1. The Elimination Of Metaphysics

There appears to be a rhythm in the history of philosophy. Periods of great metaphysical system building are followed by brief periods of antimetaphysical reaction. The great constructive systems of seventeenth‐and early eighteenth‐century metaphysics were followed by Hume's damnation and Kant's drastic restriction of metaphysics (even though we rightly deem both to have elaborated metaphysical systems). The early nineteenth century saw the revival of grand systematic metaphysics among the post‐Kantian German Idealists, only to be followed by the destructive reaction of nineteenth‐century positivism. But this too was a passing phase, followed by such metaphysical system builders as Bergson, Bradley, McTaggart, Alexander, and Heidegger. And this again bred a reaction in the inter‐war years.

The Manifesto of the Vienna Circle, a pamphlet entitled ‘The Scientific World Conception’, published in 1929, declared that metaphysical assertions are not false propositions, but nonsensical pseudo‐propositions. All true or false propositions are either analytic or empirical. The former yield no knowledge of matters of fact. The latter are the totality of cognitively significant propositions. These have a meaning in as much as they are verifiable in experience. Metaphysical assertions are not analytic, for they are not true in virtue of the meanings of their constituent terms, are not derivable from the laws of logic and explicit definitions alone, and do not express conventions for the use of terms (these alternative formulations surprisingly being taken to be roughly equivalent). But they are not empirical propositions either, for they are not verifiable in experience, they have no determinable truth‐conditions, and make no difference to any possible experience (these formulations being similarly taken to be equivalent).

The argument was essentially a rerun of Hume's condemnation of (p.346) divinity and school metaphysics as sophistry and illusion. The logical positivists' elimination of what they called ‘metaphysics’ differed from Hume's primarily in the manner in which they harnessed the new logic and the techniques of logical analysis derived from Russell and Wittgenstein to their rationale for the condemnation of metaphysics, and secondarily in their perfunctory willingness to assimilate metaphysical utterances to works of poetry or music. For they were willing to concede that metaphysical utterances might be taken to express an attitude towards life, a Lebensgefühl.¹ Their criticisms of metaphysics differed from Kant's in as much as they rejected the intelligibility of Kant's category of synthetic a priori propositions and hence of his conception of transcendental metaphysics as a description of synthetic a priori conditions of possible experience, and repudiated his claim that certain propositions of transcendent metaphysics, though not knowable, are nevertheless intelligible as ideas of reason.

It is noteworthy that their primary inspiration was Wittgenstein's Tractatus. For they took the Tractatus to have established that logical propositions in general, and analytic propositions in particular (including propositions of mathematics), are vacuous tautologies that say nothing about reality.² All cognitively significant propositions are empirical. All empirical propositions—that is, all propositions with a sense, are reducible to truth‐functional combinations of elementary propositions. Elementary propositions were equated, by members of the Vienna Circle, with descriptions of the given—that is, of immediate experience, which they called ‘protocol sentences’. Partly as a consequence of conversations with Wittgenstein in 1929, they took verifiability to be the criterion of significance for empirical propositions, and took the meaning of any empirical proposition to be given by its method of verification. Without warrant, they read this principle of significance and the principle of verification back into the Tractatus.³ Interestingly, they (p.347) disregarded the fact that Wittgenstein, in the Tractatus, had committed himself to the ineffability of metaphysical truths. For, although he had argued that metaphysical truths cannot coherently be stated in language, he claimed that they are shown (but cannot be said) by wellformed propositions of natural language. So his Weltanschauung was profoundly different from theirs. He drew the bounds of sense not in order to eliminate metaphysics tout court, but rather in order to make room for ineffable metaphysics.

The main proponent of the anti‐metaphysical doctrines of the Circle was Carnap. Already in his 1928 paper ‘Pseudo‐Problems in Philosophy: The Heteropsychological and the Realism Controversy’ he argued that disputes between realism and idealism regarding the external world or between realism and solipsism regarding other minds are bogus, since no possible experience can decide between the respective alternatives. In ‘The Elimination of Metaphysics through the Logical Analysis of Language’ published in 1931/2, Carnap argued that metaphysical utterances are mere pseudo‐propositions that fail to meet the requirement of verifiability that is a condition for possession of cognitive significance. And in The Logical Syntax of Language, published in 1934, he tried to show that the Tractatus constraints upon what can be said do not, pace Wittgenstein, make room for ineffable metaphysics and that the pseudo‐propositions of the Tractatus metaphysics are not attempts to say what can only be shown. Rather, what can be salvaged from such pseudo‐propositions are syntactical statements about words. The pseudo‐propositions of the Tractatus are pseudo‐object sentences in the material mode that can be recast as genuine syntactical sentences in the formal mode.⁴

Although Carnap was among the most fervent anti‐metaphysicians in the Vienna Circle, others, especially Schlick and Neurath, participated eagerly in the crusade. These doctrines of the Circle were eloquently transmitted to the English speaking world by A. J. Ayer's pugnacious Language, Truth and Logic, published in 1936, which became the classic English statement of the central tenets of logical positivism. Like Carnap, Ayer argued that putatively metaphysical propositions are neither tautologies nor empirical propositions. No possible sense‐experience is (p.348) relevant to determining their truth or falsehood. Consequently, they are neither true nor false but rather are literally senseless. Transcendent metaphysics is nonsense, for no statement that refers to a putative reality that transcends any possible sense‐experience can have cognitive meaning. In short, metaphysicians produce sentences that fail to conform to the conditions under which alone a sentence can be literally significant. The primary respect in which Ayer differed from Carnap was over the relatively trivial concession Carnap made in holding that the utterances of metaphysicians should be treated as on a par with the writings of poets. Ayer observed that the sentences of poetry are not normally forms of tacit nonsense—they may be false, but they typically have a (cognitive) meaning. By contrast, the sentences of the metaphysician are, inadvertently, sheer nonsense.

Apart from the logical positivists, there was in the 1930s another powerful source of animus towards traditional metaphysics. While the positivists assailed metaphysics with the blunt tool of the principle of verification, Wittgenstein himself was lecturing at Cambridge. By 1932 he had altogether abandoned his Tractatus doctrines concerning the ineffability of metaphysics and the impossibility of making metaphysical statements with any sense, rejected verifiability as a criterion of meaningfulness, and repudiated the principle of verification. Nevertheless, he launched a new onslaught on metaphysics. He now claimed that what appear to be metaphysical statements are at best expressions of rules for the use of the constituent terms of the putative statement in the misleading guise of descriptions of reality, and that traditional metaphysical doctrines, such as solipsism, idealism, and realism, are nonsense. His views were not published until after his death, and were known only to the select circle of his pupils and friends. Nevertheless, they undoubtedly contributed to the general animosity of analytic philosophers towards metaphysics in the 1930s and even more markedly in the post‐war years.

2. The Rehabilitation of Metaphysics

Throughout the 1930s in Britain analytic philosophy, originating in the pre‐war work of Moore and Russell, made progress. The earlier style of logical atomist analysis gave way in Cambridge to a new form of analytic philosophy, inspired largely by Wittgenstein's teachings, which spread from Cambridge to other philosophy departments. Logical positivism made inroads among the young Turks of academic philosophy, and Oxford awoke from its slumbers, guided by Ryle and his circle. The (p.349) metaphysical systems of the previous generation of Bradley, Alexander, and McTaggart were out of fashion. Surveying the scene in 1933, the young Braithwaite, influenced by Russell, Moore, and Wittgenstein, held that the new style of analytic philosophy showed the futility of supposing that it is possible to construct a deductive metaphysical system to which experience must conform. Hence, he wrote, ‘we can be certain beforehand that a system professing to derive by logically necessary implications from logically necessary premisses interesting empirical propositions is wrong somewhere. We in Cambridge have been fortunate in having The Nature of Existence of J. E. McTaggart as an awful example.’⁵ Similarly, in Oxford, as Ryle later wrote, ‘In the 1930s the Vienna Circle made a big impact on my generation and the next generation of philosophers. Most of us took fairly untragically its demolition of Metaphysics. After all we never met anyone engaged in committing any metaphysics; our copies of Appearance and Reality were dusty; and most of us had never seen a copy of Sein und Zeit.’⁶

Collingwood accepted the positivists' contention that metaphysical utterances, such as ‘Every event has a cause’ or ‘There is a God’, are neither true nor false. In An Essay on Metaphysics, published in 1940, he argued that such assertions are indeed not genuine propositions, but expressions of the absolute presuppositions of the thought of a given historical epoch. On his view, a proposition is essentially an answer to a question, and every question rests on a presupposition from which it arises. Presuppositions may be relative or absolute. Relative presuppositions are both presuppositions of one question and answers to another. Absolute presuppositions are presuppositions of questions but never answers to one. So they do not express propositions, can be neither verifiable nor falsifiable, and are neither true nor false. The study of metaphysics, however, is not the argumentative defence of absolute presuppositions. It is a historical discipline that articulates the absolute presuppositions of the science or history of an epoch, and it results in true or false historical statements, which are indeed metaphysical propositions, characterizing the absolute presuppositions of, for example, Greek science or history, Newtonian mechanics, or modern physics.

(p.350) The idea of identifying something that can be called ‘the presuppositions of thought’ and that can be said to characterize the science or history of an epoch is appealing to the historian of ideas. But Collingwood's conception of a proposition, and hence too of truth and falsehood, was idiosyncratic, open to numerous objections, and won no adherents. His conception of absolute presuppositions, which are not propositions, are not supportable by argument, and which lack truth values, provided no explanation of the fact that traditional metaphysicians supported their assertions by extensive argument, and endeavoured to prove the falsity of their denials. His account of metaphysics does not correctly represent either the intentions or the productions of traditional metaphysicians, who conceived of their claims as true or false descriptions of how things necessarily are, and not of how things are transiently presupposed to be.

A different defence of metaphysics, which likewise accepted the positivist claim that metaphysical utterances are neither true nor false, but denied that they are nonsense, was given by H. H. Price. In his Presidential address to the Mind Association in 1945, in which he surveyed the current state of philosophy in Britain, he argued against the trend of analytic philosophy of the previous decade, insisting that ‘clarity is not enough’. Under the influence of Wittgenstein and of the logical positivists, analytic philosophers had argued that clarification is the fundamental task of philosophy, that philosophy can yield no new knowledge but only make clear what we already know. The sole task of philosophy is analysis of the statements of science, history, common sense, and ethics—the clarification of the meanings of statements that generate philosophical puzzlement. A corollary of the analytic task, it was argued, is the therapeutic task of eliminating perplexity and conceptual confusion and of disclosing the latent nonsense in much philosophical thought. Price defended the new style philosophy against many of the accusations that were being directed at it. He was willing to concede that analysis, in some generous sense of the term—that is, conceptual clarification—is the large part of philosophy of logic, epistemology, and even moral philosophy. But, he demurred, it had been a mistake to condemn metaphysics as nonsense and to consign the venerable subject to the rubbish heaps of intellectual history. Echoing Braithwaite, he agreed that it is illegitimate to argue to a conclusion concerning matters of fact if there are no matters of fact among one's premisses. But, he claimed, the only metaphysical argument that would be eliminated by this principle is the ontological argument for the existence of God. And it would be parochial to accept the verdict that all the great systematic speculative metaphysical systems of past philosophers are nothing but nonsense.

(p.351) How then should we view such systems? Price, like Ayer, rejected the Carnapian suggestion that they are akin to works of art and poetry, expressing attitudes towards the world. Rather we should think of them as ‘alternative modes of conceptual arrangement by which the body of empirical data is systematically ordered’, like alternative maps of the same territory with different methods of projection.⁷ In this sense they are no more true or false, right or wrong, than Mercator's projection is true or false, right or wrong. They are possible or not possible, adequate or inadequate to the task of representing what we wish to represent by their use, illuminating or unilluminating in highlighting relationships that are important to us for the purposes we have. Hence the choice between different metaphysical systems is not between the true and the false but between the less good and the better, or between the good in such‐and‐such respects and the good in such‐and‐such different respects. The task of the speculative metaphysician is indeed not to add to our knowledge of facts. But it is not analysis either. It is rather ‘to produce a unified conceptual scheme under which all known types of empirical fact may be systematically arranged’.⁸ Price's plea for such metaphysics went unheeded. As far as I know, there was no debate over whether this conception of what he had called ‘speculative metaphysics’ was correct.⁹

From 1945 until the end of the 1950s analytic philosophy evolved in Britain and elsewhere without any metaphysical pretensions, and, on the whole, without much attempt to aspire to the degree of generality characteristic of the ontological and metaphysical pronouncements of the great system‐builders of the past. In 1959, however, Strawson published his rightly renowned book Individuals. It operated at dizzying heights of generality hitherto unknown among postwar British analytic philosophers, and it professed unashamedly to be an exercise in metaphysics—it was, as its subtitle announced, ‘an essay in descriptive metaphysics’. In the elegant but all too brief introduction to the book, Strawson distinguished between revisionary and descriptive metaphysics. Descriptive metaphysics, he explained, ‘is content to describe the actual structure of our thought about the world’, while revisionary metaphysics ‘is concerned to produce a better structure.’¹⁰

(p.352) Descriptive metaphysics does not differ from conceptual analysis in intent. Like conceptual analysis or, as Strawson was later to characterize it, ‘connective analysis’, it is concerned with describing and clarifying the concepts we employ in discourse about ourselves and about the world, and in elucidating their relationships—their forms of relative priority, dependency, and interdependency. Descriptive metaphysics differs from connective analysis in general primarily in that the concepts that it investigates are characteristically highly general, irreducible, basic, and, in a special sense, non‐contingent. The generality of the concepts it studies is manifest in any list of concepts that have attracted the attention of metaphysicians—concepts of material object, of property and relation, of causation, and of space and time. Some of these, such as material object or property, are quasi‐technical—regimentations of concepts available in ordinary speech. They are general in being categorial, or at least in subsuming numerous more specific concepts under them, in the sense in which the concepts of a material object subsumes numerous material object concepts such as the concept of a chair, of a lump of sugar, or of a mountain.¹¹ They are irreducible, not in being simple and unanalysable, but rather in not being eliminable without circularity in favour of other concepts that wholly define them. They are basic in as much as they are members of a set of general, pervasive, and irreducible concepts or concept‐types that together form a structure that constitutes the framework of our ordinary thought.¹² And they are non‐contingent in the sense that they are necessary constituents of any conception of experience that we can make intelligible to ourselves—essential to our conception of the experience of self‐conscious beings.

Descriptive metaphysics, Strawson declared, also differs from conceptual analysis in general in its method. Although examination of the use of words is the only sure way in philosophy (p. 9), the discriminations (p.353) we can make and the connections we can thus establish are neither sufficiently general nor far reaching enough to meet ‘the full metaphysical demand for understanding’. For the uses of expressions of natural language ‘are apt to assume, and not to expose, those general elements of structure which the metaphysician wants revealed. The structure he seeks does not readily display itself on the surface of language, but lies submerged. He must abandon his only sure guide when that guide cannot take him as far as he wishes to go’ (p. 10).

Strawson took pains to distinguish the kind of investigation he was advocating from Collingwood's conception of metaphysics as a historical science. Descriptive metaphysics is not the description of the changing absolute presuppositions of thought. Nor is it an instrument of conceptual change. For ‘there is a massive central core of human thinking which has no history—or none recorded in histories of thought; there are categories and concepts which, in their most fundamental character, change not at all’ (p. 10). These form the indispensable core of the conceptual equipment of all human thought.

Strawson took as his point of departure the correlative pair of linguistic functions of reference and predication. Picking out some individual item and saying or thinking something about it are fundamental functions of thought and speech. The categories of expressions that are employed in order to fulfil these basic functions are singular referring expressions—names, singular definite descriptions, or indexicals—and predicates. The question he addressed is what are the most general conditions for identifying reference to and reidentification of particulars, and, in the light of the requirements of identifying reference and reidentification, what are and must be the most fundamental, primitive, or basic objects of reference and subjects of predication. He argued that the basic particulars in any conceptual scheme capable of describing experience and its objects are necessarily of two general types of individuals of a relatively substantial and enduring sort located in a unified spatio‐temporal framework—material objects and persons (pp. 11, 246). Material objects are the subjects of a class of predicates that he denominated ‘M‐predicates’, and persons are the subjects of both M‐predicates and P‐predicates—that is, predicates that presuppose their subject's possession of states of consciousness. Given that these are the basic particulars of any such conceptual scheme, he proceeded to investigate various forms of non‐basic or dependent particulars, identifying reference to which depends upon the more fundamental reference to and identification of independent, basic particulars. Among these are such items as individual experiences, which he claimed to be identifiability‐dependent upon the subject whose experiences they are, and individual (p.354) events that are identifiability‐dependent upon the particular undergoing change. Furthermore, although the basic particulars are and must be material objects and persons, universals, abstract, and intensional objects can likewise be objects of reference and subjects of predication. With a characteristically relaxed attitude to ontological commitments, Strawson pleaded that we should abandon whatever natural but ill‐founded nominalist qualms we might suffer from and recognize the existence of such objects of thought.

Descriptive metaphysics is to be contrasted with revisionary metaphysics. Revisionary metaphysics is concerned not with describing the actual structure of our thought about the world, but with producing a better structure. The productions of revisionary metaphysics, at its best, are ‘both intrinsically admirable and of enduring philosophical utility’ (p. 9). But they are so only in so far as revisionary metaphysics is ‘at the service of descriptive metaphysics’ (ibid.). Strawson conceded in Individuals that perhaps no philosopher has ever been, both in intention and in effect, wholly a descriptive or wholly a revisionary metaphysician. But one can allocate Descartes, Leibniz, and Berkeley to the class of revisionists, and Aristotle and Kant to the descriptivists. In Individuals Strawson did not elaborate further on what he meant by ‘revisionary metaphysics’, and did not explain in what sense revisionary metaphysics is ‘at the service of descriptive metaphysics’ or what enduring philosophical utility it has. But he elaborated somewhat further in his paper ‘Analysis, Science and Metaphysics’.¹³ Some metaphysics, he there wrote cautiously, ‘is best, or most charitably’ seen as consisting in imagining how things might be viewed through the medium of a different conceptual scheme. And he conceded that, even when such metaphysics can be so interpreted, it is not presented thus, but rather as a picture of how things really are, as opposed to how they delusively seem to us to be. This presentation, Strawson admitted, with its contrast between esoteric reality and our daily delusion, involves and is a consequence of the unconscious distortion of ordinary concepts—that is, of the ordinary use of linguistic expressions. So revisionary metaphysics, ‘though it can sometimes be charitably interpreted [thus], in fact always involves paradox and perplexities . . . and sometimes involves no rudimentary vision, but merely rudimentary mistakes’.¹⁴

(p.355) 3. Is Descriptive Metaphysics a Form of Metaphysics?

It is clear enough, and should have been evident at the time, that the positivists’ characterization of what are to be deemed ‘metaphysical concepts’ was unsatisfactory. If it is deemed a requirement of any genuine concept that it have determinate criteria of application, and if metaphysical concepts are characterized exclusively by reference to the fact that they lack such criteria, then it is hardly an astounding discovery that they are pseudo‐concepts. Strawson's proposal that we should view as metaphysical only such concepts as are, in the manner described, very general, basic, irreducible, and non‐contingent gives us at least a rough and ready criterion. Moreover, it is a criterion the application of which produces a set of concepts or types of concept that partially coincides with the preoccupations of metaphysics as traditionally conceived. There can be no licit positivist objection to the task of providing connective analyses of such concepts and categorial concepts as space, time, material object, property, relation, cause, person, and so forth. Nor indeed can there be any objection to delineating the complex forms of dependency and interdependency among these concepts or types of concept, which conjunctively form the framework of our thought about the world and ourselves in it. It is not wholly clear whether very general concepts of ethics and aesthetics, such as moral goodness and beauty, or very general concepts of the philosophy of action, such as act and omission, ability, intention, reason for action and purpose, should be counted as metaphysical or not. Perhaps they should count as concepts of the metaphysics of morals, of aesthetics and of agency. Or perhaps they should be excluded, on the ground that only concepts presupposed by any language in which a distinction can be drawn between experience and its objects are to be deemed metaphysical.¹⁵

(p.356) Either way, descriptive metaphysics—in so far as it is concerned merely with the analysis of such concepts—is just more conceptual analysis, albeit at a very high level of generality. What might make it distinctive in a further and deeper sense would be the arguments that purport to show that these concepts are, in some deep sense, necessary concepts. If, in addition, such arguments were to result in propositions that purport to describe necessary features of reality, descriptive metaphysics would make contact with one salient feature of traditional metaphysics. For it was the hallmark of traditional metaphysics that, like physics, it investigated the nature of reality, but, unlike physics, it did not investigate the contingent features of the world but its allegedly necessary features. Physics tells us how things are in reality, what contingent laws of nature describe the behaviour of things in space and time. Metaphysics aspires to tell us how things must (unconditionally) be. Metaphysical truths, if there are any, will be non‐logical, necessary truths. So the principle that every event has a cause was held to be a truth of metaphysics, for it is allegedly inconceivable that there be uncaused events in nature. The principle that every attribute inheres in some substance was a truth of metaphysics in as much as it was held to be unthinkable that there be attributes that are not essential or accidental properties of some thing. And, descending to a much lower level of generality, it was held to be a metaphysical truth that nothing is simultaneously red and green all over, for it is unimaginable that there be any such thing. Propositions of metaphysics, so conceived, are quite different from the propositions of physics. They must be truths that can be discovered by the use of pure reason alone, independently of experience. But they are neither logical or analytic truths, nor empirical truths. Kant famously characterized some of them as a subclass of synthetic a priori propositions.

Are the concepts to be analysed by the descriptive metaphysician necessary concepts in any deep sense? Perhaps not in the sense in which seventeenth century metaphysicians thought that we are born equipped with either actually or virtually innate ideas, or in the sense in which Kant thought of the categories as a priori concepts, unifying a manifold of intuition in judgement to yield experience. The general, basic, and irreducible concepts that concern the descriptive metaphysician are necessary in the sense that they are an essential structural element in any conception of experience that we can render intelligible to ourselves. Since the salient concepts that Strawson discusses are formal or categorial concepts, such as substance, property, cause, experience, and so on, it seems evident that it is not necessary that any language in which we can distinguish between experience and its objects must contain these formal concepts—rather, it is necessary that any such language contain concepts that belong to these types—that is, substance concepts, concepts of properties of substances, causal concepts, and so on. Moreover, the necessity for such concepts is relative—relative to the possibility, in such a language, of distinguishing in general between experience and its objects. It seems evident that we can entertain the (p.357) idea of a language without a word equivalent to ‘substance’. But, more importantly, we can surely also entertain the idea of a language without substance names. For there could be a language consisting only of orders ‘Come!’, ‘Go!’, ‘Eat!’, ‘Drink!’, and so on, together with words expressing willingness or unwillingness, approval and disapproval. It would be an impoverished affair, to be sure, but I doubt whether we would have qualms in characterizing it as a rudimentary language. And it is at least arguable that we can imagine a language without any causal expressions, which consists only of substance names, colour and number words—as Wittgenstein suggested in the evolved second language‐game in the Investigations. The necessity of the select concepts or concept types that are of concern to the descriptive metaphysician is relative to the concepts of experience and its objects, or, more precisely, relative to any language and thought in which a distinction can be drawn between experience and what it is experience of. What that means is that such concepts are partly constitutive of what we call ‘experience’ and ‘objects of experience’. We would not denominate anything ‘an experience’ unless it could be ascribed to a sentient being, and would not characterize anything as ‘conceptualized experience’ (that is, ‘experience’ in the weighty sense that is Strawson's concern) unless it were ascribable to a person. And we cannot intelligibly characterize anything as a person unless it is a space‐occupying sentient being tracing an autobiographical route through a unified spatio‐temporal world consisting of relatively enduring material objects. In short, the weighty concept of experience is located at the centre of a vast web of concepts of space and time, of substance and causation, of cognition and volition. The investigation of these conceptual involvements and commitments is of great philosophical interest, in part because so much philosophical reflection failed lamentably to grasp precisely how weighty and how extensive are these involvements and commitments, and how they are interwoven. But it is not obvious that the elucidation of this web of connections, dependencies, and interdependencies is anything more than connective analysis of a select range of general concepts partly constitutive of what we call ‘a description of an objective world’. It may indeed be called ‘metaphysics’—but only in this attenuated sense.

Are the propositions of descriptive metaphysics as presented by Strawson a kind of metaphysical proposition as traditionally understood? One difference is evident. Where traditional metaphysicians conceived of themselves as limning the ultimate structure of the world, the descriptive metaphysician will conceive of himself as sketching the basic structure of our conceptual scheme—of the language we use to (p.358) describe the world and our experience of it. Or, more ambitiously, of delineating the structure of any conceptual scheme that can be employed to describe a world and a subject's experience of it. Hence it does not aim to describe the necessary, super‐physical, structure of reality—about which it may well remain altogether sceptical. Rather, as Strawson writes, it aspires ‘to establish the connections between the major structural features or elements of our conceptual scheme—to exhibit it, not as a rigorous deductive system, but as a coherent whole whose parts are mutually supportive and mutually dependent, interlocking in an intelligible way’.¹⁶ So conceived, descriptive metaphysics breaks with the metaphysical tradition, which purported to give us insights into the necessary structure of reality.

So far, so—soberly—good. The conception of a form of necessity that is not logical, but no less adamantine than logical necessity, that is an objective, language‐independent form of necessity that can nevertheless be apprehended a priori by reason alone is, surely rightly, dismissed as a fiction.¹⁷ Nevertheless, descriptive metaphysics results in an array of propositions that are held to be necessary truths constitutive of our conceptual scheme. So, for example, it is argued that it is a conceptual truth that places are defined by the relations of material bodies, that material bodies provide the framework for spatial location in general, that they are basic from the point of view of referential identification and reidentification of all other particulars of different categories, that persons have bodies, that the experiences of a person are identifiability‐dependent on the identity of the person whose experiences they are, that a condition for the intelligibility of self‐ascription of experience is the legitimacy of other‐ascription of experience on the basis of logically adequate behavioural criteria, and so on.

Again, the interest of such propositions is not in dispute. They endeavour to articulate fundamental structural features of our conceptual scheme and arguably of any conceptual scheme in which the distinction between experience and its objects can be drawn. But the moot question is: what is the status of such propositions? Obviously, they are not and are not intended to be empirical propositions. They purport to be necessary truths. They are not obviously analytic propositions, assuming that we have a tolerably clear grasp of that problematic category. Are they then synthetic a priori? Strawson did not address the (p.359) question in Individuals. But in his later book The Bounds of Sense, he repudiated the idea of the synthetic a priori. He argued that ‘Kant really has no clear and general conception of the synthetic a priori at all’.¹⁸ For once the apparatus of transcendental idealism is abandoned, as Strawson argued it should be, the category of synthetic a priori propositions is, he claimed, merely a residuum of propositions that are neither analytic nor empirical. The kinds of proposition that Kant deemed to be metaphysical are such propositions as: experience essentially exhibits temporal succession; there must be such unity among the members of a diachronic series of experiences as is required for the possibility of self‐consciousness or self‐ascription of experience; experience must include awareness of objects that are distinguishable from the experiences of them; there must be one unified spatio‐temporal framework embracing all experience and its objects; and the Principles of the Analogies of Experience—that is, of the permanence of substance, of causality, and of causal reciprocity. Strawson saw no explanatory value in characterizing such propositions as ‘synthetic a priori’. What then are they? And what is the nature of the apparent necessity that is associated with them?

At this point, it seems to me, Strawson's account falters. The Kantian propositions do, he concedes, ‘have a distinctive character or status’. Nevertheless, he sees ‘no reason why any high doctrine at all should be necessary here’. The conceptual scheme we employ reflects our nature, our needs, and our situation. It is subject to change as our needs and situation change. But such changes are conceivable only as variations within a fundamental general framework of ideas. With regard to some of the propositions of descriptive metaphysics that he elicits in Individuals, he observes that ‘it does not seem to be a contingent matter about empirical reality that it forms a single spatio‐temporal system’. For, if someone told of a kind of thing and of events that occurred to it, but insisted that that object was not located at any distance from here, and that those events stood in no temporal relation to now, since they did not belong to our spatio‐temporal system, we should take him to be saying that the events had not really occurred and that the thing in question did not really exist. In so saying, we show how we operate with the concept of reality. ‘We are dealing here’, he concludes, ‘with something that conditions our whole way of talking and thinking, and it is for this reason that we feel it to be non‐contingent’ (p. 29).

This does not seem to me to be altogether satisfactory. The fact that something conditions our whole way of talking does not obviously (p.360) suffice to explain why we should think of it as non‐contingent. Our size conditions at least much of our way of talking and thinking too, but there is nothing non‐contingent about it. The fact that we are sexual beings with a determinate gender conditions our social existence and relations and our thought about it, but it is easy enough to imagine the very different lives of asexual beings otherwise akin to us. Although we may feel certain features of the world to be non‐contingent, such as the pervasiveness in it of causal regularities, the relative permanence in it of three‐dimensional material objects, it does not follow from our so feeling that there is anything non‐contingent about such features. On the other hand, non‐contingency is surely rightly associated with the thoughts that any event is either earlier, later, or simultaneous with any other event, that every material object is spatio‐temporally related to every other material object, that every experience is the experience of some sentient being, and that subjects of experience have bodies. Even if no ‘high doctrine’ is necessary here, some modestly low doctrine is surely needed to satisfy the requirements not of metaphysical understanding but of philosophical understanding. For, if the necessity of such propositions is not merely a misguided projection of our feelings of necessity (as Hume thought that our ascription of necessary connection to any causal relation was), then we crave some explanation of the nature of such non‐logical, yet non‐empirical necessity that is not evidently analytic.

‘In order to set limits to coherent thinking’, Strawson concluded his discussion of Kant's synthetic a priori propositions, ‘it is not necessary, as Kant, in spite of his disclaimers, attempted to do, to think both sides of those limits. It is enough to think up to them.’¹⁹ This nicely echoes Wittgenstein's similar observations in the Preface to the Tractatus. But it is noteworthy that the propositions of the Tractatus were themselves ultimately condemned as nonsense, since they did not meet the criteria of significance of that book. That was, as Wittgenstein later realized, an unhappy solution to the problem of the status of propositions delineating the bounds of sense. It is evident that Strawson would not wish to venture down that cul de sac. But then some explanation of their status is required. The thought that any event must be earlier, later than, or simultaneous with any other event does not look like a statement about our conceptual scheme. Nor does the proposition that space and time form a unity. They look like insights into the essential nature of reality. So do such propositions vindicate the project of traditional meta‐physics? Or are appearances misleading? Surely more needs to be said.

(p.361) The way out of the quandary was signposted by Wittgenstein in his later reflections on the same kinds of proposition as those he had condemned as nonsensical attempts to describe the bounds of sense in the Tractatus. We should treat such propositions, which overtly or covertly describe the conceptual connections between the major structural features of our conceptual scheme, as expressions of norms of representation. The proposition that every event is spatio‐temporally related to every other event is in effect a rule for the use of the word ‘event’. And we manifest our adherence to this rule precisely in our agreeing with Strawson's verdict that a description of an object and of the changes that it underwent, which were not spatio‐temporally related to whatever is here and now, would and should be taken to be a description of a non‐existent object and a non‐occurrent event. The proposition that every event has a cause, if taken to be a metaphysical proposition rather than an empirical hypothesis, and hence licensing us to claim of any observed event that it must have a cause even if none can be discerned, is in effect a norm of representation that entitles us to infer from any event‐identification that there is a cause of that event, which may or may not be known (and may or may not be discovered). The proposition that every experience is necessarily someone's experience, that every thought is necessarily someone's thought, is an expression of one's commitment to a norm of representation—namely, that, pace Hume and James,²⁰ no sense is attached to such forms of words as ‘There is a pain in the room, although no one has it’ or ‘There is a thought in the quad, although no one is thinking it’. The necessity we associate with such propositions is fully explained by their normative status. For they are what Wittgenstein, a little provocatively, called ‘grammatical propositions’—expressions of rules of representation in the misleading guise of statements about reality. Any putative reference to the occurrence of a pain that denied its ascribability to some sentient being is not to be counted as a genuine reference to the occurrence of a pain. For, in so far as we accept these propositions, we rule out, a priori, any use for the relevant terms—for example, ‘uncaused event’, ‘subjectless pain’—as we rule out any use for the term ‘checkmate’ in draughts—there is no such thing in this game.

It is important to realize that, in ruling out the intelligibility of (p.362) uncaused events, of events spatio‐temporally unrelated to other events, of actual experiences unascribable to subjects of experience, of colour coinstantiation, we are not delimiting nature. We are not laying down possibilities that nature cannot realize. We are laying down the limits of description, not describing possibilities that are impossible. We are characterizing the bounds of sense. For we have given no use to the forms of words ‘simultaneously red and green all over’, ‘an occurrence of a pain that no one felt’, ‘an event that was neither earlier than, later than, nor simultaneous with such and such present event’. We are not expressing insights of pure reason into what nature can or cannot do, but reminding ourselves that we attach no significance to such expressions. To be sure, we could attach significance to them—after all, nothing is stopping us. All we have to do is lay down additional rules for the use of the relevant terms. But, if we do, we are thereby changing their meanings and speaking of something different. It was a confusion of traditional metaphysics to project insights into the structure of our conceptual scheme onto the objects described by our employment of it. In so doing, it confused rules determining the correct use of words and the licit inferences that their application licenses, which conjunctively define the essences of things, with objective, language‐independent necessities and necessary connections in the world.

The putative necessities in reality are merely the shadows cast by rules for the use of words in our language that are partly constitutive of their meanings. The question of whether the propositions in question are analytic or synthetic (with all the unclarities associated with these categories) can be sidestepped. For the status of such propositions is clarified in recognizing that they are rules of representation, not descriptions of reality. They can be said to be true only in the sense in which it is true that the chess king moves one square at a time. The possibility of knowledge of such propositions is relatively unproblematic, since it is knowledge of, or recognition of, the rules we follow in using the relevant words of our language correctly. Why then is it not trivially easy to attain? Largely because of the generality at which we operate in this domain and because of the ramifications and interrelations of the rules, which are anything but easy to survey. Hence, to the extent to which we wish to speak of attaining knowledge here (as opposed to attaining understanding), it takes the form of realization rather than discovery. Metaphysics thus construed yields no insight into reality, but only into our forms of description of reality. So it is just more grammar, in Wittgenstein's extended sense of the term.

The generality at which we operate in the domain of Strawsonian descriptive metaphysics is manifest not only in the generality of the (p.363) concepts and concept types that are the focal point of the investigation but also in the fact that they constitute, as he surely rightly points out, structural elements of our conceptual scheme. Precisely because such concepts as space and time, substance and property, cause and effect are, roughly speaking, categorial, their rule‐governed connectedness ramifies throughout our conceptual scheme. The formal or categorial concepts of substance and attribute, or of cause and effect, subsume thousands of material concepts that are in constant employment in our daily discourse. And their forms of connectedness determine our thinking and inferring in all our description, reflection, and action. Small wonder that at least some of these categories seem non‐contingent, for without them we would not engage in the thought and intentional action of the kind that characterizes our form of life. We could no more decide to abandon these categories of concepts than we could decide to cease to be human beings.

What then of ontology, which Strawson, in a moment of exuberance, dignifies by the name of ‘the general theory of being’?²¹ It seems to me that ontology is no more than an investigation of what is meant by saying of items belonging to a certain general category that they exist, obtain, occur, or go on. We say that material things come into existence and pass away, we refer, unashamedly, to general characteristics of things or to the occurrence of events and the obtaining of states of affairs; some philosophers insist—perfectly reasonably under some interpretation—that there are objective values, and we assert without qualms that there are such‐and‐such many primes between m and n. Ontology, if it is anything, is surely the elucidation of what is meant, from case to case, by such existence claims. Its task is not to draw up inventories of the contents of the universe, nor can it be illuminatingly described as the study of being qua being. So it is not so much a general theory of being, but rather a matter of connective analysis, concerned in particular with existence claims and with the elucidation of conceptual dependencies involved in such claims.

It might be thought, although I do not think that Strawson ever suggested as much, that Strawson's descriptive metaphysics yields transcendental arguments that prove the existence of the external world or of other minds. But that is, I think, mistaken. It would be absurd to argue from conceptual connections in thought to existential truths about the world, or, in Wittgensteinian idiom, from grammatical propositions to empirical ones. Strawson argued that a condition for the intelligibility of criterionless self‐ascription of experience is the (p.364) adequacy of the behavioural grounds for other‐ascription of experience. This conceptual connection does not prove that there are other experience‐enjoying beings—what it proves is the incoherence of scepticism about other minds that admits self‐ascription of experience and simultaneously denies the adequacy of the criteria for other‐ascription. And, it can be argued, although I shall not attempt to argue it here, both that the demand for a proof of the existence of the ‘external world’ or of other minds is itself incoherent and that we have a vast hoard of genuine knowledge about objects in the world around us and about our fellow human beings and the experiences they enjoy, without per impossibile possessing any such proof.

It should be noted that, although some of Strawson's conclusions echo Kantian synthetic a priori propositions—for example, concerning causal regularities and the existence of substance—these are not transcendental deductions of the necessity that every event have a cause or of the existence of an external world. They are rather observations upon the background conditions for the exercise of such concepts that enable a subject of experience to ascribe experience to himself. For, Strawson argues persuasively, the concept of subjective experience gets a grip only to the extent that the concepts of independently existing objects of experience get a grip, and both obtain a purchase only to the extent that such objects of experience are generally connected by causal regularities. But this is no proof of the existence of an external world or of the principle of sufficient reason. Scepticism is to be refuted by showing that it makes no sense—not by producing a proof of the existence of objects. We do indeed know of the existence of multitudinous objects around us, as we know of innumerable causal connections between substances and the events they make happen—but not on the grounds that the existence of substances and of widespread causal regularity is a condition for the employment of concepts of experience and its objects, nor on the grounds that we do enjoy subjective experiences.

Does descriptive metaphysics differ in its method from connective analysis in general? The examination of the use of words, Strawson averred, is the only sure way in philosophy, but the structures that the descriptive metaphysician wishes to reveal are not displayed on the surface of language and the connections he wishes to establish are too far‐reaching to be discernible by scrutiny of the use of words. So he must abandon his only sure guide when that guide cannot take him to the peaks of abstraction that he aims to scale. What method should he then use? Strawson offers us disappointingly little: ‘I know of no procedure or recipe for getting at the answers except to think about those (p.365) ideas and questions as hard as you can.’²² It is true that there is a sense in which ordinary usage offers few hints or clues to philosophical insight when it comes to such concepts as space and time, substance and accident, subject and object of experience. The terms do not offer that variegated field of subtle distinctions that are to be found, as Austin noted, in such domains of discourse as excuses. Nevertheless, it seems to me that the connections Strawson seeks—and finds—are not in any sense submerged beneath the surface of ordinary usage. The use of the term of art ‘substance concept’ is not likely to offer the philosopher much help. The use of the term ‘cause’ may be positively misleading, since ‘cause’ and ‘reason’ are, over a range of contexts, interchangeable, while the insightful philosopher interested in causation will wish to differentiate causes from reasons—and will indeed find ample reason to do so from a more careful examination of the use of these two terms. The clues the descriptive metaphysician seeks and the only tribunal before which his claims can be adjudicated is the general pattern or patterns of use of the multitudinous substance concepts and causal concepts that abound in natural language. Those patterns are in full view, even if it takes uncommon skill to discern them.

Strawson, it therefore seems to me, saved the letter of traditional metaphysics, but abandoned its spirit. Descriptive metaphysics is distinctive, and unlike other philosophical endeavours, in so far as it strives to disclose the most general forms of connectedness that permeate our conceptual scheme and to reveal the conceptual involvements of the most general kinds of speech functions that characterize our use of language—indeed, not only of our language and our conceptual scheme, but of any language and any conceptual scheme in which certain kinds of distinctions are drawn and certain fundamental kinds of speech acts performed. But it is also like other philosophical endeavours within the field of connective analysis and unlike the aspirations of traditional metaphysics. It yields no knowledge of reality, let alone insight into the necessary structure of reality—but only insight into the forms and structures of our thought about reality. It might indeed be said to be the legitimate heir to what used to be conceived of as metaphysics, but a dethroned heir, deprived of the ancestral crown and orb. It is metaphysics without its nimbus.

(p.366) 4. Is there Any Such Thing as Revisionary Metaphysics?

Price pleaded for the preservation of metaphysics understood as ‘alternative modes of conceptual arrangement’, and recommended that philosophers should continue to be engaged upon devising different unified conceptual schemes. And he suggested that the works of the great metaphysicians of the past should be viewed as directed at such a goal. This seems akin to Strawson's conception of revisionary metaphysics in Individuals. The first question to address is: was this the project of the great system‐building philosophers of the past? Were Descartes, Leibniz, and Berkeley—to mention only those whom Strawson characterized as revisionary metaphysicians—trying to construct a different language, with a fundamentally different grammar, from the natural languages of mankind? I can see no trace of such an intent in their works. Descartes, for example, was not recommending that we adopt a new form of language. He was trying to describe the world and the fundamental kinds of substance that exist in it, to characterize their essential natures and their modes of interaction. Leibniz's monadology did not advocate a change in notation, but elaborated a putative insight into the constitution of reality. Berkeley was not recommending that we adopt a language without names of material substances, which would merely be better for certain needs we have than our existing language. He too was trying to specify the nature of the world and of what exists in it. And he thought that any talk of material substance, understood as he took Locke to understand the term, was not an inoffensive part of our ordinary conceptual scheme, but incoherent nonsense lying at the heart of Locke's mistaken metaphysics of the scientific revolution of the seventeenth century. Indeed, the putative proof that it was nonsense, and not just ‘an alternative conceptual scheme’, was a crucial step in Berkeley's endeavour to confound sceptics and atheists.

If that was not what traditional metaphysicians were trying to do, was that in effect what they actually did? Did Descartes or Berkeley or Leibniz construct a language, different from our natural languages but equally fitted for describing the world and our experience of it? Surely not. Cartesian, Leibnizian, or Berkeleian metaphysics does not provide a novel grammar, constitutive of a novel conceptual scheme. Rather, these metaphysical systems, thus interpreted, would result in an incoherent grammar that, at crucial points, specifies kinds of particular without any associated criteria of identity, introduces psychological predicates that would have a sense only if they were, per impossibile, (p.367) definable by private ostensive definition, and retains many of our concepts while simultaneously severing them from the web of connections to other concepts that alone makes them intelligible. Of course, it does not follow that the attempt to reconstruct these grand systems in this fashion may not be instructive.

If one cannot reasonably interpret the metaphysical systems of the past as recommendations to adopt a new conceptual scheme, can it be argued, as Price suggested and Strawson, in Individuals, intimated, that devising alternative conceptual schemes is a task that metaphysics might reasonably undertake? Is there any sense in which such an endeavour might, as Strawson suggested, be at the service of descriptive metaphysics? It is noteworthy that the only philosopher who comes to mind in association with the idea of a philosophical programme of constructing alternative languages that could be deemed to constitute alternative conceptual schemes was the most fervent anti‐metaphysical crusader of the twentieth‐century, namely, Rudolf Carnap. He would rightly have been surprised to learn that doing so was a form of metaphysics.²³ It is difficult to see the point of such endeavours within philosophy, unless they are aimed at constructing fragments of a conceptual scheme, which can then be held up as a useful object of comparison highlighting features, uses, and forms of contextual dependencies, of the corresponding fragment of our own. It is only in this sense that something akin to revisionary metaphysics could be thought to be part of the task of philosophy and indeed at the service of descriptive metaphysics—that is, general connective analysis.

But part of the usefulness of revisionary metaphysics, thus conceived, runs counter to the Strawsonian vision, at least to a degree. For part of the point of devising an alternative grammar for, say, colour description (for example, adoption of colour adverbs rather than adjectives), or ascription of experience (elimination of the first‐person pronoun, as in ‘There is pain’ instead of ‘I have a pain’), is precisely to show that our concepts and their articulations in these domains are not the only possible ones. For we are prone to think that our modes of conceptualization are uniquely correct or true to the facts, or that our concepts are necessary ones in as much as they uniquely match the logico‐metaphysical forms of reality. But over a wide range of concepts, we can envisage different grammars to fulfil analogous tasks. And we can readily imagine such (p.368) changes in us or in the world as would render such‐and‐such concepts useless, and such‐and‐such other novel concepts more useful for our purposes in the novel contexts envisaged. The necessity we imagine associated with our conceptual scheme is a necessity internal to our conceptual apparatus—not a form of objective, language‐independent necessity. In this manner, the invention of fragments of a different conceptual scheme, which fulfil roles akin to fragments of our own, can be useful in disabusing us of some of the illusions of traditional metaphysics that incline us to think that our conceptual scheme pays homage to the objective metaphysical nature of the world. Nevertheless, it does not follow that, for us, with the kinds of conceptually moulded interests and purposes we have, there are serious alternatives to those major structural features of our conceptual scheme that lie at the heart of Strawson's investigations. For they constitute, as he says, ‘the massive central core of human thinking’ (p. 10). It is not merely that they are ‘the indispensable core of the conceptual equipment’ (ibid.) we deploy. Rather they are partly constitutive of our nature as self‐conscious human beings, involving concepts and categories that we could not abandon without ceasing to be human.

Philosophy is in general a matter not of concept formation, but of concept description. Hence there is little role in it for the invention of (fragments of) conceptual schemes. Mathematics is concept formation, and the mathematician does indeed invent new forms of description that may be put to use by physicists in describing spatial relations and in the description and transformation of propositions about magnitudes, quantities, velocities, and so on, and their relations—as Riemannian geometry proved fruitful for relativity theory, and as the calculus proved indispensable for Newtonian physics. However, the task of philosophy is not to devise alternative conceptual schemes, but to describe and elucidate our own. Part of that task is to elucidate the most general forms of connectedness that permeate our conceptual scheme and that are partly constitutive of our very general conceptions of substance, causation, person and personal experience, space and time, reference and predication, and so on—which Strawson has pursued with his characteristic elegance, economy, and profundity.

With regard to revisionary metaphysics, Strawson's second thoughts were, I think, more accurate than the view so briefly sketched in Individuals. Traditional metaphysics, though it can sometimes be charitably interpreted as involving a recommendation to adopt a new conceptual scheme, ‘in fact always involves paradox and perplexities . . . and sometimes involves no rudimentary vision, but merely (p.369) rudimentary mistakes’.²⁴ So there is not really any such subject as revisionary metaphysics—although scientists are free to devise fragments of alternative conceptual schemes for their purposes, if the new scheme is more fruitful in the generation of explanatory and predictive theories than the existing one. Nevertheless, there is a connection between the idea of devising a fragment of a novel conceptual scheme and the products of the metaphysical tradition. For traditional metaphysics—for example, representational realism, idealism, or solipsism—present their doctrines as if they were correct descriptions of reality, as if it were truer to the facts to say ‘There is pain’ rather than ‘I am in pain’, since the self is not a constituent of the experience of pain, or to say ‘Grass looks green (presents an idea or representation of green) to normal observers in normal conditions’ rather than ‘Grass is green’, since objects are not in themselves coloured, but only have a power to produce a representation of colour in human observers. But, as Wittgenstein pointed out, this is to suppose that a form of representation could say something false even when the proposition expressed says something true. The only way in which ‘I am in pain’ can be false is by ‘I am not in pain’ being true, and the only way in which ‘Grass is green’ can be false is by grass not being green but some other colour. So ‘the one party attack the normal form of expression as if they were attacking a statement; the others defend it as if they were stating facts recognized by every reasonable human being.’²⁵ In this sense, one might interpret some of the writings of metaphysicians as recommendations to adopt a different conceptual scheme. But then it is noteworthy that these metaphysicians fail to carry through the idea, and conflate elements of the new notation with elements of the existing one, and conclude that other people do not really have pain, or that the objects around us are not really coloured.²⁶

The conception of metaphysics characteristic of modern (post‐Cartesian) philosophy, the idea of attaining a priori insights into the objective language‐independent essences of things, was deeply rooted in the culture of seventeenth‐ and eighteenth‐century Europe. The uprooting of this fiction required the labour of many thinkers, and took many decades. It is Strawson's signal achievement to have salvaged from the wreck of the traditional enterprise a form of general connective analysis, which he has called ‘descriptive metaphysics’. What it aspires to is not knowledge of the essential nature of the world, but understanding (p.370) of the general structure of our thought about the world. And that can be achieved. Since Strawson wrote Individuals, however, a new form of the old disease has broken out, and the mythology of metaphysics has been revived. Its roots lie deep in our contemporary culture—in the science and scientism of the late twentieth century. To eradicate it will, I fear, be as difficult as it was to eradicate its more august ancestor.