Truth (Princeton Foundations of Contemporary Philosophy)

by Burgess, John P. ; Burgess, Alexis G.

  On the other hand, how seriously does the verifiability thesis conflict with deflationism? Deflationism predicts that substantive theses of metaphysics, epistemology, semantics, and so on that involve the truth predicate in their formulation will turn out on examination merely to be using the truth predicate to achieve formulation as a single generalization of what otherwise could be conveyed only by examples. We have seen in §5.3 that Lewis argues something like this point in the case of the claim “Every truth has a truthmaker.” Soames has suggested something similar in the case of the claim “Every truth is verifiable.” Perhaps the verifiability thesis, properly understood, no more clashes with deflationism than does the truthmaker principle, properly understood.

  For the argument is not that, whether or not (17) holds, still

  (19) If it is true that there is intelligent life in other galaxies, then it is at least in principle verifiable that there is.

  must hold simply because “true” just means “verifiable.” Rather, the argument is that (19) holds because (17) holds (and then, because there is nothing special about this one example, the further generalization holds that anything true is verifiable). This argument only requires the equivalence principle to get from (17) to (19), and so, it seems, does not require an inflationist notion of truth.

  Insofar as deflationism implies that the concept of truth just isn't as important a notion as it has often been taken to be, any victory for deflationism must in a sense be pyrrhic. In the contest for recognition as the true theory of truth, if deflationism wins, then the very fact that it has won in a sense suggests that the gold medal may be only gilt. Lewis and Soames between them suggest that much that avowed realists and antirealists care about could still be argued about even if deflationism were conceded to be the correct account of truth. Such a concession might not have to be viewed as surrender on the points that are really important to the realist or antirealist. But such irenic reflections have not so far noticeably reduced realist and antirealist hostility to deflationism.

  6.5 HOLISM

  The twists and turns of the dialectic include one last radical reversal: Dummett argues that truth-conditionalism could not give the correct account of meaning for any possible language; but Dummett does not and cannot claim that it is verification-conditionalism instead that gives the correct account of our actual language. For as Dummett certainly holds, meaning is what guides use, and the intuitionist/verificationist meaning (13b) is not what is guiding our use of disjunction, for instance.

  If it were, forensic scientists would not say things like “The DNA evidence shows the culprit was one of the identical twins X and Y, but we have no way of determining which.” Nor would mathematicians accept intuitionistically unacceptable existence proofs, as they systematically do. In the face of this systematic usage, it would make no more sense to claim that (13b) gives the correct account of what English speakers really mean by “or” than to claim that black is what English speakers really mean by “white.” Verification-conditionalism is a nonstarter as a description of what we have. It is, rather, a prescription to replace it with something radically different. And while Dummett does not exactly highlight this point, he does not conceal it, either.

  But if this is so, and if also Dummett is right in his criticism of truth-conditional semantics, then it must be some third alternative that gives the correct account of the meaning of our actual language. Dummett does not endorse any specific such third account, but he does seriously discuss one, the picture of the language of science that he associates with the name of Quine, and that is sometimes called “holism.”

  On this view, meaning is not given by conditions for truth or verification, at least not once one gets beyond the elementary part of language that is used for recording and predicting empirical observations. The meanings of theoretical terms are constituted in a different way, by clusters of basic laws posited to hold for them. These laws are used together with some empirical observations to predict other empirical observations. By the success or failure of such predictions we decide whether to retain the basic laws and the theoretical terms they involve, or revise them, or reject them altogether, as phlogiston has been rejected.

  The predictions of empirical observations are derived using logic, raising the question of what constitutes the meaning of logical vocabulary. The alternative Dummett considers to truth conditions or verification conditions, as in (13ab) or (18ab), is that meaning is given directly by introduction and/or elimination rules. For disjunction and the conditional, these would read as follows:

  (20a) to infer “_____ or _ _ _ _ _” from “_______”

  (20b) to infer “______ or _ _ _ _ _” from “_ _ _ _ _”

  (20c) having seen “......” to follow from “______” and from “_ _ _ _ _,” to infer “.......” from “______ or _ _ _ _ _”

  (21a) having seen “_ _ _ _ _” to follow from “________,” to infer “if ______, then _ _ _ _ _”

  (21b) from “if ______, then _ _ _ _ _” and “_______” to infer “_ _ _ _ _”

  The holist picture may be especially congenial to deflationists, who suggest the meaning of “true” is constituted by T-biconditionals or by T-introduction and T-elimination; but this is not the place to elaborate a picture Dummett himself only adumbrates. It is enough that we have sketched one alternative beyond David-sonianism “realism” and Dummettianism “antirealism”; there are others. Indeed, the vast literature inspired by Dummett is full of other “arealisms” and “irrealisms” and “quasirealisms.”

  6.6 PLURALISM

  Though Dummett may concentrate on mathematics as an example, he repeatedly indicates that the considerations he is advancing are supposed to apply generally. By contrast, a number of recent writers, beginning with Crispin Wright (long a sympathetic though not uncritical commentator on Dummettian anti-realism) and continuing with Michael Lynch, have been prepared to contemplate the possibility that truth may amount to something like verifiability for some topics—or for some “domains of discourse”—but not for others.

  This view is now called (alethic) pluralism. (Wright initially, and unfortunately, gave it the same label, “minimalism,” as Horwich's very different view.) The view has points both of agreement and of disagreement with deflationism, as with (various forms of) inflationism. It is easiest to state the point of agreement with deflationism: There is no one, single, interesting, substantive property coextensive with truth across all domains of discourse. The points of disagreement are several.

  We remarked in §2.2 that for small enough fragments of language, truth of sentences may be coextensive with something else interesting (adding in §2.5 that for Euclidean geometry, truth is coextensive with provability, and noting in §5.5 that correctness for subway diagrams is coextensive with a certain physical relation). Certainly for any fragment to which Tarski's method can be applied, there will always be a set-theoretic property coextensive with truth, namely, Tarski-truth. Perhaps a holist of the kind considered in the preceding section will want to combine deflationism with the view that truth for observation sentences is coextensive with verifiability or even some physical property.

  The pluralist, however, suggests that it is not just for small fragments, or for especially simple sentences, but for large blocks of language that truth is coextensive with some more substantive property: maybe one property in ethics, and another in mathematics, and yet another in science. This is already something that most deflationists (sensitive as they are to the diversity of the things we talk about even within, say, such a medium-sized chunk as physical science) will be reluctant to accept. And they will be even more reluctant to accept the pluralist's further claim that, for this or that large block, the substantive property truth is coextensive with is just the sort of thing that this or that inflationist theory has traditionally taken truth in general to be: perhaps something like verifiability in one domain, and coherence in another, and correspondence in yet another.

  On top of this, the plur
alist wants to claim that truth is not just coextensive for this or that large block with this or that traditional inflationist notion, but that for each large block truth literally is (or in more obscure metaphysical jargon, is “constituted” or “realized” by) that substantive property. Here pluralism directly contradicts the deflationist thesis that once one has explained what it means to call something true by citing the equivalence principle, there is no room for further questions about what it is for something to be true.

  Though there are philosophers who maintain that we may always posit that “there is more to a property than our concept of it” (an extremely dubious claim if intended to apply even to mere pleonastic properties), we have suggested on the contrary that whether, having been told what “_______” means, there is room for a further question about what ______is, depends on the nature of the meaning. On this view, if “pain” means a certain distinctive kind of sensation known by acquaintance, then there is no room for a further question about what pain is. If, by contrast, “pain” means, as some forms of materialism maintain, “whatever natural phenomenon it is that plays such-and-such a causal role” (where the role in question may involve, say, being characteristically caused by injury and characteristically causing aversive behavior), then there is room for the question “And what natural phenomenon is that?” and for such an answer as “Firing of C-fibers,” and also perhaps for such an answer as “One thing for earthlings, another for Martians.” Lynch explicitly compares his pluralism to this form of materialism, generally called functionalism, by way of explaining how it is supposed to be possible for truth to be (in his almost mystical formulation) “both one and many.”

  Pluralists oppose the deflationist account of the meaning of the truth predicate, in favor of an account on which that meaning is something like “whatever substantive property it is that plays such-and-such a role” or “having some substantive property or other that plays such-and-such a role,” with the understanding that different properties may play the role in different domains. The characterization of the “truth role” that pluralists offer does include (perhaps a bit grudgingly) the equivalence principle, but Wright and Lynch both go on to include more, beginning with being the norm of assertion, which we have seen (in §5.6) deflationists may suggest is part of the meaning of “assertion” rather than of “true.” There is then a further longish list of “platitudes” or “truisms” any deflationist is bound to claim are part of the meaning of other notions such as “honest” or “sincere,” rather than of “true.”

  One worry about alethic pluralism is the threat of pluralism creep. Pluralism about truth is less innocuous than some other forms of pluralism just because the notion of truth is routinely used in the explication of other concepts of philosophical interest (though according to deflationists the truth predicate may be serving only to enable us to state in a single generalization what otherwise could only be suggested by examples). If the notion of truth is implicated in the analyses of honesty or sincerity, assertion, knowledge, implication, and so on, then presumably if there are many “truth properties” or things that truth is, there will for each of these other items be many things that it is, too.

  In particular, alethic pluralists are often naturally led to logical pluralism, the view that different logics are appropriate for different domains of discourse: perhaps classical logic for a domain where truth is correspondence, but the weaker intuitionistic logic for one where it is something like verifiability or coherence. Pluralism seems to raise drastically the chances that arguments involving inferences across domains where truth is differently “constituted” or “realized” may be fallacious because of equivocation.

  Consider, for instance, the Puritan syllogism:

  (22a) Whatever causes pleasure without hard work is a vice.

  (22b) Whatever is a vice is ultimately deleterious to health.

  (22c) Ergo, whatever causes pleasure without hard work is ultimately deleterious to health.

  If truth for the premises (22a) and (22b), which involve evaluative notions, is a matter of something like verifiability or coherence, but truth about descriptive matters, as in the conclusion (22c), is a matter of correspondence, what grounds have we for thinking that the syllogism is truth-preserving? More fundamentally, if truth for “Driving Hummers is expensive” is correspondence, while truth for “Driving Hummers is reprehensible” is something else, what is truth for their conjunction or disjunction supposed to be? There have been several rounds of debate between pluralists and antipluralists about this problem of mixed discourse.

  There has even, perhaps, been too much attention to this one apparent weakness of pluralism, and too little to others. Even if pluralists only claimed that truth is coextensive with correspondence in some domains, and something like verifiability or coherence in others, there would still be (i) a need to specify the boundaries of these domains and what truth is supposed to be coextensive with in each, and (ii) for each substantive property (correspondence or coherence or verifiability or whatever) with which truth is supposed to be coextensive in some domain, a need to answer standard objections to the thesis identifying truth with that property, insofar as these still apply when the association is restricted to that domain and downgraded from identity to coextensiveness. Existing pluralist accounts generally do little towards satisfying needs (i) and (ii). Even sympathizers have to admit that existing pluralist discussions are, despite containing many suggestive ideas, still “programmatic.” The less sympathetic might say “sketchy.”

  Perhaps this is inevitable given that, at this writing, pluralism is the nouvelle vague in the theory of truth, still under development. (The antithetical view, advocated by Mark Richard, that truth simply isn't the right dimension of evaluation in many domains, is newer, but not yet a wave.) With pluralism we come to the end of our survey of contemporary theories of truth, insofar as the issue of the nature of truth can be discussed without taking into account the paradoxes. It is time to take them into account.

  CHAPTER SEVEN

  Kripke

  TARSKI WAS THE MOST PROMINENT ADVOCATE of the inconsistency theory of truth, notoriously maintaining that the intuitive notion of truth is self-contradictory. (Holding, like many positivistically inclined philosophers of his day, that languages come equipped with “meaning postulates” or “semantic rules,” he expressed his view that those governing “true” permit the deduction of contradictions by saying that natural languages like English are inconsistent, a claim that some later commentators, in real or feigned ignorance of the historical background, have professed to find unintelligible.) His aim was to define rigorously a restricted substitute for the intuitive notion of truth, safely usable for mathematical purposes, and to demonstrate its utility. Many later writers have, by contrast, written as if they took it to be the job of a philosophical account of the paradoxes to vindicate the intuitive notion of truth, and show that despite all apparent paradoxes the notion involves no real antinomies.

  In what is surely the most influential work since Tarski on the paradoxes, Saul Kripke presents a mathematically rigorous, paradox-free treatment of truth for certain formal languages, different in spirit from Tarski's. He adds hints about how his formal construction might model some features of natural language, but his hints steer a path between an inconsistency view and a vindicationist one. Reserving discussion of Kripke's philosophical position for the next chapter, we describe in informal terms some main features of his construction in the first half of this one, adding in the second half, whose sections are starred as optional, technical details about the constructions of Kripke and some of his successors.

  7.1 KRIPKE VS TARSKI

  Tarski's reconstructed notion of truth is severely restricted. He starts with an object language L0 not mentioning truth at all, and defines a truth predicate T 0 for L0 in a metalanguage L1. His method could be applied again to define a truth predicate T 1 for L1 in a metametalanguage L2, and so on, to create a hierarchy of truth pr
edicates T n for a hierarchy of languages Ln, but we never get a language L containing its own truth predicate T.

  The natural-language analogue of Tarski's hierarchical approach would be a language with no unrestricted truth predicate “is true,” but only a series of restricted ones: “is true0“ or “is a true sentence not mentioning truth,” then “is true1“ or “is a true sentence mentioning at most truth0,” then “is true2“ or “is a true sentence mentioning at most truth0 and truth1,” and so on. (And similarly for “false,” with falsehood being identified, as always, with truth of the negation.) Tarski never advocated that we should add such subscripts to natural language; nonetheless, Kripke finds it worthwhile to rehearse why the use of audible or visible subscripts in speech or writing would be unworkable, and would leave us far short of what we can say, usually without encountering any contradictions, in natural language.

  Kripke's work dates from the Watergate era, and he considers things John Dean and Richard Nixon might have wanted to say about each other's veracity in Watergate-related matters. Suppose Dean wants to say, on inductive grounds,

  (1) Nothing Nixon said about Watergate up to the time of his resignation is true.

  Dean would have to put a subscript on his “true,” and the subscript would have to be higher than the subscript on “true” in anything Watergate-related that Nixon said up to the time of his resignation. Finding an appropriate subscript would be infeasible in practice, unless Dean had a complete recording of everything pertinent Nixon said during the relevant time period. But as is well known, the Nixon tapes contain an eighteen-minute gap.