sentences in which they occur, thus giving us singular propositions that contain the object

JUSTIFICATION AND RELATIVE APRIORITY Heimir Geirsson Abstract There is obviously tension between any view which claims that the object denoted is all that names and simple referring terms contribute to propositions expressed by sentences in which they appear and the apparent a posteriority of identity statements containing different but codesignative names. Frege solved the tension by adopting a description theory of names. The direct designation theorist cannot do the same, for that would amount to abandoning the theory. Instead, she has to provide one of two solutions; (a) argue that although Hesperus is Hesperus and Hesperus is Phosphorus express the same proposition their epistemic status differs such that one s justification of the proposition expressed by the former but not the latter is a priori, or (b) argue that both Hesperus is Hesperus and Hesperus is Phosphorus express a priori truths. I will argue for a version of option (a), and that while coreferential names can be freely substituted in simple belief contexts, they cannot be freely substituted in contexts involving justification. The direct designation theory maintains that proper names and other simple referring terms are nondescriptive in content. On most accounts it furthermore includes the claim that these names and simple referring terms only contribute their referents to the propositions expressed by the sentences in which they occur, thus giving us singular propositions that contain the object denoted. One of the alleged consequences of the direct designation theory is that identity sentences such as 1. Hesperus is Phosphorus and

2 2. Cicero is Tully express necessary but a posteriori truths. 1 To claim that the class of necessary truths and the class of a priori truths are not extensionally equivalent is in itself remarkable and goes against a well established philosophical tradition reaching, at least, back to Kant. To claim that identity statements provide examples of necessary a posteriori truths is even more remarkable, since it is those kinds of statements Gottlob Frege used when he established his descriptive view of proper names as being infinitely superior to a view that embeds the object denoted by a name occurring in a declarative sentence in the proposition expressed by that sentence. 2 If proper names only contribute the object denoted to the propositions expressed by the sentences in which they appear, Frege asked, how can it be that the proposition expressed by (1) differs in epistemological status from 3. Hesperus is Hesperus which is certainly a priori? There is obviously tension between any view which claims that the object denoted is all that names and simple referring terms contribute to propositions expressed by sentences in which they appear and the apparent a posteriority of identity statements containing different but codesignative names. Frege solved the tension by adopting a description theory of names. The direct designation theorist cannot do the same, for that would amount to abandoning the theory. 3 Instead, she has to provide one of two solutions; (a) argue that although (1) and (3) express the same proposition their epistemic status differs such that one s justification of the proposition expressed by (3) but not (1) is a priori, or (b) argue that (1) as well as (3) expresses an a priori truth.

3 I have argued for a version of option (a). 4 Nathan Salmon has argued for option (b) 5 saying, for example, that Recognition of the fact that Cicero is Tully is a priori simpliciter is crucial to finding a philosophically satisfactory solution to Frege s Puzzle. 6 Since Salmon s focus is on the semantic content of singular propositions he allows for substitutions of coreferential names in all belief contexts and claims that identity statements are uninformative and trivial. Against that, I will argue that since the justification of one s belief of singular propositions depends on one s epistemic access to the proposition, substitutions of coreferential names in epistemic contexts involving justification (and thus knowledge) generally fail, even though we can freely substitute coreferential names in simple belief contexts. The bulk of the paper will focus on explaining what it is about singular propositions that forces one to abandon the orthodox account that propositions either are or are not a priori and why the relevant factor only affects contexts involving justification. The last part of the paper will show how Salmon s view is similar to the orthodox account, raise a serious problem for Salmon s view, and show how the problem can be avoided given the relative account of apriority that I advocated in the earlier part of the paper. I. There are those who claim that true identity statements are not necessarily true. Let us assume that proper names are both direct and rigid designators. Consider a world in which Hesperus does not exist, the objection goes, and we have a world in which neither (1) nor (3) is true. But this can easily be fixed. We can consider a conditional, namely 4. If Hesperus exists, then Hesperus is Phosphorus,

4 which gives us a necessary truth. If Hesperus does not exist the antecedent is false, and if Hesperus does exist, then both parts of the conditional are true. But, the objection might go, if Hesperus does not exist at a world, then it is not at all clear how to treat existential statements involving Hesperus at that world. So we have not solved a problem, we have just passed it on to an even more persistent problem of existential claims involving nondenoting terms. But this objection is based on the misunderstanding that when we evaluate the truth value of a proposition at a world we have to be at that world and there use the proposition. That is not what we do. Instead, we use the English language as we always do, having already established that Hesperus is a denoting name. Given that I will assume in the remainder of the paper that true identity statements are necessarily true. II. The orthodox account of a priori justification has found a strong supporter in Frege s description theory of names. The basic idea behind the orthodox account is stated as OA below. OA: A proposition is a priori iff it can be justified without empirical evidence. It is important to note that the focus is on justification, not how one acquires the belief. One might need experience to acquire the concepts bachelor and unmarried males, but once one has acquired the concepts one needs no further empirical evidence to justify the belief that all bachelors are unmarried males. This account fits nicely with the descriptive picture Frege gave us. For Frege, propositions are composed of conceptual entities, senses, that reside in the third realm and can be apprehended or grasped by thought. Such an apprehension is direct, so speakers of different languages and different cultures can grasp the same proposition. Furthermore, if two

5 persons grasp the same proposition, then they have the same thought or the same belief. A general proposition, unlike singular propositions, cannot be grasped in more than one way. 7 If the sense of Kasparov is the highest rated chess player ever, then the sentence 5. Kasparov is a grandmaster expresses a proposition more perspicuously expressed by 6. The highest rated chess player ever is a grandmaster. And anyone who sincerely assents to (5) (or a proper translation of (5)) would believe the proposition expressed by (6). Most direct designation theorists would balk at this result. According to them the object denoted by a referring term in a simple declarative sentence becomes a part of the proposition expressed by the sentence, giving us singular propositions. The proposition expressed by (5) is properly represented as an ordered pair consisting of the property being a grandmaster and Kasparov himself. (5) expresses the singular proposition 7. <being a grandmaster, Kasparov> and the proposition expressed by (1) is 8. <identity, <Hesperus, Phosphorus>>. How we think of the objects in the propositions plays a role in our thoughts. Contemporary Russellians, such as Nathan Salmon and Scott Soames acknowledge that much. Nathan Salmon talks about one grasping a proposition under an appearance or a guise. 8 Scott Soames claims that Propositions are contents of various intermediaries with which we are intimately related--sentences, belief states, and other modes of presentation. 9 For the purpose of this paper it is not important what exact form mental representation needs to take for a contemporary Russellian. 10 What is important is that one can grasp a proposition in different

6 ways depending on how one mentally represents the object in the proposition. We need to consider whether and how the possibility of grasping singular propositions in different ways affects belief contexts. Let us assume for a moment, going along with the Fregean that names have descriptive meanings, that the names Kasparov and Garry have the same descriptive sense and that the embedded sentences in (9) - (14) express the same general proposition. As long as we are dealing with a single general proposition we can assume that if it is true that 9. Sue believes that Kasparov is a grandmaster then it is also true that 10. Sue believes that Garry is a grandmaster and if it is true that 11. Sue is justified in believing that Kasparov is a grandmaster then it is also true that 12. Sue is justified in believing that Garry is a grandmaster, and, if it is true that 13. Sue knows that Kasparov is a grandmaster it is furthermore true that 14. Sue knows that Garry is a grandmaster. The explanation of this is that as long as the proposition remains the same one is grasping it. One cannot grasp the same general proposition in two or more different ways. Consequently, if Sue is justified in believing the proposition, then she cannot grasp the proposition in a new and different way such that she is not justified in believing it, or she doesn t know it, as she now grasps it. 11 Different belief content is explained, by the Fregean, with reference to a different

7 proposition being believed, so if two belief contents differ they result from beliefs of different propositions. 12 If one accepts the direct designation theory and singular propositions and tries to accept the same kind of reasoning, then the direct designation theorist is saddled with the view that because the ancients were justified in believing that Hesperus is Hesperus they were justified in believing that Hesperus is Phosphorus, and because they knew that Hesperus is Hesperus, they also knew that Hesperus is Phosphorus; highly improbable scenarios. So, what is a direct designation theorist to do? Suppose Sue sees Kasparov being introduced as being a grandmaster when crowned as the world chess champion. She has then acquired the belief that Kasparov is a grandmaster, her belief is justified, and she knows that Kasparov is a grandmaster. Later Sue sees Kasparov playing volley ball on the beach and hears his teammates call him Garry. She fails to realize that Garry is Kasparov. She has then apprehended 7. <being a grandmaster, Kasparov> in a different way and does not recognize that (7) and 15. <being a grandmaster, Garry> are the same proposition. The fact that Sue does not recognize that (7) and (15) are the same proposition does not prevent us from saying that she nevertheless believes (15) since, after all, (15) and (7) are the same proposition, and one plausible way to account for the belief relation is simply in terms of one standing in the appropriate relation to a proposition. 13 The situation is different with regard to justification. Here we need to take into account how the proposition is apprehended. Sue is obviously justified in believing (7) when she apprehends the proposition as involving Kasparov the chess player. It is equally obvious that

8 she is not justified in believing (7) (=(15)) when she apprehends the proposition as involving Kasparov (=Garry) the volley ball player. 14 The reason has to lie in one of two things; how the proposition is presented to Sue, that is, via what sentence she is acquainted with the proposition, or how Sue mentally represents the object in the proposition. It cannot lie in how the proposition is presented to Sue for the same problem can be presented with sentences using only one name of the object in the proposition. 15 So, the reason Sue is not justified in believing that Garry the volley ball player is a grandmaster has to lie in how she mentally represents Kasparov. So, justification, and hence knowledge of singular propositions is tied to mental representation of the object in the proposition. Consider identity statements again. Suppose someone points at Kasparov and informs Sue that this is Kasparov. Sue immediately forms the belief that Kasparov is Kasparov and, given that she only has one representation of Kasparov, her belief is justified without further empirical investigation. Her belief that Kasparov is Kasparov, given how she believes it, is justified a priori. Suppose now that someone points at Kasparov playing on the beach and tells Sue that this is Garry. She already knows who Kasparov is and she knows that Kasparov is Kasparov, but she wonders whether Kasparov is Garry. Given how she now believes the proposition it will take some empirical investigation to justify her belief that Kasparov is Garry. No amount of a priori pondering will justify her belief given how she now believes the proposition. We now see why the orthodox account of the a priori does not apply to singular proposition as it did to general (or conceptualized) propositions. The orthodox account assumes there is only one epistemic access to propositions and that everyone who believes a general

9 proposition grasps it in the same way. Consequently the proposition either can or cannot be justified a priori. But this is only true of general propositions. Since there is more than one epistemic access to singular propositions they can be believed in more than one way. Consequently, depending on how the singular proposition is apprehended it can be justified either a priori or only a posteriori. Instead of (OA) we get a relative account of a priori justification of singular propositions RA: A singular proposition is a priori relative to a way of believing it iff when so believed its truth can be justified without empirical evidence. And the corresponding account for a posteriori justification is RAP: A singular proposition is a posteriori relative to a way of believing it iff when so believed its truth cannot be justified without empirical evidence. Finally, since identity statements of the type we have been working with can easily be transformed into necessary truths without its affecting their epistemic status, we have examples of relative necessary a posteriori truths. Keith Donnellan has suggested, but not developed, an account which relativizes a priority to sentences. III. Given that it is true that Cicero is Tully (and whatever we need about what the relevant sentences express) Cicero is Cicero and Cicero is Tully express the same proposition. And the proposition is necessarily true. But looking at the proposition through the lens of

10 the sentence Cicero is Cicero the proposition can be seen a priori to be true, but through Cicero is Tully one may need an a posteriori investigation. 16 In Relative and Absolute Apriority Nathan Salmon argues against Donnellan s suggestion that Cicero is Cicero and Cicero is Tully differ in epistemological status. Since Salmon is the only direct designation theorist as of late to argue against relative apriority I will discuss his arguments below. 17 As I have already noted, it is not with what sentence a proposition is presented that makes the crucial difference to whether the proposition is a priori or a posteriori. Rather, it is how the believer mentally represents the object in the proposition that makes the crucial difference. It is therefore a little surprising that both Salmon and Kai-Yee Wong take apriority to apply to sentences as well as to propositions. 18 Where S ranges over true sentences, Salmon provides the following definition of sentence S being a priori. SA: S is a priori (simpliciter) = df S is [could be] a priori with respect to some way of taking a proposition. As a result, Kasparov is Garry is a priori on Salmon s account because there is a way of taking 16. <identity, <Kasparov, Garry>> in such a way that it is a priori. Identity statements, according to Salmon, are analytic, uninformative and trivial, and a priori, for any identity statement N is M, where N and M stand for names, is nothing more than the logical truth that N is itself (e.g., Cicero is Tully is nothing more than the logical truth that Cicero is himself). 19

11 But Salmon s account runs into problems with the following example. Let S be the sentence Salmon is in Santa Barbara on March 13th, 1996, at 9:25am, then consider the (supposedly true) proposition I am entertaining expressed by 17. Salmon is in Santa Barbara on March 13th, 1996, at 9:25am. This proposition is the same as the proposition expressed when Salmon thinks to himself in Santa Barbara on March 13th, 1996, at 9:25am 18. I am here now. Anytime the subject in the proposition expressed by (18), the one denoted by I, entertains the proposition his belief cannot be wrong and he does not need any empirical investigation to justify his belief. Hence, (18), when entertained by the subject in the proposition, is a priori. 20 And from this and (SA) it follows that S is a priori; a highly implausible conclusion, for it is utterly impossible that I, I not being Salmon, can know without empirical investigation the whereabouts of Salmon. Notice the similarity between Salmon s position and the orthodox position on apriority. In the spirit of the orthodox account Salmon claims that if a sentence can be a priori, then it is a priori. This is a reasonable claim to make if one does not accept singular propositions. General propositions, remember, are made up of senses that one either grasps or doesn t grasp. There are no ways of grasping general propositions. Consequently, any two persons who grasp the same general proposition grasp it in the same way. If one of the two persons has access to an a priori justification of the proposition, so does the other. And, since one cannot grasp a general proposition in this-or-that way, one cannot believe a general a priori proposition in such a way that, given how one believes it, one can only justify it a posteriori. Because there is only one way in which the proposition can be grasped, it is either a priori or a posteriori. But given that

12 singular propositions can be believed in various ways one has to ask why Salmon supports the orthodox account of apriority. The reason Salmon sides with the orthodox view is that he focuses on the semantic content of singular propositions and the semantic contents of, e.g., the propositions expressed by Cicero is Tully or Kasparov is Garry are the same as those expressed by Cicero is Cicero and Kasparov is Kasparov, respectively. The reason Salmon opts for focusing on the semantic content is to be able to account for the analyticity of identity statements. Catsup is Ketchup, he says, is unquestionably analytic, and tomatoes are tomatoes (pronounced tomae-toes and to-mah-toes, respectively) however it is pronounced,...has the logical form of a valid sentence. 21 But this reasoning is not very convincing, for the very issue at hand is whether identity statements, which arguably have the logical form of valid sentences, are unquestionably analytic and thus a priori. Granted, in Relative and Absolute Apriority, but not in Frege s Puzzle, Salmon is a little sympathetic with the relative a priori, allowing propositions to be speaker-relative. But he argues that even the speaker-relative notions do not support the view that Cicero is Tully is a posteriori. More importantly, [the relative notions] do not support the claim that Cicero is Tully is a posteriori, any more than the proposed absolute notions of s-apriority and s- aposteriority do. As we saw above, although Cicero is Tully is a posteriori with respect to some ways of taking its content, a Millian philosopher who both understands the sentence and knows that it is true is liable to take its content in a way with respect to which the sentence is a priori rather than a posteriori. The fact that the sentence is a priori

13 with respect to at least one way of taking its content is sufficient for the sentence to be a priori (simpliciter)--otherwise even Tomatoes are tomatoes and Paderewski is Paderewski should be counted a posteriori. 22 I don t find Salmon s examples convincing. Keep in mind that when it comes to singular propositions the road to the semantic content lies through modes of presentation or mental representations. Salmon s Tomato example involves a Santa Barbaran whose limited experience of tomatoes consists of seeing them sliced in salads in the US and in the form of a sauce in England. The Catsup example (previously introduced) involves someone learning the names by ostension or by reading labels or tasting the condiment. And the Paderewski example is, of course, of the same type as my Kasparov example where I argued that Kasparov is Kasparov could, given the right epistemic access, only be justified a posteriori (and notice that my conclusion that Kasparov is Kasparov could, given the right epistemic access, only be justified a posteriori is a far cry from Salmon s attempt to show by reductio that, given his scenario, the sentence or proposition Paderewski is Paderewski should be counted a posteriori, for that claim carries with it a commitment to absolute (not relative) notions of apriority and aposteriority). All of the examples above involve a person having learned the names in the identity statements on different occasions and having different mental representations of what is named. It is precisely because we need to access singular propositions through modes of presentation that vary from person to person that no simpliciter account of apriority will do. Compare this, again, with the traditional account of apriority for general propositions. General propositions are made up of senses and the propositional content is fully disclosed in the senses. You grasp the

14 general proposition by grasping the senses. If you change the sense of a name in a sentence that expresses a general proposition, then you change the proposition as well. That is, if you change a mode of presentation, you change the proposition. Not so for singular propositions. When it comes to singular propositions you can change the appearance (the mode of presentation) of an object in a singular proposition without changing the proposition. In short, you cannot change the epistemic access to a general proposition in the way you can change the epistemic access to a singular proposition. Because we cannot change the epistemic access to general propositions, they either are or are not justifiable without empirical evidence. They are either a priori simpliciter or a posteriori simpliciter. And because the epistemic access to singular propositions, such as identity statements, can change, their justification can be either a priori or only a posteriori depending on how the proposition is believed. So, they are not a priori simpliciter and not a posteriori simpliciter. As long as we accept Salmon s assumption that we should only be concerned with the semantic content of propositions when we assess their information content and epistemic status it might seem difficult to avoid his conclusion. But, as I have argued, once we get beyond mere belief and start dealing with justification and knowledge, modes of presentation of the objects in singular propositions affect the epistemic status of the proposition. That allows us to reject Salmon s highly counterintuitive claim that sentences such as Kasparov is Garry are uninformative, trivial, and knowable without experience. Consider again the propositions expressed by (17) and (18) for they provide an excellent reason why we should prefer the relative account of apriority. According to Salmon s account (17) (the sentence) is a priori because what (18) expresses is justified a priori. But to say that (17) or what (17) expresses, when entertained by me, is an a priori truth is highly implausible.

15 There is absolutely no way that I, I not being Salmon, can have an a priori justification of what (17) expresses. My epistemic access to the proposition prevents that. Salmon s epistemic access to the same proposition, via (18), shows that there is a way of believing the proposition such that it can be justified a priori. My epistemic access to the same proposition shows that there is a way of believing the proposition such that, given how it is believed, it can only be justified a posteriori. Relativizing apriority accommodates this. Hanging on to the orthodox account, or variations thereof such as Salmon s simpliciter account, saddles us with the highly unintuitive view that my belief about Salmon s whereabouts is justified a priori. Salmon claims that finding a philosophically satisfactory solution to Frege s puzzle requires a recognition of, e.g., Cicero is Tully being a priori simpliciter, for that allows us to say that, in a relevant sense, Cicero is Cicero and Cicero is Tully do not differ in epistemic status. But, as I have argued, Salmon s account faces problems of its own, is counterintuitive, and does not mesh well with singular propositions. Once we accept singular propositions and the fact that they require an epistemic access, allowing us to believe them in more than one way, relativizing justification to how a proposition is believed certainly seems to be the alternative one should take. What stands in the way is only the philosophical tradition of embracing absolute apriority; a tradition that is out of step with recent developments in philosophy of language and metaphysics. IV. I have argued that while we can accept the view that if one believes that, e.g., Hesperus is Hesperus, then one believes that Hesperus is Phosphorus, we cannot assume that the same is true in epistemic contexts, i.e., contexts involving justification and knowledge. If one is justified in believing that Hesperus is Hesperus when thinking of Hespeus under one mode of

16 presentation, then it does not follow that one is justified in believing that Hesperus is Hesperus when thinking of Hesperus under different modes of presentation (as represented with, e.g., Hesperus is Phosphorus ). I have argued that singular propositions, unlike general propositions, require an epistemic access via modes of presentation that allows one to grasp the same proposition in more than one way, and that the epistemic access affects the justificatory status of singular propositions in such a way that they require a relative a priori/a posteriori status. The main strength of a view that requires that justification applies to singular propositions as they are believed is that such a view allows us explain why it is that sentences such as Cicero is Tully and Kasparov is Garry are informative and not trivial. It also helps us to explain why a sentence such as (17) does not express an a priori proposition. Two thorny problems for the direct designation theorist are thus taken care of. 23 Department of Philosophy and Religious Studies Iowa State University Ames, IA 50011 USA geirsson@iastate.edu

17 1 I argue for the necessary a posteriori in Necessity, Apriority, and True Identity Statements, Erkenntnis 1994: 227-242, and examples of contingent a priori truths in The Contingent A Priori: Kripke s Two Types of Examples, Australasian Journal of Philosophy 1991: 195-205. 2 On Sense and Meaning, Translations from the Philosophical Writings of Gottlob Frege, eds. P. Geach and M. Blank (Oxford: Basil Blackwell, 1952). 3 In Necessity, Apriority, and True Identity Statements I argue that the direct designation theorist cannot assign nondenoting descriptive senses to names. 4 Geirsson, Necessity, Apriority, and True Identity Statements. 5 Salmon, Frege s Puzzle (Cambridge: The MIT Press, 1986, and Relative and Absolute Apriority, Philosophical Studies 1993: 83-100. 6 Salmon, Relative and Absolute Apriority, p. 81. 7 Although one can of course express it through two or more different media, such as different languages. 8 See for example Frege s Puzzle, chapter 8. 9 Scott Soames, Substitutivity, On Being and Saying: Essays for Richard Cartwright, ed. Judith Jarvis Thomson (Cambridge: The MIT Press, 1987) pp. 99-132, p. 125. 10 I give an account of the nature of mental representations of singular propositions in Necessity, Apriority, and True Identity Statements. 11 Note that I am not claiming that one cannot have an a posteriori justification for an a priori general proposition. For one can certainly learn an a priori truth by hearing it from an authority; one might, e.g., learn from an authority of the English language that all bachelors are unmarried males, and one might learn mathematical a priori truths from a mathematician. But

18 the proposition nevertheless remains a priori, for it can be justified a priori and (unlike singular propositions) one cannot believe the proposition in such a way that the way one believes it lends itself only to a posteriori justification. 12 We might find an exception in those Fregeans who admit of non-linguistic, or de re, beliefs of the kind Gareth Evans, Graeme Forbes, and John McDowell have advocated, for they might allow for more than one epistemic access to a proposition. If in fact they do, I expect that the argument can be extended to include their position. The key to the argument s success, after all, is not the direct designation theory, but rather that it allows us to have more than one epistemic access to a proposition. 13 Vision provides a good analogy. If you see object X at two different times and you do not recognize it as being the same object, you nevertheless see object X on both occasions. Like Salmon, I think that the belief relation is a binary relation between a person and a proposition. A belief report simply relates a person to a proposition without making any claims about how she believes the proposition. I discuss belief reports in Partial Propositions and Cognitive Content, Journal of Philosophical Research, 1996: 117-128. It is important to note that even though I accept belief as being a binary relation, as simply standing in the appropriate relation to a proposition, I nevertheless recognize the importance of how a proposition is believed when it comes to its justification. 14 Consider me asking Sue Is this person a grandmaster, pointing at Kasparov (a) at the ceremonies and (b) at the beach. Sue would confidently answer with a yes at the ceremonies, but not at the beach. 15 Kripke s well-known Paderewski puzzle.

19 16 Donnellan, K. Kripke and Putnam on Natural Kind Terms, in C. Ginet and S. Shoemaker, eds., Knowledge and Mind (Oxford University Press, 1983), pp. 84-104, at p. 88n. 17 Only two philosophers have, to my knowledge, argued for relative apriority. K-Y Wong, in A Priority and Ways of Grasping a Proposition, Philosophical Studies 62, 1991: 151-164, has argued that N. Salmon should welcome a relative notion of apriority within his theoretical framework in Frege s Puzzle. For Salmon s reply to Wong see How Not to Become a Millian Heir, Philosophical Studies 62, 1991: 165-177. See also Singular Propositions and the A Priori in Journal of Philosophical Research, 21, 1996: 107-116. I have argued more broadly for relative apriority, first in Names and Beliefs (1988), a dissertation written at the University of Nebraska, Lincoln. Subsequent versions of the arguments were presented at the 1990 meeting of the Central States Philosophical Association and the 1991 APA Pacific Division meeting, and in Necessity, Apriority, and True Identity Statements, Erkenntnis 1994: 227-242. My arguments are not directed specifically against Salmon but are rather intended to apply to any view that accepts untainted singular propositions as objects of belief. 18 Wong, K-Y Sentence-Relativity and the Necessary A Posteriori, Philosophical Studies 81 1996: 53-91. 19 Salmon, Relative and Absolute Apriority p. 85. 20 Wong and Salmon discuss this example in their exchange, but there the focus is on whether the proposition expressed by I am here now is or is not a logical truth. Salmon doubts it is a logical truth, and hence that it is a priori. The example that seems to have changed Salmon s mind on the issue (for in Frege s Puzzle he allowed for the proposition to be a priori) is G. Vision s example of the standard answering-machine message I am not here now. But the

20 discussion, as well as Salmon s doubt, are misguided. The issue does not concern the supposed logical truth of I am here now, but whether it is a priori. And assertions in absentia, like in the answering-machine example, do not show that the proposition is not a priori, for the I in the proposition is presumably not entertaining the proposition at the time the machine plays its tape. 21 Salmon, p. 86. 22 Salmon, p.91. 23 I thank Rod Bertolet and Richard Mendelsohn for comments on an earlier version of this paper.