Published online by Cambridge University Press: 01 January 2020
Every student of English-speaking analytical metaphysics is taught that the early twentieth century philosophical debate about truth confronted the correspondence theory, supported by Russell, Moore, the early Wittgenstein and, later, J.L. Austin, with the coherence theory advocated by the British Idealists. Sometimes the pragmatist conception of truth deriving from Dewey, William James, and C.S. Peirce is regarded as a third player. And as befits a debate at the dawn of analytical philosophy, the matter in dispute is normally taken to have been the proper analysis of the concept.
No doubt this conception nicely explains some of the characteristic turns taken in the debate. Analysis, as traditionally conceived, has to consist in the provision of illuminating conceptual equivalences; and illumination will depend, according to the standard rules of play, on the analysans’ utilizing only concepts which, in the best case, are in some way prior to and independent of the notion being analyzed — or, if that's too much to ask, then concepts which at least permit of some form of explication which does not in turn take one straight back to that notion.
A version of this paper was originally written for delivery as a lecture in the series “Unsere Welt: gegeben oder gemacht? Wissensproduktion zwischen sozialer Konstruktion und Entdeckung” held at the Johann Wolfgang GoetheUniversität in Frankfurt in the spring of 1996. It is published in German in Vogel Matthias and Wingert Lutz (eds.), Unsere Welt gegeben oder gemacht? Menschliches Erkennen zwischen Entdeckung und Konstruktion (Frankfurt/M.: Suhrkamp, 1999). Thanks to the discussants on that occasion and also to participants at colloquia at University College, Dublin; the University of Kent at Canterbury; Columbia University; and the 1998 Austin J. Fagothey S.J. Philosophy Conference on Truth at Santa Clara University; and to Bob Hale, Fraser MacBride, Stewart Shapiro, and Charles Travis.
1 Two loci classici of coherentism are Joachim, H. H.The Nature of Truth (Oxford: Oxford University Press, 1906)Google Scholar and Bradley, F.HEssays on Truth and Reality (Oxford: Oxford University Press, 1914)Google Scholar. Ralph Walker has argued that coherentism is implicit also in the forms of anti-realism canvassed by Michael Dummett and Hilary Putnam (at least, the Hilary Putnam of Reason, Truth and History). See Walker, RalphThe Coherence Theory of Truth: Realism, Anti-Realism, Idealism (London: Routledge, 1989)Google Scholar. Myself, I doubt this — for further discussion, see my critical study of Walker's book in Synthese 103 (1995): 279-302.
2 How this constraint may be made to consist with the requirement that analysis be illuminating is, of course, the heart of Moore's paradox of analysis. But the sort of objection about to be noted need read no more into sameness of content than sameness of truth-conditions.
3 Plantinga, Alvin “How to be an Anti-Realist,” Proceedings and Addresses of the American Philosophical Association 56 (1982): 47–70CrossRefGoogle Scholar. Plantinga's point also engages certain formulations of the coherence theory. For instance to suppose that “true” means would be believed by a subject who had arrived at a maximally coherent and comprehensive set of beliefs is again implicitly to surrender the means to construe the truth of the thought: no-one holds a maximally coherent and comprehensive set of beliefs. The problem is a special case of the so-called conditional fallacy: any analysis in terms of subjunctive conditionals is potentially in trouble if its intended range comprises statements which are incompatible with the protases of the relevant conditionals.
4 For discussion, see Carruthers, Peter “Frege's Regress,” Proceedings of the Aristotelian Society 82 (1981): 17–32CrossRefGoogle Scholar. See also the useful account in Walker's, Ralph C.S. survey article “Theories of Truth” in the Blackwell Companion to the Philosophy of Language, ed. Hale, Bob and Wright, Crispin (Oxford: Blackwell, 1997), esp. Section 6Google Scholar.
5 Horwich's, PaulTruth (Oxford: Blackwell, 1990)Google Scholar provides a detailed defence of the deflationary tradition and a useful bibliography of its literature. While Field's, “The Deflationary Conception of Truth” (in Fact, Science and Morality, ed. Macdonald, G. and Wright, C. [Oxford: Blackwell, 1986), 55–117)Google Scholar eventually suggests that there are purposes for which a correspondence conception is needed; his more recent “Deflationist Views of Meaning and Content” (Mind 103 , 249-85) takes a more committed deflationary line.
6 Sentences, token utterances, statements, beliefs, and thoughts are some among the other content-bearing items which we ordinarily think of as apt for truth.
7 Excepting, of course, the case where a proposition is itself about propositions.
8 I do not know that anyone has ever seriously proposed an intrinsicist conception of truth quite generally.
9 Horwich is more guarded on this than many writers in the deflationist tradition. But although he seems unwilling expressly to deny that truth is a property, it is not, he contends, a “complex property“- not “an ingredient of reality whose underlying essence will, it is hoped, one day be revealed by philosophical or scientific analysis” (Truth, 2). Thus there is, for Horwich, nothing to say about what truth really consists in, no real question for, e.g., correspondence and coherence accounts to address themselves to.
11 This claim, too, is advanced in Dummett's “Truth.“
12 All these directions of criticism are usefully referenced and reviewed in Horwich's Truth.
13 See Chapter 1 of my Truth and Objectivity (Cambridge, MA: Harvard University Press, 1992).
14 The classic treatment of this phenomenon is, of course, Grice's, H.P. “Logic and Conversation,” reprinted in his Studies in the Way of Words (Cambridge, MA: Harvard University Press, 1989).Google Scholar
15 This would be less than a commitment to the idea that “true” means epistemically justified. There is a distinction between holding that a word expresses no property but is used to commend items for their possession of a certain property and holding that it expresses that very property.
16 There is scope for some skirmishing. Rumfitt, Ian has responded (in “Truth Wronged,” Ratio 8 [New Series] : 100-7)CrossRefGoogle Scholar that the divergence in the behaviour of “true” and “assertible” just noted may straightforwardly be accommodated in a fashion entirely consonant with the purposes of deflationism, without admission of a distinctive norm of truth, provided the deflationist is prepared to allow primitive norms of warranted denial to operate alongside those of warranted assertion. Rather, that is, than restrict his distinctive deflationary claims to the word, “true,” the deflationist should contend “that ‘is true’ and ‘is not true’ function purely as devices for endorsing and rejecting assertions, beliefs and so on … and which therefore register no norms distinct from justified assertibility and justified deniability” (“Truth Wronged,” 103; compare my Truth and Objectivity, 30). How would this help to explain the commutativity of truth and negation? Rumfitt is not entirely explicit, but the point may seem clear enough. Since denying a statement is asserting its negation, a primitive warrant — an anti-warrant is Rumfitt's term — for the denial of P, registered by a claim of the form, it is not true that p, will be eo ipso a warrant for asserting the negation of P, so — via the Disquotational Scheme — for asserting that it is true that not-P. So the problematical direction of commutativity is secured, while the invalidity of the corresponding principle for assertibility is vouchsafed, as before, by the possibility of states of information in which one has neither warrant nor anti-warrant for P.
However, the problem recurs. Consider again the problematical equivalence, It is not true that P if and only if it is true that not-P.
and the result of negating both its sides:
It is not not true that P if and only if it is not true that not-P.
Supposing that the role of “(is) not true” were merely to register the presence of an anti-warrant, there seems no way of shirking the transition to
It is not anti-warranted that p if and only if it is anti-warranted that not-p. But that, of course, is no less unacceptable when neutral states of information are possible than is
It is not warranted that P if and only if it is warranted that not-P.
In short, for any discourse in which neutral states of information are a possibility, the Equivalence Schema imposes a contrast both between “is true” and “is assertible“; and between “is not true” and “is anti-warranted.” Rumfitt's proposal that the deflationist should recognize anti-warrant as primitive — whatever its independent interest — is thus of no assistance with her present difficulty.
18 This simple observation is a partial response to a recent tendency of Richard Rorty's, viz., to dismiss those features of our practice with “true” which are recalcitrant to “pragmatist” interpretation as mere reflections of the concept's absorption of a misguided representationalist metaphysic. See, for instance, his “Is Truth a Goal of Enquiry? Davidson vs. Wright,” Philosophical Quarterly 45 (1995): 281-300. But it is to be expected, of course, that Rorty would refuse to hear any but a metaphysically inflated reading of “an objective subject matter.“
20 For truth as a property of sentences, the rubric might naturally be applied to issue in something along the lines: for any sentence s, an utterance of s in a particular context is true just in case there is a proposition, that p, which such an utterance would express, and which is true.
21 Compare the remarks of Horwich quoted in note 9.
22 Russell, Bertrand “On the Nature of Truth,” Proceedings of the Aristotelian Society 7 (1906-7): 33-4.Google Scholar
23 This moral is repeatedly emphasized in Walker's excellent study The Coherence Theory of Truth (see note 1 above).
24 The explicit argument has been against a response to the original Bishop Stubbs objection - the privileging manoeuvre -which was canvassed as an alternative to relativism about truth. Briskly, then, to review how a similar difficulty afflicts the relativistic move: the relativist proposal has it that truth is always coherence with a system, but that there are thus as many versions of the truth as there are coherent comprehensive systems. Thus the proposition that Bishop Stubbs was hanged for murder, while unfit to participate in any comprehensive coherent system which is controlled by what we actually believe, may — presumably will — participate in other comprehensive and coherent systems. Well, we should now immediately press the question: What account has this relativism to offer of the truth of contingencies about belief — of propositions of the form “S is believed“? Again, it should cohere with any particular coherent comprehensive system to suppose that it is in fact believed — so such a proposition should be true relative to each particular system. So now the fact of actual belief seems fugitive. Suppose there is a single comprehensive and coherent system, S, incorporating (most of) what we actually believe, and that the proposition that Bishop Stubbs was hanged for murder is not a participant. Consider by contrast such a system, S', in which that proposition is a participant. Add to each the proposition that it is believed by most human beings. Clearly a Martian, presented only with axiomatizations of each system, would have no way of telling, just on the basis of facts about coherence, which, if either, we did believe. So the truth of the proposition that it is S we believe, if constituted just in facts about coherence, must reside in other such facts. The relativistcoherentist will offer, presumably, that it will be a matter of coherence with the Martian's own beliefs. But that is to appeal to a non-reconstructed notion of what is in fact believed by the Martian - and it was exactly the counterpart fact about us that the proposal seems to have no means to construe. So there is no progress.
25 Except in cases, naturally, where the proposition is actually about other propositions.
26 That is, in favour of the view that truth everywhere consists in the same thing. (This kind of “monism” about truth contrasts, of course, with that of Bradley and Joachim, for whom the thesis of monism is rather that reality is an intrinsically unified whole which is distorted when conceived as a totality of individual states of affairs, each apt to confer truth on a single proposition considered in isolation.)
27 The limitation to a priori cases effects, of course, a restriction on the standard lay use of “platitude,” which applies to anything which no-one would dispute (and also carries an unwanted connotation of tedium).
28 One possible addition is reviewed in Section VII below.
29 For elaboration of this claim, see my Truth and Objectivity, 24-7.
30 Readers familiar with Michael Smith's work will note a point of contact here with the conception of a network analysis which he derives from Ramsey and Lewis (see in particular Chapter 2, Section 10, of Smith's, The Moral Problem [Oxford: Basil Blackwell, 1994])Google Scholar. The principal contrast with the approach to truth here canvassed is that a network analysis has to be based on a comprehensive set of platitudes whose conjunction so constrains the target concept that the replacement within them of all expressions for that concept by a variable and its binding by the description operator results in a definite description which is at the service of an analytically true identity,
Φ-ness is the property, F, such that ( … F … & … F … & …)
which thus effectively supplies a reductive analysis of the concept ell. An analytical theory, by contrast, need not — though it may — subserve the construction of such an analytically true identity.
31 For exploration of one local case, arithmetic, see the Appendix to this paper.
32 Truth and Objectivity Chapter 2; an earlier discussion is in Chapter 14, “Can a Davidsonian Meaning-Theory be Construed in Terms of Assertibillty,” of the second edition of my Realism, Meaning and Truth (Oxford: Blackwell, 1993).
33 For relevant details see the Appendix to this paper.
34 How does it follow that a satisfier of the platitudes will be definable on such contents? Very straightforwardly. First, if we are dealing with a range of genuine contents — to the extent ensured by the hypothesis of discipline — for which we have the conditional construction, then nothing can stand in the way of the definitional introduction of a predicate, or operator, which is subject to the Equivalence Schema:
That P is Φ if and only if P.
As noted, that will then suffice for versions of Transparency, Contrast, the minimal degree of Opacity that attends contrast, and a Correspondence Platitude for Φ. It will further be open to us to insist that Φ be defined for all combinations of specified kinds of the contents in question and thereby secure Embedding. Assuming that the contents in question allow of tensed expression, Trmelessness — effectively the principle that whatever may truly be thought or expressed at any particular time may, by appropriate variations of tense, be truly thought or expressed at every time — may be secured by stipulating that Φ is to be governed by analogues of the usual truth-value links between differently tensed counterparts. (If the contents in question are tenseless, then Trmelessness will hold by default.) Absoluteness, for its part, will hold by default in any case unless we explicitly fix the use of a comparative.
36 Of course, an assertion may be insincere. For an utterance to be a profession of a certain state means that one who accepts its sincerity must be prepared to ascribe that state to the utterer.
38 Michael Smith himself eventually takes such a view in The Moral Problem.
39 See Blackburn, Spreading the Word, passim but especially Chapters 5 and 6.
40 For parallel discussion, see my Truth and Objectivity, 25-7.
41 See my Truth and Objectivity, Chapter 2, Section V.