Truth (Princeton Foundations of Contemporary Philosophy) Read online

  The answers offered by traditional views tended to make truth a notion of metaphysics or epistemology or ethics. That tendency, and all the traditional views, are opposed by the cluster of positions commonly labeled “deflationist.” The question “What do the different views called ‘deflationist' all have in common, to make them all deflationist?” itself admits no easy answer. Deflationists are, however, typically committed to three theses about the phrase “is true,” usually called the natural language truth predicate. (That label also covers the phrase's synonyms “holds” and “is so” and “is the case,” along with corresponding expressions in other languages.)

  First, applying the truth predicate to something is equivalent to just saying it. One version of this equivalence principle is embodied in Tarski's T-scheme, but there are others. Different deflationists, besides holding different views on whether the “something” in question should be taken to be a sentence or a proposition, give different accounts of what the “equivalence” here amounts to.

  Second, the equivalence principle is a sufficient account of the meaning of the truth predicate. There is nothing more to understanding the truth predicate than recognizing the equivalence principle, and that by itself ultimately suffices to account for our usage of the predicate and its utility. Different deflationists give different accounts of what the “recognition” here amounts to.

  Third, an account of the meaning of “true” is a sufficient account of the nature of truth. There is nothing to be said about what it is for something to be true once one has said what it means to call something true. Commitment to this last thesis is implicit in the practice of the typical deflationist, who begins by promising an account of the nature of truth (often quoting Pilate's question), but in the end offers only an account of the meaning of “true.” Explicit enunciation of the principle is less common.

  But the different varieties of deflationism may be united more by family resemblance than shared essence, and our initial characterization doubtless accommodates some varieties less comfortably than others. What follows is a quick survey of the main types of deflationism, offered in hopes that as it proceeds the theme behind the variations will gradually become more apparent. The main division among deflationists is between two groups we will call radicals and moderates. The radicals maintain that what is conventionally called the truth predicate is not truly a predicate, or is one “only grammatically and not logically.” They implicitly or explicitly contradict Tarski's view that if two speakers successively say

  (1) The love of money is the root of all evil.

  (2) It is true that the love of money is the root of all evil.

  then the second speaker has made a statement about the first speaker's statement.

  3.1 REDUNDANCY

  The prehistory of deflationism vs inflationism can be traced back to the tension between the two Aristotelian formulations “to say of what is that it is” and “agreement of thought with its object,” but what is usually accounted the earliest version of deflationism was the redundancy theory of Ramsey. His view is radical. He holds that (1) and (2) are equivalent in the very strong sense of expressing exactly the same proposition. The difference between them is merely stylistic, and the truth predicate in (2) is redundant.

  Note, however, that if someone has just said (1), then the effect of repeating (1) as if taking no notice of its having just been said and the effect of saying (2) can be quite different. (The effects would be closer if we inserted “as was just said” after “that” in (2).) To say that the difference between (1) and (2) is stylistic is not to say that it is unimportant.

  The truth predicate appears in many forms other than the simple, present-tense affirmative found in (2). Consider for instance the following:

  (3a) It is not true that the love of money is the root of all evil.

  (3b) Is it true that the love of money is the root of all evil?

  (3c) It used to be true that the love of money was the root of all evil.

  (3d) If it is true that the love of money is the root of all evil, then it is easier for a camel to go through the eye of a needle, than for a rich man to enter into the kingdom of God.

  If the truth predicate is to be truly redundant, it must be eliminable from such formulations as well. How is it to be eliminated? Well, presumably if (1) and (2) express the same proposition as each other, then their negations express the same proposition as each other, the corresponding interrogatives express the same question, and so on. Thus Ramsey's account provides for the equivalence of (3abcd) to “The love of money is not the root of all evil” and “Is the love of money the root of all evil?” and so on.

  Also we may ask how the truth predicate is to be eliminated from items like the following:

  (4) That's true.

  (5a) That's not true. (5b) Is that true?

  when said in response to another's assertion of (1). But on almost any theory these are presumably equivalent to (2) and (3ab). The same brevity as is achieved by (5ab) can be achieved without “true” as well, thus:

  (6a) It isn't. (6b) Is it?

  Here the anaphoric pronoun “it” and pro-verb “is” have as antecedents the subject “the love…” and predicate “is the root…” of (1).

  Dispensing with “true” is more difficult in the case of what is called blind affirmation or denial or querying, when “true” is applied to a proposition that is not exhibited by displaying a sentence of our language expressing it, but specified in some other way, thus:

  (7a) Radix malorum est cupiditas is true.

  (7b) If 1 Timothy 6:10 is true, then Mark 10:25 is true.

  These may, however, be construed as equivalent to generalizations:

  (8a) Whatever Radix malorum est cupiditas says, it is true.

  (8b) Whatever 1 Timothy 6:10 and Mark 10:25 say, if the former is true, so is the latter.

  though the conversion reads a bit awkwardly in the case of questions.

  But how is “true” to be dispensed with in these or other generalizations? Consider, say,

  (9) If Moses says something, then it is true.

  There are various ways one might claim to be able to express (9) without the use of “true.”

  It is sometimes said that (9) amounts to the huge conjunction of the ascriptions of truth to all propositions expressed by Moses. But this is not so, since one cannot infer that huge conjunction from (9) or vice versa without knowing exactly what things Moses did and didn't say.

  It may be said that what (9) directly expresses can be indirectly suggested by a list of a few hypothetical examples on the order of plus an “et cetera.” But indirect suggestion is not direct expression.

  (10a) If Moses says that Cain slew Abel, then he did.

  (10b) If Moses says that Lilith slew Eve, then she did.

  (10c) If Moses says that there were giants in the earth, then there were.

  (10d) If Moses says that there were angels on the moon, then there were.

  It is sometimes said that (9) amounts to the huge conjunction of all conditionals of type (10abcd) for all sentences of English. But this cannot be so if, as it seems, one can understand (9) without understanding all sentences of English.

  It may be said that much at least of the content of (9) can be expressed using anaphoric expressions as in

  (11) If Moses says that someone did something, then he or she did.

  But (11) does not express the whole content of (9). It covers (10ab) but not (10cd), since “someone did something” and “he or she did” cover only (propositions expressed by) subject-predicate sentences with singular, personal subjects and active, past-tense predicates.

  Ramsey would represent (9) symbolically, and then apply the supposed identity between a proposition and the proposition that it is true to eliminate the truth predicate, thus:

  (12a) p(Moses says p → p is true)

  (12b) p(Moses says p → p)

  But putting (12b) back into words produces nonsense:

 
(13) If Moses says something, then it.

  The problem is grammatical. “For any” demands a noun like “proposition” as complement, but “if…then” demands a sentence like “it is true.” Ramsey thought this a defect of English and other natural languages rather than of his notation, but the long and short of it is that Ramsey does not show that “true” is redundant in English. He only shows that it would become redundant if English were supplemented by his grammar-defying quantifiers. That will be seriously significant only if Ramsey can claim that, though he was speaking English and using the word “true” long before he invented his kind of quantifiers, nonetheless he has an understanding of those quantifiers that is independent of any prior understanding of “true.” Such a claim is easy to make, but difficult to prove; it may be equally difficult to disprove, but the burden of proof is arguably on its proponents, not its opponents.

  Another way to render “true” redundant would be to add to our language prosentences, anaphoric expressions standing to sentences as anaphoric pronouns stand to nouns—in effect, expressions like “he or she did,” but without the limitation of covering only sentences of a certain specific form. Arthur Prior actually coined expressions to serve as prosentences, thus:

  (14) If Moses says that somewhether, then thether.

  (Compare: If Moses says to go somewhither, then go thither.) These expressions even permit a definition of truth: The proposition that somewhether is true iff thether. But again there would be a question whether one can have an understanding of the new “somewhether” and “thether” independent of any understanding of the old “true.” The closest approximation to “somewhether/ thether” in English is perhaps “things are some/that way,” using which the Priorese definition of truth becomes: The proposition that things are some way is true iff things are that way. The definition only works, however, on the understanding that any proposition whatsoever, even one about how things used to be or might have been or ought to be, counts as being about “how things are.”

  3.2 OTHER RADICAL THEORIES

  Historically, the next version of deflationism was the performative theory of P. F. Strawson, a leader of the ordinary language school, which dominated British philosophy in the decades immediately after World War II. It is a professed development of Ramsey's views, and an explicit rejection of Tarski's. On Strawson's view, (2) does not, pace Tarski, express a proposition about the proposition expressed by (1), nor does it, pace Ramsey, express the same proposition as (1). Rather, it does not express a proposition at all. To utter (2) is not to say something about (1), but to do something to (1), namely, to endorse it. In jargon, it is not constative, but performative. (Other ordinary language philosophers said about “good” what Strawson said about “true,” that to say “That's good” about something is not to say anything about it, but to do something to it, namely, to commend it.)

  Strawson gives little attention to examples beyond the simplest. By dropping Ramsey's view that (2) expresses the same proposition as (1), he loses the redundancy view's ability to account for examples like (3abcd). Needless to say, these cannot be regarded as instances of endorsement, though it is still open to Strawson to say that they are cases of doing something to, rather than saying something about: repudiating in the case of (3a), querying in the case of (3b), and so on. Perhaps with (9) one may speak of “blind” and “blanket” endorsement.

  But the need to find different kinds of “performances” (endorsement, repudiation, querying, blanket endorsement) to go with the different logical types of sentences involving “true” (affirmative, negative, interrogative, universal) is an unattractive feature of the theory, and performativism seems incapable of dealing with examples that are really logically complex. Consider, for instance, the following, in which truth is attributed not just generally but so to speak fractionally:

  (15) I don't know what percentage of the things Berlusconi said in his last speech were true, but I suspect it's well under fifty.

  What are we doing to the things Berlusconi said when we assert (15)? (Change “said in his last speech” to “did during his last term in office” and “true” to “good” to see the corresponding problem with performative theories of “good.”)

  For Strawson, (2) endorses rather than saying anything about (1), but one may well wonder why one cannot endorse a proposition by saying something about it. (Isn't that how one endorses most other kinds of things?) In jargon, one may well wonder whether the constative and performative are exclusive. J. L. Austin, who invented the jargon, eventually came to conclude that they are not. In debate with Strawson, he defended a version of the correspondence theory (to be described in §5.2).

  Departing somewhat from historical sequence, the later view closest to Ramsey's has been the prosententialism of Dorothy Grover and associates at the University of Pittsburgh, which dates from the 1970s. Where Prior proposed adding prosentences to English, Grover maintains that English already has them: We don't need “thether,” because, despite the way it is written, “it is true” or “that's true” is such a prosentence already. On this view, applying the truth predicate to a given sentence is equivalent to repeating the sentence itself in the same sense in which using a pronoun whose antecedent is a given noun is equivalent to repeating the noun itself.

  Grover can handle examples like (3abcd) simply by elaborating on the (postulated) analogy between prosentences and pronouns. When the pronoun “he” has “Whistler” as antecedent, saying the inflected and compounded form “he and his mother” of the pronoun is equivalent to saying the correspondingly inflected and compounded form “Whistler and Whistler's mother” of the noun. So likewise, according to prosententialism, if “That's true” has as antecedent “Beer is sold in grocery stores,” then saying “That used to be true, but no longer” is equivalent to saying “Beer used to be sold in grocery stores, but no longer.”

  Grover's account is still dependent, however, on the not especially plausible construal of “blind” assertions like (7ab) as generalizations like (8ab). And prosententialism shares the fundamentally implausible claim of earlier radical theories that “is true” is not only not a genuine predicate in the fullest sense, but not even a separately significant unit, anymore than “soever” in “whatsoever” or “tofore” in “heretofore.”

  These features are avoided by Robert Brandom's neoprosententialism. According to Brandom, “is true,” though not a predicate, is a separately significant unit of a special kind: a prosentence-forming operator. Attached to a phrase denoting a sentence, such as “Proverbs 16:18,” it forms, not a disguised generalization, but an anaphoric expression, “Proverbs 16:18 is true,” expressing the same proposition as the sentence denoted by the original phrase (namely, that pride goeth before a fall).

  Neoprosententialism, however, besides attributing to prosentences a kind of behavior without obvious precedent in the behavior of pronouns, seems to have an implausible implication of its own, namely that

  (16) The most famous conjecture in number theory is true.

  has the same content as Fermat's theorem, which seems wrong, since (16) says something about fame, but Fermat's theorem does not. Further bells and whistles can be added, but nonetheless (neo)prosententialism remains a minority view among deflationists.

  3.3 DISQUOTATION

  Historically, the next important view after Strawson's was the disquotationalism of W. V. Quine, the most influential American philosopher of the 1950s and 1960s. Quine's view counts as “moderate” in the sense that he took the apparent grammar of truth-talk at face value. On his view, quotation turns a sentence into something noun-like, while adding the truth predicate gives back a sentence, and one equivalent to the original, so that we understand what it is to say a sentence is true as well or as poorly as we understand the sentence itself. (If blind people cannot fully understand what sighted people mean by “Snow is white,” the same will be the case for “‘Snow is white’ is true.”) The back-and-forth is needed to express generaliz
ations because quantifier expressions in English demand noun-like expressions as complements. The indispensable role of the truth predicate in expressing generalizations that otherwise could only be suggested by lists of examples plus an “and so on”—compare (9) and (10abcd)—accounts for the appearance of “true” in theses of metaphysics, epistemology, and ethics (among other areas), without our having to suppose that the concept of truth itself is metaphysical or epistemological or ethical.

  Quine distinguishes immanent from transcendent use of the truth predicate, its application “at home” to sentences of our own language, and its application “abroad” to foreign languages. Initially, “at home” has to be understood quite narrowly. Quine was a militant antipropositionalist, and on Quine's account the truth predicate specifically undoes direct, rather than indirect, quotation. Quine thus gives priority to the first over the second of the following:

  (17a) “_______” is true iff ______.

  (17b) It is true that _____ iff _____.