Kripke's puzzles and de dicto

Kripke’s puzzles and de dicto André Bazzoni LOGOS, University of Barcelona andrebazzoni@gmail.com Abstract This paper examines two puzzles about belief ascriptions due to Kripke, namely the London puzzle and the Paderewski puzzle. Regarding the former, I argue that the deductive step in the derivation of the puzzle that makes use of a translation principle of truth preservation is illicit. I then explain the sense in which such principle can be false as applied to the London fiction, while still being a reasonable principle in general. I proceed to argue that the Paderewski case gives rise to no puzzle in the first place, and is thus so to speak an ‘aborted’ puzzle. Both arguments rely on discussions about the de dicto/de re distinction concerning attitude reports. Keywords: Kripke’s puzzles ¨ belief ascription ¨ translation ¨ disquotation ¨ de dicto/de re 1 Introduction: attitude-ascription puzzles One of the most discussed topics in the philosophy of language regards substitutivity principles in attitude-report contexts. As Frege [?] observed, substitution of coreferential terms does not seem to preserve truth value in belief (and related) contexts, as opposed to what is verified in extensional contexts—such as 1 (??) below. Indeed, adapting one of Quine’s [?] famous examples, it intuitively seems unwarranted (at least under a de dicto interpretation of the first and third sentences below; more on this later) to infer the truth of (??) from the truths of (??) and (??): (1) Ralph believes that Bond is a spy. (2) Bond is Ortcutt. (3) Ralph believes that Ortcutt is a spy. Frege’s solution was to introduce, apart from Bedeutung—which accounts for referential, or extensional level of semantic value—an additional intentional level that he called Sinn—which on Frege’s terms includes the ‘mode of presentation’ associated with the referent of the respective expression. He then proceeded to postulate that in belief contexts what is denoted is not the customary Bedeutungen of the terms belonging to the reported belief at stake, but rather their Sinne. In this way, whereas in (??) the referents of the names ‘Bond’ and ‘Ortcutt’ are their customary Bedeutungen, in the cases of (??) and (??) their customary Sinne are instead denoted. Substitutivity then fails to apply according to Frege simply because their respective Sinne are different—even though both names have the same customary Bedeutung. On the other hand, the thesis that proper names have Sinn is controversial and seems to be supported but by a minority of philosophers since Kripke’s Naming and Necessity [?]. In that work Kripke presented serious objections to Frege’s theory and, in line (although not equivalently) with certain ideas in Mill [?], argued that proper names are rigid designators, that is, terms that have the same referent in all possible worlds (in which they exist). This together with Kripke’s claim that the rigidity of names is not accidental (which means 2 in his terms that names are rigid de jure, not merely de facto) seem to entail that coreferential names should preserve truth value under substitution in any sentence containing them, including attitude ascriptions. If Kripke is correct about the semantics of names, therefore, the Fregean puzzle1 strikes again with all its force. In his ‘A puzzle about belief’ [?], Kripke’s deep project was to argue that Frege’s theory is in reality no less immune to issues of attitude ascription than his own theory. It is not, therefore, that Kripke intended to deliver any solution to the Fregean puzzle; he rather aimed at showing that Frege’s theory is also subject to analogous puzzles. To this end, he presents a new puzzle about belief ascription that uses no substitutivity principle and affects Frege’s theory as well, with the consequence that problems related to belief contexts are in fact more complex than mere failure of substitutivity. Kripke’s puzzle is in fact two puzzles, namely the London puzzle and the Paderewski puzzle. I shall present the puzzles in section ??, and then proceed in section ?? to argue that the translation principle that Kripke uses in the derivation of the London puzzle can be in fact false while keeping its intuitive appeal. In particular, we shall see that the principle is true, but only with respect to the de re interpretation of the translated sentence. I shall also argue in section ?? that the Paderewski case is of a different nature, and in that case there is no puzzling aspect whatsoever but only an outright mistake in the assumptions of the ‘puzzle’—ones that attribute de dicto certain beliefs that could only be truthfully attributed de re. The overall conclusion will be that Kripke’s puzzles are derived on the basis of incorrect belief attributions de dicto. 1I prefer the terms ‘Fregean puzzle’ or ‘Frege-like puzzle’ rather than the more common ‘Frege’s puzzle’ because this type of substitutivity issues was never presented as a puzzle by Frege. 3 2 Getting puzzled It is commonly assumed that propositional attitude reports give rise to two different readings typically called de re (or transparent, or relational) and de dicto (or opaque, or notional). Semantically speaking, the contrast between them can be articulated in terms of substitutivity: de re allows for substitution of coreferential expressions salva veritate, whereas de dicto typically does not. For example, according to the de re reading of ‘Bond’ in (??) above, which can be made explicit by (4) About Bond, Ralph believes that he is a spy. Ralph’s belief is about a certain individual, so that the name (or description, or whatever identifying terms) by which such individual is referred to is irrelevant to the truth conditions associated with (??). As a consequence, substitution of ‘Ortcutt’ for ‘Bond’ (thus yielding the sentence “About Ortcutt, Ralph believes that he is a spy”) cannot affect the truth value of the sentence; hence substitutivity holds and no puzzle is derived from (??)–(??). It is thus a background assumption in any analysis of substitution puzzles in belief contexts that only the de dicto reading of the relevant sentences is to be considered. In this line, Kripke is careful to stress right from the outset2 [?, p. 128, his emphasis] that the de dicto “is the only reading, for belief contexts as well as modal contexts, that will concern us in this paper.” 2 As well as in several other places—thus shortly after, fn. 7, p. 129: “de re uses are simply not treated in the present paper;” and fn. 8, p. 130: “all ‘believes that’ contexts should be read de dicto unless the contrary is indicated explicitly.” 4 2.1 The London puzzle The London puzzle is based on a piece of fiction featuring Pierre, a Frenchman who lives in France and speaks only French. The original story contains some details about Pierre’s life, but the following summary will suffice for present purposes. By interacting with people who visited a foreign city called in his language ‘Londres’, Pierre becomes eventually disposed to assent to the following French sentence: (5) Londres est jolie. In order to turn Pierre’s disposition to assent to (??) into a belief report involving Pierre and (??), we make use of what Kripke calls the Disquotational Principle [?, p. 137]:3 Disquotational Principle. If a normal English speaker, on reflection, sincerely assents to ‘p’, then he believes that p where ‘p’ stands for an English sentence. The principle is supposed to be general enough to hold regardless of any particular language, thus by applying it to (??) we conclude that the following French sentence is true de dicto: (6) Pierre croit que Londres est jolie. Kripke’s next step is to apply to (??) a Translation Principle [?, p. 139] according to which: 3 The following and any other subsequent principle stated below assumes that no ambi- guity or context-sensitive items such as personal pronouns and demonstratives appear in the sentences involved in the principles—cf. Kripke’s caveats on this, e.g. [?, p. 137]. (See section ?? below for the importance of this assumption with respect to ambiguity.) 5 Translation Principle. If a sentence of one language expresses a truth in that language, then any translation of it into any other language also expresses a truth (in that other language) Since the following English sentence: (7) Pierre believes that London is pretty. is in fact (let us assume) a correct translation of (??) into English, the translation principle guarantees that (??) is true—de dicto, the only relevant reading of (??). According to Kripke’s fiction, at some point of his life Pierre moves to London, for some reason ignoring that London is in fact the city he refers to in French by ‘Londres’. In London, Pierre learns English by ‘direct method’ (i.e., with no recourse to any translation from or into his native language). Furthermore, he happens to live in a most unpleasant area of the city, which (again for some reason) he never leaves. In this way, he eventually becomes disposed to assent to: (8) London is not pretty. Now using the disquotational principle, we infer from (??) that the following English sentence is true de dicto: (9) Pierre believes that London is not pretty. This is the puzzle. From apparently self-evident principles, we derive both (??) and (??) from Pierre’s lifestory. What is more, no substitutivity principle is at work now, although the puzzle is apparently similar to the Frege-like one derived from (??)–(??) above. 6 2.2 The Paderewski puzzle Kripke offered a second puzzle about belief attribution that is supposed to yield the same sort of problem as the London case. The new construction features a new character Peter and his troubled relation with the name ‘Paderewski’, which happens to refer (uniquely) to a certain politician and musician. The issue is that Peter ignores that ‘Paderewski’ actually refers to one single man; rather, he associates with the name two different men, one being a politician and the other a musician. Peter also happens to entertain (for some reason) the general belief that no politician can possibly have any musical talent. As a consequence, Peter is disposed to assent to both, “Paderewski had musical talent” (in connection with Paderewski the musician, according to Peter’s perspective) and, “Paderewski had no musical talent” (in connection with Paderewski the politician, again following Peter’s perspective). We then infer, using the disquotational principle: (10) Peter believes that Paderewski had musical talent. and (11) Peter believes that Paderewski had no musical talent. As in the London case, we come upon two belief ascriptions reporting contradictory beliefs held by our supposedly rational belief agent. On the other hand, the purported strength of the Paderewski puzzle as compared to the London puzzle is that it uses only the disquotational (hence not the translation) principle, so that still fewer assumptions are required in the present case for the apparent conclusion that, as Kripke suggests, our customary way of attributing beliefs to rational agents has a deep-rooted puzzling nature. 7 2.3 The structures of the puzzles Taking inspiration from Sosa [?], I shall next lay down the argumentation lines involved in the two puzzles, beginning with the London puzzle: London Puzzle L1. Pierre is rational [Assumption 1 ] L2. Pierre is disposed to assent to: ‘Londres est jolie’ [Assumption 2 ] L3. Pierre is disposed to assent to: ‘London is not pretty’ [Assumption 3 ] L4. ‘Pierre believes that London is pretty’ is a correct English translation of ‘Pierre croit que Londres est jolie’ [Assumption 4 ] L5. ‘Pierre croit que Londres est jolie’ is true de dicto [L2, Disquotational Principle] L6. ‘Pierre believes that London is not pretty’ is true de dicto [L3, Disquotational Principle] L7. ‘Pierre believes that London is pretty’ is true de dicto [L4, L5, Translation Principle] L8. ‘Pierre believes that London is pretty and Pierre believes that London is not pretty’ is true de dicto [L6, L7, conjunction] L9. Pierre is not rational [L8 ] L10. Contradiction [L1, L8 ] We now turn to the Paderewski case: Paderewski Puzzle P1. Peter is rational [Assumption 1 ] 8 P2. Peter is disposed to assent to: ‘Paderewski had musical talent’ [Assumption 2] P3. Peter is disposed to assent to: ‘Paderewski had no musical talent’ [Assumption 3 ] P4. ‘Peter believes that Paderewski had musical talent’ is true de dicto [P2, Disquotational Principle] P5. ‘Peter believes that Paderewski had no musical talent’ is true de dicto [P3, Disquotational Principle] P6. ‘Peter believes that Paderewski had musical talent and Peter believes that Paderewski had no musical talent’ is true de dicto [P4, P5, conjunction] P7. Peter is not rational [P6 ] P8. Contradiction [P1, P7 ] 3 The London puzzle and translation Once a contradiction is derived, one is typically forced to revise some (at least one) of the steps that led to the undesirable conclusion. To begin with, it would be pointless to drop the rationality assumptions (L1; P1), for denying them amounts to denying virtually any philosophical puzzle altogether (after all, arguments are in general puzzling precisely when they challenge common reason). Relatedly, it would be also unavailing to reject the de dicto assumptions (L2–3; P2–3), for otherwise as Kripke stressed no puzzle about belief attribution could possibly arise. Yet it is worth noting that such rejection is unwarranted only if one accepts the plausibility of the situations described by the puzzles. Burgess [?, ?] is an example of an author who resists the plausibility of the de 9 dicto assumptions and hence of the London construct itself, on the basis that Pierre cannot count as a competent bilingual speaker of French and English— and if so, indeed, we cannot truly report de dicto the beliefs that the London construction ascribes to him. Different philosophers have blamed different argumentative moves in the structure of the London puzzle, and perhaps the most common strategy is to blame (Kripke’s version of) the disquotational principle. Thus Marcus [?,?] proposes a reformulation of the principle based on her distinction between believing and claiming to believe. According to her, the disquotational principle is false for believing, but true for claiming to believe, and it is the former that gives rise to the London puzzle. Salmon [?] defends the truth (and even analyticity) of the disquotational principle, but he includes a move in his version of the puzzle’s structure to the effect that having contradictory beliefs does not necessarily imply irrationality. His strategy can thus be seen as blocking the puzzle between L8 and L9, where another unwarranted move would be at work—the puzzle should then be understood as a reductio of the following hypothesis [?, p. 250]: “If Pierre is rational, then he does not have contradictory beliefs.” Sosa [?] also includes a (different from Salmon’s) hypothesis in the derivation of the puzzle, one that in fact implies the Millian thesis on proper names, from which he concludes that the puzzle is a reductio of that hypothesis, hence of Millianism. I shall not pursue any detailed analysis of such strategies (or of any others; cf. also [?]; [?]; [?]; [?]). My hope is that my own treatment of the matter will eventually appear to the reader as less vulnerable than the existing ones, in the sense that it presents the problem with the puzzle in a quite direct way and does not rely on any additional hypothesis, or theories of names, or distinctions (apart from the ones stated in the puzzle) that would themselves be subject to at best subtle (and at worst undecidable) debates, but rather only on the 10 already well-established de dicto/de re dichotomy, and the fact that the use of the translation principle to derive L7 in the structure of the London puzzle is illicit.4 3.1 ‘Yes’ and ‘No’: Answering Kripke One point is of immediate relevance concerning Kripke’s articulation of the translation principle. We have seen that he was careful enough to indicate— even after many reiterations to the same effect (cf. fn. ?? above)—that the disquotational principle regards only belief reports as understood de dicto. One can find no such explicit qualification, however, concerning the translation principle. This leaves us with two possibilities: either the principle involves only attributions de dicto and hence follows Kripke’s several warnings in that respect (in which case the fact that Kripke does not state the qualification explicitly might merely suggest that he considered it unnecessary to repeat the point once again); or the translation principle does not (necessarily) involve attributions de dicto. If the latter, however, the puzzle immediately vanishes, for such an option would force us to remove the de dicto specification in L7, with the consequence that the puzzle could feature a belief sentence taken de re, contrary to Kripke’s fundamental background assumption. I thus assume that L7 is a correct rendition of Kripke’s line of argumentation. Now this is my first point: even though (??) (“Pierre croit que Londres est jolie”) is true de dicto and (??) (“Pierre believes that London is pretty”) is a correct translation of it, we have no a priori grounds to conclude therefrom that (??) is also true de dicto—hence (??) is not guaranteed to be true de dicto 4 Richard [?] also identifies the problem with the puzzle as a problem with translation, although the details of his account remain largely undeveloped. At any rate, the present analysis could be taken as an elaboration on Richard’s intuitions. 11 merely in virtue of its being a correct translation of a true-de dicto report. The translation principle only states that the translated sentence is true, period; but at least in principle we could have it (as attitude ascriptions in general) true de re and false de dicto. Also and foremost, the context supplied by Kripke’s London fiction implies that (??) is in fact false de dicto. First notice that as a sentence of English, it may be true or false (de dicto or de re) in that context regardless of any consideration about its being a translation from French or any other language. Suppose that a certain Mary is a monolingual English speaker and that all we wish to know is whether it is true or false de dicto that, “Mary believes that London is pretty.” This is a perfectly legitimate question, and the same should apply to Pierre. We ask then: Is, “Pierre believes that London is pretty” true de dicto? To see that the sentence is false de dicto, we need but assumptions already contained in the structure of the puzzle, plus basic logical rules—and importantly, regardless of any application of the translation principle: §1. Pierre is rational [(L1)] §2. ‘Pierre believes that London is not pretty’ is true de dicto [(L6)] §3. ‘Pierre believes that London is pretty and Pierre believes that London is not pretty’ is false de dicto [§1 ] §4. ‘Pierre believes that London is pretty’ is false de dicto [§2, §3, conjunction] Strictly speaking, we do not even need the implicit assumption of disquotation in §2 (L6). The reason is that (??) (“Pierre believes that London is not pretty”) being a belief report that may be true or false de dicto, one can always find a situation (or possible world) in which it is true de dicto—even if in the end the London fiction does not describe one such situation. In fact, the 12 disquotational principle only plays a methodological role in the formulation of the puzzle. It is used as a tool for producing true-de dicto belief reports from a context in which the agent is disposed to assent to the complement clause of that report. We can thus simply stipulate an arbitrary situation in which (??) is true de dicto, and derive the falsity (de dicto) of (??) using the straightforward line of argumentation displayed above. The point about the London fiction is that we need a second stipulation, namely that (??) is also true de dicto, and such stipulation might conflict with the first one, in which case they would jointly give rise to an impossible situation. The London fiction is deliberately conceived as an example of a possible situation involving both stipulations, and the way in which the plausibility of such a construction is achieved is by the introduction of a bilingual character who happens to ignore that ‘Londres’ and ‘London’ are names for the same city in the two languages (French and English) at issue. This indeed makes it plausible that he may be willing to assent to the complement clauses of both (??) and (??). The disquotational principle is then the final move in this preliminary step of the construction yielding the truth de dicto of (??) and (??). So far, everything is in order.5 5 Even if the disquotational principle were not valid in the end, one might still pursue a similar argumentation line leading to the puzzling conclusion, provided that an alternative case is made for the plausibility of those two stipulations. This is to my view a significant weakness regarding strategies for dissolving the London puzzle by blaming the disquotational principle, for what one must show for disarming the puzzle is more general than the supposedly illegitimacy of disquotation: one must show that (??) and (??) cannot be both true de dicto (assuming the rationality of the related agent), and in particular that Pierre’s situation does not give rise to a possible situation. The difficulty in showing this becomes critical as one recognizes the plausibility of Pierre’s ignoring the fact that ‘Londres’ and ‘London’ refer to the same city. Moreover, this is precisely the paradigmatic type of situation that motivated 13 The problem is that the translation principle then intervenes (in L7) to generate (together with the rationality assumption and standard logic) the puzzling conclusion. At that point, however, as we have just seen from the argument above, we are already in a position to conclude that the product of the application of the translation principle is false de dicto. This provides us with a reductio of (Kripke’s version of) the translation principle. Let me first lay down the argumentative steps of the reductio, and then I will propose a more general version of the principle that retains truthpreservation intuitions and at the same time does not yield any puzzling result with respect to Pierre’s situation. Finally, I shall elaborate on the intuitive basis of the new proposed version of the principle. Let us thus spell out a reductio argument for L7: London Unpuzzled L*1. Pierre is rational [Assumption 1 (L1)] L*2. ‘Pierre croit que Londres est jolie’ is true de dicto [Assumption 2 (L2)] L*3. ‘Pierre believes that London is not pretty’ is true de dicto [Assumption 3 (L3)] L*4. ‘Pierre believes that London is pretty’ is a correct English translation of ‘Pierre croit que Londres est jolie’ [Assumption 4 (L4)] L*5. ‘Pierre believes that London is pretty’ is true de dicto [L*2, L*4, Translation Principle] L*6. ‘Pierre believes that London is pretty’ is false de dicto [§4 ] the de dicto/de re distinction in the first place—recall Frege’s examples, and Quine’s Ortcutt case in which it is precisely the fact that Ralph ignores that two names are coreferential that motivated his notional/relational interpretations of belief reports. 14 L*7. Contradiction [L*5, L*6 ] This argument involves no conceptual material that is not already contained in the original structure of the puzzle. The contradictory conclusion is derived merely from our previous conclusion that (??) is false de dicto (i.e., L*6 aka §4) and the application of Kripke’s translation principle to (??). Since the only step that uses material other than structural assumptions is L*5 (i.e., our old L7), I take London Unpuzzled to be a reductio of that step. Moreover, since such step involves two structural assumptions and one principle, I finally take the argument to be a reductio of that principle, namely the translation principle. The point, to repeat, is that in any situation (involving a rational agent) in which (??) is true de dicto, (??) must be false de dicto, hence if we use some principle(s) to derive the truth (de dicto) of the latter, then there must be something wrong with such principle(s). Now Kripke’s translation principle guarantees truth preservation, and we have indeed strong intuitions supporting that. That, however, is not what the reductio above is targeting. Rather, what is thereby rejected is the fact that translation preserves truth de dicto. This opens the way for accommodating both the reductio argument and truth-preservation via translation: what we need is simply a version of the translation principle such that, as applied to belief (and in general, attitude) reports, it becomes a principle regarding preservation of truth de re. Hence the following: Specified Translation Principle. If a sentence in one language expresses a truth de re in that language, then any translation of it into any other language expresses a truth de re (in that other language) In the particular case in which the source sentence is a simple (i.e., nonattitudinal) statement, then the ‘de re’ qualification of the specified translation principle becomes void, and the formulation of that principle nails down to 15 the unspecified one advanced by Kripke. The specified version accounts for the specificities of sentences such as attitude reports, which feature a type of ambiguity (namely between de dicto and de re) that is absent from simple statements and that in fact happens, as shown by the argument spelled out above, to introduce an important semantic subtlety into the general mechanism of translation. To be sure, the specified translation principle yields that both, “Pierre believes that London is pretty” and, “Pierre believes that London is not pretty” are true de re, since both the latter and, “Pierre croit que Londres est jolie” are clearly true (not only de dicto but also) de re according to the London construct. This is as expected, however (and this is the deep reason why de re readings do not generate puzzles of the type discussed here), for Pierre is mistaken about the identity of the city that he saw on photos back in France, and the one where he is currently settled. One could indeed report his beliefs by saying with amusement, “You see, Pierre is totally confused about the identity of London, he thinks it’s pretty and not pretty at the same time!” As far as Pierre’s beliefs are concerned, there is no puzzle involved in such conclusion. Furthermore, in the situation described in the London construct, (??) is indeed true de re, as predicted by the specified translation principle. To see why, first recall that we may isolate the de re interpretation of an attitudeascription sentence containing a name (or any other referring term) by freeing that name from the scope of the attitude verb. Then consider the following: (12) About the city that Pierre once saw in pictures back in France, he believes that it is pretty. This sentence is uncontroversially true, for it is just a reformulation of the very narrative that constitutes the London fiction. Now since (??) is the equivalent of the de re reading of, “Pierre believes that the city that he once saw in 16 pictures back in France is pretty,” then substitution of a coreferential term for ‘the city that Pierre once saw in pictures back in France’ preserves truth value. ‘London’ provides just such a term to yield (??), which is thus true: (13) About London, Pierre believes that it is pretty. Now the latter is the equivalent of the de re reading of (??), which entitles us to conclude that (??) is true de re.6 This shows that the London construct does not threat our intuitions regarding translation and truth preservation, provided that we adopt the specified translation principle and recognize that what is relevant to the derivation of the puzzle (and of any belief puzzle as repeatedly emphasized by Kripke) is the truth de dicto of the translated sentence. To sum up, what the observations above suggest is that the move to L7 in the structure of the puzzle constitutes quite an illicit argumentative step: we are not allowed to automatically derive a true-de dicto report using the translation principle. Moreover, the context supplied by the London fiction implies that (??) is actually false as interpreted de dicto, although it is taken (on the basis of an illicit move) as true de dicto in the structure of the puzzle. When considering possible reactions to the effect that his puzzle is at bottom no genuine puzzle, Kripke insistently challenges his imaginary detractor with a question [?, p. 147; p. 148]: “Does Pierre, or does he not, believe that London is pretty?” The twofold answer that is on offer here is, “Yes, Pierre believes that London is pretty—de re;” and “No, Pierre does not believe that London is pretty—de dicto.” 6 It is worth noticing that accounting for the plausibility of the translation principle in these terms does not require a very elaborated articulation of the de dicto/de re distinction (cf. also section ??), but only a rather general and purely semantic one in terms of substitution of coreferential terms salva veritate. 17 4 The Paderewski case is no Paderewski puzzle In this new case, we again come upon two belief ascriptions reporting contradictory beliefs held by a supposedly rational agent, namely Peter. The purported strength of the Paderewski puzzle as compared to the London puzzle is that it uses only the disquotational (hence in particular, not the translation) principle. This seems to be crucial to our discussion, because we have been treating the London puzzle precisely as an illegitimate product of (Kripke’s unspecified version of) the translation principle. On the other hand, as Kripke himself pointed out [?, p. 154], “the original ‘two languages’ case had the advantage that it would apply even if we spoke languages in which all names must denote uniquely and unambiguously.” This observation is important for at least two reasons. First, it shows that, although Kripke identifies [?, p. 154] that “[t]he situation is parallel to the problem with Pierre and London,” he also recognizes a weakness of the Paderewski construct in comparison with the London case—namely: that the former, contrary to the latter, would not arise in a context in which the referential relation would be one-one. Second, Kripke explicitly mentions the issue of ambiguity in connection with the Paderewski case. This is the starting point of Sosa’s [?] analysis of Kripke’s construction. As he observes, one of Kripke’s assumptions (as we have seen above) regarding the disquotational principle is that the complement clause of the respective belief report is to be free of any ambiguity. It is ambiguity, however, that seems to be at work in the Paderewski case, as Kripke himself acknowledges in the quotation above.7 7 Sosa [?, p. 395] is careful, though, not to conclude that this observation is enough to dismiss the applicability of the disquotational principle to the Paderewski case, but only to make it suspect. (Sosa also stresses that his argument is independent of whether the principle 18 Nevertheless, it seems to me that the issue is less transparent and more nuanced that such an analysis would in fact suggest. As it happens, the name ‘Paderewski’ in Kripke’s disquotational principle is not ambiguous; it denotes the unique (at least as long as Kripke’s fiction is concerned) individual who bears the name ‘Paderewski’, namely the politician and musician Paderewski. What Kripke says in the quotation above is only that the Paderewski fiction is set against a background of naming practices according to which more than one individual may bear the same (generic or particular, depending on one’s views) name—and such background assumption is what makes plausible a situation in which Peter believes that the name is ambiguous. In any case, Kripke does not say that ‘Paderewski’ is in fact ambiguous. Importantly, the mere fact that Peter believes that ‘Paderewski’ is ambiguous does not count as ambiguity in a sense that could invalidate the application of the disquotational principle. In particular, we could think of Pierre as believing that ‘Londres’ is ambiguous, in which case we could not any longer use disquotation to derive the London puzzle. But clearly, even though Pierre believes that ‘Londres’ is ambiguous—say, between London, UK and London, ON8 —it is still reasonable to conclude that if Pierre assents to, “Londres est jolie,” then it is true de dicto that, “Pierre croit que Londres est jolie”—the significant point being that ‘Londres’ is not ambiguous in the complement clause, “Londres est jolie.” Similarly, there does not seem to be any real issue about ambiguity in relation to disquotation in the Paderewski case. On the other hand, as I shall attempt to show next, the fact that Peter believes that ‘Paderewski’ is ambiguous is not irrelevant to the real issue involved in the Paderewski construction; and the real issue is again about de dicto. is true or false; it is really a matter of applicability of disquotation to the case in point.) 8 London, ON is not translated in French, thus ‘Londres’ refers in that language unam- biguously to London, UK. 19 Now how could one deny that the belief attributions in the Paderewski fiction are taken de dicto, if they are simply stipulated by Kripke to be so? Notice first a crucial distinction in this regard between the London and the Paderewski constructs. In the former, Pierre is disposed to assent to both, “Londres est jolie” and, “London is not pretty.” If he were urged to react to the following sentence: (14) “Londres est jolie” is true and “London is not pretty” is true. he would certainly assent to it without hesitation or any other special reaction. On the other hand, suppose that Peter is confronted to: (15) “Paderewski had musical talent” is true, and “Paderewski had no musical talent” is true. It seems that now Peter will at least take a moment to react. Perhaps he will laugh and say, “You’re trying to put contradictory words into my mouth!” Perhaps he will hesitate, “Well, not exacly;” or still make it more precise (i.e., ‘disambiguate’ it), “Yes, if ‘Paderewski’ names different guys in each sentence.” But what is most certain is that he will not crudely reply, “Yes” without any hesitation and further qualification to his answer. This is a relevant difference between the London and the Paderewski cases, and the culprit of such distinction is that Peter is not disposed to assent to (??), hence he is not disposed to assent to both conjuncts of (??). He would only on the condition that the name ‘Paderewski’ do not refer uniquely. Thus suppose that he reacts by saying, “Yes, but only if ‘Paderewski’ names different guys in each sentence.” Would such but-only-if proviso count him as assenting to (16) Paderewski had musical talent. and to 20 (17) Paderewski had no musical talent. taken separately? The answer can only be negative, for in (??) and (??) ‘Paderewski’ is not ambiguous: this name is a name for Paderewski the musician/politician, and as such ‘Paderewski’ does not name different guys in those sentences. If we look at logical forms, we can formulate the point by observing that when assenting to two sentences, Peter is using two different names,9 say ‘Paderewski1 ’ and ‘Paderewski2 ’. Therefore, if one would be inclined to affirm that Peter is disposed to assent to each conjunct of (??) while still rejecting (??), it would be only because one is mistaken about which statements Peter is in fact willing to endorse; and these are different sentences from (??) and (??), hence ultimately not the conjuncts of (??). More specifically, what Peter is disposed to assent to are the following: (18) Paderewski1 had musical talent. and (19) Paderewski2 had no musical talent. while rejecting (??) (in which ‘Paderewski’ is uniquely indexed), which is thus perfectly consistent. But this does not count as being disposed to assent to both (??) and (??) while rejecting (??), since the latter is not the conjunction of the former two statements. If Peter is not disposed to assent to both (??) and (??), however, the disquotational principle fails to apply, and no puzzle is derived.10 9 This is in fact Kripke’s strategy for dealing with homonymy, as he elsewhere tacitly adopts “the practice of calling homonyms distinct ‘words’, according to which uses of phonetically the same sounds to name distinct objects count as distinct names” [?, p. 8]. 10 In connection with my comment on Sosa’s article above, it is not ambiguity that is block- ing the applicability of disquotation in the Paderewski case, but rather Peter’s indisposition 21 As in the London case, moreover, disquotation is not needed in the argument. For the Paderewski puzzle to get off the ground what is required is the more general plausibility of the situation in which both (??) (“Peter believes that Paderewski had musical talent;” cf. P4 in the structure of the Paderewski puzzle) and (??) (“Peter believes that Paderewski had no musical talent;” P5 in the structure) are true de dicto, be this via disquotation or any other adequate means. But for the reasons just exposed, this gives rise to an impossible situation (with respect to Kripke’s fiction), one according to which the following are both true de dicto: (20) Peter believes that Paderewski1 had musical talent. (21) Peter believes that Paderewski1 had no musical talent. It is important in the above sentences to index the name ‘Paderewski’ uniquely, for the de dicto/de re distinction is based precisely on the fact that the former is related (in some way, which need not concern us here) to the complement clause (the dictum), and in the respective complement clauses that are relevant to the puzzle—as pointed out above in relation to (??) and (??)—the name ‘Paderewski’ is not ambiguous. The fundamental difference in this respect between the London and the Paderewski construct is that the former supplied a further contextual element that accounted for the plausibility of Pierre’s situation, namely the fact that Pierre ignores that ‘Londres’ and ‘London’ refer to the same city; whereas in the present case, such element is clearly unavailable. The reason is that stipulating that Peter ignores that ‘Paderewski1 ’ and ‘Paderewski1 ’ refer to different individuals would directly conflict with Peter’s rationality—it would indeed amount to believing either that a uniquely referring term is not uniquely referring, or in to assent to the two sentences on which the principle is supposed to operate according to Kripke’s fiction. 22 a contradictory statement of the general form (for P a predicate constant and a an individual constant) P paq ^ P paq. What does give rise to a possible situation is the postulation that the following are true de dicto: (22) Peter believes that Paderewski1 had musical talent. (23) Peter believes that Paderewski2 had no musical talent. As we have seen, however, there is nothing puzzling about such situation except for the fact that at least one of the names ‘Paderewski1 ’ and ‘Paderewski2 ’ would (according to the associated context) fail of reference. For instance, one might then say that Peter ignores that ‘Paderewski2 ’ refers to no individual.11 The possibility of ambiguity is crucial to the understanding of why the Paderewski construct gives rise to no puzzle, and besides Sosa Kaplan is another author who is sensitive to this aspect, as he remarks that “Peter is making this kind of error [i.e., thinking that there were two different common currency names] in Saul Kripke’s ‘Paderewski’ case [. . . ]. It is my belief that the analysis in terms of word individuation is valuable, and perhaps critical, in understanding that fascinating case” [?, p. 108]. Kaplan [?] does not say any more on the reasons why he believes that Peter’s mistake is relevant to the understanding of the case, but in his [?] he mentions (again only in passing, his primary goals being related to quite other topics) that [Peter’s] misunderstanding was due to a failure to disambiguate all occurrences of the generic name ‘Paderewski’ correctly. Since the 11 Notice also that the de re interpretation of (??) cannot possibly be true with respect to the Paderewski construct. Indeed, since ‘Paderewski2 ’ fails of reference in such context, the sentence, “About Paderewski2 , Peter believes that he had no musical talent” cannot be true—it is either false or neither-true-nor-false. 23 stated (and natural) preconditions of disquotation require that no linguistic mistake is made, we were not justified in making de dicto reports of his beliefs. [?, p. 157] This is not what I am suggesting is the source of the Paderewski illusion, but Kaplan’s passage states explicitly the critical point, namely the illegitimate attribution-de dicto to Peter of the belief that Paderewski had musical talent and that Paderewski had musical talent. What opposes Kaplan’s diagnosis and my own is that according to Kaplan it is the fact that Pierre committed a linguistic mistake (of taking one name for two) that blocked the application of the disquotational principle. But that mere mistake is insufficient to halt the application of the principle, because strictly speaking there are in fact two sentences that Peter is disposed to assent to, namely (??) and (??). The problem is rather that those sentences are not the ones that give rise to the puzzle. The reason why disquotation cannot be applied is that Peter is not disposed to assent to the ones that would give rise to the puzzle, namely (??) and (??). Such proviso notwithstanding, the present view can be understood, it seems to me, as a development of Kaplan’s (and perhaps Sosa’s, to a lesser degree since he does not speak in terms of de dicto) intuitions on the case. In relation to our discussion on the central place of the de dicto/de re distinction in the derivation of the London puzzle, recall that the disquotational principle is explicitly used by Kripke to derive true de dicto reports. As a matter of fact the belief reports concerning Peter that give rise to the Paderewski puzzle are indeed true, but only if taken de re, for under this interpretation it is certainly true that Peter believes that Paderewski had musical talent, and that Peter believes that Paderewski had no musical talent (it suffices to operate substitutions of coreferential terms as we did in connection with the London puzzle; I leave the exercise to the reader). 24 One final note about the use of the disquotational principle in the Paderewski case. It seems that one aspect of our analysis of this case is weaker than the one carried out in connection with the London case, in the following sense. In the latter case, in order to disarm the puzzle we only needed an argument using assumptions already contained in the structure of the puzzle. The dialectic can be thus outlined: take any way of making sense of Pierre’s believing de dicto that London is not pretty; it then follows from Pierre’s rationality that it is false that he believes de dicto that London is pretty; from which we conclude that L7 is an illicit argumentative step, hence the unspecified translation principle fails to hold in general. The crucial point here is that even if disquotation were shown to be misapplied in the London case, the argument would remain unaffected, for all that is needed is some way of making sense of Pierre’s believing de dicto that London is not pretty; and of course, one can always find such a situation—that is, we can just stipulate that Pierre believes de dicto that London is not pretty, no matter how the situation in which he does comes to be specified. Once the de dicto/de re distinction is acknowledged (and in a quite general way; cf. section ?? below), the conclusion that it is false that Pierre believes de dicto that London is pretty seems to be inescapable. This is not, however, how the rationale works in the case of the Paderewski puzzle, in which disquotation plays a more central role. Indeed, the argument in that case was that disquotation is not applied to the sentences that would generate the puzzle, but rather to other sentences that would not. We crucially used in the argument the fact that Peter would not be disposed to assent to the relevant sentences. Now the weakness of this treatment is that it only applies in principle to the particular situation described by Kripke. Specifically, it does not prima facie account for other possible ways of making sense of the relevant puzzling beliefs without using disposition to assent and disquotation. I grant 25 the point, but suffice it to say for present purposes that the burden is now on those who do believe such line to be worth exploring. 5 Summary and concluding remarks Kripke presented two puzzles about belief, the London and the Paderewski puzzles. I argued that none of them are genuine puzzles. The reasons in the two cases are different, but they share a common aspect: both puzzles crucially involve true-de dicto belief attributions that could only be true de re according to the respective contexts of the puzzles. In the London case, the structure of the puzzle makes critical use of a translation principle that is in fact, as we have seen, only legitimate (in general) for deriving truths de re. In the Paderewski case, Kripke’s applications of the disquotational principle to the specific sentences relevant to the derivation of the puzzle only hold, as we have also seen, if the associated belief reports are taken de re. In both cases, therefore (since de re is immune to attitude-ascription puzzles), no genuine puzzle arises. I first suggested that the structure of the London puzzle points to a reductio of step L7 (i.e., the statement that, “Pierre believes that London is pretty” is true de dicto). I further pursued the analysis of the relationship between the translation principle and the de dicto/de re dichotomy to argue that it is illicit to use the translation principle to derive L7. Finally, I suggested how then the translation principle can be at the same time false and appealing: it is actually false for de dicto, but true for de re. We then saw that the reason why the Paderewski puzzle is illegitimate is that it relies on the mistaken assumption that the sentences, “Peter believes that Paderewski had musical talent” and, “Peter believes that Paderewski had no musical talent” are both true de dicto. We saw that such conclusion is unwarranted if one assumes that Peter is not disposed to assent to (??) (“Paderewski 26 had musical talent and Paderewski had no musical talent”), which is precisely the case according to the Paderewski fiction—the fact that Peter assents to both conjuncts in isolated contexts being due to Peter’s understanding that the name ‘Paderewski’ was ambiguous—although, as we have also seen, it is not ambiguity per se that blocked the application of disquotation. Kripke’s primary motivation in presenting his puzzles was to show that the problems raised for his theory of names in connection with the interchangeability of coreferential names in attitude contexts salva veritate stems from “our normal practices of translation and disquotation” [?, p. 143]. As we have seen, however, our normal practices of translation and disquotation are perfectly in order, problems arising only when those principles are misapplied, as in the two cases presented by Kripke. Now one interesting feature of our analysis is that its central element involves the de dicto/de re distinction, yet we have not treated that distinction with any level of detail. In particular, we have not said much about what characterizes each type of reading except for the fact that the latter, but not the former, always allows for interchangeability salva veritate of coreferential terms. This, however, should not strike the reader as a simplistic treatment of the London and Paderewski cases, for in fact the reasons why the puzzles are not genuinely puzzling are so fundamental, that we do not really need a deeper examination of the de dicto/de re distinction in order to appreciate the arguments presented here. In a word, this is the fundamentality aspect of our analysis: it is based on the very methods that Kripke uses for attributing true-de dicto reports, and only on them. This said, one might feel puzzled after all. For suppose that we adopt a specific account of the de dicto interpretation of belief (and in general attitude) reports, according to which a de dicto belief report involves the proposition 27 expressed by the complement clause of the report. Thus to say that Pierre believes-de dicto that London is pretty amounts to saying that the proposition expressed by, “London is pretty” is part of Pierre’s belief state. Now the name ‘London’, so goes the reasoning, plausibly contributes the city of London, and only the city of London, to the proposition expressed by, “London is pretty.” But the latter proposition is arguably the same proposition as the one expressed by, “Londres est jolie.” On the other hand, we have acknowledged throughout our discussion that, “Pierre croit que Londres est jolie” is true de dicto. This is then the puzzling consequence: if we uphold the plausible theses that (i ) names contribute only their referents to the propositions expressed by sentences containing them; and (ii ) that de dicto reports are about propositions, then we are forced to conclude that Pierre must believe-de dicto that London is pretty, for the proposition expressed by “London is pretty” and the one expressed by, “Londres est jolie” are the same proposition, and the belief report that we acknowledged as true de dicto is about the latter. In the face of this, my first reaction would be quite direct: if so, the burden is on theses (i ) and (ii ). Both are controversial, and in fact (i ) is the very thesis that Kripke wishes to defend12 in the first place, and which he wishes to shield from difficulties raised by Frege-like puzzles. This joins Sosa’s argument concerning tacit assumptions of Millianism in the derivation of the puzzle; if dropping Millianism is required for dissolving a puzzle of rationality, then it seems that Millianism has to go away rather than the assumption that belief agents are rational. The arguments provided here are so direct, that if some controversial thesis has to go away in the light of them, so they should. However, I do not think in this particular case that the rejection of Millianism is a necessary consequence of our analysis, and this brings us to thesis (ii ). It 12 There are significant differences between (i), Kripke’s views and Millianism, but for present purposes these theses can be assimilated without loss. 28 seems that the general failure of Kripke’s unspecified translation principle is directly connected to the presence of proper names in the relevant sentences. If we disregard such sentences, it appears indeed (as the reader might want to check) that the old unspecified translation principle holds without further qualifications. As a consequence, it would seem that the conception of de dicto sketched in (ii ) must be refined (while keeping Millianism) accordingly in order to accommodate such peculiarity—perhaps (and this is only a rough indication) de dicto attributions must somehow take into account the linguistic form of proper names, not only their referents; this would be in any case (if successful at all) a good prospect for accommodating the more basic facts revealed by the arguments presented in this paper, while retaining a Mill-like semantics for names. On the other hand, the peculiarity of the role of proper names as to the translation principle does not per se have any problematic consequences to our general discussion, for the same can be said of Kripke’s puzzles as well: the latter only (illusory, as we might presently put it) arise in connection with sentences containing proper names in a crucial way (in our cases, as the primary sources of confusion in relation to the respective belief agents). In particular, no other kind of referring term can be the primary source of a Kripkean puzzle. These are certainly quite exciting topics for discussion, and ones that I have not addressed in this paper. But the reason why I have not is simply that any treatment of such and related issues must accommodate the more basic facts derived from the arguments put forward above, namely that according to the London construct it is false de dicto that Pierre believes that London is pretty; or that Paderewski1 had and did not have musical talent. That is, before putting forward controversial theses that jointly conflict with those facts, one should explain why the latter are not to be derived. 29 References [1] Burgess, J.P. (2005). Translating names. Analysis 65(3): 196–205. [2] Burgess, J.P. (2013). Saul Kripke. Cambridge: Polity Press. [3] Donnellan, K. (1989). Belief and the identity of reference. Midwest Studies in Philosophy 14(1) : 275–288. [4] Frances, B. (2000). Disquotation and substitutivity. Mind 109: 519–25. [5] Frege, G. (1892). On sense and reference. Translated by Max Black. In P. Geach & M. Black (eds.), Translations from the Philosophical Writings of Gottlob Frege. 1952. 56–79. Oxford: Blackwell. [6] Kaplan, D. (1990). Words. Proceedings of the Aristotelian Society 64: 93– 119. [7] Kaplan, D. (2012). An idea of Donnellan. In J. Almog & P. Leonardi (eds.), Having In Mind: The Philosophy of Keith Donnellan. Oxford: Oxford University Press. 122–175. [8] Kripke, S. (1979). A puzzle about belief. In S. Kripke, Philosophical Troubles – Collected Papers. Vol. I. 125–161. Oxford: Oxford University Press. [9] Kripke, S. (1980). Naming and Necessity. Cambridge, MA: Harvard University Press. [10] Marcus, R.B. (1981). A proposed solution to the puzzle about belief. Midwest Studies in Philosophy 6(1) : 501–510. [11] Marcus, R.B. (1983). Rationality and believing the impossible. The Journal of Philosophy 80(6) : 321–338. 30 [12] Mill, J.S. (1843). A System of Logic. Honolulu: University Press of the Pacific, 2002. [13] Moore, J. (1999). Misdisquotation and substitutivity: When not to infer belief from assent. Mind 108: 335–365. [14] Over, D.E. (1983). On Kripke’s puzzle. Mind 92(366): 253–256. [15] Quine, W.V. (1956). Quantifiers and propositional attitudes. The Journal of Philosophy 53(5): 177–187. [16] Richard, M. (2011). Kripke’s puzzle about belief. In A. Berger (ed.), Saul Kripke. 211–234. Cambridge: Cambridge University Press. [17] Salmon, N. (1983). Frege’s Puzzle. Cambridge, Mass.: The MIT Press. [18] Salmon, N. (2011). A note on Kripke’s puzzle about belief. In A. Berger (ed.), Saul Kripke. 235–252. Cambridge: Cambridge University Press. [19] Sosa, D. (1996). The import of the puzzle about belief. The Philosophical Review 105(3) : 373–402. 31