Understanding Realism - page 2
If we adhere to Dummett's formulation, in which acceptance of bivalence and knowledge-independence of truth-value commits us to the existence of a knowledge-independent reality, we have indeed thereby identified what that acceptance entails as to its theoretical underpinning. But there is no converse entailment. It may be natural, but it is not obligatory, for someone to hold the metaphysical and epistemological doctrines in question, and also to hold the view about truth which Dummett says is the mark of realism. But it is open to someone to hold those metaphysical and epistemological doctrines and to make different moves over truth: for one example, to deny that truth is a property conferred on what we assert by the reality so conceived; or, for another, to deny bivalence. And someone might do this latter even if he accepted that truth is recognition-independent and is so because it is conferred by a knowledge-independent reality. At the same time, it is clear that what a philosophical doctrine of realism seeks to preserve from our ordinary beliefs about the nature of reality is precisely what is conveyed by the metaphysical and epistemological theses which give natural but not inevitable rise to the view of truth in question. Given this, it is hard to resist the view that the metaphysics and especially epistemology of the matter are fundamental.
One explanation of the apparent pressure to think otherwise comes from the idea that any adequate conception of truth has to rely on stipulative features. In particular, it might be held that a correspondence principle, for one of the two main senses of 'correspondence', functions regulatively in our account of truth, prompting commitments about the nature of the corresponding relata. At its roughest a correspondence theory of truth says that a proposition, statement or belief is true if it corresponds to 'the facts' or 'how things are', false otherwise. Three difficulties immediately present themselves: what is the correspondence relation? What are the linguistic or psychological entities which stand in the correspondence relation, whatever it is, to something else? And what is this something else, here vaguely denoted by 'the facts' or 'ways things are'? For all the superficial plausibility of the proposal that truth is correspondence between sayings (or beliefs) and facts, the debate it has generated has not yielded any satisfactory defence of it. Accordingly it has been suggested that in response to our need for a regulative conception of truth, we simply lay it down as a minimum feature of truth that it consist in a correspondence - leaving open the question of what in detail this is - between suitably characterised relata.
There might accordingly be a demand for an independence clause in a specification of truth - that is, one which asserts that 'the facts' exist, and have the character they have, independently of our investigations of them - precisely to serve our need to be able to mark off true beliefs and utterances from those which are not so. Given that it matters so much in practice whether what we say and believe is true, just that objectivist division is forced, and in the heat of the moment an utterance's failing to be true invites no closer scrutiny as to how it does: we do not stop to ask whether it fails to be true because it is false, or for some other reason (because, say, it is meaningless, or has some third truth-value, or is not a candidate for truth-value at all). The view that falsehood exhausts ways of failing to be true is doubtless a natural one to have arisen in the history of ordinary uses of language, a fact which might yield a moral for anyone inclined to believe that ordinary usage is sacrosanct.
But the need which prompts stipulations about the nature of truth is precisely an epistemic need (all practical needs are such, although the converse is not true). Viewed as a strategic commitment, a correspondence principle entails the allied but further strategic commitment to the independence of truth-conferring states-of-affairs from knowledge of them, because this objectivist attitude alone sustains what is required by the urgencies of practice - that is, exhaustive classification into 'true' and 'not-true'. Considerations of practice of course lead on to the drawing of distinctions among ways beliefs and utterances can fail to be true. But at the outset our view of the character of truth is determined by the controlling influence of our metaphysical and epistemological concerns, namely, those which constitute our commitment to there being a knowledge-independent realm of entities. The power of these concerns can be seen in the fact that they give rise to the very intuitions which are offended by counterfactual conditionals that appear not only incapable of determinate truth-value, on the grounds that there is nothing 'in virtue of which' they could be either true or false ('if God had created such-and-such beings, they would have done so-and-so' is a familiar example), but that they are not even capable of being either true or not true. Here the lack of something 'in virtue of which' an objective truth-value can be assigned suggests that there is nothing to be committed to antecedently which would sustain a notion of truth-value for the cases in question.
The tension in Dummett's account is not far to seek. His reason for characterising realism as a thesis about the truth of statements rather than as an ontological (still less an epistemological) thesis, is, as noted, that 'certain kinds of realism, for instance realism about the future or about ethics, do not seem readily classifiable as doctrines about a realm of entities' . Yet he immediately goes on to define realism for any subject matter in expressly ontological and epistemological terms: 'The very minimum that realism can be held to involve is that the statements in the given class relate to some reality that exists independently of our knowledge of it' . This is inconsistent, so one of these views must give way. It is not hard to say which. If the notion of truth which constitutes the Dummett hallmark of realism depends for its content on an antecedent commitment to there being an independently existing reality, and if, as already quoted, 'the fundamental thesis of realism ... is that we really do succeed in referring to external objects, existing independently of our knowledge of them', it follows that what we should say about those 'realisms' which are not readily classifiable in terms of entities is, simply, and on Dummett's own reasoning, that they are not realisms. Disputes concerning them are disputes of a different kind: and insofar as they raise questions about what concept of truth is applicable to them, that concept cannot involve considerations about the knowledge-independent existence of entities. And it is accordingly no longer clear whether the concept of truth at stake in these disputes is objectivist. When we find that a theory of meaning rests on a semantics to which that concept of truth is central, commitments of the metaphysical and epistemological kind at issue have therefore already been made.
One point, then, is that whatever else 'realism' might denote, it at least denotes a thesis about a realm of entities. This should hardly be surprising; even in traditional debates about universals and the external world this much is a common feature. But as we see it follows that if ethics and mathematics and talk of other times - especially the future - are not about realms of entities, then controversies over the concepts of truth and knowledge applicable to them are not realism-anti-realism controversies. On this conception, although we recognise that, in ethics, the debate is between cognitivists and those who disagree with them, and that in mathematics it is between espousers of different understandings of what makes for the truth of mathematical statements, we also recognise that in neither debate is it just that there is no obligation to talk about the existence of entities (the respective candidates might be 'moral properties' and 'structures'); it is, as Dummett himself suggests, positively misleading. For if, respectively, cognitivist and Platonist theses turn on claims about the existence of certain sorts of moral properties or mathematical structures, the question immediately arises as to how we can reduce the metaphorical character of such claims, given that their sense is imported from the one case (the 'external world' case) which alone has unmetaphorical content .
The absence of an answer to this question is precisely Dummett's motive for switching attention to the problem of truth. But as shown, doing so brings too much under one label. The solution is not to find a different reason - one given in terms of truth - for classifying all these controversies together as realism controversies, but instead to recognise that they are controversies of quite different kinds. So we do well to restrict talk of realism to the case where controversy concerns unmetaphorical claims about the knowledge-independent existence of entities or realms of entities - namely, the 'external world' case - and employ more precise denominations for the different debates which arise in other, different, domains.
The most important point to be made about the nature of realism, however, is that what crucially differentiates it is the epistemological thesis that the realms or entities to which ontological commitment is made exist independently of knowledge of them. It is vital to note that existential commitment without this epistemological independence claim is not realism. For it is no-one's view that if the existence of something x can only be understood in terms of what it is for x's existence to be known or detected, therefore x is unreal. Obviously enough, an anti-realist metaphysics is a metaphysics of existing entities. What distinguishes such a view from a realist one is that unlike the realist, the anti-realist can make no sense of metaphysical claims without a supporting epistemology that yields grounds for them. If something is asserted to exist, in other words, it is because something counts as validating, supporting or making sense of that claim; in short, something counts as evidence for the claim, grasp of which plays its part in constituting the claim's sense. The anti-realist in this way regards the relations between existing things and the relevant kinds of epistemic access to them as internal ones - from which it does not follow, as in cruder characterisations of anti-realism it is supposed to follow, that the existing things are 'dependent' (perhaps even causally dependent) on cognition of them. This is a hangover of misunderstandings of Berkeley, whose denial of the existence of material substance is too often read as a denial of the existence of the external world . Viewed from this perspective, what is at issue between realists and anti-realists is the epistemological thesis that what exists does or can do so independently of any thought, talk, knowledge or experience of it. To make out his case the realist has to show that this claim is intelligible. The chief anti-realist point is that the claim is unintelligible .
Are we helped by recognising that the realism dispute is primarily an epistemological one? In giving an affirmative answer one must begin by stressing that ordinary discourse is, without question, realist in character. We assume that the entities we refer to exist independently of our cognising them, and we assume the same about the states of affairs which, we further assume, make true or false our assertions about them. Our realism at the level of ordinary thought and talk, the 'first order' level, is indeed rather promiscuous: we take literally a sense, to be informatively compared with the case of fictional discourse, of there being something we are talking about when we talk. Various ways of cashing this thought suggest themselves, one of which is that it would render explanation of our first order linguistic practice incoherent if we did not or could not attribute to speakers beliefs about the existence, independently of them, of the entities constituting the domain over which their discourse ranges.
The clue lies in the fact that these realist commitments are commitments and that they are fundamental to first order practice. We might now - to wax schematic - distinguish between realism, assumed at the first order, and what I shall for present purposes label 'Transcendentism', the second order thesis that realism is literally true. On this way of putting matters, anti-Transcendentism is the thesis that it is mistaken to claim that realism is literally true, since nothing can, on the realist view itself, establish that it is either true or false, for the content of realist claims exceeds the possibilities of verification of them. Rather, says the anti-Transcendentist, realism is a fundamental assumption of our practice at the first order. It is therefore not true but assumed to be true. The dispute between these positions is accordingly a second-order controversy about the correct understanding of our practice and the logical and metaphysical presuppositions of it. Second order commentary might show that there is need to revise a first order practice wherever the commentary reveals the practice to be wrong. A second order thesis like this would constitute an error theory with respect to first order practice, as in the case, say, of Mackie's view about ethics.  But it depends on cases, for it can be that second order interpretation of first order practice leaves the latter as it is.
At first blush the difference between the Transcendentist and the anti-Transcendentist positions looks vanishingly small. But the consequences for a range of issues, including our understanding of truth and knowledge, are great. On the Transcendentist view the relations between speakers and what they speak about are external ones, so it is at very least natural to treat truth as a property which our utterances have conferred on them by knowledge-independent states-of-affairs, and our notion of knowledge as having it that whatever consequences, if any, our knowing something about the world has for the world, they are contingent ones only. In particular, coming to know things about the world is a process of discovery, one which lies under the austere constraint of our inherent epistemic limitations. (It is in this sense, to use Crispin Wright's phrase, that realism is 'modest'.) Taken together, these theses about truth and knowledge entail a commitment to there being a sharp distinction between a statement's truth-value and our having grounds for assigning it one. This epistemological commitment is sometimes identified by Dummett as fundamental to Frege's views and - quite rightly, as the argument here has it - as the crux of Transcendentism. For the anti-Transcendentist that distinction exists for us only at the first order, as a matter of epistemic strategy.
Either way, therefore, the nub of the matter at the second order concerns the question whether metaphysical commitments at the first order can be regarded as literally true (or false) as Transcendentism claims, or, as anti-Transcendentism claims, as having an irreducibly strategic character, constituting assumptions of our discourse which we hold true as a framework for organising experience fruitfully. It is a debate primarily about the role of epistemic constraints in understanding our thought, not a debate about what logical principles our practices should embody, nor a debate about what is taken to exist in our first-order scheme of things (or the science by which they are explained and, to the extent possible, manipulated). In this sense the debate leaves everything as it is, and therefore if anti-Transcendentism is correct, no revisions to logic, linguistic practice or mundane metaphysics are called for.
These ideas are taken up and explored in more detail elsewhere. But before we leave the matter of identifying accurately what is at stake in the realism-anti-realism debate - or Transcendentist-anti-Transcendentist debate, as I think we should call it - we should keep the following in mind.
Earlier it was noted that current orthodoxy defines the debate as being primarily about meaning. But despite the fact that, if the foregoing discussion is right, the debate is correctly to be seen as one which primarily concerns thesis B, that is, epistemology, and one which moreover applies only where questions about the existence of entities are taken to be already settled, there is nevertheless no suggestion that semantic questions are irrelevant. Far from it. This is because a decision about the role of epistemic constraints one way or the other has immediate results for our view about what sort of theory of meaning we can have; and that in turn will tell us where to look for a detailed understanding of our concepts of truth, reference and the nature of valid inference.
Dummett is explicit in his opposition to this kind of approach. 'An attack from the top down tries to resolve the metaphysical problem first, and then to derive from the solution to it the correct model of meaning, and the appropriate notion of truth, for the sentences in dispute, and hence to deduce the logic we ought to accept as governing them.'  But the disadvantage of this approach, he says, is that we have no way of resolving the metaphysical dispute because, despite centuries of debate, we cannot give it a clear content; we cannot reduce the metaphorical character of the terms in which it is posed. Therefore we should proceed bottom-up, starting with the question of the correct model of meaning for statements of the disputed class, 'ignoring,' he says, 'the metaphysical problems at the outset'.  When the correct model of meaning has been devised the metaphysical problems will thereby be solved, because there is nothing more to a metaphysics than its being the 'picture of reality' that goes with a particular model of meaning. 
My response is to say, first, that metaphysical problems will only be statable in metaphorical or pictorial terms if we think they arise in connection with ethics, mathematics, the future, or similar subject matters. One is much inclined to agree with Dummett that no real content can be attributed in such cases; or indeed in any case, for this is just the difficulty noted in connection with thesis A. But as argued above, this is precisely the reason for saying that disputes over these cases are not therefore realism disputes: they are not candidates for evaluation in terms of Transcendentist commitment, that is, commitment to the knowledge-independent existence of entities. The only subject matter where this makes non-metaphorical sense is the external world case.
But in any event the top-down strategy does not start with the metaphysical problem, not even the unmetaphorical one: it starts with an epistemological one, namely, the question whether epistemic constraints are necessary for the intelligibility of our metaphysical claims and indeed for our discourse in general. Once that issue is resolved, ipso facto one rather than another basis for a choice of model of meaning has been laid. So much is implicit in Dummett's own characterisation of the notion of truth he identifies as the source of realism. The point here is to insist on the dependence of that notion on epistemological considerations, and therefore to urge a redistribution of emphasis in what follows.
1. Dummett, M.A.E. 'Realism' Synthese 52 (1982) pp 56-7 (references to this paper henceforth 'RS') The Logical Basis of Metaphysics London 1991 (henceforth 'LBM') pp 9-10, 325-6. Everything that can be said about the ultimately epistemological character of realism as Dummett conceives it can be said of Putnam's view of (therefore mislabelled) 'metaphysical realism' too; in Putnam's terminology, the property — 'Independence' — that captures the core of realistically-conceived truth is its freedom from epistemic constraints (cf. Frege's sharp distinction between truth and grounds for truth); the claim that truth is Independent is the claim that 'the world could be such that the theory we are most justified in accepting would not really be true ... rational acceptability is one thing, truth is another' ('Model Theory and the Factuality of Semantics' Reflections on Chomsky ed. George, A., Blackwell 19189 p 214). Any view that opposes this (any anti-realism) is ipso facto a thesis that truth (and our view of the world in general therefore) lies under the government of epistemic considerations. For economy's sake I restrict attention to Dummett here.
2. Dummett 'Realism' in Truth and Other Enigmas London 1978 p45. References to this paper henceforth 'RT'.
3. ibid and RS p55, LBM Ch. 1 passim.
4. RS ibid.
6. Dummett 'What Is A Theory Of Meaning II' in Evans and McDowell (eds) Truth and Meaning Oxford 1976. Also see, for example, Wright, C. Realism Meaning and Truth Blackwell 1987 pp 13 et. seq.; Grayling, A.C. An Introduction to Philosophical Logic London 2nd Ed. 1990 Ch.8 passim.
7. RS p55.
8. ibid p104. See also LBM pp 9, 345.
9. See Grayling A.C. The Question of Realism Oxford University Press forthcoming passim esp. Chapter 2.
10. See e.g. McDowell J. 'Truth-Values, Bivalence and Verification' in Evans and McDowell op cit. and Wright C. 'Realism, Truth-Value Links, Other Minds and the Past' Ratio XXII (1980) pp112 et. seq.
11. RS p55.
13. Wiggins, D. 'Moral Cognitivism, Moral Relativism and Motivating Beliefs' Proceedings of the Aristotelian Society Vol. XCI (1990/91) p62.
14. The beginnings of the difference would be marked by the response of the man on the Clapham omnibus to attempts at running them together in a putatively univocal way: 'tables and numbers exist' instead of 'tables exist' and 'numbers exist'.
15. See Grayling A.C. Berkeley London 1986 passim.
16. It is to Colin McGinn's credit that he has recognised this obligation and tried - heroically but unsuccessfully - to meet it; see McGinn The Subjective View for one attempt, and 'Can We Solve The Mind-Body Problem?' Mind 1990 for another, premised on an abandonment of the first; and my respective replies in Berkeley Ch.4 and The Question of Realism Ch.3.
17. RS pp 105 et seq.
18. cf RS pp 68-9 et. seq. and Berkeley pp 104, 197.
19. This is why Berkeley was in no sense a phenomenalist. In his view the possibility of something's being perceived turns entirely on the actuality of its existence; so, possibilities of perception relate only to finite minds, and the truth-value of counterfactuals is determined by the truth-value of nonconditional statements. See Berkeley op. cit. pp 95 et seq.
20. Tennant Anti-Realism and Logic Oxford 1987 p12.
21. ibid and M. Luntley Language Logic and Experience London 1988.
22. LBM pp 9, 319-20.
23. cf Tennant op. cit. pp 185 et. seq., Luntley op. cit. pp 100 et. seq.
24. C. Wright 'Anti-realism and Revisionism' in Realism Truth and Meaning Blackwell 1987 p 317.
25 see S. Rasmussen and J. Ravnkilde 'Realism and Logic' Synthese 52 (1982) pp 379-80.
26. Wright op. cit.
27. J.L. Mackie Ethics London 1977 Ch. 1 passim.
28. LBM p 12.
30. ibid p 15.