizing a status similar to that of scientific theorizing, but the similarity is only superficial. A judgement that the world actually exhibits a given metaphysical feature—for instance, that it contains substances or that time is real—will indeed be an a posteriori judgement, being responsive to the evidence of experience. But the content of the judgement still retains its modal character as expressing a genuine metaphysical possibility, albeit one judged now to be actualized. This view of the epistemological status of metaphysical claims, as being at once a posteriori and modal, is obviously akin to the view commonly associated with Kripke. He claims, for instance, that some truths of identity and constitution are metaphysically necessary and yet knowable only a posteriori. 23 What can be known a priori, according to Kripke, is only that if an identity between objects a and b obtains, then it obtains of metaphysical necessity—but that it obtains can only be known a posteriori. As it happens, I am not entirely convinced by this particular Kripkean claim—or, at least, by his argument for it—nor by his connected claim that the original constitution of an object is metaphysically necessary. 24 But I wholeheartedly endorse his insight that metaphysics can at once be concerned with modal truths and yet deliver answers to questions about actuality which are, and have to be, a posteriori in character.
Kant, of course, thought otherwise. He thought that to the extent that metaphysics can tell us anything about actuality, it must issue in a priori judgements. It must tell us, independently of any recourse to the evidence of experience, that the world must be thus‐and‐so. And its deliverances must not be mere analytic truths, which would reveal only logical connections between certain of our concepts, without any guarantee of those concepts answering to reality. It must then deliver a priori, substantive, and necessary truths—a tall order. No wonder that Kant concluded that (p.24) the ‘actual’ world of which metaphysics speaks can only be the phenomenal world as humanly experienced, not reality ‘as it is in itself’. But if we give up the vain hope that metaphysics can provide absolutely certain and unalterable insights into the fundamental nature of things, we can retain the conviction that metaphysics itself is a viable a priori discipline, and moreover one which deals in real possibilities—that is, possibilities for reality ‘as it is in itself’.
Kant would object here that the ‘categories’ are only capable of legitimate employment when restricted to objects in space and time, which he regarded as phenomenal. But such a restriction is wholly unwarranted and the attempt to impose it leads Kant into self‐contradiction. Whether space and time are ‘phenomenal’, and thus not features of reality independently of the way in which we experience it, is itself an important metaphysical question—so that, to the extent that Kant himself attempts to offer a reasoned answer to this question, he cannot consistently claim that metaphysics is or should be concerned only with how things are in the phenomenal world. When we examine his arguments for the phenomenal status of space and time, we find that he says, for instance, that if space were real, it would have to be a real ‘non‐entity’, apparently because it could not be a substance nor a relation between substances. 25 But this is (p.25) just a metaphysical argument: not one which I find very compelling, indeed, but still one which conforms to the conception of metaphysics which I have been defending. And the telling point is that it does not respect the official Kantian restriction to speak only of objects in the phenomenal world of human experience. (How could it respect that restriction, given that it is an argument for the phenomenal status of that world?) It is indeed possible that space and time are ‘unreal’, in the sense that our best judgement as to how in actuality reality is fundamentally structured will not find an application for these notions. But if so, that judgement will have been arrived at in part through metaphysical argument in which the categories (not precisely Kant's categories, of course) have been deployed in a way unhampered by Kant's attempted restriction.
The question remains as to precisely how, on my view, experience may enable us to advance from a judgement of metaphysical possibility to a claim that such a possibility is actualized. But there is no general algorithm to be discovered here. Each case must be assessed individually, on its merits. This is one reason why the Dummettian vision of resolving the outstanding problems of metaphysics by the general strategy of examining the applicability of the bivalence principle seems somewhat facile. The sorts of empirical consideration which will bear on the question of whether time is real, and the way in which they will bear upon it, can be expected to have little similarity with the corresponding issues to do with the question of whether the self is real, for instance. However, a simple example will illustrate how, in a particular case, empirical considerations can interact with a priori metaphysical argument to motivate a claim about actuality.
Consider, as a case study, David Lewis's claim that persisting objects perdure rather than endure—that is, that they persist by virtue of having successive temporal parts at successive moments of time. He argues for this on the grounds, inter alia, that only thus can we explain the possibility of such objects undergoing intrinsic change, that is, undergoing a change of their intrinsic or non‐relational properties, such as a change (p.26) from possessing a bent shape to possessing a straight shape. 26Lewis holds that it is different temporal parts of the persisting object which must possess the different shapes, so that the persisting object only possesses them derivatively, by virtue of consisting of a succession of such parts some of which possess different shapes. To this it may be replied that if all objects capable of undergoing intrinsic change are ultimately composed of intrinsically unchanging objects—‘atoms’ of some sort—upon whose properties the properties of those changeable objects supervene, then we need not after all suppose that any persisting objects have temporal parts. 27 The atoms need not have, because, ex hypothesi, they do not undergo intrinsic change. Nor need the intrinsically changeable objects which are composed by the atoms, because on the proposed account an intrinsic change in the properties of a composite object is ultimately just a change in the relations between its atomic constituents. Now, whether one is convinced by this reply is not relevant here (though we shall be looking into the issue much more deeply in Chapter 5 ). What is of interest now is that it presents us with an opportunity to combine metaphysical argument with empirical scientific theory in order to reach a judgement as to whether or not a certain metaphysical view is plausibly true in actuality—in this case, the view that persisting objects have temporal parts. If the metaphysical argument just given is sound and it is also the case that empirical science gives us good reason to believe that atomism (in some appropriate form) is correct, then we shall have grounds—partly a posteriori ones—for claiming that persisting objects do not in fact have temporal parts.
Oddly enough, many philosophers feel distinctly uneasy about combining empirical and metaphysical considerations in this way: they make remarks about the dangers of ‘giving hostages to fortune’—that is, opening oneself up to the possibility that one's claims about the metaphysical features of actuality will be undermined by developments in empirical scientific theory. This, I think, can only be because they still hanker after the impossible ‘rationalist’ dream of being able to determine the fundamental structure of reality wholly a priori and with absolute certainty. Kant taught us that this is indeed a dream. But, regrettably, rather than continuing to trade in reality, he opted for the comforts of certainty and empirical inviolability. The message that I have been trying to convey in (p.27) this chapter—and hope to reinforce throughout the book—is that metaphysics can indeed be about reality, and can avoid collapse into empirical scientific theory, provided we can learn to be content with the fact that, as far as actuality is concerned, metaphysics cannot provide us with certainties.
Notes:
(20) See, for example, Alvin Plantinga, The Nature of Necessity (Oxford: Clarendon Press, 1974 ), 2, and Graeme Forbes, The Metaphysics of Modality (Oxford: Clarendon Press, 1985 ), 2. As we shall shortly see, however, some philosophers—such as Bob Hale, in a paper to be discussed in the nextsection—do not follow in this tradition.
(22) Here it may be objected that circularity—or at least relativism—is threatened by my concomitant claim that experience itself is, as it were, metaphysically laden. But in fact this need no more create a problem than does the theory‐laden nature of observation in the case of empirical scientific theory. At most, all that needs to be abandoned is any simple‐minded foundationalism in either realm of inquiry, and an acceptance of some form of fallibilism.
(23) See Kripke, Naming and Necessity, 97 ff.
(25) See Kant, Critique of Pure Reason, A 39–40, B 56–7. That such a ‘non‐entity’ [Unding] cannot exist in reality is, of course, itself a metaphysical claim, whose truth requires to be established for the success of Kant's argument. The relevant passage of the Critique reads, in Kemp Smith's translation, as follows: ‘Those . . . who maintain the absolute reality of space and time, whether as subsistent or only as inherent . . . [either] have to admit two eternal and infinite self‐subsistent non‐entities (space and time), which are there (yet without there being anything real) only in order to contain in themselves all that is real [zwei ewige und unendliche für sich bestehende Undinge (Raum und Zeit) . . . welche dasind (ohne dass doch etwas Wirkliches ist), nur um alles Wirkliche in sich zu befassen] . . . [or] are obliged to deny that a priori mathematical doctrines have any validity in respect of real things (for instance, in space).’
Kant's targets are, of course, the views of Newton and Leibniz respectively. His objection here to the relational theory of space presupposes the correctness of his view that geometrical truths have an a priori synthetic status—a view which can no longer be taken seriously. But even if that view of geometry were correct, it cannot be valid to argue for the phenomenal status of space on such purely epistemological grounds; one must in addition at least establish that it is impossible for us to have substantive a priori knowledge of a mind‐independent reality, and this is a metaphysical claim. Elsewhere, of course, Kant uses the argument from incongruent counterparts against the relational theory, though this is, once again, a straightforwardly metaphysical argument in my sense. (See, especially, Kant's pre‐critical work, Concerning the Ultimate Foundation of the Differentiation of Regions in Space, in Kant: Selected Pre‐Critical Writings, trans. G. B. Kerferd and D. E. Walford, Manchester: Manchester University Press, 1968 .)
I have, incidentally, included the original German at one point in my quotation above from theCritique, where Kemp Smith's translation is not entirely perspicuous. By wirklich I take Kant to mean here not just ‘real’ in the sense of being objectively existent, for that would make his relative clause [welche dasind (ohne dass doch etwas Wirkliches ist)] express something little short of a contradiction. (There can't be something ‘there’—not even a ‘non‐entity’—without there being something existent.) Thus—here, at least—I take Kant to mean, by wirklich, ‘real’ in a more substantial sense, perhaps one which implies being efficacious (wirksam, in German). It is, of course, a familiar complaint against the Newtonian conception of absolute space that it is hard to see how the existence of space thus conceived could make any difference to the behaviour of physical things. (This, again, is a metaphysical point.) For a related discussion of Frege's use of the word wirklich, see Michael Dummett,Frege: Philosophy of Mathematics (London: Duckworth, 1991 ), 80–1.
