J E R R O L D J. K A T Z

THE DOMINO THEORY*

This paper presents a domino theory of the criticisms of intensionalism over the last three decades of Anglo-American philosophy. Admittedly, domino theories are risky: one rarely finds objects arranged like dominos so that a slight push on the first topples them all. But those criticisms of intensionalism are arranged in precisely this fashion. The foundational domino in this construction is Quine's argument in "Two Dogmas of Empiricism". ~ Standing just behind the first domino is Quine's argument for the indeterminacy of translation. Directly behind these two is Davidson's argument for replacing the intensionalist paradigm of analysis with his extensionalist Tarskian paradigm, the argument for a purely extensional possible worlds semantics, Putnam's arguments against intensionalist semantics, and finally Burge's arguments against semantic individualism. These are by no means the only arguments that have been given against intensionalism during this period, but the others are by and large derivative from one or another of them. The domino theory's picture of the intensionalism/extensionalism controversy differs radically from the standard picture. On the standard picture, intensionalism is portrayed as beset by a host of independent arguments which have converged to completely swamp it. On this picture, there is so much over-kill that philosophers who notice that one or another of the anti-intensionalist arguments is not all it is cracked up to be nonetheless feel that to criticize it would amount to little more than an academic exercise. The domino theory challenges the principal assumption of this picture, namely, that the anti-intensionalist arguments are largely independent of one another. The domino theory claims that there is a dependency structure to the set of anti-intensionalist arguments wherein one of them, Quine's argument that no sense can be made of concepts in the theory of meaning, is fundamental. All the others stand or fall with Quine's argument. Davidson's dedication to Quine in his book Inquiries into Truth and Philosophical Studies 58: 3--39, 1990. 9 1990 Kluwer Academic Publishers. Printed in the Netherlands.

4

JERROLD

J. K A T Z

Meaning, which reads "Without Whom Not", explicitly acknowledges the dependency of his arguments on Quine's. With broader scope, this motto could also serve as the epigraph for the present paper, since the thesis of the paper is that Quine's arguments not only underlie the criticisms of intensionalism on which Davidson's extensionalist program rests, but also the criticisms on which the other principal extensionalist programs -- the possible worlds program, the scientific essential program, and the semantic communitarian program -- rest. In this paper, I try to establish the domino theory, that is, that the fall of Quine's criticism of intensionalism in "Two Dogmas of Empiricism" leads to the fall of those other criticisms of intensionalism. In earlier works, I spelled out the reasons why this argument of Quine's fails, and why its failure does away with his argument for indeterminacy.2 After briefly summarizing the criticisms in those works, I shall concentrate on showing that the other arguments against intensionalism fail without Quine's. Quine's argument in "Two Dogmas of Empiricism" has the form of a proof by cases. The cases he considers are the areas of definition, logic, and linguistics. The cases seem to be exhaustive, as required, since these areas seem to be the only ones where we might reasonably expect to find a way of explaining logico-linguistic concepts. Thus, if there is no way to explain concepts of the theory of meaning in any of them, we must give up trying to make objective sense of such concepts. Quine has no trouble showing that they cannot be explained either in the area of definition or logic. His argument in the case of linguistics seems equally forceful, for, on apparently plausible assumptions, he is easily able to show that any attempt to explain synonymy or analyticity is viciously circular. But, in fact, one of the assumptions is false. This is the assumption that the proper way to explain concepts in linguistics is to define them on the basis of substitution criteria. Substitution criteria were the preferred form of definition at the time Quine looked into linguistics. At that time, linguistics had been under the influence of a neopositivist conception of science which advocated operationalist definition. But, shortly thereafter, the Chomskian revolution took place, replacing the operationalist paradigm of substitution criteria with the theory construction paradigm of generative grammar? On this paradigm, the definition of grammatical concepts takes the

THE DOMINO THEORY

5

axiomatic or recursive form familiar from definitions in logic and mathematics. A set of axioms or rules defines a concept by relating it to other concepts in its own family. Familiar examples are systems of propositional logic, or Peano arithmetic. The explanation afforded by such definitions comes from the light they shed on the internal relations of a concept to other concepts within its family and, in virtue of revealing those relations, from the way the definitions show the bearing of the concept on a wide range of phenomena in the domain. There is no requirement that a theoretical definition specify the concept in terms of concepts outside its family, as there is with substitution criteria. This is because there is no presumption that the concept to be defined requires vindication based on independent concepts whose validity is taken for granted. Successful theoretical definition is all the validation a concept needs. Given the option of theoretical definition, there is nothing viciously circular in defining concepts in the theory of meaning in terms of others concepts in the theory of meaning, so the vicious circularity in defining them in terms of substitution criteria looses its significance. Hence, the emergence of theory construction as the preferred form of definition in linguistics robs Quine's argument in "Two Dogmas of Empiricism" of all force. Note that the logic of this criticism of Quine does not require that we already have well-developed theoretical concepts in the theory of meaning. Since Quine's argument claims to show that there is no serious possibility of making objective sense of such concepts, merely exhibiting a definitional paradigm in linguistics that he overlooks, by itself, establishes such a possibility, and hence, eliminates his argument. The second domino is Quine's argument for the indeterminacy of translation, that is, for his claim that ... rival systems of analytical hypotheses can fit the totality of speech behavior to perfection, and can fit the totality of dispositions to speech behavior as well, and still specify mutually incompatible translations of countless sentences insusceptible of independentcontrols.4 The argument for this claim depends on the argument in "Two Dogmas of Empiricism" against the theory of meaning. To exhibit this dependecncy, and hence, the way in which the possibility of theoretical definitions for concepts in the theory of meaning undercuts the argument for indeterminacy, we first separate two cases where the indeter-

6

JERROLD J. KATZ

minacy claim can be made. On the one hand, there is the case of radical translation which Quine uses to illustrate indeterminacy. In radical translation, it seems clear that we are "'incapable of deciding among 'rabbit', 'rabbit stage', and various other terms as translations of 'gavagai' ". The reason is also clear: any evidence in favor of one translation could be matched with essentially the same evidence for the others, s On the other hand, there is actual translation, that is, translation as it actually occurs or could occur. ! submit that it is not at all clear that the case of actual translation is, as Quine thinks, the same as that of radical translation. It could be that radical translation is simply Quine's creation, constructed is so that nothing beyond referential considerations provide evidence for translation. Since referential considerations can be the same for incompatible but referentially equivalent translations, it is hardly surprising that, in the absence of any evidence about the senses of expressions, radical translation contains an evidential symmetry that leads directly to indeterminacy. But what about the case of actual translation? Couldn't actual translation situations contain evidence bearing on sense differences among analytical hypotheses? If so, then the necessary "independent controls" might exist in actual translation, and no evidential symmetry would be found in the philosophically significant case. There is, then, a step in Quine's argument for indeterminacy from reflections about radical translation to a conclusion about actual translation. I will argue that Quine's justification for taking this step is the argument from ""Two Dogmas of Empiricism". Quine says that radical translation is merely the extreme case of actual translation in which differences in historical and cultural context are maximal. ~ The case, as he sees it, is simply the one we want to look at because the historical and cultural similarities in other cases tend to confuse the philosophical issue about translation. But why think radical translation is a case of actual translation at all? To suppose this without argument would beg the question of whether there could be evidence about senses of words in the extreme case of actual translation. Without an independent study of actual translation, how can we know there is no intensional evidence to provide the controls necessary for determinate translation? Thus, it seems we ought to set out to examine actual translation without an a priori vieW about its relation to radical translation, and the best way to do this would clearly be to try to construct a

THE DOMINO THEORY

7

theory of actual translation, using the most successful theories in the sciences as our model. But Quine protests that such a project misjudges the relation between intensional theories of translation and the scientific theories that would serve as a model. He claims that there is a fundamental difference between them, namely, that there is no fact of the matter in connection with the former theories. 7 He claims that there are no 'Yree-floating, linguistically netural meanings", s Indeed, the extreme divergence of linguistic history and cultural setting in radical translation shows us that there is really nothing over and above the verbal forms themselvesY This claim obviously needs backing up. Given that the possibility of appealing to facts about senses in actual translation would surely constitute a difference sufficient to block any inference from radical translation to actual translation, in the present context, there is a need to show that actual translation does not differ relevantly from radical translation. A bare assertion that there is no semantic fact of the matter in actual translation would beg the question, since whether there are linguistically neutral senses is the crux of the issue between Quine and the intensionalists. H~ Furthermore, asserting that there are no linguistically neutral meanings introduces an ontological skepticism over and above the epistemological skepticism Quine has already put forth. Failure to be able to break evidential symmetry is an epistemological failure, and as such, it is compatible with there being meanings. Meanings, like Kant's noumena, could exist but be unknowable. Hence, Quine's introduction of an ontological skepticism, rather than helping to support his epistemological skepticism, only increases the burden of proof. But Quine thinks the support for his ontological skepticism is already in place and will be widely known to his readers. This is why he takes no steps to support it at this point. As he himself indicates, the skepticism is intended to rest on the earlier argument from "Two Dogmas of Empiricism"J ~ Quine's reasoning is this: if the earlier arguments have shown that the concepts belonging to the theory of meaning, especially the all-important concept of synonymy, cannot be made objective sense of, then there will be no serious theory that quantifies over meanings, and hence, parsimony will oblige us to take them not to exist. But, as we have seen, the required support from the argument in

8

JERROLD J. KATZ

"Two Dogmas of Empiricism" is not forthcoming. As a consequence, there is nothing to back up Quine's protest against understanding the theory of translation on the model of scientific theories. Thus, we are at liberty to assume there is a fact of the matter in actual translation in the same spirit in which scientists, when beginning an investigation, assume there is regularity in nature. We may construct theories of natural languages which refer to linguistically neutral meanings and evaluate such theories on the basis of the account they give of the intensional as well as the extensional structure of those languages. An evaluation of an account of intensional structure will be based on judgments of speakers about sense properties and relations of sentences. Assuming bilingual fluency on the part of the informant or linguist, the linguist can ask the informant outright whether "gavagai" is synonymous with "rabbit", "rabbit stage", or "undetached rabbit part'U 2 Or, the linguist can a s k whether "gavagai" bears the same relation to some expression that "finger" bears to "hand" or "blade" to "knife". Or, the linguist can ask whether "gavagai" is closer in sense to "infancy" and "adolescene" than to "infant" and "adolesent", or the other way around. Then again, the linguist can ask whether the expression formed by modifying "gavagai" with an adjective that means 'detached' is contradictory in the way, say, "married bachelor" is. Or, the linguist can ask whether the expression formed by modifying "gavagai" with an adjective that means 'being a part of' (or 'being a stage of') is redundant in the way that, say, "free gift" is. The evidence necessary to choose between extensionally equivalent analytical hypotheses can, in principle, be obtained form such interrogations. Of course, practical attempts to acquire evidence may be frustrated. The linguist may fail to come up with a good hypothesis, the informant may be incompetent in one or the other language, be uncooperative, etc. But this sort of thing happens even in the best of sciences. Hence, a skepticism based solely on the logical possibility of persistent frustration of this sort is useless for philosophical purposes. It is applicable to all sciences, and hence, it proves too much. Thus, there is a prospect of intensional evidence in the case of actual translation, and this prospect by itself confines indeterminacy to the case of radical translation. Accordingly, the second domino falls. The third domino is Davidson's argument for replacing the inten-

THE D O M I N O T H E O R Y

9

sionalist p a r a d i g m of analysis, 's m e a n s p', with the extensionalist p a r a d i g m (T). This, (T)

s is T if a n d o n l y if p.

a r g u m e n t is the linch-pin of D a v i d s o n ' s p h i l o s o p h i c a l p r o g r a m . T h e a r g u m e n t runs as follows: Anxiety that we are enmeshed in the intensional springs from using the words 'means that' as filling between description of sentence and sentence, but it may be that the success of our venture depends not on the filling but on what it fills. The theory will have done its work if it provides, for every sentence s in the language under study, a matching sentence (to replace 'p') that, in some way yet to be made clear, 'gives the meaning' of s. One obvious candidate for matching sentence is just s itself, if the object language is contained in the metalanguage; otherwise a translation of s in the metalanguage. As a final bold step, let us try treating the position occupied by 'p' extensionally: to implement this, sweep away the obscure 'means that', provide the sentence that replaces 'p' with a proper sentential connective, and supply the description that replaces 's' with its own predicate. The plausible result is [(T)I.~ T h e r e a r e t h r e e aspects of this a r g u m e n t to consider. First, t h e r e is what D a v i d s o n has in mind when he refers at lhe beginning of the p a s s a g e to the "anxiety that we a r e e n m e s h e d in the intensional", It is o b v i o u s that what he has in m i n d is largely, if not c o m p l e t e l y , Q u i n e ' s arguments. H e is referring to the fear that Q u i n e ' s a r g u m e n t s are e x p e c t e d to strike in the h e a r t s of those who might b e c o n t e m p l a t i n g the use of c o n c e p t s f r o m the t h e o r y of meaning, particulary, the c o n c e p t of s y n o n y m y on which the intensionalist p a r a d i g m of analysis rests. Textual s u p p o r t for this i n t e r p r e t a t i o n , if a n y is n e e d e d , c o m e s f r o m a n o t h e r discussion o f D a v i d s o n ' s . In examining relations like ' m e a n s that' which r e p o r t t h r o u g h i n d i r e c t d i s c o u r s e , D a v i d s o n writes that 9 indeterminacy shows . . . that if there is one way of getting it right there arc other ways that differ substantially in that non-synonymous sentences are used after 'said that'. And this is enough to justify our feeling that there is something bogus about the sharpness questions of meaning must in principle have if meanings are entities.~4 If Q u i n e ' s a r g u m e n t s a r e all that D a v i d s o n has in mind, then we c o u l d i m m e d i a t l e y c o n c l u d e that t h e r e is no cause for anxiety a b o u t the intensional, a n d hence, that t h e r e is no reason to r e p l a c e the traditional intensionalist p a r a d i g m with c o n v e n t i o n (T). But b e f o r e we can d r a w this conclusion, we s h o u l d b e sure Q u i n e ' s a r g u m e n t s a r e the only thing that D a v i d s o n has in mind. T h a t he might have s o m e t h i n g else in m i n d is suggested b y r e m a r k s such as " . . .

10

JERROLD J. KATZ

wrestling with the logic of the apparently non-extensional 'means that' we will encounter problems as hard as, or perhaps identical with, the problems our theory is out to solve". ~-~ Now the only thing Davidson mentions in this connection is problem of substituting into opaque contexts. The difficulty, as he puts it, is " . . . the standard problem, which is that we cannot account for even as much as the truth conditions of [belief sentences] on the basis of what we know of the meanings of the words in them". 1~' Let us suppose that Davidson's grounds for thinking that we cannot account for the truth conditions of sentences involving "believe" and other verbs creating an opaque context is not simply a direct appeal to Quine. If Davidson were directly invoking Quinean objections to the Fregean move of using synonymy as a condition for substitiution, we would already have refuted Davidson's argument against the intensionalist paradigm. So if "the standard problem" is independent of Quine, it must be the problem Mates raises in his paper "Synonymity".~7 On this assumption, Davidson's grounds for replacing the intensionalist paradigm of analysis with convention (T) are that doing so avoids that allegedly unsolvable problem. But what, according to Davidson, is supposed to make it unsolvable? In particular, why isn't Frege's move itself, either as he formulates it or as developed by later intensionalists, a solution? It might strike one as peculiar that Davidson does not consider the principal intensionalist solutions to the problem. To be sure, he does consider Carnap's operationalist explication of belief in terms of affirmative response, and he nicely puts his finger on Carnap's confusion, l~ But Carnap's explication is really beside the point. What is needed is not a criticism of such an explication, but a rebuttal of the classical Frege-Church solution. On their solution, inferences like that from "Everyone believes that bachelors are bachelors" to "Everyone believes that bachelors are unmarried men" can have no counter-example because belief in the premiss but not the conclusion ipso facto demonstrates a linguistic confusion or deficiency which disqualifies the counter-example. ~) The expressions "bachelors" and "unmarried men" are being taken, for the sake of argument, as synonymous in English. If they were not taken a synonymous, Frege and Church would not claim the inference goes through, and hence, there would be nothing against which to try to raise a counter-example. But if there is a claim about an

THE DOMINO THEORY

11

entailment, then a belief of a speaker stemming from a linguistic defect can no more be taken as evidence for the invalidity of an inference than a belief of a reasoner stemming from a logical defect could be taken as evidence for the invalidity of one. Genuine evidence for the validity or invalidity of inferences, whether linguistic or logical, does not derive from incompetence. But it is not the case that Davidson simply neglects the FregeChurch solution. If we look beyond "Truth and Meaning", we can see that Davidson has a reason for ignoring their solution. As emerges clearly in later papers, Davidson recognizes that their solution depends on the notion of synonymy and the analytic-synthetic distinction, and he is one hundred percent convinced that Quine has already gotten rid of them. In "On saying that", Davidson explains that "the lesson" of indeterminacy which he gives us in the preceding quote, was implicit in Mates's problem. Davidson writes. Mates claimed that the sentence "Nobody doubts that whoever believes that the seventh consulate of Marius lasted less than a fortnight believes that the seventh consulate of Marius lasted less than a fortnight" is true and yet might well become false if the last word were replaced by the (supposed synonymous) words "period of fourteen days", and that this could happen no matter what standards of synonymy we adopt short of the question-begging "substitutable everywhere salva veritate". Church and Sellars responded by saying the difficulty could be resolved by firmly distinguishing between substitutions based on the speaker's use of language and substitiutions coloured by the use attributed to others. But this is a solution only if we think there is some way of telling, in what another says, what is owed to the meanings he gives his words and what to his beliefs about the world. According to Quine, this is a distinction that cannot be drawn.2o Thus, it is evident that Davidson thinks that Mates's problem is the death blow to intensionalist semantics because he thinks that Quine has shown that the notions required for its solution are unavailable. Therefore, Davidson's grounds for thinking that there is cause for anxiety about being enmeshed in the intensional, as we might all along have concluded from his dedication to Quine, come down to Quine. Hence, as soon as the fallacy in Quine's argument is revealed, the first aspect of Davidson's rationale for replacing the intensionalist paradigm with convention (T) disappears. The second aspect of Davidson's rationale is his claim that a semantic theory for a natural language will have "done its work" if it just provides the appropriate instances of (T). But this claim presupposes

12

JERROLD J. KATZ

that there is no work to be done of pairing sentences with meanings in the traditional sense. Frege, for example, thought that there is intensional work to be done, one task being to characterize synonymous expressions in the case of substitution into opaque contexts and another to distinguish analytic identity sentences from synthetic identity sentences. If Frege is right, a semantic theory which only provides appropriate instances of (T) will not have done all its work. So, a further argument is needed to show that the intensionalist tasks that cannot be done if a semantic theory is restricted to specifying instances of convention (T) are not properly the work of a such a theory. Since Davidson offers no argument of his own to show this, again, we have to interpret him as falling back on Quine's arguments. But, on this interpretation, the failure of Quine's arguments is again fatal for Davidson's argument: the failure of Quine's arguments removes the second aspect of Davidson's justification for replacing the intensional paradigm of analysis with (T). The third aspect of Davidson's rationale is the presumption of his move of "sweep[ing] away the obscure 'means that' ", namely, that the relation 'means that' is obscure in some damaging sense. But why is 'means that' supposed have this vice? Admittedly, it is obscure in the innocuous sense that every complicated concept is obscure before it receives careful study. However, such obscurity properly motivates careful study of the relation 'means that' not its replacement. Hence, an argument is needed to show that the obscurity is something inherent in the relation, something which cannot be removed with careful study. But, again, we encounter the familiar pattern: an argument is needed for Davidson's criticism, Davidson offers none of his own, and so he must he understood as appealing to Quine's argument that no objective sense can be made of the notion of meaning. Accordingly, obscurity, too, disappears as a reason for accepting Davidson's proposal to replace the intensionalist paradigm of analysis with the extensionalist paradigm (T). 2~ At this point, we can conclude that there is no reason for accepting the proposal. But, in fact, we are in the position to draw an even more damaging conclusion. F o r once there are no reasons for accepting the proposal, there is nothing to neutralize counter-intuitive consequences of extensionalist analyses, and then those consequences are reasons

THE DOMINO THEORY

13

against Davidson's proposal. In order to exhibit such counter-intuitive consequences, let us employ convention (T) as our paradigm for the semantic analysis of the sentence (1). (1)

Snow is white

Davidson himself notes that (3) represents as good an account of the meaning of (1) as (2). 22 In addition to this counter-intuitive consequence, it also (2) (3)

'Snow is white' is T i f f snow is white 'Snow is white' is T iff grass is green

follows that English contains only two meanings, since the truth condition for every true sentence of English can be given on the basis of any true sentence, and the truth conditions for every false sentence of English can be given on the basis of any false sentence. Davidson attempts to mitigate such consequences on the grounds that ... the grotesqueness of [(3)] is in itself nothing against a theory of which it is a consequence, providing the theory gives the correct result for every sentence.., if [(3)] followed from a characterization of the predicate 'is true' that led to the invariable pairing of truths with truths and falsehoods with falsehoods -- then there would not, I think, be anythingessential to the idea of meaning that remained to be captured.23 Now Davidson could claim that there is nothing remaining to be captured if he had Quine's arguments to rely on, since, if they had gone through, there would be no possibility of a fine-grained notion of meaning to offend against in putting (3) on all fours with (2) as an account of the meaning of (1). But once Quine's arguments are taken away, Davidson's attempt at mitigation fails. In saying that the "grotesqueness" is nothing against his theory, he is arguing from his own position. His position says that the work of a semantic theory is simply to pair true sentences with true sentences and false ones with false ones, so, on that theory, nothing essential remains to be captured once a theory does this. But, without Quine's countervailing arguments, the counter-intuitive consequences themselves constitute an argument that such a position is too limited. These consequences show that it is also necessary for semantic theories to capture genuine synonymy to more finely discriminate among putative semantic analyses of (1). Hence,

14

JERROLD J. KATZ

trying to mitigate such consequences by arguing that nothing essential remains to be captured begs the question. Thus, without Quine's arguments, the grotesqueness of such consequences is strong reason against replacing a theory of meaning with a theory of the truth predicate. Recently, Davidson has argued that such criticism fails to note "the essential proviso" in "Truth and Meaning" requiring theories of the truth predicate to give the correct results for every sentence. 24 Davidson thinks that sentences like (3) can be ruled out on the requirement of a "reasonably simple theory that also [gives] the right truth conditions for 'That is snow' and 'This is white' ,,.a~ Davidson's point seems to be that, for such sentences, we will need to employ a non-grotesque instance of (T), and consequently, the shift to grotesque instances of (T) in the case of sentences where we have the option will produce a more complex overall theory. I am doubtful about this, since it seems reasonable to count a theory that replaces "p" with the same single true sentence in the case of all invariably true sentences as simpler -- or, by a trade off of complexities, as no more complex -- than one which replaces "p" with a different true sentence in each case. But, for the sake of argument, let us suppose Davidson is right. Let's suppose that there is something about the way in which theories handle the interpretation of occurrences of an expression in different sentences which makes the theories simpler when constructed the way Davidson intends. Even so, Davidson doesn't get off the hook. There are two reasons. First, Davidson cannot assume such a systematic approach to the interpretation of sentences without begging the question. The question at issue is whether a purely extensional theory of natural language is adequate, and the grotesque consequences are a sign of the inadequacy of purely extensional constraints on the analyses of sentences. Another sign are the cases in which the extensional interpretations of the parts of a complex expression do not combine to form a correct interpretation for the whole expression. An example of such a case is the failure of the interpretations of "John is good" and "John is a dancer" to give the correct interpretation of "John is a good dancer". Davidson mistakenly minimizes the linguistic and philosophical significance of such problems. 26 In the first place, such problems are very general: not only do they arise when "good" applies to other nouns such as, say, "doctor",

THE DOMINO THEORY

]5

"lawyer", etc., but they arise for many other words, e.g., extensionalist composition of "John is a clever" and "John is a boxer" does not give the correct interpretation for "John is a clever boxer". In the second place, such composition problems arise because extensional analyses can take no account of the way specific components of the sense of the modifier interact with specific compoents of the sense of the head, e.g., only the components of the sense of "doctor" which express the concept of medical duties to patients interact with the sense of "good". 27 But if such problems are merely another manifestation of the same underlying inadequacy of ignoring intensional constraints from which consequences like (3) arise, then Davidson cannot simply a s s u m e there is a way, in an extensional theory such as his, of systematically relating the interpretations of occurrences of an expression in different sentences. It is just this assumption that is being challenged in every one of these cases. The second reason is that, even if we allow Davidson to make this assumption and suppose that, as a result, he can avoid grotesque consequences like (3) being counted on a par with (2), still there will indefinitely many equivalent counter-examples which cannot be gotten rid of in this' way. Two strains that are resistant to "the essential proviso" are the following. One is the logical and mathematical parallels of (3) Such as the example we get by replacing "Snow is white" by "Two is less than three" and "Grass is green" by "The even prime is less than three" in the sentences (1)--(3). 28 Or, replace "Snow is white" by "There is no largest integer" and "Grass is green" by "There is no even prime greater than two". The other resistant strain is found among the sentences Quine uses to illustrate indeterminacy. Replace "Snow is white" by "Lo, a rabbit" and replace "Grass is green" by "Lo, a rabbit stage" or ,Lo, an undetached rabbit part". Davidson has also introduced a new requirement on instances of (T) designed to eliminate grotesque consequences like (3), namely, that instances of (T) "support appropriate counterfactuals".29 But this is a feeble requirement which, even if it eliminates counter-examples like (3), does little more. Taking the counterfactuals to have the form "If 's' were true, then it would be the case that p", let us grant for the sake of argument that (3) can be eliminated on the grounds that counterfactuals like "If 'Snow is white' were true, then grass would be green" are not

16

J E R R O L D J. K A T Z

appropriate. Presumably, the point is that forming counterfactuals from instances of (T) ought not give rise to false counterfactuals. But it is child's play to construct equally grotesque consequences which satisfy the requirement. For example, the instance of (T) "'Snow is white' is true if and only if natural selection produces white polar animals" gives us the proper counterfactual "If 'Snow is white' were true, then natural selection would produce white polar animals". Or, we can construct counter-examples simply by choosing non-synonymous expressions which are logically or mathematically equivalent. For example, one is "If 'Two occurs at the trillionth place in the decimal expansion of pi' were true, then it would be the case that the even prime occurs at the trillionth place in the decimal expansion of pi'. We have now seen how the fall of the first domino, Quine's arguments in "Two Dogmas of Empiricism", causes the fall of Quine's argument for indeterminacy and Davidson's argument for abandoning the intensionalist paradigm. The next domino is what I shall call "extensionalist possible world semantics". There have been two species of possible worlds semantics. Both characterize the notion of meaning and proposition in terms of extensions in possible worlds, but their accounts of possibility differ and so yield different notions of meaning and proposition. Carnap's semantics, the best known example of the intensionalist species of possible worlds semantics, derives from Frege. Frege had defined analytic statements as consequences of logical laws plus definitions. Carnap streamlined Frege's definition by eliminating the reference to definitions in favor of a new type of logical law, viz., meaning postulates, which expresses the extensional force of definitions. This resulted in the fullest possible expansion of the logical vocabulary. Not only modals but all other predicates standardly considered extra-logical vocabulary become logical vocabulary. 3~ Predicates like "is a bachelor" and "is married" occur as logical vocabulary within meaning postulates. Like logical posulates, meaning postulates constrain the model-theoretic interpretation of sentences so that there is no possible world in which something is a bachelor and married. As Quine points out, Carnap's possible worlds semantics is, accordingly, a form of intensionalism, and hence, will be vulnerable to the same objections as other forms of intensionalism? 1

THE DOMINO THEORY

17

Extensionalist possible worlds semantics, which receives its clearest expression in the work of David Lewis, 32 is possible worlds semantids tailored to conform as much as possible to Quine's scruples about meaning. Its account of possibility and proposition is based on a distinction between logical and extralogical vocabulary: predicates like "is a bachelor" and "is married" remain extra-logical vocabulary. But Quine's influence is not absolute here as it is with Davidson's semantics. Extensionalist possible worlds semanticists give modal concepts a central role in the logical form of sentences in natural language. Quine's objections to modal concepts are seen as tempered by recent achievements in modal logic. Extensionalist possible worlds semanticists think that, by allowing modals into logical vocabulary but otherwise treating the semantics of natural language in conformity to Quine scruples about meaning, they escape Quine's criticism of Carnap's position while gaining a semantic position more powerful than Davidson's. Thus, the particular virtue they claim for their position is that its notion of meaning and proposition succeeds in being less coarse-grained than Davidson's without being as vulnerably fine-grained as Carnap's. Quine's criticisms of meaning are essential to extensionalist possible worlds semantics because they provide it with the argument it requires in order to escape the objection that its notion of meaning and proposition is not fine-grained enough. Without Quine, critics could object that the finer-grained notion that comes from using synonymy as an identity condition does more semantic work. Quine enables these possible worlds semanticists to claim there is no finer-grained notion of meaning or proposition than the one that comes from using extensional equivalence over possible worlds as an identity condition. Thus, the failure of Quine's criticisms of meaning has essentially the same effect on extensionalist possible worlds semantics as it has on Davidson's semantics. This ought not be surprising, since the only substantial effect of having modal concepts is that extensionalist possible worlds semantics is able to distingusish propositions expressed by different contingent truths and propositions expressed by different contingent falsehoods. But, since possible worlds are extensionally defined, grotesque consequences of essentially the same sort as arose

18

JERROLD J. KATZ

for Davidson's semantics arise for extensionalist possible worlds semantics in the domain of necessary truths and necessary falsehoods. For example, necessary truths such as (4) and (5) must be counted the same (4) (5)

One is less two There is no decision procedure for quantificational validity

in meaning. This is surely as counter-intuitive as any of the grotesque consequences of Davidson's semantics. No one would paraphrase or translate (4) and (5) in the same way. No one would take them to contribute the same semantic content to sentences in which they occur as constituents. Inter alia, the logical equivalence of (4) and (5) is not sufficient for the logical equivalence of (6) and (7). Further, on extensionalist possible worlds semantics, each (6) (7)

Even small children know that one is less than two Even small children know that there is no decision procedure for quantificational validity

necessary truth asserts the same thing as its conjunction with all its logical consequences, and each necessary falsehood asserts everything. Finally, all necessary truths have the same meaning, and all necessary falsehoods have the same meaning. In consequence, there are only two meanings among the English sentences expressing logical and mathematical statements. 33 Hence, without Quine's criticism of meaning to neutralize such grotesque consequences, extensionalist possible worlds semantics goes the way of Davidsonian semantics. Without being able to appeal to Quine's criticisms to show that their notion of meaning and proposition is as fine-grained a notion as we can possibly have, extensionalist possible worlds semanticists have no way to meet the objection that the counter-intuitive consequences of their notion of meaning and proposition show it to be unacceptably coarse-grained. Thus, when a finergrained notion becomes a possibility, extensionalist possible worlds semantics can no longer defend its coarser-grained notion as an austerity measure required because there is nothing better. The anti-intensionalist arguments of Putnam and his followers seem prima facie to depend on no more than a simple exercise of the imagination, and hence, to be independent of Quine's arguments against

THE DOMINO THEORY

19

meaning. In the remainder of this essay, I shall show that their arguments depend on Quine as much as do Davidson's and those of the extensionalist possible worlds semanticists. Putnam's own arguments against intensionalism will be the fifth domino. The last domino will be Burge's related arguments against semantic individualism. Putnam's arguments have a common form. From "It Ain't necessarily So" to "The Meaning of 'Meaning' ", he employs science fiction type situations to try to show that sentences cannot have the sense structure that intensionalists suppose them to have. Putnam's idea is that such situations raise unforeseen possibilities which show that allegedly analytic and synonymous sentences express nothing stronger than contingent connections. 34 Taking analyticity and synonymy to be the basic concepts of the theory of meaning, Putnam first directs his argument against analyticity,35 and then against synonymy. Let us take (8), Putnam's original case of an allegedly analytic sentence, (8)

Cats are feline animals

as our example. For the sake of argument let us assume that intensionalists take (8) to be analytic. This is to asume that intensionalists claim that the sense of the subject of (8) contains the sense of its predicate. Putnam's argument against analyticity begins with the observation that it is perfectly imaginable that the things picked out in all prior referential uses of "cat" are in actuality Martian robots, not feline animals. From this correct observation, he tries to show that (8) cannot be analytic. His idea is that, if (8) were analytic, there could no possible case in which (8) is false, but the situation we have just imagined is precisely such a case. But does Putnam's thought-experiment really produce a case in which (8) is false? To answer this question, we have to be clear about what exactly it is that the experiment shows. Putnam's own discussion is open to two quite different interpretations on this point. On one, the experiment shows something tantamount to (9). On this interpretation, we have the potential for an (9)

It is possible that the referents in all of the statements people have made with (8) are not animals

20

JERROLD

J. K A T Z

argument of considerable force because even intensionalists must grant the possibility that the things people have been referring to in their uses of "cat" are not animals. Therefore, if Putnam can go on to show that (9) is incompatible with the analyticity of (8), then, since (8) is being taken as a representative analytic sentence, he has a knock-down argument against analyticity. On the other interpretation, the experiment straightoff shows that some cats are not animals. 3c' However, a flat assertion that the experiment directly exhibits things which are cats but not animals simply invites the response, "Yeah, sure, and next we'll hearing about round squares". On this interpretation, then, Putnam's argument forfeits the initial plausibility it has on the first interpretation. As the "Yeah, sure" response suggests, on this interpretation, Putnam begs the question: he concludes that the case is a case in which cats are robots without showing that what has been imagined are cats which are robots. We can clarify matters by considering a hypothetical ancestor of Putnam's in the Salem of 1692 who sets out to criticize the claim that (10) is analytic. (10)

Witches possess magical power

Ebenezer Putnam proposes a thought experiment in which the magical powers of suspected witches have a natural explanation. Surely, it will not do for Ebenezer Putnam to simply assert that the imagined case is one of something which is both a witch and an ordinary woman with no magical powers. A flat assertion begs the question of whether such a case is possible. Ebenezer's intensionalist opponents have the easy and plausible reply that the invented situation is merely one in which there are no witches. Correspondingly, Hilary Putnam's claim that his case is a case of something that is both a cat and a robot is subject to the easy and plausible reply that his invented situation is merely one in which there are no cats. This point can be reinforced as follows. Presuming synonymy, the analyticity of (8) and the analyticity of (11) come to the same thing. Now (11)

Feline animals are feline animals

(11) is a straighforward logical truth, so we aren't inclined to say

THE DOMINO THEORY

21

anything is a counter-example to it. A case of something which is both a feline animal and not a feline animal is impossible. But what goes for (11) goes for the synonymous (8), and hence, Putnam's thoughl-experiment cannot be taken as exhibiting something which is both a cat and a non-animal. Thus, Putnam's argument must start where it is on firm ground. The only such place is the undeniable (9). The question now is whether (9) can be used to refute the intensionalist's claim that (8) is analytic. Putnam's reason for thinking that it can is that intensionalists who accept the analyticity of (8) must deny (9) because their acceptance of the analyticity of (8) together with their position that sense determines reference commits them to (12). Thus, he (12)

It is not possible that the referents in all statements people have made with (8) are non-animal

argues that, since (9) is true, (12) is false, and so by modus tollens, the claim that (8) is analytic is false. This argument goes through if, as Putnam believes, intensionalism is committed to the position that sense determines reference. But is it? It is certainly true that Frege, the father of modern intensionalism, took this position. Frege defined the sense of an expression as a mode of determining its referent, and accordingly, the sense of an expression, on his definition, contains the information necessary to fix its reference in statements. 37 It is also true that Church, Carnap, and other influential intensionalists have followed Frege's lead. But regardless of how widely held, Frege's position is not essential to intensionalism. Elsewhere, l have argued at length that intensionalists ought to abandon Frege's position on the relation between sense and reference. 3s It is widely believed that intensionalists cannot abandon Frege's position because doing so would leave them without a definition of sense. This is a critical point because, to a great extent, the pressure on intensionalists to go along with Frege, as well the rationale of antiintensionalists for thinking that intensionalists have to go along, is a consequence of believing that Frege's way of defining sense is the only way open to intensionalists. Except perhaps for Quine's arguments themselves, this belief has been the single most important factor in the recent misfortunes of intensionalism.

22

JERROLD J. KATZ

To see that Frege's way of defining sense is not the only way, note that, in approach, Frege's definition of sense is, curiously, like the substitution criteria approach that Quine insists on for semantics, Both seek to define a concept from a family of concepts in terms of a relation between the concept and concepts outside the family. Thus, theoretical definition provides an alternative to Frege's definition of the concept of sense in terms of the concept of reference. On this approach, the concept of sense can be defined, independently of reference, in terms of its relations to other concepts in the theory of meaning: sense is the

aspect of the grammatical structure of sentences which is reponsible for their synonymy relations, their meaningfulness (or meaninglessness), their degree of ambiguity, their analyticity, and so on. Thus, sense structure is understood as the locus of grammatical properties and relations like synonymy and analyticity, in analogy to the way that syntactic structure is understood as the locus of grammatical properties and relations like same part of speech and well-formedness. Like syntactic structure, sense structure is to be explicated on the basis of a formal representation of an aspect of grammatical structure, but in the case of semantic representation, the aim is to reveal the configurations of relations that constitute the grammatical properties and relations like meaningfulness, ambiguity, synonymy, analyticity, etc. 3~ On this definition, sense is understood in terms of relations internal to the grammatical system of the language, not in terms of relations of language to the world. The relation of sense to reference is left as a further question to be taken up once significant progress is made in the theory of sense. Such a definition is, therefore, an even more consistent way to develop Frege's original distinction between sense and reference than his own definition. If the possibility of such a theoretical definition didn't exist, intensionalists would be Fregeans by default. In that case, Putnam would be right in thinking that they were committed to the position that sense determines reference, and consequently, his argument against them would go through. Hence, Putnam's argument requires the further premise that there can be no theoretical definition of sense in terms of concepts like analyticity and synonymy. But this premiss is available to Putnam only if it is illegitimate to use concepts like analyticity and synonymy in defining sense. The need for an argument to show that the

THE DOMINO THEORY

23

use of such properties and relations is illegitimate makes Putnam's antiintensionalist argument depend on Quine's. In order to see the nature of the dependency more clearly, let us briefly consider the lines along which a theoretical definition of analyticity can be given. Being analytic is no longer to be thought of as Frege thought of it, as being a consequence of laws of logic and substitutions sanctioned by defintion. 4~ Rather, being analytic is to be thought of as having a certain sense structure. Historically, this way of thinking of analyticity receives its clearest expression in Locke's doctrine of trifling propositions, and its most influential expression in Kant's account of concept containment. 4j Both Locke and Kant conceived of analytic propositions as ones whose subject concept includes its predicate concept. The only thing wrong with this -- apart from the completely informal character of containment -- is the restriction to prospositions of subject-predicate form. My reconstruction extends their characterization, not as Frege's reconstruction does by making definitional truth a species of logical truth, but by generalizing the notion of containment to cover the relations that occur in sentences like "John marries those he weds", "Mary walks with those with whom she strolls", "John and Mary borrow books from anyone who lends books to them", etc. 42 A sentence is analytic, on my reconstruction, just in case the sense of its subject, its direct object, or any of its constituents which function as a term literally contains the sense of its full predicate together with the senses of the other constituents which function as terms. On the basis of such a theoretical definition, it is apparent that the claim that (8) is analytic is only a claim about the sense structure of (8). There is no assertion about its reference. To be sure, when we turn to the relation of sense to reference, we will want to correlate analytic sense structure with the referential property that analytic sentences which make statements make only true ones. But, in such a correlation, we are connecting a feature of a sentence in the language, i.e., a feature of a type, with a feature of its utterances in a context, i.e., tokens. Thus, the claim that (8) is analytic, in and of itself, implies nothing about whether this, that or the other utterance or token of "cat" refer to feline animals or Martian robots. Hence, intensionalists, as opposed to Fregean intensionalists, need

24

JERROLD J. KATZ

face no difficulty in connection with Putnam's imaginary situation. They have the option of theoretically defining sense, and if they do so, their claim that (8) is analytic is only a claim about the structure of the sense of the English word "cat" and the sentences in which it occurs. Intensionalists are then free to take the non-Fregean position that utterances of words take on reference as a function not only of its sense but also of the conditions of utterance. As a consequence, intensionalists can avail themselves of Donnellan's insight into reference under a false description, and say that, in cases like Putnam's, tokens of "cat" refer to robots under the false (semantic) description 'feline animal'. 43 This is a very natural account of such cases because, until the news breaks, the utterance conditions involve the belief on the part of all speakers that apparent cats are real ones. Thus, tokens of "cat" have referred to robots under the false description 'feline animal' because speakers have been living in a collective fool's paradise about the objects that appear to be cats. Therefore, Putnam cannot take it for granted that, when intensionalists claim that (8) is analytic, they accept (12). As a consequence, Putnam's thought-experiment which establishes the truth of (9) provides no basis for inferring that (8) cannot be analytic. Putnam has a powerful argument against Fregean intensionalism, but without Quine's arguments to block the use of sense properties and relations in a theoretical definition of sense, Putnam has no argument against intensionalism p er se.

In "The Meaning of 'Meaning' ", Putnam recasts his argument as an argument against synonymy, but, as with the previous version of the argument, his aim remains that of refuting the intensionalist view of meaning. 44 H e r e he is completely explicit that his argument rests on an intensionalism which involves Frege's conception of the relation between sense and reference. Putnam says that the traditional concept of meaning has come to rest on two assumptions:

O) (Ii)

That knowing the meaning of a term is just a matter of being in a certain psychological state. That the meaning of a term (in the sense of "intension") determines its extension (in the sense that sameness of intension entails sameness of extension). 4~

THE DOMINO THEORY

25

Given this, he claims that his argument shows that "these two assumptions are not jointly satisfied by a n y notion, let alone any notion of meaning", thereby proving that "The traditional concept of meaning is a concept which rests on a false theory". 4~' In order to show this, Putnam introduces a new science fiction case: Twin Earth is very much like Earth... In fact, apart from differences we shall specify the reader may suppose that Twin Earth is exactly like Earth. he may even suppose that he has a doppelganager -- an identicalcopy -- on Twin Earth.47 9

Putnam gives the argument against synonymy in three forms. In the first form, there is no H 2 0 on Twin Earth and in its place is stuff chemically different from HzO, namely, XYZ, which is indistinguishable from Earth water under normal conditions. Since my doppelganger has the same psychological states as I do, but the extension of my English word "water" is H 2 0 while the extension of his Twin Earth English word "water" is XYZ, (I) and (II) can't both be true. In the second form, my words "aluminum" and "molybdenum" refer, respectively, to aluminum and molybdenum, but my doppelganger's language switches the terms extensionally, so that his "aluminum" refers to molybdenum and his "molybdenum" refers to aluminum. 4's Again, we have sameness of psychological states and difference of extensions, and hence, (I) and (II) cannot both be true. In the third form, "elm" and "beech" take the place of the chemical terms, with the slight twist that Putnam puts the reasoning in terms of the fact that his own concepts of elms and beeches are the same, but the extensions of "elm" and "beech" in his idiolect differ in the way they do for other speakers of the language. 4t) Again, same psychological states and different extensions, and hence, same conclusion. We may grant that this argument, in all of its forms, shows (f) and (II) are not jointly satisfiable. We may also concede that it shows that any theory resting on them is a false theory. We may concede as well that Frege's theory rests on them, and therefore, is shown to be a false theory. In making these concessions. I accept (I) as an informal statement of the conditions for knowledge of the meaning of a term, taking the psychological state in question to be one in which the agent grasps the meaning of a term and takes the meaning grasped to be a sense. 5~ Putnam's argument works against Frege because (II), his principle

26

J E R R O L D J. KATZ

connecting sense and reference, connects them so strongly that possibilities concerning the world refute intensionalist claims about the language. But P u t n a m claims that his argument does m o r e than refute just one explication of the traditional c o n c e p t of meaning. H e claims to have refuted the traditional c o n c e p t of meaning itself by showing that it "rests on a false theory". This stronger claim is mistaken. Since the traditional c o n c e p t itself does not rest on Frege's principle (II), P u t n a m has not shown it to rest on a false theory. T h e fact that (I) and (II) cannot both be true is irrelevant to an explication of the traditional c o n c e p t based the n o n - F r e g e a n intensionalism described above. O n such an intensionalism, joint unsatisfiability is entirely a matter of the truth of (I) and the falsehood of (II). Recently, P u t n a m has considered a version of the criticism that his a r g u m e n t doesn't work if intensionalists d r o p the claim that sense determines reference. 51 P u t n a m replies: if we go the way I went in "The Meaning of 'Meaning' ", and take the extension of a term as one of the components of its 'meaning vector', then we have a clear reason for saying that 'grug' has a different meaning in north Ruritanian and in South Ruritanian. Since the word ~grug'clearly has a different reference in the two dialects of Ruritanian, it also has a different meaning9 Once we decide to put the reference (or rather difference in reference) aside, and to ask whether 'grug' has the same 'content' in the minds of Oscar and Elmer, we have embarked upon an impossible task. Far from making it easier for ourselves to decide whether the representations are synonymous, we have made it impossible . . . . 'Factoring out' differences in extension will only make a principled decision on when there has been a change in meaning totally impossible. 5z 9

Putnam's claim that the task of determining sameness of content without recourse to reference is impossible is wrong. O u r n o n - F r e g e a n definition of sense automatically provides such a determination of sameness of content, namely, two expressions have the s a m e content, are s y n o n y m o u s , j u s t in case any sense property or relation o f one is a sense property or relation o f the other. In the case of particular expressions, the task of determining sameness and difference of sense will be a matter of inference to the best explanation of the sense properties and relations of the expressions, where the best explanation is the one which most simply accounts for the available evidence concerning properties and relations. F o r example, it would be evidence for the s y n o n y m y of "chair" and "piece of furniture serving as a seat for

THE DOMINO THEORY

27

one" that "chair which is a piece of furniture serving as a seat for one" is redundant, and it would be evidence against their synonymy that "chair" and "stool" differ in sense and that "stool which is a piece of furniture serving as a seat for one" is also redundant. Putnam is quite sure that determining synonymy without recourse to reference is impossible -- he says this three times in the above passage. Since he gives no reasons, we have to assume that he thinks it is unnecessary to give them because the impossibility of such determination is obvious on the face of it. But why should Putnam should think it obvious? The only thing around that could make someone like Putnam think this is Quine's arguments against sense properties and relations, for such properties and relations are required for a nonFregean intensionalism. This being the case, the fall of the first and second dominos undercuts Putnam's reply to the rejoinder that intensionalists can drop the principle that sense determines reference. But, without the reply, the rejoinder stands, and the rejoinder shows that he has no argument against intensionalism itself. The Putnam's "The Meaning of 'Meaning'" has stimulated a number of arguments which develop its ideas. The most interesting and influential of them are Tyler Burge's arguments against semantic individualism, i.e., the view that the semantics of language is independent of sociological relations within the linguistic community and other environmental factors. 53 Burge's arguments are anti-intensionalist because intensionalism, as the position is undertstood here, is individualistic. It embodies the doctrine that senses are autonomous grammatical objects and that knowing the senses of an expression is a matter of grasping an aspect of its grammar and nothing more. In "Other Bodies", Burge imagines a Twin Earth scenario where the focus is not on meanings p e r se but "belief contents", i.e., the conceptual or semantic aspects of what is believed, or more generally, the proposition associated with the clause functioning as the complement of a verb expressing a propositional attitudeY Burge's argument can be sketched as follows. Compare my belief contents involving the notion of water with those of my doppelganager on Twin Earth where there is no H20 but only XYZ (where there is no water but only twater, to use Burge's way of putting it). Even though there is no difference between my doppelganager and me with respect to what's in our heads (includ-

28

J E R R O L D J. KATZ

ing non-intensional p h e n o m e n a l experiences, dispositions to respond, etc.), n o n e of the belief contents with the notion of water in them which can be ascribed to me can be ascribed as well to m y doppelganger. F o r example, my d o p p e l g a n a g e r in his Twin Earth bar will not have my belief that his drink has been watered, but, rather the belief that the drink has been twatered. Since such differences in content can only be a function of the difference in our enviroments, Burge concludes that semantic individualism is false. Burge takes P u t n a m ' s reasoning in an interesting direction: When Adam says or consciously thinks the words, 'There is some water within twently miles, I hope', Adamte says or consciously thinks the same word forms. But there are differences. As Putnam in effect points out, Adam's occurrences of 'water' apply to water and mean water, whereas Adamte's apply to twater and mean twater. And, as Putnam does not note, the differences affect oblique occurrences in 'that'-clauses that provide the contents of their mental states and events. Adam hopes that there is some water (oblique occurrence) within twenty miles. Adamte hopes that there is some twater within twenty miles. That is, even as we suppose that 'water' and 'twater' are not logically exchangeable with co-extensive expressions salva veritate, we have a difference between their thoughts (thought contents). 5~ Thus, f r o m the premiss that A d a m ' s w o r d "water" means water and A d a m t e ' s w o r d "water" means twater, Burge draws the conclusion that the m e a n i n g / t h o u g h t contents of the propositions which express their hopes are different. Since Burge's premiss is that the belief contents for A d a m and A d a m t e and the meanings of their c o r r e s p o n d i n g words are different, his a r g u m e n t d e p e n d s on (II) much as Putnam's Twin Earth a r g u m e n t does. W i t h o u t (II), there is no way to go f r o m the supposition that A d a m ' s utterances of "water" refer to water and A d a m t e ' s utterances of "water" refer to twater to the conclusion that "water" means w a t e r and "twater" means twater. Only if Burge's inferences to meanings or belief contents take (II) as a premiss can he project up from a difference in the referents of "water" and "twater" to a difference in the meaning or belief content of these words. But, given m y n o n - F r e g e a n intensionalism, there are semantic individualists, "rugged individualists", as I shall call them, w h o reject (II). Hence, Burge's a r g u m e n t is vulnerable in the same way as P u t n a m ' s original T w i n - E a r t h argument. W i t h o u t that premiss, there is no reason for thinking that the meanings of "water" and "twater" are different or that the belief contents for A d a m and

THE DOMINO THEORY

29

Adamte are different. Hence, in the absence of Quinean arguments to secure (II), this one of Burge's criticisms of semantic individualism fails. Let's examine the point in more detail. On the rugged individualist's conception of meaning, it can well be that "water" and "water" apply to very different substances but, nonetheless, are fully synonymous. One way this could happen is for "water" and "water" to be exact translations, for example, both have the sense (13). Thus, "water" and "twater" both pick out H20 on (13)

naturally colorless, odorless, and tasteless liquid that descends from the sky as rain

Earth, where there is only H20 , and these words both pick out XYZ on Twin Earth, where there is only XYZ. In each case, the substance is picked out because it is a naturally odorless, colorless, and tasteless liquid that descends from the sky as rain. Assuming the synonymy of "water" and "twater", Putnam's and Burge's imaginary situations correspond to certain real situations. For example, in the actual world, the sense of "anesthetic", namely, 'something which causes temporary loss of sensation in a part of the body'. picks out chemically different substances such as ether, chloroform, novocain, etc. The sense does this because each substance has the defining causal property. This reflection shows that Burge's and Putnam's use of (II) in inferences about meaning leads to a reductio of those inferences, since we can conclude, reasoning as they do, that "anesthetic" means different things in application to ether, chloroform, novocain, etc. "Anesthetic" would be highly ambiguous, constantly increasing in degree of ambiguity with the discovery of new anesthetics. It thus seems clear that the use of (II) as a principle constitutive of meaning is just another way of confusing meaning and reference. At some points in their discussion, Putnam and Burge claim intuitive support for their arguments. This claim is incorrect: their position is in no way supported by the intuitions in question, since those intuitions are easily accommodated by the intensionalist position. Putnam may well be correct that "it is intuitively plausible that earthling space explorers reporting back will say that Twin Earth has no water instead of saying that it has water with the chemical structure XYZ. 5~ But without the

30

JERROLD J. KATZ

reasoning based on (II), such intuitive evidence about our use of "water" is easily interpreted consistent with the intensionalist position. Intensionalism p e r se says nothing concerning what particular words mean in the language. It does not concern descriptive lexicography; it is a meta-theory for grammars, concerning itself with the form of the semantic component in grammars and how semantic evidence is brought to bear in choosing among semantic hypotheses. Accordingly, intensionalism doesn't say that the expression "chemical anesthetic" will be judged redundant, but only that if it is, then that fact is evidence for a definition of "anesthetic" which excludes something like hypnosis. Hence, the intensionalist can accept the intuition that the space explorers will report that there is no water on Twin Earth. Such intuition can simply be taken to reflect the lexical fact that the English word "water" means 'substance with the composition H20'. It may be that there has been a language change due to the chemical nature of water becoming so widely known, so that now "water" no longer means (13). Or perhaps there has just been an addition to the dictionary, so that "water" now has two senses, a technical, scientific sense referring to its molecular structure and an ordinary sense like (13). The use of "water" in the former sense on the part of space explorers can be explained as a natural preference for a technical sense in a technical context. So much for Putnam's appeal to intuition. Let us turn to Burge's. He writes: We now consider whether the same natural kind terms should occur obliquely in attributions of propositional attitudes to Adam and Adamte. Let us assume, what seems obvious, that Adam has propositional attitudes correctly attributed in English with his own (English) natural kind terms in oblique position. He hopes that there is some water within twenty miles; he believes that sailboat masts are often made of aluminium, that elms are deciduous trees distinct from beeches, that shrimp are smaller than mackerel. Does Adamte have these same attitudes, or at least attitudes with these same contents? As the case has been described, I think it intuitively obvious that he does not.57 Burge's phrase "as the case has been described" can cover two quite different things, but in neither does intuition show what Burge supposes. On the one hand, the case has been described as one in which Adamte's word "water" means something different from Adam's word "water". U n d e r this description, Burge is correct to say that Adamte does not have the same attitudes with the same contents as Adam. But here intuition is idle, since it serves merely to reiterate what is already

THE DOMINO THEORY

31

in the description. The description must carry the weight of the argument. But, as seen above, it rests on (II), and hence, cannot support the conclusion. On the other hand, the description may not include a specification of the alleged semantic dfference between "water', and "twater". But, under such a neutral description, intuition cannot decide whether Adamte has or does not have attitudes with the same contents. The case itself is underdetermined: we can construe the case along the lines of whatever semantic view we come to it with. Suppose we come to the case with the view that "water" means water and "twater" means twater, either because we share Putnam and Burge's assumptions or because, although we do not, we know about the meaning change brought about by widespread scientific sophistication among the populations o f Earth and Twin Earth. Then we will "intuit" that Adamte and Adam have attitudes with different contents. Alternatively, suppose we come with the view that "water" and "twater" have the same meaning, say, (13), but that they refer to different stuff depending on what there is on Earth and Twin Earth (as "anesthetic" can refer to different stuff). Then Adamte and Adam have attitudes with the same content. Therefore, intuition -- pure and simple -- plays no role in deciding whether the same or different concepts occur obliquely in attributions of propositional attitudes to the Adams. However, Burge does not intend his argument to rest on intuition. He says that "two broad types of consideration back the intuition" that different concepts occur obliquely in such attributions. The first consideration is that "it is hard to see how Adamte could have acquired thoughts involving the concept of water without contact with water, or someone having contact with water". 58 There are two replies to this consideration. First, Burge's presupposition that Adamte has no contact with water can be challenged. We can claim that Adamte d o e s have contact with water insofar as he has contact with a naturally colorless, odorless, and tasteless liquid that descends from the sky as rain. Burge has not yet shown that the rugged individualist would be wrong to take any liquid that satisfies the definition (13) as the referent of "water". Second, even if we allow that Adamte does not have contact with water, it is quite implausible to suppose that intelligent beings cannot acquire concepts of things that neither they nor anyone they know have

32

JERROLD J. KATZ

had contact with. Science fiction stories, scientific treatises, fairy tales, tall stories, novels, etc. are full of concepts without contact. There are space-ships which travel across galaxies, teletransportation, death rays, dragons, witches, and even a ten dimensional universe consisting of tiny strings forming vibrating loops. Adamte could acquire the concept of water in the way that Scientists develop a concept of something whose existence is only a matter of sheer speculation. It would be no use replying that, in order to imagine abstract concepts, scientists must start in contact with natural phenomena and build up theories step by step. The natural phenomena with which scientific theories are in contact may be quite different from ,the theoretical objects they conceptualize, and very indirectly connected~with them. Hence, there is no reason to think that the natural phenomena with which Adamte may have contact, even though quite different from water, would not suffice to enable him to conceptualize water. The second type of consideration which Burge thinks backs the intuition is that, if we were to say Adamte had beliefs involving the concept of water, then "a large number of his ordinary beliefs will be false . . . But there seems no reason to count his beliefs false and Adam's beliefs true (or vice versa)'. 59 Neither of these points has force. As long as an intensional notion of concept has not been ruled out, it is not sufficient for counting a large number of Adamte's beliefs false that they involve the concept of water. Whether they are counted as false depends on what concept of water it is that figures in Adamte's beliefs. If it is (13), S e n his beliefs, say, those about the stuff boats sail on and mackerel swim in, will be largely true. Of course, if Adamte's concept of water is the concept 'stuff with the composition H20' , then a large number of his beliefs will be false. However; in this latter case, there is, contra Burge, a good reason for asymmetry. Counting Adamte's beliefs largely false and Adam's beliefs largely true follows from the fact that what Adamte designates as stuff with the composition H20 does not have that composition, while what Adam designates as stuff with the composition H20 does have that composition. What makes an asymmetry s e e m misguided is that Adamte gets along in ordinary circumstances as well as Adam in spite of the fact that he has a bunch of false beliefs while Adam's corresponding beliefs are true. But there is really no problem about why Adamte gets along

THE DOMINO THEORY

33

equally well. The case of Adamte with his false beliefs about X Y Z corresponds to the case of people with their many false beliefs about cat-like robots in Putnam's counterfactual. The only reason it might seem misguided to someone to count a large number of Adamte's beliefs false is that they overlook the fact that false beliefs, if shared, can be just as good as true ones. If people in Putnam's counterfactual want to rid their house of mice or stop the horrible meowing from the back fence, they can use 'Tll buy that cat" or "Shoot the cat making that horrible noise". Such utterances will achieve reference under the false description 'feline animal', and hence, they will suffice for the purpose. A functioning robot will produce the desired absence of mice in the one case, and a non-functioning robot will produce the desired absence of horrible noise in the other. Correspondingly, if A d a m t e wants to obtain something to drink or wants to wash up before dinner, he can use of "A glass of water, please" or "Where is the water faucet?". Such utterances will do the trick because they will achieve reference to X Y Z under the false description 'stuff with the composition HzO'. Burge has other forms of the argument against semantic individualism in which the notion of arthritis takes the place of the notion of water. 6~ People have various and sundry "arthritis thoughts", i.e., thoughts in which the notion of arthritis figures as the object of the propositional attitude. Now suppose that, not knowing the lexical fact that "arthritis" is defined as a condition of the joints, a patient thinks that his or her arthritis has spread to the thigh, perhaps overgeneralizing in the way that infants do in applying "daddy" to the mailman. Burge contrasts this case of a patient who thinks the painful inflamation in his or her thigh is arthritis with the hypothetical case of a patient in a community where everyone uses "arthritis" in the way the lexically ignorant patient does. Perhaps arthritis has not as yet been isolated as a disease of the joints in this Twin Earth community. Burge wants to say that, whereas the first patient has arthritis thoughts, his or her doppleganger in the TwinEarth community would lack arthritis thoughts. Burge explains: We suppose that in the counterfactual case we cannot correctly ascribe any content clause containing an oblique occurrence of the term 'arthritis'. It is hard to see how the patient could have picked up the notion of arthritis. The word 'armritis' in the counterfactual community does not mean arthritis. It does not apply only to inflammations of the joints. We suppose that no other word in the patient's repertoire means arthritis. 'Arthritis', in the counteractuai situation, differs both in dictionary definition and in

34

JERROLD J. KATZ

extension from 'arthritis' as we use it. Our ascriptions of content clauses to the patient (and ascriptions within his community) would not constitute attributions of the same contents we actually attribute. For counterpart expressions in the content clauses that are actually and counterfactually ascribable are not even extensionally equivalent. However we describe the patient's attitudes in the counterfactual situation, it will not be with a term or phrase extensionally equivalent with 'arthritis'. So the patient's counterfactual attitude contents differ from his actual ones.6~ The moral of the contrast is supposed to be that the semantic differences between the patients can only be chalked up to the social differences in their environments. But why assume there is any contrast in the first place? True enough, the term "arthritis", as we use it, expresses the concept of a rheumatic ailment of the joints, but Why suppose that that the linguistically deviant patient in our community should be described differently from his or her doppelganger in the Twin-Earth community? Burge gives no argument for supposing this. Morever, it seems quite natural to say that someone who is linguistically deficient in lacking the concept of arthritis, as a consequence, would lack the attitudes which involve content clauses containing the concept. Thus, the linguistically deficient patient should lack arthritis attitudes every bit as much as does his or her Twin-Earth doppelganger. Additional support for saying this comes from the analogy between the linguistically deviant patient who overgeneralizes the sense of "arthritis" and the infant who overgeneralizes the sense of "daddy". We wouldn't say that the infant who calls every man "daddy" has attitudes whose description requires the concept of being a father, so by parity of reasoning, we oughtn't say that the patient who calls every rheumatic ailment "arthritis" has attitudes whose description requires the concept of being a rheumatic ailment of the joints. 62 Further, Burge is wrong to say that "It is hard to see how the patient [in the Twin-Earth community] could have picked up the notion of arthritis". It is easy to see how this could happen. Consider one example. There is a type of patient who glories in illnesses. The more the merrier. Such a patient invests much time and energy thinking about new disease possibilities, and delights in adding them to the list of their afflictions. Surely, even moderately imaginative and medically informed patients of this type could hit on the happy thought that, instead of an inflammation spreading to the thigh -- a piece of news unlikely to attract much attention - - they have, in fact, contracted a new disease, a

THE DOMINO THEORY

35

distinct ailment consisting of inflammation of the joints. If patients are presumed not to be in the proper position to identify a new disease, we can imagine that, rather than a neurotic patient, our subject is a hero of Twin-Earth medical science. Finally, Burge is also wrong to suggest that the absence of a word for arthritis in the patient's repertoire means that the notion will not play a role in his or her thoughts. But how could absence of a word be sufficient to prevent someone in the counterfactual community, no matter how neurotically motivated, medically informed, and scientifically imaginative, from coming up with the arthritis concept? If absence of a word were sufficient, we would face extreme conceptual stagnation. There are many counter-examples. The notion of a bad smelling house was part of many people's thoughts well before Madison Avenue coined the term "housitosis". Further, English speakers have no trouble forming the thought that dying from lack of water would be worse than dying from lack of food despite the fact that English has no word meaning 'death from lack of water' parallel to "starvation". NOthing more is required of an English speaker is such circumstances than to recognize that the meaning of "death", "lack", "water", and the causal preposition combine to form the compositional meaning parallel to "starvation". Burge's suggestion is, therefore, both paradoxical and arbitrary. It is paradoxical because it would make it impossible to understand how the formation of novel concepts occurs. It is arbitrary because it doesn't explain why words should be accorded a special status. Why should absence of a word meaning 'arthritis' be critical when, thanks to compositionality, other grammatical units, like pharases or clauses, are available to express the meaning? Since, obviously, there is no rationale for giving this special status to one category of syntactic signs rather than another, it can't matter how the patient refers to the condition. Having no word meaning arthritis, the patient can coin a new term or construct a phrasal or clausal description to express the meaning. There can be no objection to assuming that there are descriptions synonymous with "arthritis", since we have removed the arguments for supposing synonymy cannot be made objective sense of. It seems quite straightforward, then, for us to describe the patient as informing everyone that he or she suffers from the newly identified disease arthritis. Such a

36

JERROLD

J. K A T Z

description is no different from our describing early science fiction authors as having thought about ray guns. The purpose of this paper has been to show that the principal antiintensionalist arguments over the last three decades are not a collection of distinct criticisms, each standing or falling on its own merits, but have a structure in which Quine's early argument against the theory of meaning underlies all the others. If my criticisms of Quine's argument, summarized at the beginning of the paper, are correct and if my domino theory is also correct, intensionalism emerges unscathed from the host of arguments that have been thought to refute it many times over. NOTES * The first draft of this paper was presented to the Eleventh Annual Philosophy Symposium, "Thought and Reference", of the University of North Carolina at Greensboro, April 10--12, 1987. The author wishes to thank Ned Block, William Fisk, and Virginia Valian for helpful suggestions. Quine, W. V. "Two Dogmas of Empiricism", in From a Logical Point of View, Harper & Row, New York, 1953, pp. 20--46. 2 The most detailed presentation is found in Katz, J. J. "The Refutation of Indeterminacy", The Journal of Philosophy, Vol. LXXXV, No. 5, May 1988, pp. 1--26. 3 Chomsky, N. Syntactic Structures, Mouton & Co., The Hague, 1957. 4 Quine, W. V. Word and Object, M.LT. Press, Cambridge, 1960, p. 72. 5 Ibid., pp. 26--79. 6 Ibid., pp. 71--72. 7 Ibid., pp. 73--79, and pp. 205--206.

8 Ibid.,p. 76. 9 Ibid., pp. 76--79. lo Ibid., pp. 205--206. ~N For example, Quine, W. V. "The Problem of Meaning in Linguistics", in From a Logical Point of View, pp. 56--64, and also Quine, W. V. "Indeterminacy of Translation Again", The Journal of Philosophy, Vol. LXXXIV, No. 1, 1987, pp. 5--10. See my discussion in "The Refutation of Indeterminacy", pp. 9--11. 12 Quine claims that appealing to bilinguals assumes a mistaken mentalism, but this claim seems to confuse an appeal to a bilingual's overt judgments about expressions with an explanation of them which invokes mentalistic notions. In any case, the appeal to bilinguals in translation is no more a difficulty than the appeal to speakers of a language in ordinary grammar construction. See my discussion in "The Refutation of Indeterminacy". ~-~ Davidson, D. "Truth and Meaning", Synthese, Vol. XVII, No. 3, 1967, pp. 304-323; reprinted in Readings in the Philosophy of Language, J. F. Rosenberg and C. Travis, eds., Prentice-Hail, Inc., Englewood Cliffs, 1971, p. 455 (references are to this reprint). ~4 Davidson, D. Inquiries into Truth and Interpretation, Clarendon Press, Oxford, 1985, p. 101. 15 "Truth and Meaning", p. 455.

THE DOMINO THEORY

37

16 Ibid., pp. 453--454. 17 Mates, B. "Synonymity", in Semantics' and the Philosophy of language, L. Linsky, ed., University of Illionois Press, Urbana, 1952, pp. 201--136. 18 "Truth and Meaning", p. 461. ~9 Church, A. "Intensional Isomorphism and Identity of Belief", Philosophical Studies', Vol. 5, 1954, pp. 65--73. 2o Inquiries into Truth and lnterpretation, pp. 101--102. 21 Thus, not only is Davidson's constructive program a direct implementation of Quine's recommendation of the Tarskian paradigm at the end of his paper "Notes on Theory of Referece", in From a Logical Point of View, p. 138, but Davidson's rationale is Qnine's rationale that the theory of truth is preferable to the theory of meaning on the grounds that truth is free of the ills that allegedly plague meaning. 22 "Truth and Meaning", p. 457 23 "Truth and Meaning", p. 457. 24 In Davidson, D. Inquiries into Truth andlnterpretation, p. 26, In. 10

-,'~ Ibid.,p. 26, fn. 10. 26 Davidson, D. "The Logical Form of Action Sentences", in Essays on Action and Events, Oxford University Press, New York, 1980, pp. 81--120. _,7 Katz, J. J. Leacock, C., and Ravin, Y. "A Decompositional Approach to Modification", in Action and Events, B. McLaughlan and E. LePore, eds., Basil Blackwell, Oxford, 1986, pp. 2~ Katz, J. J. "Logic and Language: An Examination of Recent Criticisms of Intensionalism", in Language, Mind, and Knowledge, Minnesota Studies in the Philosophy of Science, Vol. VII, K. Gunderson, ed., University of Minnesota press, 1975, pp. 63--76. 2L) In "Inquiries into Truth and Interpretation", p. 26, fn. 11. .~0 Camap, R. "Meaning Postulates", in Meaning and Necessity, University of Chicago Press, Chicago, Enlarged Edition, 1956, pp. 222--229. 31 Quine points out (From a Logical Point of View, p. 23) that Carnap's broader notion of possibility falls within the scope of his criticisms because the specification of possible worlds assumes knowledge of the incompatibility of predicates like "bachelor" and "married". 32 Lewis, D. "General Semantics", in Semantics of Natural Language, D. Davidson and G. Harman, eds., D. Reidel Publishing Co., Dordrecht-Holland, 1972, pp. 169--218. 33 The revision of possible worlds semantics know as "situation semantics" improves matters somewhat but it doesn't address the basic problem. Barwise, J. and Perry, J. Situation Semantics, M.I.T. Press, Cambridge, 1983. Situation semantics introduces the idea that the interpretation of a sentence is a situation type, a set-theoretic construction out of pieces of actual situations, which may or may not reflect an actual situation. The idea avoids the Fregean move of taking sentences to designate truth values, and as a result, enables those who adopt this position to avoid having to say that there are only two necessary propositions. But this doesn't help much with the general problem of having an exclusively extensional characterization of the notion of a proposition. The same situation type of John being a bigamist has to interpret the non-synonymous sentences "John has two wives" and "The number of John's wives is equal to the even prime". Further, having cut themselves off from anything like the Fregean device of referring to the sense of a sentence in an opaque context, Barwise and Perry have to allow substitutions on the extensional condition of sameness of situation type, leading to familiar counter-intuitive results in connection with verbs of propositional attitude. Barwise and Perry bite the bullet, taking the position that there are no semantic reasons for a reluctance to make inferences on the condition of sameness of situation type, but merely pragmatic ones (pp. 199--200). They claim that the semantic notion of a proposition that underlies the appeal of those familiar counter-intuitive results is misguided, based on our having lost our blessed state of semantic innocence in which

38

JERROLD

J. K A T Z

expressions stand directly for objects in the world and sentences for situations. But this claim flies in the face of a wide variety of observations about our ordinary linguistic talk which show it to be shot full of intensionalist idiom. See Soames, S. "Lost Innocence: Comments on Situations and Attitudes", Linguistics and Philosophy, Vol. 8, No. 1, 1985, pp. 59--71. Also, see my "Why Intensionalists Ought Not Be Fregeans', pp. 90-91, where I discuss the way that Frege's narrowing of the scope of a theory of meaning has encouraged extensionalists to think there are very few pre-theoretic semantic phenomena that theories must do justice to. 34 Putnam, H. "It Ain't Necessarily So", The Journal of Philosophy, Vol. LIX, No. 2, 1962, pp. 658-671, "Is Semantics Possible?", Metaphilosophy, Vol. I, 1970, pp. 187-201, "The Meaning of 'Meaning' ", in Language, Mind, and Knowledge, Minnesota Studies in the Philosophy of Science, pp. 131-- 193. 35 Thg~ is to say, analytic in the full-blooded intensionalist sense. Putnam introduces his own notion of analyticity in "The Analytic and the Synthetic" in Minnesota Studies in the Philosophy of Science, vol. Ill, H. Feigl and G. Maxwell, eds., University of Minnesota Press, Minneapolis, 1962, pp. 369--397, but this is a pale copy which offers no basis for necessary truth. Putnam's claim to be replying to Quine seems to misunderstand the Kantian and Fregean focus of Quine's criticisms. Putnam's notion of analyticity would be quite acceptable to Quine. 36 "It Ain't Necessarily So", p. 659. 37 Frege, G. "On Sense and Reference", in Translations from the Philosophical Writings of Gottlob Frege, M. Black and P. Geach, eds., Basil Blackwell, Oxford, 1952, pp. 56-78. I use the term "determine" rather than the more usual "present" as closer to Frege's intent. See Sluga, H. Gottlob Frege, Routledge & Kegan Paul, London, 1980, pp. 150-153. 3s "Why Intensionalists Ought Not Be Fregeans'. 39 Frege's failure to develop his distinction between sense and reference has mistakenly encouraged critics of intensionalism to think that a theory of sense is somehow deficient if it doesn't contain an account of truth and reference. For example, David Lewis makes this objection in Lewis, D. Convention, Harvard University Press, Cambridge, 1969, pp. 170--171. Such a criticism would be correct if the intensionalist theory being criticized were based on Frege's definition of sense. The criticism fails to recognize that the fundamental move in creating the theory in question is to replace Frege's definition of sense with one that makes no appeal to reference, and consequently, to reject Frege's notion of analyticity as a mistaken unification of definition and logical law. 4~ Frege. G. Foundations of Arithmetic, Basil Blackwell, Oxford, 1953, pp. 3--4. 41 Katz, J. J. Cogitations, Oxford University Press, Oxford, 1986, pp. 41 --97. 42 This is the conception in Katz, J. J. Semantic Theory, Harper & Row, New York, 1972, pp. 171--182, and developed further in Cogitations and Smith, G. E. and Katz, J. J. Supposable Worlds, Harvard University Press, to appear. 4.~ Donnellan, K., "Reference and Definite Descriptions", The Philosophical Review, Vol. LXXV, 1966, pp. 281--304. 44 "The Meaning of 'Meaning' ", p. 136. 45 Ibid.,pp. 135--136. 46 Ibid., p. 136. 47

Ibid.,

p. 1 3 9 .

48 Ibid., pp. 142--144. 49 Ibid.,pp. 134--139. 5t~ See also Ibid., pp. 134--139. 5~ Putnam neglects the earliest and most complete form of the criticism, see "Logic and Language: An Examination of Recent Criticisms of lntensionalism', especially, p. 98. The form of the critictism Putnam chooses to respond to makes it easy to come up with a response. This form of the criticism that the principle that sense determines reference

THE DOMINO THEORY

39

can be dropped is not accompanied with any account of how relations among senses can be determined independently of reference. Without such an account, Putnam can respond, as the does, that it is impossible to determine such relations if the principle is dropped. See the text for a discussion of what happens to such a response when there is an account of how sense relations can be determined independently of reference. 52 Putnam, H. "Computational Psychology and Interpretation Theory", in Realism and Reason, Philosophical Papers, Volume 3, Cambridge University Press, Cambridge, 1983, p. 149. 53 The argument is presented in a number of places, with a number of twists. I shall consider two papers: Burge, T, "Individualism and the Mental", in Midwest Studies in Philosophy, Vol. IV, P. A. French, T. E. Uehling, Jr., and H. K. Wettstein, eds., Unievrsity of Minnesota Press, Minneapolis, 1979, pp. 73--122, and Burge, T. "Other Bodies", in Thought and Object, A. Woodfield, ed., Clarendon Press, Oxford, 1982, pp. 97--120. 54 "Other Bodies", pp. 100-- 10 i. ss Ibid., p. 101. Burge rejects Putnam's claims about the indexicality of common nouns like "water", which I believe he is right in doing. I shall ignore this issue because, as Burge himself says, it "tends to divert attention from major implications of [Putnam's] examples". (p. 107) 56 "The Meaning of 'Meaning' ", pp. 139--142. 57 "Other Bodies", p. 109. 5~ Ibid.,pp. 109--110. 59 IBM., p. 110. 6o "Individualism and the Mental", pp. 77--79. 6I Ibid., p. 79. 6z We might say that the infant has certain de re attitudes involving the concept of a father, but, then, we can also say that the patient has certain de re attitudes involving the concept of arthritis. Also, in some exceptional cases, we might describe the first patient's attitude as involving the as yet unacquired conceptual restriction, as when he or she speculates, "Could my condition be a special rheumatoid aliment which just afflicts joints?". But, of course, there are parallel cases for the second patient. How such cases can arise is explained is the text below.

Graduate Center, C.U.N. Y., Ph.D. Program in Philosophy, 33 West 42nd Street, N e w York, N. Y. 10036, U.S.A.