Substance

3. A new role for substance: realism versus conceptualism

Humean empiricism revived in the twentieth century. For A.J. Ayer, for example, the philosophical notion of substance reflects only a ’primitive superstition’ about names, since nothing is independently there to be named except sense-data, the equivalent of Hume’s impressions (see Ayer 1936). But much present-day philosophy, although influenced by both Hume and Kant, in general denies both that there is anything ineluctable about the way we think about the world, and that anything is simply there to be named. W.V. Quine (1953) expresses this denial in the terms of Russellian logic holding that to be is to be the value of a bound variable. For Quine, ontology is a matter of deciding which types of entity to postulate and quantify over – material objects, events, properties, ’stages’, ’fusions’, ’sense-data’ or what you will – a decision that is not determined for us by reality. The difference between types of entity is a difference between criteria of identity – between what we count as the same individual again (1960). Here a rational decision is simply a pragmatic one (see Quine, W.V. §5; Ontological Commitment). Material objects may fill the fundamental place in evolved natural language, but that is a matter of practical convenience or utility. Indeed, the folk-physics embodied in natural language has been supplanted, unless for everyday purposes, by a physics with no room for the ’bodies’ canonically assigned to the category of substance.

Is the traditional notion of substance therefore dispensable, the manifestation of an illusion engendered by contingent features of natural, everyday language? To suppose so would arguably be to set aside a storehouse of philosophical insight capable of suggesting to us just what can most cogently be brought against a conceptualism like Quine’s. Conceptualism is a ’top-down’ philosophy, taking the structure of our thought and language to be the structure of an imposed theory of interpretation of sensory input such that (contrary to Humean empiricism) even input itself must be conceived of in terms of the theory. Concepts are holistically inter-related, slicing out entities as theoretical postulates of the whole scheme. There are no independently given individuals. Not all conceptualists approve the analogy between natural language and scientific theory. Wittgenstein (1953), for example, invites us to see the structure of language as a product of its employment in, above all, social interaction. But all conceptualists have agreed, since the first modern idealists, that we cannot compare our scheme of notions or beliefs directly with reality. An equally old, still popular conclusion is that we can therefore only aim for coherence, and the total scheme that best anticipates experience.

The stiffest opposition to this aspect of pragmatism has come largely from two directions. Drawing on Kant, Peter Strawson (1958) has constructed a transcendental argument aimed at demonstrating that the distinction between experience of independent objects and merely subjective experience – in effect the distinction between oneself and other things which even the sceptic takes for granted – presupposes that both the individual objects and the individual subject of experience exist, endure, are mobile, and interact in space (see Strawson, P.F. §8; Transcendental Arguments). Whether or not this argument rebuts scepticism (would it not be enough for these distinctions that the subject should seem to itself to be so related to bodies in space?), it does offer an explanation of the primacy of the category of substance in natural language, and of the unreality of the suggestion that, but for utility, entities of a different category – events, thing-stages, sense-data or whatever, might have played the basic role. Yet although Strawson disavows an idealism like Kant’s, and presents a powerful alternative to Quine’s relativism, he does not clearly depart from a ’top-down’, conceptualist picture. Indeed he has talked of his thesis as if it concerned the necessary structure of any conceptual scheme that could be the vehicle of objective experience – the general scheme or order which sensory input must fit, as it were, for there to be experience of objects. Individual substances are necessarily basic, but are picked out only by means of sortal concepts applied to input.

A different line of argument, pioneered by Saul Kripke (1980) and Hilary Putnam (1975), might be taken to refute the idea that our conceptual scheme structures the world for us, rather than being structured by the world. This proposes that, like proper names, names of natural kinds owe their meaning not to some concept in our heads but to their naming what they name. Consequently, ’gold’ on the lips of a present-day physical chemist means the same as an ancient Chinese character, whatever differences exist between the users’ conceptions of, or theories about, the natural kind of stuff that both name. That their words mean the same could not be determined by an inspection of their ’ideas’ or definitions, but would depend on the identity, that is the actual (and, for at least one of them, unknown) common nature, of what each names (see Reference §3). Yet even if this point is taken to undermine the analogy between language and theory, and to demonstrate the possibility of a pre-theoretical naming of things, it does not speak directly to the question of the distinction between what is naturally, and what is only conceptually distinct. For it could be made almost as readily about the general names of natural events, processes or properties as about the names of substances. The term ’freezing’ has the same pre-theoretical grip on a natural kind of process as ’horse’ has on a natural kind of substance, yet the freezing of some water in a glass is not a naturally discrete whole in the way that a horse or, for that matter, a glass is. Although no process is properly freezing unless it shares the common nature of freezing, unknown to most of us, yet a particular instance of freezing is individuated and bounded by the concept, not by nature. Every event or process is part of indefinitely many wider events or processes, and there are no natural, only conceptual, parts and wholes. But a horse is a natural whole, and its hooves and tail are natural parts. Because it has nothing directly to say about boundaries, unity, individuation and identity, the Kripke–Putnam thesis is compatible with conceptualism, as Putnam himself has claimed. David Wiggins (1980) has presented a ’conceptualist-realist’ theory of the identity of substances drawing on Kripke’s insights, but remains firmly within the conceptualist fold.

To rebut conceptualism, therefore, more is needed than a Kantian transcendental argument or the recognition of natural kinds in the Kripke–Putnam sense. It must be shown that reality itself contributes to the way we take the world to be articulated or divided, and this contribution must be distinguished from the contribution of the mind as receptor or interpreter or speculator – from the contingencies of sense-experience, language or theory. The canonical doctrine of substance took it that primitive logical subjects are apprehended as naturally unitary, naturally whole, naturally enduring individuals, prior to their classification by us. Such ‘bodies’ are standardly contrasted with merely notionally distinct and unitary individuals in other categories. Any philosophy of experience, identity and classification must either endorse such a distinction or rule it out: that is, must be either realist or conceptualist.

What reasons are there for adopting a realist, ‘bottom-up’ theory of identity and classification? A non-substantial entity such as an event is not determinately individuated except in virtue of some general concept, commonly a correspondent to nominalized predicate, adjectival or verbal. If standard conceptualist theories of identity were correct, the same would apply to substances. Only in virtue of a quasi-theoretical concept of an insect, or insect of a certain type, would it be possible to single out the individual which is in turn egg, larva, pupa and imago. Yet it seems evident that a material object can be picked out without our knowing what kind of thing it is, and that we can then learn from observation what metamorphoses such an object will normally undergo, or can survive. The possibility of refined classification comes with sustained observation of previously identified individuals. To classify substances is to group given individuals. Genera are groups of these groups. There is therefore no question of the individuals of the genus being different from the individuals of the species. This peculiarity of substances lies behind the Aristotelian principle that there are no ultimate species or individuals except in the category of substance. If a class of non-substances, say, public assemblies, is divided for some purpose into peaceful ones and violent ones so as to include, say, demonstrations and riots, then the class of riots is as good a species as the class of assemblies, and the individuals of these classes have different principles of individuation. An assembly exists while people are assembled, but a riot exists only as long as they riot. If an assembly is composed of a demonstration followed by a riot, the assembly, the demonstration and the riot are three distinct individuals. With substances, however, species can be divided without creating new species or individuals, as the species of human beings is divided almost indefinitely by natural languages. Bakers, adolescents, albinos, diabetics and rioters constitute neither distinct species nor individuals distinct from human beings. Rioters do not cease to exist when they cease to riot, nor children when they grow up. This is not an accidental feature of natural language, modifiable by stipulation. To stipulate identity-conditions for an entity, (say) a bakens, such that a bakens exists just as long as a baker is a baker, is only to individuate a non-substantial entity such as a baker’s practice of baking. The notion of a temporal part or ’stage’ has played a central role in twentieth-century conceptualist theories of identity, but what comes to an end when a kitten becomes adult is its kittenhood, a part or stage of its life, not a temporal part of a cat – there is no such thing. ‘Kitten’ is a compound predicate satisfied by naturally unitary individuals whose identity through time consists in their continuing so united, not in their satisfaction of that or any other predicate. Things which satisfy primitive noun-predicates such as ‘cat’ do so throughout their existence, but that is because these predicates mark membership of more or less natural groups by origin, morphology and structure. The individuality of the members of the kind is prior to such classification. The unity upon which their continuity depends is the natural discreteness and coherence of a material object: materiality and the possibility of such natural unity go together. In contrast, individual non-substances are notional unities sliced out by general terms; which is why modification of the terms in question breeds new individuals, and new species.

Other considerations relating to identity confirm the contrast between the category of substance and other categories of entity. For example, just as we can literally place a material object in new circumstances, so we can consider what would have happened to it if circumstances had been different. The very same individual might have had very different attributes – that is implicit in our ordinary understanding of a material thing as possessing indefinitely many potentialities. An event or state of affairs or action could not in this way be supposed qualitatively different and yet numerically identical: a possible Battle of Hastings won by the Saxons cannot be the same event as the battle actually won by the Normans.

Realism’s strength lies not only in its philosophy of language but in an epistemology which explains how substantial individuals can be ’given’ prior to conceptualization. It has traditionally been supposed that the deliverances of the different senses present only qualities which have somehow to be tied together conceptually or theoretically in order that we should conceive of the things or substances that possess them. In the Second Mediation Descartes saw this process as the intellect’s employing an innate idea of extended substance in order to think the reality behind appearances, but idealists and some empiricists have seen it as an act of quasi-theoretical construction (see Kant, I. §3; Phenomenalism). Yet the model itself is wrong. That we have several senses does not mean that they are not integrated in their operation, presenting a unified field of bodies in space oriented in relation to our own perceived and perceiving body (that is, ourselves). Indeed, the briefest reflection on the phenomenology of perception supplies instances of such integration not only of sense with sense, but of sense with action. That we can hear a sound close to our ear does not mean that we hear our ear, or infer a relation to our ear from a quality of the sound, but that hearing is integrated with bodily awareness. That prismatic spectacles can invert all the objects in the visual field equally demonstrates that the way we feel our body to be, which includes our feeling it to be upright, can frame the deliverances of another sense. Eye and hand are coordinated in normal subjects because sight, bodily awareness and touch present a single sense-field integrated with the field of action. Here the philosophy and empirical psychology of perception converge, allowing us to understand the possibility of our pre-theoretical, preconceptual apprehension of just those natural individuals, traditional ’substances’ that are logically fundamental in natural language. Although such objects are not also, as Aristotle supposed, fundamental in natural science, their unity and structure are there for physics to explain. It is false, then, that the world as we experience it, conceive of it and act within it in ordinary life owes its articulation to the structure of language, theory or ’conceptual scheme’. Rather, thought is shaped by the experienced world, and language owes the possibility of its deepest features, reference and predication, to the senses’ grasp of really discrete individuals. That is a lesson that traditional theory of substance can still help to teach us.