paul horwich, truth and meaning, chapter 3

22
7/21/2019 Paul Horwich, Truth and Meaning, chapter 3 http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 1/22 Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 35 3  A Defense of Minimalism My aim in this paper is to clarify and defend a certain ‘minimalist’ thesis about truth: roughly, that the meaning of the truth predicate is fixed by the schema, ‘The proposition that p  is true if and only if p’.¹  The several criticisms of this FN:1 idea to which I wish to respond are to be found in the work of ¹  For a thorough discussion of the relationship between  deflationism  and minimalism about truth, see ‘‘Varieties of Deflationism’’—Ch. 2 of the present volume. Buthere, for convenience, is a summary. Deflationism isthe somewhat vagueideathattruthisnota‘substantive’property,thatnoreductivetheoryofit should be anticipated, and that our grasp of the truth predicate comes from our appreciation of the trivial way that each statement specifies its own condition for being true. But philosophers who sympathize with this general point of view disagree amongst themselves about how best to elaborate it. Minimalism is one such strategy—the one defended here and previously articulated in my Truth  (2nd edn., Oxford University Press, 1998). Besides minimalism, the main alternative forms of deflationism about truth are: (1)  disquotationalism, according to which sentences (rather than propositions) should be regarded as the bearers of truth, and the schema, ‘‘p’’ is true  ↔ p, will be the core of what defines the truth predicate; (2)  prosententialism, which denies that ‘‘true’’ is a genuine (logical) predicate, and which stresses instead the analogy between a pronoun and ‘‘That is true’’ (insofar as both inherit their content from some other contextually salient expression); (3) the redundancy theory ,  whereby ‘‘The proposition that p is true’’ means exactly the same as ‘‘p’’; (4)  the sentence-variable analysis , which analyses truth-talk in terms of quantification into sentence positions—‘xistrue’means‘( p)(x =<p>&p)’;and(5) Tarski’s theory , which explains the truth of each sentence of a language in terms of the referential properties of its components (characterized diquotationally) and the logical structure in which they are embedded. The relative advantages of the minimalist version of deflationism cannot be fully spelled out here. But, in a nutshell, its merits are that (a) it deals with our  actual  concept of truth, rather than some allegedly superior one; (b) it does not attempt to explain truth in terms of notions that should themselves be explained in terms of truth (e.g. substitutional quantification); (c) it recognizes that there is no call for an

Upload: deborahfenster

Post on 12-Apr-2018

233 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 1/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 35

3

 A Defense of Minimalism

My aim in this paper is to clarify and defend a certain

‘minimalist’ thesis about truth: roughly, that the meaning of the truth predicate is fixed by the schema, ‘The propositionthat p  is true if and only if p’.¹  The several criticisms of this

FN:1

idea to which I wish to respond are to be found in the work of 

¹  For a thorough discussion of the relationship between   deflationism   andminimalism about truth, see ‘‘Varieties of Deflationism’’—Ch. 2 of the presentvolume. But here, for convenience, is a summary. Deflationism is the somewhatvague idea that truth is not a ‘substantive’ property,that no reductivetheory of itshould be anticipated, and that our grasp of the truth predicate comes from ourappreciation of the trivial way that each statement specifies its own conditionfor being true. But philosophers who sympathize with this general point of 

view disagree amongst themselves about how best to elaborate it. Minimalismis one such strategy—the one defended here and previously articulated in my Truth   (2nd edn., Oxford University Press, 1998). Besides minimalism, themain alternative forms of deflationism about truth are: (1)  disquotationalism,according to which sentences (rather than propositions) should be regardedas the bearers of truth, and the schema, ‘‘p’’ is true  ↔  p, will be the coreof what defines the truth predicate; (2)   prosententialism, which denies that‘‘true’’ is a genuine (logical) predicate, and which stresses instead the analogy between a pronoun and ‘‘That is true’’ (insofar as both inherit their contentfrom some other contextually salient expression); (3)  the redundancy theory , whereby ‘‘The proposition that p is true’’ means exactly the same as ‘‘p’’; (4) the sentence-variable analysis , which analyses truth-talk in terms of quantificationinto sentence positions— ‘x is true’ means ‘(∃p)(x =<p>& p)’;and(5) Tarski’s theory , which explains the truth of each sentence of a language in terms of thereferential properties of its components (characterized diquotationally) and thelogical structure in which they are embedded. The relative advantages of the minimalist version of deflationism cannot be fully spelled out here. But,in a nutshell, its merits are that (a) it deals with our actual  concept of truth,rather than some allegedly superior one; (b) it does not attempt to explaintruth in terms of notions that should themselves be explained in terms of truth(e.g. substitutional quantification); (c) it recognizes that there is no call for an

Page 2: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 2/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 36

36   A Defense of Minimalism

Davidson, Field, Gupta, Richard, and Soames, and in a classicpaper of Dummett’s. But before addressing these criticisms letme begin by saying something more about the thesis itself.

Consider biconditionals like

<snow is white> is true ↔ snow is white

and

<lying is wrong > is true ↔ lying is wrong ²FN:2

—that is, instances of the equivalence schema 

<p> is true ↔ p

It can be argued that such biconditionals are  epistemologically  fundamental —we do not arrive at them, or seek to justify ouracceptance of them, on the basis of anything more obvious ormore immediately known. It can be argued, in addition, thatour underived inclination to accept these biconditionals is thesource of  everything else   we do with the truth predicate. Forexample, from the premises

 What he said is that he was abducted

and

 What he said is true

 we are prepared to infer

He was abducted

This particular use of the word ‘‘true’’ is explained by supposing that we first employ Leibniz’s Law to get from our pair of premises to

<He was abducted> is true

explicit  definition of truth; (d) it can countenance the attribution of truth topropositions whose logical forms we do not know; and (e) it does justice to therole of truth as a device of generalization.

²   ‘‘<p>’’ abbreviates ‘‘the proposition that p’’; and ‘‘↔’’ is the   material biconditional.

Page 3: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 3/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 37

 A Defense of Minimalism   37

and then invoke the relevant instance of the equivalenceschema. And, more generally, it can be made plausible that nofurther fact about the truth predicate—nothing beyond ourallegiance to the equivalence schema—is needed to explainany  of our ways of using it. It is for this reason that we areentitled to conclude that the meaning of ‘‘true’’ is determinedby that schema. For, plausibly, the property of a word thatconstitutes its having the particular meaning that it has shouldbe identified with the property that explains the   symptoms of its possessing that meaning—and these symptoms are the

various characteristic ways in which the word is used.³  ThusFN:3

my minimalist thesis is the product of two prior claims: first,that our underived endorsement of the equivalence schema is explanatorily fundamental with respect to the overall useof the truth predicate; and second, that the meaning of any 

 word is engendered by the fact about it that explains itsoverall use.

This line of thought can be challenged at various points andno doubt stands in need of considerable further support.⁴ But

FN:4

my main aim here is not to defend my  route  to the minimalist

conclusion, but rather to defend that conclusion itself: namely,that the meaning of ‘‘true’’ stems from the equivalence schema.

³  This view of how meaning-constituting properties are to be identified isan instance of the general idea that an underlying property U constitutes a relatively superficial property S when U’s being co-extensive with S explains why possession of S has the symptoms that it does. My speaking of S as being constituted  by U in those circumstances, rather than as being  identical   to U,requires a fine-grained conception of ‘property’ whereby two predicates standfor the same property only when they have the same meaning. This way of speaking does not preclude also deploying a more coarse-grained conception whereby, if fine-grained property U constitutes fine-grained property S, theneach of them is associated with a single coarse-grained property.

⁴  For justification of the claim that the equivalence schema is explanatorily fundamental, seeTruth, pp.50– 1.For justificationof the use theoryof meaning,see my  Meaning  (Oxford: Clarendon Press, 1998, ch. 3), Reflections on Meaning (New York: Oxford University Press, 2004, ch. 2), and ‘‘Regularities, Rules,Meanings, Truth Conditions, and Epistemic Norms’’—ch. 7 of the presentvolume.

Page 4: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 4/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 38

38   A Defense of Minimalism

For most of the recent objections to this thesis do not targetany particular rationale for it, but purport to demonstrate thatthe thesis itself cannot be correct. However, before addressing these objections, let me help to prepare the ground for my replies to them by saying a little more to clarify just what theproposal is, and is not, intended to encompass.

Several different kinds of theory, with very different ex-planatory objectives, might appropriately be labeled ‘‘theoriesof truth’’. So it is important to be clear about what sort of the-oretical work the minimalist proposal is not meant to do and

should not be blamed for failing to do. In the first place, it is notintended to provide an   explicit definition  of the word ‘‘true’’,neither descriptive nor stipulative; it does not offer a way of re-articulating the contents of sentences containing the word;indeed, it implies that no such reformulations are possible.⁵ In

FN:5

the second place, the proposal does not amount to a substantivereductive theory  of the property of being true—something inthe style of ‘water is H2O’—which would tell us how truthis constituted at some underlying level. Again, it suggests thatthe search for such a theory would be misguided. And in the

third place, it is not a ‘theory of truth’ in the sense of a setof fundamental theoretical postulates on the basis of whichall other facts about truth can be explained.⁶  Its immediate

FN:6

⁵  To claim that ‘‘x is true  ↔ x is F’’ is the explicit definition of the truthpredicate (where ‘‘F’’ might be replaced with ‘‘in correspondance with reality’’,‘‘verifiable’’, ‘‘useful’’, etc.) is to claim that our acceptance of such a principle isexplanatorily fundamental with respect to our overall use of that predicate. Butsuch a claim is incompatible with the minimalist thesis according to which it israther our endorsement of the equivalence schema that is explanatorily basic.

⁶  There is, however, a plausible theory of truth itself that is closely affiliated with the minimalist account of the meaning-constitution of the word, ‘‘true’’. Arguably, the axioms of the fundamental theory of truth itself —those that will provides the best explanation (i.e. simplest derivation) of all facts abouttruth—are instances of ‘<p> is true ↔ p’. For (a) such axioms would appearto suffice (in conjunction with theories of other matters) to explain every otherfact about truth; and (b) it is hard to imagine a simpler body of principles onthe basis of which those instances could themselves be explained. For furtherdiscussion see Truth, pp. 25–31, 50–1.

Page 5: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 5/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 39

 A Defense of Minimalism   39

concern is with the word  ‘‘true’’ rather than with truth itself. Itpurports to specify which particular non-semantic fact aboutthat word is responsible for its meaning what it does; and thefact it so specifies, roughly speaking, the role of that word inthe equivalence schema.

Now let me turn to an array of objections. I will look atone difficulty raised by Hartry Field, three devised by AnilGupta, one due to Mark Richard, a couple that I put to myself,an old but still influential objection of Michael Dummett’s,and two posed by Donald Davidson. My discussion of each

of these problems will be brief—merely indicating the linesalong which I think the response should be given, rather thangiving it in full. In some cases these responses will be somwhatconcessive—involving certain significant adjustments of theminimalist thesis.

Objection 1: The minimalist proposal would leave it mysterious how  we are able to attribute meanings to sentences that predicate truthof  untranslatable   foreign statements. If an utterance V is knownto mean (say)   that dogs bark , then (according to the proposal)the sentence ‘‘V is true’’ (or ‘‘V expresses a truth’’) might be

interpreted as saying roughly   that dogs bark . However, given anutterance, U, that cannot be translated into our language, theproposal enables us to attach no meaning at all to ‘‘U is true’’—eventhough it surely  would  be meaningful. So the minimalist proposal isdefective.

(Field⁷)FN:7 

But, as we have stressed from the outset, the minimalistform of deflationism does not offer an explicit or contextualdefinition of ‘‘true’’: it does not purport to provide a way of reformulating or re-articulating the content of each sentence

⁷   Field, H. ‘‘Critical Notice: Paul Horwich’s   Truth’’,   Philosophy of Sci-ence   59 (1992) 321–30. The importance of being able to attribute truthto untranslatable statements has been emphasized by Stewart Shapiro. Seehis ‘‘The Guru, the Logician, and the Deflationist’’,   Noûs   37 (2003),113–32.

Page 6: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 6/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 40

40   A Defense of Minimalism

containing the word ‘‘true’’. On the contrary, it insists that nosuch a thing is possible. So one cannot reasonably complainthat the minimalist proposal fails to yield a conceptual analysisof the sentence ‘‘U is true’’.

One certainly  could  complain if the proposal implied thatthis sentence lacked meaning—but it has no such implication.It aims to specify the underlying use-property in virtue of 

 which the truth predicate means what it does. To that end, itidentifies certain deployments of that predicate as explanatorily fundamental and hence meaning-constituting— namely, those

that appear in instances of the equivalence schema. But othertokens of the word may perfectly well have the very samemeaning, as long as their deployment is partially explained by the fundamental ones. Thus someone who reasons inductively to ‘‘U expresses something true’’ on the basis of the factthat the other assertions of the speaker—those that   can   betranslated—have turned out to be true (where that dependson the equivalence schema), uses the truth predicate with a constant meaning, one that is engendered by the schema.

Objection 2: The equivalence schema is not strong enough to

identify the meaning of the truth predicate, because exactly parallelschemata are satisfied by predicates that do not mean the same as‘‘true’’. For example, instances of the schema, ‘<p> is true-and-not-red↔ p’, are just as obviously correct as instances of the equivalenceschema (since, obviously, no proposition is red). But ‘‘is true and notred’’ is not a strict  synonym of the truth predicate. More generally,the schema, ‘<p> is f ↔ p’, will be endorsed relative to a variety of predicates, ‘‘f’’, that posssess somewhat different meanings from oneanother; therefore, for no given ‘‘f’’, can it be that our acceptance of instances of ‘<p> is f ↔ p’ is what fixes ‘‘f’’ ’s meaning.

(Gupta ⁸

)FN:8

⁸   This objection was put to me by Gupta in October 1992, and appearsin n. 17 of his ‘‘Deflationism, the Problem of Representation, and Horwich’sUse Theory of Meaning’’, Philosophy and Phenomenological Research 67, (Nov.2003), 654–66.

Page 7: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 7/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 41

 A Defense of Minimalism   41

Indeed. However, according to minimalism, what fixes themeaning of the truth predicate is   not   merely our   allegiance to the equivalence schema but, in addition, the fact that thisallegiance is the use that is explanatorily fundamental—i.e.the fact that our endorsement of the equivalence schema isthe   basic law of use   for ‘‘true’’, accounting for its overalldeployment. Certainly there are parallel schemata, constructed

 with other predicates in place of ‘‘true’’, that are no lessacceptable. However, in every such case our commitmentto the schema is not what explains the predicate’s overall

deployment; rather, that commitment is itself explained by thepredicate’s meaning-constituting use.

For example, our acceptance of instances of ‘<p>  is glub↔   p’— where ‘‘glub’’ abbreviates ‘‘true and not red’’— isa consequence of three more basic commitments: (i) ourendorsing the equivalence schema for ‘‘true’’, (ii) our accepting ‘‘No proposition is red’’, and (iii) our accepting ‘‘x is glub ↔x is true and not red’’. The meaning of ‘‘glub’’ is given by (iii),rather than by the ‘glub’-schema. Only in the case of the  truthpredicate does the schema capture what is explanatorily basic

in our usage of the predicate. So only in that  case is the schema meaning-constituting.

Objection 3: The minimalist proposal implies  either  that the word‘‘true’’ will never be fully understood  or   that the meaning of eachperson’s truth predicate depends on, and varies with, whateverelse is in his vocabulary. For the proposal is tantamount to thedefinition:

x is true   ≡   [x =<dogs bark > & dogs bark; orx =<pigs fly > & pigs fly; or. . . and so on]

Therefore, if the ‘‘and so on’’ is intended to cover  all   propositions,then— since some of them must involve concepts that no-onepossesses—the meaning of ‘‘true’’ will not be fully known toanyone. And if, alternatively, the definition of each person’s truth

Page 8: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 8/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 42

42   A Defense of Minimalism

predicate is supposed to cover only those propositions he himself can currently grasp, then, as new concepts are deployed and new 

terms coined, his definition of ‘‘true’’ will change. But neither of these alternative implications of minimalism is acceptable. Surely our understanding of the truth predicate is both complete andconstant.

(Gupta ⁹)FN:9

 Agreed. But my proposal is perfectly consonant with suchintuitions, because it is not at all equivalent to the abovealleged definition. As already emphasized, the minimalist thesisdoes not offer anything like an explicit definition. Rather itpurports to specify the fact of usage that provides the truthpredicate with its meaning. That fact of usage, it claims, is theexplanatory role of our inclination to accept instances of theequivalence schema—a fact that remains fixed as the rest of our language evolves. So, for example, at the moment that theterm, ‘‘tachyon’’, enters our language, we become inclined toaccept

<tachyons go backwards in time> is true↔ tachyons go

backwards in timeBut this is merely one more application of a single andinvariable regularity—our inclination to accept any instanceof the schema that we understand. That inclination precededthe introduction of the term ‘‘tachyon’’ and was in no way altered by it. Thus the minimalist thesis implies neither thatthe meaning of the word ‘‘true’’ can’t be fully grasped, nor thatit changes with expansions of our vocabulary.¹⁰

FN:10

⁹  Gupta, A. ‘‘A Critique of Deflationism’’,  Philosophical Topics  21 (1993)57– 81; and his ‘‘Minimalism’’, Philosophical Perspectives  7 (1993) 359–69.

¹⁰  A further objection of Anil Gupta’s—one that I do  think is correct—isthat our underived endorsement of the equivalence schema will not explainour refusal to apply truth to things such as Julius Ceasar (i.e. our confidentacceptance of sentences like ‘‘Julius Caesar was not true’’). To accommodate thispoint we can suppose that the explanatorily-basic, meaning-constituting factsabout ‘‘true’’ include, not merely our underived allegiance to the equivalence

Page 9: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 9/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 43

 A Defense of Minimalism   43

Objection 4: Our reliance on the equivalence schema will not sufficeto explain our commitment to  general  facts about truth. Consider,

for example, ‘‘All propositions of the form, <p→ p>, are true’’. Nodoubt our particular  logical convictions of that form, together withour acceptance of the equivalence schema, can explain,  for any single 

 proposition, why we take it to be true that this proposition impliesitself. Thus we can explain, given our logical commitment to ‘‘dogsbark → dogs bark’’, why we also accept ‘‘The proposition that dogs bark → dogs bark  is true’’. But we have not thereby explained how the above  generalization is reached. Therefore our allegiance to theequivalence schema does not really suffice to account for  all   usesof the truth predicate. So that practice does not fix the meaning of ‘‘true’’, contrary to what the minimalist maintains.

(Gupta, Soames¹¹)FN:11

Granted, if all uses of ‘‘true’’ are to be accounted for, it willnot be enough merely  to cite our allegiance to the equivalenceschema. Further explanatory premises will be needed. Butthis is an obvious and familiar point. Note, for example, thereference above to our acceptance of ‘‘dogs bark → dogs bark’’in explaining why we accept ‘‘The proposition that dogs bark →

dogs bark  is true’’. That is perfectly consistent with minimalismbecause the further explanatory factor makes no mention of the word ‘‘true’’. So perhaps the present objection can be defusedin the same way?

In other words, it remains to be seen whether or not the extra factors needed to account for our acceptance of  generalizations concern the word ‘‘true’’. Only if they do can one conclude

schema, but also our underived acceptance of the principle, ‘‘Only propositionsare true’’.

¹¹  Gupta, A. ‘‘Minimalism’’. This objection has also been forcefully articu-lated by Scott Soames in his ‘‘The Truth About Deflationism’’, E. Villanueva (ed.) Philosophical Issues  8, Atascadero, Cal., Ridgeview Publishing Company,1997. See also ‘‘Some Remarks on Deflationism’’ (unpublished) by PaoloCasalegno (University of Milan). A version of the problem was raised by Tarskiin sect. 5 of ‘‘The Concept of Truth in Formalized languages’’, Logic, Semantics, Metamathematics: Papers from 1923 to 1938 , (Oxford: Oxford University Press,1958).

Page 10: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 10/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 44

44   A Defense of Minimalism

that the explanatorily adequate regularities of use governing ‘‘true’’ must go beyond our underived acceptance of ‘‘<p>is true  ↔ p’’—so that minimalism is defective. But we havebeen given no reason to think that this is so. On the contrary,there is good reason to suspect that it isn’t so.

Suppose it were the case that whenever anyone is disposedto hold, concerning each F, that it is G, then he comes, on thatbasis, to believe that every F is G. Our disposition to accept, foreach proposition of a certain form, that it is true would thensuffice to explain our acceptance of the generalization, ‘‘Every 

proposition of that form is true’’.Now this particular response to the objection can’t be right

as it stands, because the proposed extra explanatory premiseis glaringly incorrect. It is   not   always the case that having shown, for each F, that it is G, one will inevitably come tothe belief that all Fs are G. After all, such demonstrationsmay well coincide with the mistaken conviction that notall the Fs have been considered. Imagine, for example, thatsomeone mistakenly suspects that there are mountains higherthan Everest. He might nevertheless be able to show, of every 

actual mountain, that it is no higher than Everest; but hedoes not believe the generalization that all mountains have thisproperty.

However, a modified version of this strategy is much morepromising. Let us restrict the proposed extra premise to kindsof entity, F, and properties, G, that satisfy the following condition:

 We cannot conceive of there being additional Fs—beyondthose Fs we are disposed to believe are G—which we

 would not have the same sort of reason to believe areGs.

This restriction is satisfied when (a) it is essential to ourconception of the Fs to maintain that all Fs result fromthe application of certain operations to certain basic Fs(—thus Fs might be propositions, numbers, or sets); and

Page 11: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 11/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 45

 A Defense of Minimalism   45

(b) given any such F, there is a uniform way of proving it tobe G.

Thus a more plausible version of our extra premise wouldrun along the following lines:

 Whenever someone is disposed to accept, for any propo-sition of structural type F, that it is G (and to do so foruniform reasons) then he will be disposed to accept thatevery F-proposition is G.

 And this will do the trick. We are indeed disposed to accept,

for any proposition of the form,  <

p →

 p>

, that it is true.Moreover, the rules that account for these acceptances arethe same, no matter which proposition of that form is underconsideration. So it is now possible to infer that we accept thatall such propositions are true, and hence to explain why wedo so.

Thus we have a sound explanatory premise which, in con- junction with our commitment to ‘<p>   is true  ↔  p’, willenable us to explain our acceptance of generalizations abouttruth. And since that premise does not mention the word,‘‘true’’, the need for it does nothing to suggest that the basic(hence, meaning-constituting) use of it must exceed the boundsset by minimalism.¹²

FN:12

Objection 5: Certain people (mostly philosophers) do not  in fact havea completely general inclination to endorse the equivalence schema.Some hold that ethical pronouncements fail to yield acceptableinstances. Others take that view of contingent statements about thefuture, or of applications of vague predicates to borderline cases, or of sentences containing empty names. But all these people neverthelessmean the same thing as we do by the word ‘‘true’’. After all, wemight argue with them about ‘‘Whether ethical pronouncements

can be true?’’—yet each of us expresses the issue in just that way.Consequently, it cannot be that to understand the English truth

¹²  For an extended response to the above ‘generalization’ problem, see ‘‘A Minimalist Critique of Tarski’’— Ch. 5 of the present volume.

Page 12: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 12/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 46

46   A Defense of Minimalism

predicate one must have an inclination to accept  every  instance of the schema.

(Richard¹³)FN:13

Notice that the present objection is not that certain instancesof the equivalence schema actually  are  incorrect, or that someof them are unhesitatingly rejected by everyone. The pointrather is that even if someone—perhaps, mistakenly—hasno inclination to  accept  a certain class of instances he mightnonetheless   understand   them exactly as we do. And so we

have to conclude, it would seem, that an endorsement of the general  schema is  not  what provides the word ‘‘true’’ with itsmeaning.

But this conclusion is unwarranted. For we can invoke  social externalism in order to reconcile our minimalist thesis with thefacts under consideration. The rough idea is that a word is a social entity with a certain public meaning, and each member of the linguistic community can (if he wants) deply the word withthat meaning—even if his own usage of it diverges radically from everyone else’s. What fixes that shared meaning is thebasic use that  predominates  within the community (or, in thecase of a technical term, within the sub-community in whichit is deployed). After all, it is such ‘predominant basic use’ thattranslation mappings are attempting to preserve.

Now, our word ‘‘true’’ is not a merely technical term. Andmost people’s use of it is unaffected by philosophical fretting, sotends to be governed by the full  equivalence schema. Therefore,its shared meaning derives from that characteristic. And eventhose few who are inclined to reject certain instances of it arenonetheless using the truth predicate with that shared meaning.

Objection 6:   But what about the notorious paradox 

-inducing in-stances of the schema—e.g. the one that results from applying it to

¹³   Richard, M. ‘‘Deflating Truth’’, E. Villanueva (ed.), Philosophical Issues  8,op. cit. See his book, When Truth Gives Out , (Oxford University Press, 2008),for further elaboration of his anti-minimalistic perspective.

Page 13: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 13/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 47

 A Defense of Minimalism   47

the statement that this very statement is not true ? It is easy to provethat such instances lead to contradiction. As a consequence, many 

of us have learned not to accept them. Indeed, most people,  were they to grasp the relevant proofs,  would  agree that those instancesare unacceptable. Thus our meaning what we do by ‘‘true’’ doesnot require us (or most of us, or the ‘experts’ among us) to have a disposition to accept every  instance of the equivalence schema.

It seems to me that this point is correct, but not devastat-ing.¹⁴  One way of accommodating it is to concede that the

FN:14

meaning of ‘‘true’’ does not derive from an   entirely   unre-

stricted equivalence schema. Rather, it might be said, the basicmeaning-constituting practice is merely to accept patently non-paradoxical instances— instances concerning propositions, like<snow is white> and <lying is wrong >, that make no men-tion of truth. Given these, we are then tempted to generalizeand to accept all  instances of the schema. But when this con-clusion is found to lead into contradiction, the retreat back toa more more constrained schema need involve no revision of the meaning-constituting use of ‘‘true’’.

However, there’s an alternative strategy—one that seems to

me more natural. Instead of linking the meaning of ‘‘true’’ withthe disposition to wholeheartedly accept a certain restrictedclass of instances, we might link it with the defeasible inclinationto accept any  instance. We might suppose that, in paradoxicalcases, this inclination is over-ridden; but that it nonethelesscontinues to exist —sustaining the sense of paradox.

Objection 7: Mightn’t there be a linguistic community in which thevery existence of truth is widely debated—in which some peopleare disposed to accept some (or all) instance of their equivalenceschema—‘‘<p>   is schmoo  ↔  p’—but some reject it altogether,

and most are simply not sure? And isn’t is natural to report them (as

¹⁴   Although one might instead follow Hartry Field in supposing that ‘liar’paradoxes are best avoided by revising clasical logic rather than restricting theequivalence schema. See his  Saving Truth From Paradox , (Oxford: OxfordUniversity Press, 2008).

Page 14: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 14/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 48

48   A Defense of Minimalism

I just did) as questionisng the reality of  truth? Wouldn’t their word‘‘schmoo’’ mean exactly the same as our ‘‘true’’?

If so, then the social externalism of meaning will not sufficeto rescue minimalism. The moral will be, rather, that theremust be some way of using ‘‘true’’ that (a) is implicit in, butweaker  than, an endorsement of the equivalence schema, (b) isdisplayed by the community of truth-debaters and by ourselves,and (c) constitutes what we communitites mean by that word.

 And this conclusion would seem to be at odds with minimalism.

However, there is a natural minimalist way to implement it. We must appreciate that even the truth-sceptics are preparedsometimes to   suppose for the sake of argument   (what they donot actually believe) that certain things are ‘‘schmoo’’. And

 when they do this, they will proceed—within the contextof this supposition—to ‘accept’ their equivalence schema. Inaddition, when they suppose, again merely for the sake of argument, that   some  schema of the form, ‘<p>   is    ↔  p’,holds, they will be prepared to infer (relative that supposition)instances of ‘<p> is schmoo↔ p’. Moveover, we non-skepticsalso follow these rules— but using ‘‘true’’ instead of ‘‘schmoo’’.Thus we have here a way of deploying the equivalence schema that is shared by skeptics and non-skeptics alike, and this is what

 we can take to provide the truth predicate with its meaning.

Objection 8: Truth is valuable: we ought to pursueit and we ought to

avoid false belief. But these normative sentiments are not containedin (nor can they be extracted from) instances of ‘<p> is true ↔ p’, which merely tell us when beliefs possess the property of being true,and are completely silent on the question of whether its possession isdesirable. Consequently, our concept of truth is not fully capturedby the equivalence schema; so the minimalist proposal is false.

(Dummett¹⁵)FN:15

¹⁵  Dummett, M. ‘‘Truth’’, Proceedings of the Aristotelian Society  59, (1958–59), 141–62.

Page 15: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 15/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 49

 A Defense of Minimalism   49

On the contrary, the equivalence schema  is  able to account forthe value of truth. To see this, consider specific norms of belief such as

It is desirable that (one believe   that wombats fly   ↔ wombats fly)

Clearly our commitments to norms like this one have nothing to do with the concept of truth; for that concept is completely absent from their articulation. Nor is there any reason to sup-pose that the concept of truth will need to be deployed in

explaining  why  we accept them.Let us then imagine that all such specific normative com-

mitmensts are somehow explained.¹⁶ Suppose, that is, we canFN:16

account for our attachment to all norms of the form

It is desirable that (one believe  that p ↔ p)

Given our a priori knowledge of the equivalence schema, we will then be able to explain our attachment to every norm of the form

It is desirable that (one believe <p>↔ <p> is true)

that is, to every norm of the form

It is desirable that (one believe x ↔ x is true)

But this engenders (via themechanism discussed in the responseto Objection 4) a commitment to the generalization

(x) (It is desirable that (one believe x ↔ x is true)

—or, in English, to the principle

It is desirable to believe what is true and only what istrue

¹⁶  I would argue that the basis for our commitment to these specific normsis both  pragmatic —insofar as we are more likely to get what we want if weabide by them—and moral —the value of truth for its own sake—. For more,see my ‘‘The Value of Truth’’—Ch. 4 in the present volume.

Page 16: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 16/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 50

50   A Defense of Minimalism

Thus the value we attach to true belief is explained by the roleof truth as a device of generalization—which is itself explainedperfectly by the equivalence schema.

Objection 9: The minimalist proposal implies that the meaning of the word, ‘‘true’’, depends on the meaning of the expres-sions, ‘‘the proposition that . . .’’. For someone’s acceptance of,for example, ‘‘The proposition   that Hesperus rotates   is true if and only if Hesperus rotates’’ manifests a standard understand-ing of the truth predicate only to the extent that its component,‘‘The proposition that . . .’’, is being understood in the standard

 way. Thus minimalism implies that one must  already  understand‘‘that’’-clauses—i.e. one must understand sentences of the form ‘‘uexpresses (i.e. the proposition that p’’— in order to be in a position

to acquire the concept of truth. But this surely gets things the wrong way round! Surely the intimately related notions of mean-ing and proposition must be analysed in terms of truth (or better,perhaps, must be   replaced   by truth– conditional notions). Morespecifically, we must suppose that, insofar as there are any facts of the form

u means (says, expresses the proposition) that p

they consists in facts of the form

u is true (i.e. expresses a truth) if and only if p

Thus truth is conceptually prior to meaning, contrary to what isrequired by the minimalist proposal.

(Davidson¹⁷)FN:17 

Davidson gives three reasons for thinking that   is a morebasic concept than   .

First, he infers it from the idea that there are sentences (suchas ‘‘That is red’’) whose meaning-constituting, assertibility conditions are to accept those sentences only when they aretrue .

¹⁷   Davidson,D. ‘‘The Folly of Trying to Define Truth’’, Journal of Philosophy 87 (1996) 267–78.

Page 17: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 17/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 51

 A Defense of Minimalism   51

But this inference involves several mistakes:

(1) It neglects the fact that, for each specific sentence, ‘‘p’’,of that special sort, the assertibility condition is, ‘Oneshould accept ‘‘p’’ only if p’—in which   the concept of   truth is not deployed . This concept is needed only by thetheoretician who wishes to make the  general   point thatthere are sentences of that kind (i.e. to be accepted only 

 when true). It isn’t needed to formulated any particularassertibility condition.

(2) His rationale wrongly presupposes that meaning-constitut-ing regularities of use are  explicitly  known by the speakersof a language. Even if the truth-predicate were  to be neededin order to articulate the fact about a person that underliesher understanding of some word, we could not concludethat she herself would himself have to possess the conceptof truth.

(3) Davidson confuses the conditions for understanding par-ticular non-semantic sentences, like ‘‘That is red’’, with thereal issue here—namely, the conditions for understand-

ing ‘that’-clauses, for grasping the concept of ‘proposition’.Remember that his thesis was that one cannot have theconcept of meaning without possessing the concept of truth. Thus the issue is  not   whether one can have vari-ous non-semantic concepts—such as ‘red’—without theconcept of truth.

Davidson’s second stated reason for thinking that   isconceptually prior to    is that someone’s affirmationof a sentence can be justified only if he believes that thesentence is true. Thus it would seem that the concept of 

  —i.e. of when one is justified inassenting to a sentence—presupposes the concept of truth.

 And, in that case, one cannot go along with the minimalist orderof explnation:—from assertibilty conditions, to meanings, totruth.

Page 18: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 18/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 52

52   A Defense of Minimalism

But, in fact, this sequence of conceptual dependenciescoheres pefectly well with the normative constraint on publicaffirmation, to which Davidson calls attention. For note firstthat S’s believing  that ‘‘p’’ is true  is equivalent to S’s accepting the sentence, ‘‘ ‘‘p’’ is true’’, which, given deflationism, corre-lates with S’s accepting ‘‘p’’. In addition—and this is the basicnorm here—S should  assent to ‘‘p’’ only if he accepts it. It thenfollows, as Davidson says, that S should assent to ‘‘p’’ only if and only if he believes it to be true. However, the just-givenaccount of why this norm holds does not preclude that the

meanings of terms be constituted by the conditions in whichsentences containing them are (or should be) accepted.

It might be protested that what we call ‘‘accepting’’ a sentence is simply a matter, as Davidson would say, of ‘‘holding it true’’. So, relative to the deflationist’s ‘acceptance-conditions’account of meaning, it may appear that the notion of truthis, afer all, prior to the notion of proposition. But this useof the term, ‘‘holding true’’, is a somewhat misleading one,given that the intended commitment—‘holding’—is to a sentence that does not contain the truth predicate. The notion

needed here is simply that of   relying on   a sentence as a premise in inference (both theoretical and practical), and thatpsychological role may be explicated without bringing in thenotion of truth.¹⁸FN:18

Davidson’s third motivation is his attachment to a directionof explanation which goes from(i) facts about thecircumstancesthat cause the acceptance (=   ‘holding true’) of sentences, to(ii) facts about their truth conditions, to (iii) facts about theirmeanings.

But among the many obstacles to working out such an

idea there is one that has proven notoriously difficult tonavigate around. The problem is to articulate a conception of ‘truth condition’ that is strong enough. For, given a  material 

¹⁸  See my  Meaning , pp. 84–6, for a sketch of how this can be done.

Page 19: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 19/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 53

 A Defense of Minimalism   53

construal of ‘if’, ‘‘p’’ may very well be true   if and only if     q, without  meaning   that q. And stronger construals of ‘‘if andonly if’’ merely make the counterexamples slightly harder toconstruct.¹⁹FN:19

Objection 10: ‘That’-clauses cannot be regarded as referring expres-sions, because there is no way of seeing how their referents would bedetermined by the referents of their component words. Therefore,sentences like ‘‘The proposition that Hesperus rotates  is true’’, insofaras they are construed as predicating truth of the propositions to which ‘that’-clauses refer, are in fact unintelligible. But if such truth

ascriptions (so construed) are unintelligible, then the minimalistproposal cannot be correct.

(Davidson²⁰)FN:20

Davidson’s basis for maintaining that alleged referents of ‘that’-clauses would not be determinable by the referents of theparts of these clauses is that—as Frege observed— substitutionof co-referential terms (e.g. putting ‘‘Phosphorus’’ in placeof ‘‘Hesperus’’) within a ‘that’-clause occuring in some sen-tence (e.g. ‘‘Mary believes that Hesperus rotates’’) will notalways preserve the sentence’s truth value. But why does henot continue to follow Frege’s line of reasoning, and con-clude that an expression within a ‘that’-clause does not haveits   standard   referent, but instead refers to the  meaning   (i.e.‘sense’) of that expression? Why not identify the referentof ‘‘that Hesperus rotates’’ with the meaning of ‘‘Hespe-rus rotates’’ and identify the referents (in that context) of the contained words ‘‘Hesperus’’ and ‘‘rotates’’ with   their meanings?

The reason given is that the meanings of words in ‘that’-clauses are just their normal meanings. After all, we understand

¹⁹  For further disussion of this final point see  Reflections of Meaning , ch. 8(‘‘Deflating Compositionality’’, p. 201, and ‘‘Semantics: What’s truth got todo with it?’’—ch. 8 in the present volume.

²⁰   Davidson,D. ‘‘The Folly of Trying to Define Truth’’, Journal of Philosophy 93 (1996).

Page 20: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 20/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 54

54   A Defense of Minimalism

‘‘The proposition that Hesperus rotates’’ only if we understandthe isolated sentence ‘‘Hesperus rotates’’. And in the bicon-ditional, ‘‘The proposition   that Hesperus rotates   is true   ↔Hesperus rotates’’, the two occurences of ‘‘Hesperus rotates’’are clearly supposed to be understood in the same way. But inthat case—since meaning determines reference—how could

 words in ‘that’-clauses fail to have their standard referents? And if they do have their standard referents then ‘that’-clausescannot refer, since what would be determined by those stan-dard referents would be the wrong   thing (e.g. ‘‘that Hesperus

rotates’’ would acquire the same referent as ‘‘that Phosphorusrotates’’).

However, there’s a pretty obvious response to this anti-Fregean story. We can simply deny that meaning  all by itself   determines reference. We can allow—and this, of course, iscompletely uncontroversial— that the referent of a term is fixedin part by the context  in which it occurs. More specifically, wemight say that the admittedly uniform meaning of ‘‘Hesperus’’yields one referent (the planet) for standard (non-opaque)occurences of the word, and that it yields a different referent

(the meaning, or sense, of ‘‘Hesperus’’) for occurences of the word within the context of a ‘that’-clause.

It’s worth emphasizing that although our Fregean response isfar from problem-free, it is not being introduced ad hoc, merely for the sake of a certain theory of truth. After all, ordinary language is full of ‘that’-clauses (e.g. in attributions of belief).

 And there is no available strategy for dealing with them that isevidently less problematic than Frege’s—namely, to supposethat they designate entities (nowadays called ‘‘propositions’’),and that each of our ‘attitude’ attributions assserts that a certain

psychological relation (of believing, or conjecturing, or stating,etc.) holds between a person and one of these entities (sodesignated).

Moreover, the question arises, if one were  to prefer a non-Fregean semantics for ‘that’-clauses, why it could not be

Page 21: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 21/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 55

 A Defense of Minimalism   55

deployed by minimalists? Consider, for example, an approachsuggested by Davidson’s ‘paratactic’ analysis of ‘S says that p’.²¹

FN:21

 We might interpret

The proposition that Hesperus rotates  is true

as

The proposition expressed by this —Hesperus rotates—istrue

—where ‘‘Hesperus rotates’’ is uttered as an aside. Thus the

equivalence schema would become:The proposition expressed by  this  (- p -) is true ↔ p

Or, along similar lines, we might reformulate it as:

The proposition expressed by the   immediately following sentence-token is true ↔ p²²

FN:22

Therefore, even if Davidson’s squeamishness about Fregean‘that’-clasuses were correct—and my first response arguedthat it is not—the minimalist proposal can be salvaged. For

²¹  He proposes that ‘‘S says that dogs bark’’ be construed as ‘‘S’s utter-ence same-says  that . Dogs bark.’’—where the demonstrative in the speaker’ssupposed first sentence refers to the subsequent utterance of ‘‘Dogs bark’’.See Davidson’s ‘‘Saying that’’, collected in his  Essays on Truth and Interpret-ation.

²²  Clearly these formulations still trade in propositions ! However, Davidson iscareful to emphasize that his complaint about minimalismis not  its commitmentto such things. His point, rather, is that if    they exist, then the logical form of  what designates them must be ‘‘The proposition expressed by u’’ (instead of ‘‘The proposition that p’’). Perhaps he would be prepared to analyze this along the lines of ‘‘The class of utterences that same-say  u’’.

In addition to Davidson’s  critique of minimalism, which focuses on the way that propositions are designated , there are several objections to the very  existence of propositions—objections that a minimalist must be able to rebut. The mainones are (1) that propositions lack satisfactory identity conditions; (2) that false  propositions do not exist (because any actual combination of objects andproperties would amount to a  fact ); (3) that propositions are ontologically weirdand explanatorily unnecessary. For discussion of some of these issues, see theend of ‘‘Varieties of Deflationism’’—Ch. 2. For a more complete discussionsee Truth, pp, 86–97, 106.

Page 22: Paul Horwich, Truth and Meaning, chapter 3

7/21/2019 Paul Horwich, Truth and Meaning, chapter 3

http://slidepdf.com/reader/full/paul-horwich-truth-and-meaning-chapter-3 22/22

Paul Horwich chap03.tex V1 - August 12, 2009 6:20pm Page 56

56   A Defense of Minimalism

it’s truth schema may be articulated in one of the two ways just-mentioned.

The   full   minimalist picture of truth includes considerably more than the thesis I have been defending in this essay. Itinvolves, besides the present claim about how the meaning of ‘‘true’’ is constituted, an affiliated view about the  functionof the truth predicate (namely—as illustrated in response toObjection 8—that it is merely a device of generalization), anaffiliated view about the   underlying nature  of truth (namely,

that there is no such thing), and an affiliated view about thegeneral shape of the basic theory that will best explain all thefacts about truth (namely, that its postulates are instances of the equivalence schema). I have not attempted to elaborateor establish these further minimalist doctrines. However, since

 what I have been concerned with here is the central componentof minimalism, my defense of that thesis, if successful, providesimportant support for the view as a whole.²³

FN:23

²³   I would like to thank Hartry Field and Michael Lynch for helpful

discussion.This essay is a considerably revised version of a paper published, under thesame title, in The Nature of Truth, ed. M. Lynch (Cambridge, Mass.: MITPress, 2000), which was itself a revision of an earlier paper, written for theconference on Truth that took place in Leuwen in 2000, and published withthe other conference papers in Synthese  (vol. 126, Special Issue on Truth, eds.Volker Halbach and Leon Horsten, (2001) pp. 149– 165.