PIERRE
IS T H E R E REALISM
A PATH
JACOB
HALF-WAY
BETWEEN
AND VERIFICATIONISM?*
There exists a traditional distinction, long recognized in the history of philosophy, which has, I think, strong intuitive appeal: the distinction between what makes an ordinary or scientific empirical statement true and our reasons for holding such a statement true. A statement is supposed to be true in virtue of the fact that it corresponds to some state of affairs in the world which it talks about. A statement is held true if, e.g., it is deducible from, consistent with, or supported by other statements that we hold to be true. If one assumes that one cannot compare a statement with some unconceptualized bit of the real world, one will be led to conclude that one's reasons for holding a statement true are restricted to relations holding between statements (such as consistency, contradiction, deducibility, inductive support, etc.). And of course this is consistent with the view that a statement is true if it agrees with some real fact or state of affairs to which it corresponds. In Tarski's semantic conception of truth (not unlike the redundancy theory of truth, in this respect), the truth predicate is seen as an instrument of semantic ascent: to assert that "All men are mortal" is a true sentence is to assert that all men are mortal. The metalinguistic statement has the same assertibility conditions, the same acceptability conditions, the same degree of confirmation, etc. as its simple objectlanguage counterpart. The semantic conception defines truth in terms of satisfaction. Satisfaction is a relation between linguistic and nonlinguistic entities. But since anything (e.g. mental entities) can count as non-linguistic, all contemporary analytic philosophers accept that the semantic conception is consistent with both idealism (or antirealism) and realism. In the 1930's, within the Vienna Circle, there arose a dispute about whether, in order to determine the truth of an empirical statement, one can " c o m p a r e " it with real non-linguistic facts (or states of affairs) or only with other statements. 1 The semantic conception of truth afforded a way out of the dilemma, along the lines of a distinction between the semantic predicate "true" and pragmatic (or epistemic) predicates such as "verified", "refuted", "confirmed", "corrobSynthese 73 (1987) 531-547 t~) 1987 by D. Reidel Publishing Company
532
PIERRE
JACOB
orated", "disproved" and so on, with the use of which we express our attitudes towards statements to which we apply them: we tend to reject a statement we consider refuted; we tentatively accept a statement which has been confirmed by some favorable evidence. As Carnap (1949) said, unlike epistemic or pragmatic predicates (like "confirmed"), the truth of a statement is relative neither to a person, nor to a time, nor to a place, nor to background knowledge. Although it is perfectly sensible to say, e.g., of the statement "John Kennedy will run again in 1964" both that it had a high probability in 1962 and that it was refuted in 1963 (after Kennedy was shot), still it will not do to say that that statement was true in 1962 and became false in 1963. So there is a point on which Carnap agrees with the traditional view: namely, the fact that a statement is true is different from the fact that we hold it true. This distinction has recently come under attack. In what follows, I want to defend it against three criticisms: one verificationist criticism and two pragmatist criticisms. But first I want to draw your attention to one common feature shared by the realist rejoinder and the standard defense of essentialism: a feature Eliott Sober (1982) has aptly called "a declaration of independence" - from the human mind, human thought, or human language. The essentialist argues for the independence of modal features of the world from human thinking or human language. In Carnap's terminology, the realist argues for the independence of semantic facts from epistemological (he might say pragmatic) facts. 2 1. E S S E N T I A L I S M
The pragmatist's distaste for the traditional distinction between truth and rationality or between what makes a statement true and what makes it rational to believe a statement true is reminiscent of his distaste for the idea of necessary truth or that of an essential property of an object, a natural kind or substance. According to essentialism, by the numeral 'five' we refer in any possible world to a number which is odd. By the word 'water' we refer in any possible world to a liquid whose molecular structure is, if chemical thecry is correct, H20. By the word 'gold' we refer in any possible world to a metal with, if chemical theory is correct, atomic number 79. In other words, the essentialist claims that water has, if chemical theory is correct, the essential property: consisting of molecules composed of two hydrogen
REALISM
AND
VERIFICATIONISM
533
atoms and one oxygen atom. Or, what amounts to the same thing, he claims that, if chemical theory is correct, the statement 'water is 1-120' expresses a necessary truth. Now, unlike the advocate of the analytic/synthetic distinction (say Carnap), the advocate of revisionist pragmatism (say Quine) takes all statements to be revisable in principle. Some statements (e.g., logical laws) are located at the center of our conceptual scheme so that it is difficult for us to revise them, for it is difficult for us to imagine an alternative conceptual organization consistent with their negation. Suppose with revisionist pragmatism, that every statement is revisable. Then we cannot maintain that some truths are knowable a priori, that they are necessary, that they are true in every possible world. Or so it seems. In a book on the philosophy of arithmetic, we might encounter the following statement (1): (1)
9 is necessarily greater than 8.
In its desperate attempt to express a necessary truth, is that statement gibberish? The anti-essentialist might charitably offer the following construal: by asserting (1), we communicate the fact that we take the arithmetical proposition expressed by the simple statement without the modality, in virtue of its central location in our conceptual system, as difficult to revise. A person who asserts (1) might be taken to mean that, as a result of the limitations of his or her conceptual resources, he or she lacks the capacity to imagine a conceptual system including an arithmetic consistent with the negation of '9 is greater than 8'. But the revisionist's interpretation is somewhat ambiguous: Is he claiming that, by asserting (1), one is thereby stating the psychological fact that one lacks the capacity to imagine an arithmetic including the negation of '9 is greater than 8' or that it would be irrational, for a being endowed with a mind like his, to accept such an arithmetic? Be that as it may, the revisionist has provided us with only one reading - the only sensible reading, in his view - of (1), the de dicto reading, in virtue of which the property of being a necessary truth is predicated of the proposition that 9 is greater than 8. But surely, the essentialist claims, there is another reading of (1), the de re reading, in virtue of which the number 9 itself is said to be necessarily greater than 8 or to possess the essential property of being greater than 8. At this stage, there are two arguments available to the anti-essen-
534
PIERRE JACOB
tialist. First, though it is true that 9 happens to be the number of planets in the solar system, (2), which follows from (1) by replacing '9' by 'the number of planets in the solar system', on the de dicto reading, is false: (2)
The number of planets in the solar system is necessarily greater than 8.
At least, someone who would want to argue that (2), on the de dicto reading, is true would have to acknowledge that 'necessarily' does not have the same meaning in (1) and (2): the solar system contains nine planets, in virtue of physical, not arithmetical laws. According to the anti-essentialist, his charitable interpretation is the only intelligible reading of (1) for necessity resides in language and furthermore it accounts for why two coreferential expressions are not interchangeable salva veritate. Second, from the fact that, on the de dicto reading, (1) is true, it follows, by existential generalization, that '(3x) (x is necessarily greater than 8)' is true. The anti-essentialist then asks: What is that x that can make the above existential statement true? The number 9. But there are indefinitely many ways we can refer to the number 9: we can, e.g., refer to it either by thinking of its place in the sequence of integers or by the thought that it names the cardinality of the set of planets in the solar system. So we can name 9 accordingly. The singular statement thus predicating of 9 that it is necessarily greater than 8 or that it has the essential property of being greater than 8 will be true on one occasion and false on another occasion. If we cannot attribute to a number an essential property but by using some referring device or other and if the truth of the attribution is a function of the device selected, then is not the very notion of a de re reading incoherent? Except f o r the claim that the de re reading makes no sense, the essentialist has no objection to the observation that two coreferring expressions are not interchangeable salva veritate in modal contexts since (interpreted de dicto) the proposition expressed by the use of one expression is not the same as the proposition expressed by the use of the other. From the standpoint of essentialism, the second anti-essentialist argument is confused. Notice the parallel between the present defense
REALISM
AND VERIFICATIONISM
535
of essentialism and the distinction between holding a statement true and what makes it true. That human beings hold a given proposition as an indubitable truth, even in the presence of confuting experience, is a fact pertaining to the human mind. So is the fact that it might be rational to give up, in some circumstances, a given proposition long held by humans as an indubitable truth. That a proposition is a necessary truth, whatever view (if any) men take of it, is something else entirely. According to essentialism, having the molecular structure H20 might be an essential property of water; being greater than 8 might be an essential property of number 9. And still the existence of the human race might be irrelevant to this. By claiming that the de re reading of the modal statement is unintelligible, the anti-essentialist merely expresses his own prejudice. To the second argument, the essentialist will make the following reply. From the truth of '9 is necessarily greater than 8' and 'The number of planets in the solar system = 9', can one validly infer the truth of 'The number of planets in the solar system is necessarily greater than 8'? According to the anti-essentialist, this is typically an invalid argument from true premises to a false conclusion. As a matter of fact, the conclusion has two possible readings and so does the first premise: a de re and a de ditto reading. De dicto, the conclusion is, as the anti-essentialist says, and unlike the first premise, false. De re, the conclusion says of the number of planets in the solar system (which is the number 9) that it has the essential property of being greater than 8. It does follow from the premises (the first, read de re) and it is, according to essentialism, true. No doubt, it is contingent that there are nine planets in the solar system: there might have been eight. But given that there are nine, then it is an essential property of that number that it comes after 8 in the sequence of integers. In spite of the anti-essentialist protest, 9 has this property essentially, however we refer to it, whether it be by the use of 'the square of 3' or 'the number of planets in the solar system'. According to essentialism, there are two separate kinds of distinctions: on the one hand, there is the distinction between necessary and contingent truths or between essential and accidental properties of an object. On the other hand, there is the distinction between a priori and a posteriori knowledge. According to the essentialist, the revisionist critique of the analytic]synthetic distinction might have hit the Cartesian ideal of unrevisable certainty that had survived in the
536
PIERRE
JACOB
distinction between truths knowable a priori and truths knowable a posteriori. But from the essentialist perspective, unless one specifies the structure of the mind together with the sense-organs of the creature to which knowledge is being attributed, it makes no sense to claim that a particular truth can or cannot be known a priori. Perhaps the minds of Martians are such as to allow them to discover by pure reasoning t h e true laws of the universe. Perhaps Martians should be said to know those laws a priori. Kripke (1972) has claimed that physical identities such as 'Heat is molecular motion' express necessary truths that are discovered by humans a posteriori. Now back to our original distinction, namely that between what makes a statement true and having reasons for holding it true. Let us consider two kinds of cases. First, we might, in light of the fact that we have marshalled good evidence for it, rationally hold true a statement which is actually false. Second, there might exist true propositions which humans have no reason for holding true: cases in which humans cannot marshall good enough evidence for the truth of any statement expressing such a proposition, or cases in which humans never manage to represent it to themselves, through the medium of a sentence of a human language, for lack of the requisite concepts or because they never bring together the constituent concepts. 2.
ELIMINATiVE
PRAGMATISM
According to Richard Rorty (1982), whom I will take to be the typical advocate of what I will call "eliminative pragmatism", we should give up the age-old distinction between the truth of a proposition and our having reasons for believing it to be true, so that a new "postPhilosophical" tradition can emerge. 3 The eliminativist strategy, in the theory of truth, is reminiscent of that of the eliminative materialist in the philosophy of mind, who in turn draws upon the "incommensurability" thesis in the philosophy of science. The picture of scientific change derived from the incommensurability thesis is, I believe, incoherent. The analogy drawn by the eliminative materialist between the fate of words belonging to a specialized vocabulary such as 'phlogiston' or 'caloric' and that of ordinary psychological words such as 'pain', 'belief' or 'fear' is at least controversial. According to the eliminative pragmatist, substantial theories of truth (be they realist or idealist) deserve what will happen to folk psy-
REALISM
AND VERIFICATIONISM
537
chology as predicted by the eliminative materialist: elimination. The eliminative pragmatist bets on the emergence of a new philosophical culture whose members, when they look into themselves, will not share the intutition that an empirical statement is true either in case it fits the state of affairs which it corresponds to or in case it agrees with a body of accepted statements. In that new culture, philosophers will gladly restrict themselves to the redundancy-theory of truth and to concrete historical inquiries devoid of general 'philosophical scope. They will surrender the utopian task of a substantial theory of truth. For they will need the comfort of a substantial theory no more than religion. The fallacious (Platonic) ideal of objectivity, based on an ineffable relation between human language and the inhuman world will be replaced by man's anthropocentric acceptance of his community. From the point of view of the eliminative pragmatist, after many many years, truth, unlike electricity, heredity or numbers, has turned out to be a sterile property, concept or relation about which we have not managed to construct deep, interesting theories. According to Rorty, the pragmatist stands to truth as a secular intellectual stands to God: the latter cannot meaningfully deny the existence of God since he does not understand what the assertion of God's existence means. Consequently, he can only argue for refraining from using a theological vocabulary. If, as he says, the realist assertion that a statement is true if it agrees with the state of affairs to which it corresponds is meaningless, then so is presumably the idealist negation of the realist assertion. Unlike the redundancy-theorist (or disquotation theorist) of truth, the eliminative pragmatist advocates elimination of our truth concept, He is thus exposed to the question: what will b e the successor concept? Is there then no difference between the eliminative pragmatist and the relativist according to whom anything goes? Since he knows that relativism is self-refuting, the eliminative pragmatist tries hard to dissociate himself from the relativist by claiming that his is not another theory of truth (no more subjectivist than realist): he simply wants to change the subject matter. I am not going to argue that by maintaining that there is not and could not be any substantial (or interesting) philosophical theory of truth, the eliminative pragmatist is thereby offering a substantial philosophical theory of truth and hence that his claim is self-refuting.
538
PIERRE
JACOB
On my view, the eliminative pragmatist's mistake is similar in spirit to a mistake once made by Kuhn (1970). Kuhn thought his theory of incommensurability between scientific theories separated by a paradigm change was just another way to express Quine's famous thesis of the indeterminacy of translation. Kuhn's mistake was to identify his view that there is no translation between, say, Newton's and Einstein's mechanics with Quine's view that between any two languages there are infinitely many possible translations and no one is the uniquely correct one. Suppose it is a fact that there are infinitely many correct satisfactions or correspondences or truth schemes for a language, as argued in several places, notably by Putnam (cf. Putnam 1981 and 1983). Endorsing a typically verificationist strategy, the eliminative pragmatist's mistake is to infer from that putative fact the conclusion that there is n o satisfaction or correspondence relation. As Rorty (1985, p. 3) reports approvingly of James' view, "he thought that the moral of philosophers' failure to discover, as it were, the micro-structure of the correspondence relation was that there was nothing there to find". 4 3. P R O O F VS. T R U T H
The paradigm of the realist's distinction between truth and rationality is the Platonist's view of undecided conjectures. Let us consider a mathematical proposition such as Goldbach's conjecture (henceforth GC). It says that any even number greater than two is the sum of two prime numbers. It has never been proved or disproved: it has not been deduced from the axioms of arithmetic nor has a counter-example to it been brought forward. According to Platonist realism, GC has a truth-condition: there is, in the reality.of arithmetic, a state of affairs in virtue of which GC is either true or false. On that view, the truth (or the falsity) of GC is a fact that obtains independent of the human mind (or any other mind, for that matter). The general fact to which GC corresponds - the fact that any even number greater than two is the sum of two primes - obtains (if it does), whatever human mathematicians think about it. On the one hand, the Platonist realist claims that the truth or falsity of an arithmetical proposition which, like GC, is neither proved nor disproved, is independent of any rational method that would help establish it. If the human species had not been selected by evolution,
REALISM
AND
VERIFICATIONISM
539
Goldbach would not have been able to hypothesize his conjecture (perhaps it would never have been). Still the arithmetical state of affairs which makes GC true, would, if GC is true, nonetheless exist, If GC is false, then that state of affairs does not exist. We presently do not know which it is. Perhaps GC is undecidable and humans will never know which it is, unless they can prove that it is undecidable and it follows from the proof that, since there is no counter-example to GC, it is true. In any case, the set of arithmetical facts either contains the state of affairs corresponding to GC or the state of affairs corresponding to its negation. On the other hand, suppose that some day in the future a human mathematician finds a counter-example to GC, thereby proving the falsity of GC. Suppose that other mathematicians check the counterexample and accept the statement that GC is false. Then, according to the Platonist realist, in spite of the fact that the assertion that GC is false would be based on the best available reasons, the assertion might still be false and GC might be true. The alleged counter-example would no doubt be a very very big number, it might after all be the sum of two big prime numbers one of which was mistakenly believed not to be prime. Or there might be a mistake hidden in a lengthy proof of GC, a mistake never detected by human mathematicians, who therefore mistakenly held GC to be true. For Platonist realism, what it is rational to believe - even at the Peircian limit of mathematical investigation - is (or might be) distinct from what is true. An arithmetical proposition (be it singular, existential or universal) is true if it agrees with the numerical fact to which it corresponds and false otherwise. We hold a statement true if we have good reasons (a proof, in the best possible case) for so doing. We know that there are true arithmetical propositions that are neither provable nor refutable inside arithmetic. 4.
THE
VAGUENESS
OBJECTION
On the realist interpretation, although it is neither verifiable nor refutable, an ordinary empirical statement in the past tense such as 'There was an even number of tulips in the Luxembourg garden, in Paris, April 3, 1955' is either true or false: it is true if there was an even number of tulips in the Luxembourg garden, in Paris, April 3, 1955, and false otherwise - that is, if on that day there was an odd
540
PIERRE JACOB
number of tulips in the Luxembourg garden. 5 The realist claims that such a statement has a truth-condition even though, as we shall assume, there is no evidence one way or the other - there is no document left, for example, on the number of tulip-bulbs bought in 1955 by some member (in charge of gardening) of the cabinet of the President of the French Senate located in the Luxembourg garden. The anti-realist will object that the extension of the word 'tulip' is (or the boundaries of our tulip concept are) vague. Do the shoots or the bulbs that did not grow count as tulips? The anti-realist will answer that there is no fact of the matter. One might want to distinguish two issues: one of which is vagueness, the other one the lack of relevant evidence. I take it the anti-realist (say, Dummett) will also deny that bivalence holds of a statement such as 'There was an odd number of certified MDs in room 265 in the Science Center (where the door was closed) at Harvard University, in Cambridge Mass., June 13, 1977', if we have no evidence. Arguably though, even this statement is not devoid of vague predicates. Perhaps apart from pure mathematics, no statement is. In any case, the lack of relevant evidence is, according to the anti-realist, sufficient ground for asserting that there is no fact of the matter. Consider the anti-realist (or verificationist) conditional claim that bivalence does not hold of a statement S about the past, in case we have no evidence one way or the other. What to make of the antecedent of such a claim? Obviously it can only be true of the present. We cannot rule out the possibility that in the future we might discover evidence in favor of the assertion either of tile statement or of its negation. Unless we assume that the statement (A) that we presently have no evidence for or against the truth of S is itself incorrigible or unrevisable. But why assume this? Now, consider what happens if the anti-realist assumes that (A) is revisable. He then assumes that we might after all discover in the future relevant evidence bearing on the truth of S. This is, I take it, in the spirit of the verificationist maxim that truth is verifiability in principle. But verificationist so construed becomes indistinguishable from realism. So if the verificationist wants to take a position really unacceptable to the realist, he has to hold (A) to be incorrigible. But what reason is there for taking such a view? The anti-realist might object that for his view to collapse into realism the body of accepted evidence must include not only present-tense but also past-tense statements. However the
REALISM
AND VERIFICATIONISM
541
view that past-tense statements are not relevant to establishing the truth of a statement about the past (i.e., are not part of the evidence relevant to the truth of such a statement) seems to me not to have very much to recommend itself. Besides the possible lack of relevant evidence, according to him, in the tulip case, the realist runs the risk of a famous paradox - the Sorites' paradox: if a man (say Alfred) can lose one hair without thereby becoming bald, then, according to mathematical induction, no matter how many hairs he loses he will never be bald. According to the anti-realist, the paradox arises if bivalence is assumed through the induction, namely if every statement, saying of Alfred that he is bald (or not), is required to be either determinately true or determinately false at each step of the induction. In other words, the anti-realist argues from the existence of vague predicates to the non-existence of facts or states of affairs which we describe by means of statements containing vague predicates. The realist will first ask what sense there is to the anti-realist claim that there is no fact of the matter corresponding to a statement containing a vague predicate. Does the anti-realist take his own claim to be true? If he does not, the realist will feel justified in ignoring it. If he does, the realist will ask him in what sense the claim is true - what makes it true? Presumably the truth of the anti-realist claim must inherit (upwards so to speak) the vagueness of the initial statement containing the word 'tulip'. In other words, the anti-realist cannot consistently assume that his claim is true in virtue of the fact that it corresponds to some state of affairs, since his claim denies that any such state of affairs exists. So given the fact that vagueness and bivalence are, according to him, inconsistent, he should not assume that bivalence holds of his own claim that there is no fact of the matter making true or false a statement containing a vague predicate. Secondly, according to the realist, when the anti-realist concludes that there is no fact of the matter as to whether a statement containing a vague predicate is true or false, he commits a fallacy. The anti-realist concludes, from the fact that 'bald' and 'tulip' are vague predicates, that there is no fact or state of affairs which makes true the statement, at any given step of the induction, that Alfred is bald or that there was in the Luxembourg garden on April 3, 1955 an even number of tulips. The reason why the realist rejects this inference is this: according to him, to say of a predicate that it is vague is to make a statement about
542
PIERRE JACOB
the predicate's conditions of application or conditions of satisfaction. Those conditions can vary as a function of the context in which one inquires about whether a statement containing a vague predicate is true or whether the predicate is satisfied. They can vary as a function of the intentions, beliefs, and/or background knowledge of the person doing the inquiry. But from the fact that the conditions of satisfaction of a predicate are vague, it follows, according to the realist, that there exists a range of different verification procedures of the fact that the vague predicate is satisfied. So one such procedure must be selected from that range. For a botanist, what would count as a tulip on April 3, 1955 can diverge from what would so count for the man in the street. But, accordin~ to realism, it does not follow that there is no fact (or state of affairs)which makes it true (or false) that there was an even number of tulips in the Luxembourg garden on April 3, 1955. From the realist perspective, what follows is that, in order to know the state of affairs that makes true a statement containing a vague predicate, one must first fix the interpretation of the predicate. Once the interpretation is fixed, then it can be seen that there is a state of affairs corresponding to the statement containing a vague predicate in virtue of which it is true or false. The anti-realist can conclude that there is no fact of the matter as to whether there was an even number of tulips in the Luxembourg garden April 3, 1955, only if he assumes in addition that there is no fact bearing on the divergence between the botanist and the man in the street. But the realist rejects such an assumption on the ground that there are reasons why one should prefer the botanist's view of what counts as a tulip rather than that of the man in the street, in case they disagree. For the realist assumes that the botanist's knowledge of tulips is better (more rational) than the man in the street's. Much the same can be said of the 'bald' case. The realist will argue that, as a matter of fact, we have not developed a specialized knowledge of hair and baldness comparable to the botanist's knowledge of tulips, for lack of interest in the subject matter. But it is conceivable that other creatures (say Martians) might. So the realist is committed to an epistemic treatment of vagueness. 5. H A L F W A Y B E T W E E N
REALISM AND VERIFICATIONISM?
According to Putnam (1981), whom I take to be the prototypical representative of rationalist pragmatism, there are two competing
REALISM
AND
VERIFICATIONISM
543
philosophical perspectives - an externalist and an internalist perspective. They depend on two different intuitions: the realist intuition according to which a statement is made true by a state of affairs to which it corresponds and the pragmatist intuition according to which talk of comparing a statement with unconceptualized reality is meaningless. As noted above, the externalist intuition will only conflict with the internalist's if the externalist insists that what makes a statement true is not just some real state of affairs but some unconceptualized state of affairs. But the realist is not, it seems to me, bound to make that assumption. Unlike the eliminative pragmatist, his rationalist counterpart has no intention of changing the subject matter: his goal is to contribute his own substantial theory of truth located half-way between realism and verificationism. According to Platonist realism, GC is made true or false by an arithmetical state of affairs; whether or not we humans have the capacity to prove or disprove it. Although the best human mathematicians might rationally hold it to be false (alternatively true), it nevertheless might be true (alternatively false). According to rationalist pragmatism, akin in that respect to constructivism or verificationism, this step is unintelligible: if the gap between the truth of a statement and our having reasons for holding it to be true is as big as that, given the fact that we only have a handle on the latter or that our grasp of the former is derivative on the latter, then truth becomes utterly mysterious. If what is rationally verifiable (or acceptable) cannot give us a clue as to what truth is, nothing will. This is not to say that the rationalist pragmatist wants to reduce truth to justification (or define the former in terms of the latter, see note 7). Rather, truth and justification feed on each other. According to rationalist pragmatism, the realist idea (typical of the externalist perspective) that, "at the limit of scientific investigation"(in Peirce's words), the most rational theories that would be available to humans could turn out to be false is unintelligible for it presupposes a view-point which we humans cannot adopt or imagine - namely God's or a no Eye point of view. So what is the rationalist pragmatist's alternative? By Putnam's own admission, there might not be a unique answer for " ' t r u t h ' (idealized Justification) is as vague, interest relative, and context sensitive as we are". 6 It might depend on what statement one is thinking about. Presumably ordinary empirical truths should be distinguished from mathematical and other truths from theoretical
544
PIERRE
JACOB
science. If, to take one of Putnam's own examples, I say 'There is a chair in my office right now', the ideal acceptability conditions for such a statement involve being present at the time of utterance, and being in normal sensory (not drugged or asleep) conditions. The Peircian limit of scientific investigation might well be irrelevant to verifying the truth of such a statement. Still, I would argue, the fact that such a statement is true is different from the fact that I am ideally justified in accepting it. What about a theoretical statement? According to strict verificationism (say, Dummett 1978), the truth of a statement just is verification. Rationalist pragmatism, however, accepts the Carnapian argument that, unlike epistemic predicates, the truth of a statement is relative neither to a person, nor to a time, nor to background knowledge; it is a kind of property a statement cannot lose if it has it (see Putnam 1981, p. 55). So truth is not, according to rationalist pragmatism, what it is rational to hold true hic et nunc, for the latter can change with the progress of science. Therefore "truth is an idealization of rational acceptability".7 There are statements whose being true the rationalist pragmatist might identify with their being rationally acceptable at the Peircian limit of scientific (or rational) investigation. GC might be such a statement and so might other truths of physics or theoretical parts of science. Let time t be far removed in the future and not long before t' when the human species disappears from the surface of the earth. If between t and t' we humans have good reasons to believe some theoretical statement p to be true (or to assert p) and we do not acquire any reason to believe the truth of (or to assert) not-p, then, according to rationalist pragmatism, nothing rational can prevent us from considering p to be true. From a pragmatist standpoint, there would be no sense to assume, as realism would have us do, that although it is rational to hold p true, nonetheless p might be false. Has the advocate of rationalist pragmatism opened for us a viable interpretation of the truth of theoretical statements in between realism and verificationism; I would say that he succeeded only if his idealization can work. He claims that the idea of unconceptualized reality (independent of the human mind) as something against which to check the truth of our statements is unthinkable by a human mind. From that he concludes that he can make no sense of an idealization whose job would be to abstract away from the presence of the human mind and unveil reality as it is. Consequently, he favors an epistemic idealization
REALISM
AND
VERIFICATIONISM
545
designed to abstract away from non-rational aspects of the human mind and unveil rationality. His idealization is supposed to reveal, not reality as it is, but scientific rationality purged of any disturbing factor that might alter its functioning. But it is not clear how one idealization is less mysterious than the other. If the ideal of reality that is unveiled at the limit of human scientific investigation makes no sense at all, then, it seems to me, so does the idea of what we would be justified believing true at the limit of scientific investigation. If, on the other hand, what it is rational to hold true at the limit of scientific investigation makes any sense, so does the idea of reality unveiled at the limit of scientific investigation. Therefore I claim that the pragmatist's idealization on rational acceptability is no worse and no better than his realist counterpart. Unless the pragmatist can show that the human mind is in a better position to get some insight into a property of itself, such as optimal rationality, than into reality - perhaps it has direct introspective access or perhaps it is a case of self-knowledge and it has special authority. But I doubt it. For is not the mind a part of reality? Furthermore it is by no means obvious to me how the advocate of rationalist pragmatism can really deny the intelligibility of the realist view that at the limit of scientific investigation our best theories might be false. If he rejects that view, he must identify the truth of a theoretical statement (say GC) with what our best theory ultimately makes it rational to believe. But to assume that such a truth is what our best theory will say presupposes that there is one such theory. And this is just an assumption that Putnam (i981, p. 49) attributes to metaphysical realism and that according to him the internalist ought to reject - the assumption that "there is exactly one true and complete description of 'the way the world is' ,,.8 According to realism, then, the fact that we do or do not have proper reasons for holding a statement true is one thing. What makes a statement true is something else. The mistake of the eliminative pragmatist, I have argued, is that from the fact that there are many formally correct correspondence relations for a given language, he concludes that there is nothing to the idea of correspondence. For the strict verificationist to make his view a genuine alternative to realism he has to assume that his claim that there is no evidence one way or the other in favor of the truth of a particular statement is incorrigible. For the realist, insofar as he avails himself of talk of the Peircian limit
546
PIERRE
JACOB
of scientific investigation, our most rational theories could turn out to be false. From the fact that the rationalist pragmatist's epistemic idealization is neither more nor less intelligible than the realist's (if it is a fact), and from the fact that the pragmatist accepts the Carnapian argument against the identification between truth and verification hic et nunc, it follows, I think, that rationalist pragmatism collapses into realism. So there is no real difference between the former and the latter. NOTES * I am grateful to a number of people. To Ettore Casari, Maria Luisa Dalla Chiara for having invited me to the Conference on Theories of Meaning. To Ermanno Bencivenga, Richard Carter, Daniel Dennett, Robert M. Harnish, Jaakko Hintikka, Paul Horwich, Charles Larmore, Massimo Piattelli-Palmarini, Dag Prawitz, Hilary Putnam, Francois Recanati, Charles Travis and Bas Van Fraassen for comments on this paper. l Both camps in the controversy seem to have confused issues about what truth is and issues about what makes it rational to hold a statement true. 2 This kind of distinction is elegantly expressed by Putnam (1978, p. 99) in terms of a distinction between a theory of understanding a language and an account of the contribution of language to the success of global behavior: " A number of tools have this feature: that the instructions for use of the tool do not mention something that explains the successful use of the tool". So "flip the switch" says how to turn the light on without mentioning electricity. But reference to electricity will be part of the explanation of why switching the flip succeeds in turning the light on. Strawson (1976, p. 16) urges a similar distinction between the fact that "a rational speaker's grasp of his language is manifested in, inter alia, his responding in certain ways to the recognizable situations with which he is and has been confronted" and the fact that "the rational speaker's response to such situations (might) be governed by a certain kind of conception - a conception of a state of affairs, of a condition of truth, which, for one reason or another, in fact or in principle, is not, or is no longer, or is not for the speaker, accessible to direct observation or memory". Those distinctions tit realism as I understand it. 3 According to Rorty (1982, Introduction), there is a major division between pragmatism and the Platonic tradition. Philosophy (with a capital P) offers Platonic answers to Platonic questions (about the True, the Good, etc.). Pragmatists are eliminativists in that they typically try to dissolve Platonic questions by exposing them as pseudoquestions. 4 I am grateful to Ernest Lepore for drawing my attention to and providing me with a copy of Professor Rortv's (1985). 5 If zero is an even number,, then presumably the statement is true if on that day there was no tulip in the Luxembourg garden, as noted by Ermanno Beneivenga. 6 See e.g. Putnam (1983, p. 280): "In my view, truth is idealized justification (the true is what would be justified under optimal conditions, where optimal conditions depend on the particular assertion, context, and interests in complex ways)". For similar statements, see Putnam, 1983, p. xvii or p. 231. 7 Although at the conference Putnam argued that his view entailed neither that there is a unique kind of rational acceptability for all statements nor that talk of Peircian limits
REALISM AND VERIFICATIONISM
547
was (always or ever?) relevant to truth, it seems to me that the following passage does suggest that Peircian limits of rational investigation are relevant to the truth of theoretical scientific statements: "We speak as if there were such things as epistemically ideal conditions, and we call a statement 'true' if it would be justified under such conditions. 'Epistemically ideal conditions', of course, are like 'frictionless planes': we cannot really attain epistemically ideal conditions, or even be absolutely certain that we have come sufficiently close to them. But frictionless planes cannot really be attained either, and yet talk of frictionless planes has 'cash value' because we can approximate them to a very high degree of approximation" (Putnam 1981, p. 55). He then adds "Perhaps it will seem that explaining truth in terms of justification under ideal conditions is explaining a clear notion in terms of a vague one. But 'true' is not so clear when we move away from such stock examples as 'Snow is white'. And in any case, I am not trying to give a formal definition of truth, but an informal elucidation of the notion" (Putnam, 1981, p. 56). Besides the anti-reductionist flavor, I take this passage to allude to truths from theoretical science. For what could such ideal epistemic conditions be but, as the simile to frictionless planes implies, some Peircian limits of rational investigation? 8 Since Putnam has presumably an overall physical theory in mind, GC in the present paragraph should be replaced by some statement of theoretical physics. But the spirit of the argument would not be altered. REFERENCES Carnap, R.: 1949, 'Truth and Confirmation', in H. Feigl and W. Sellars (eds.), Readings in Philosophical Analysis, Appleton-Century-Crofts, New York. Dummett, M.: 1978, Truth and Other Enigmas, Harvard University Press, Cambridge, Mass. Kripke, S.: 1972, 'Naming and Necessity', in D. Davidson and G. Harman (eds.), Semantics of Natural Language, Reidel, Dordrecht. Kuhn, T. S.: 1970, 'Reflections on my Critics', in I. Lakatos and A. Musgrave (eds.), Criticism and the Growth of Knowledge, Cambridge University Press, Cambridge. Putnam, H.: 1978, Meaning and the Moral Sciences, Routledge & Kegan Paul, London. Putnam, H.: 1981, Reason, Truth and History, Cambridge University Press, Cambridge. Putnam, H.: 1983, Realism and Reason, Philosophical Papers, vol. III, Cambridge University Press, Cambridge. Rorty, R.: 1982, Consequences of Pragmatism, Harvester Press, Brighton, Sussex. Rorty, R.: 1985, 'Pragmatism, Davidson and Truth', to appear in a volume edited by E. Lepore to honor Professor Davidson. Sober, E.: 1982, 'Realism and Independence', Nods 16, 369-85. Strawson, P. F.: 1976, 'Scruton and Wright on Anti-Realism', Proceedings of the Aristotelian Society 77, 15-21. UA 1484 CNRS S6minaire d'l~pist(~mologie Comparative Universit6 de Provence 29 Avenue Robert Schuman 13621 Aix-en-Provence France