SOME REMARKS ON PERRY’S REFLEXIVE CONTENT AND COGNITIVE SIGNIFICANCE

In this paper I present and discuss the solution offered by John Perry to Frege’s Puzzle in terms of the reflexive content of utterances. I first discuss his purported solution for the indexical version of the Puzzle, and argue that reflexive content cannot explain the triviality of some utterances. Hence, it is not the sort of thing that accounts for co­­gnitive significance adequately. I then discuss Perry’s solution for the Puzzle as arising for proper names. I argue that, even if reflexive content does explain cognitive significance in this case, it does not do so in terms of the meaning of expressions, as Perry originally intended.

tial expressions can nevertheless have different cognitive significance. The Puzzle can be roughly stated as follows: how can sentences of the form a=a be trivial and sentences of the form a=b be informative if a and b refer to the same r example. It seems clear that the first is utterly uninteresting, whereas the second apparently conveys relevant historical knowledge. It may be completely silly to be told the first, while it can be very informative to be told the second. One speaker might accept the first unquestionably and vehemently deny the second. How is this possible if both names refer to the same object? How come they have different cognitive significance? This puzzle has bothered philosophers since Frege (1960aFrege ( , 1960b first discussed it. And this problem is not restricted informative despite referring to the same object twice. Frege believed that the differences in the epistemic profiles of these expressions must be explained by differences in their meaning. For this reason, he thought that any semantic theory ought to be able to account for cognitive significance. In other words, he believed that a solution to Wettstein (1986) called this idea the equacy for semantics. Frege (1960b) satisfied this criterion by distinguishing between the sense of an expression and its reference. The sense of an expression constitutes its semantic contribution to the proposition expressed 3 and it proent senses, thus explaining why their epistemic profiles are distinct. Senses, therefore, guarantee the required connection between semantics and epistemology.
Many philosophers, even endorsing the so-called direct reference theory (which is anti-Fregean by nature), share adequacy for semantics despite rejecting his theory of sense and reference. The problem is that direct reference theorists hold that the semantic content of singular terms, like indexicals and proper names, is exhausted by their reference. Therefore, any coreferential singular terms have identical semantic contents. They make the same contribution to the proposition being expressed. If this is right, how can sentences of the form a=a and a=b have different cognitive significance if a and b have the same content? In other words, why do these sentences have distinct epistemic profiles if they express the same proposition? John Perry is one of those direct reference theorists who tried to explain how this is possible through his notion of reflexive content. According to Perry, reflexive content is able to account for cognitive significance without abandoning direct reference, and it does so by appealing to certain aspects of the meaning of the relevant expressions, just as Frege recommended. In this paper, I present and discuss ent, even if it is able to solve the puzzle of cognitive significance, it cannot do so on semantic grounds, and hence does not there are reasons to believe that it does not solve the Puzzle at all.
sion of the Puzzle. But first, a word about indexicals and how direct reference theorists have traditionally dealt with Filipe Martone them. Indexicals are linguistic expressions that have different semantic contents depending on the context of their utterance. The clearest cases of indexical words are words like ing on the occasion of their use, they refer or designate different things.
For Perry (1977), indexicals have to levels of meaning. One is what he called the role of the indexical and the other is its value. The distinction between these two types of meaning is fairly intuitive. As we saw, the semantic content of an indexical shifts from context to context. However, it seems evident that despite this shift of semantic content there is something that remains stable in every use of an insomething that keeps constant despite the fact that this indexical can be used to refer to different people. This indexical is what Perry (1977) called the role of the indexical, and what Kaplan (1989) called its character (since ology in this debate, I rule of use of the relevant expression. Since linguistic rules are fixed by the linguistic conventions, it is plausible to define the character of an indexical as its linguistic meaning. Most importantly, the character is the semantic feature of indexicals that determines their value in a context and it is associated with indexicals as types Value or content, as Kaplan (1989) called it is what tent of an indexical, so I shall stick to it hereafter). The content of an indexical is the contribution it makes to the proposition being expressed. For direct reference theorists, the content of an indexical expression is what is referred to stance, the content of the indexi being demonstrated. Put another way, it is the referent of the indexical that is its semantic content, not its character. This, of course, is the fundamental difference between Fregean semantics and direct reference: it is the object itself that constitutes the content of singular terms, not some sort of qualitative or descriptive material. Indexicals, therefore, express singular propositions (propositions that have objects as their constituents) whenever they are used. Kaplan (1989), and initially Perry (1977Perry ( , 1979 as well, believed that the character is able to play the epistemic role of Fregean senses: different characters would entail different cognitive significance, and identity of character would entail identity of cognitive significance. But think of the following instance of the indexical version of the Puzspeaker is pointing to the same person twice in such a way that this utterance is informative pointing to a man depicted in a photograph and to the same man in our vicinilinguistic meaning) is the same in both of its utterances, and both of them obviously refer to the same object. Yet, this utterance is informative: I might be really surprised to learn that the man in the photograph is the same man in front of me. But why? The two levels of meaning character and content are identical in this case. This is essentially the problem Wettstein (1986) raised character. As this objection seems to show, the same character is involved twice in an utterance, so it should turn out trivial; yet, it does not 4 . Wettstein concluded that the failure of the Perry/Kaplan semantic framework to account for criterion of adequacy for semantics. In other terms, if the best direct reference semantics cannot accommodate this epistemic dimension of language, then it seems that cognitive significance is not an aspect of meaning after all. Thus, a solution to the Puzzle should not be pursued within semantics.
Perry (1988) is a response to this problem and an attempt to avoid this drastic conclusion. Perry claimed that, besides character and content, we must distinguish between the proposition expressed by an utterance and the proposition created by an utterance. He argued that the latter is what explains cognitive significance. The proposition expressed by an utterance u is its official content (or content C , in later terminology) 5 . This proposition is what we would normally regard as what is said by utterance u in the conxample, the proposition expressed is the singular proposition containing me and the property of being hungry as its ingredients. It is what we intuitively regard as what I literally express with the utter-4 ore nuanced than expounded here. Kaplan  ance.
On the other hand, the proposition created (or content M ) 6 by an utterance u is the proposition which is generated on the occasion of its production. This proposition has the utterance itself as constituent. This created proposition, which Perry calls the reflexive content of an utterance, states the conditions under which the utterance u is true. This kind of propositional content is determined solely by the linguistic meaning associated with the sentence-type of which utterance u is a token. Put another way, the reflexive content is a product of the occasion of an utterance and of the linguistic rules which are attached to the relevant linguistic expressions. These linguistic rules, as we saw, can be thought of as characters. Note also that the speaker does not need to explicitly and consciously think about the reflexive content of an utterance; since this sort of content is derived from linguistic rules, it is presumably grasped automatically and effortlessly by every competent speaker of the language.
Let us see an example. Consider a situation in which, goes on in this situation, we must distinguish between various levels of meaning. First, there is the pressions as types by the linguistic norms. Second, there is the proposition I literally express in this context, which is the official content (content C ), of my utterance: the singular proposition containing the professor and the property of being brilliant. Third, there is the reflexive content (con-tent M ), which is something along the lines of (a) There is one discriminated female x which is the referent of this utterance of x is really brilliant 7 .
The same goes for every indexical that occurs in an utterance of a complete sentence. For example, the reflexive this utterance he reflexive content of an utterthis utterance flexive content does not contain the referent as an ingredient; it contains only the utterance u itself and a general description of the referent in terms of the context and its relation to the utterance u. This is why, according to Perry, one can in some sense understand every utterance every utterance u of this sentence-type creates a proposition that states the conditions that must be met in order for utterance u to be true in the relevant context. These conditions, as said above, are derived from the linguistic meanings the characters of the relevant expressions. Thus, the hearer grasps the created proposition even if she does not grasp the official content expressed by the utterance. More importantly, since the reflexive content of an utterance has the utterance itself and its relevant parts as constituents, each utterance of an indexical will contribute a different ingredient to the created proposition. Every time is loaded into the reflexive content of the full utterance. This is why, according to Perry, utterances containing identical the man himself , the contributions they make to the reflexive content are distinct. And since the reflexive content describes the referent in terms of its relation to the context and to the utterances themselves, and since the utain two different descriptions of the referent. Put another way, the reflexive content will contain a description of the referent in terms of its relation to the first scription of the referent in terms of its relation to the second ent cognitive perspectives over the referent. In short, the different reflexive contents, and thus they differ in cognitive significance. As Corazza and Dokic (1992, p. 187) put it, although characters do not determine cognitive significance directly, they do so when applied to a particular utaspect of (applied) meaning, respects the Fregean criterion of adequacy for semantics.
However, if we individuate cognitive significance in terms of utterances and their reflexive content, numerically different utterances should always have different cognitive significance, for each new utterance will generate a new reflexive content. After all, utterances are unrepeatable and always numerically distinct (at least in non-extraordinary circumstances). However, this is clearly not the case. We have many blatantly trivial Filipe Martone static object) etc., even though their reflexive contents are distinct. In other terms, every utterance of an indexical will contribute itself to the created proposition, and so their reflexive content will always be different; hence, if cognitive significance is individuated in terms of reflexive content, different utterances should always differ in their epistemic profile. But this is obviously not the case 8 . Reflexive content is simply too fine-grained to individuate cognitive sigsolves the problem of informativeness but creates a problem for explaining triviality. And triviality, of course, is just Now, let us see how reflexive content deals with the name version of the Puzzle. The Puzzle for names, as we saw above, is the problem of explaining why sentences like (1988) and the other in Perry (1997). Let us start with the former. In that paper, Perry claims that there is a piece of they express the same proposition: the metalinguistic inthing (PERRY, 1988, p. 12). This information is expressed in a proposition such as the official or truth-conditional content, but at the level of the reflexive Put another way, a speaker who is competent with both information is expressed in the reflexive content by containing the very names is not. This is all well, except for one thing: contrary to the reflexive content of indexical sentences, this metalinguistic information is not in any sense derived from the meanings of the names in question. Perry, being a direct reference theorist, believes that all there is to the meaning of a name is its thing. Their semantic properties are completely identical. The reflexive proposition (b), therefore, is not in any sense derived from the meanings of these proper names. It expresses information about English sentences and names qua syntactic objects. This is not even a solution in terms of applied linguistic meanings, as was the solution to the indexical version of the Puzzle. In short, this sort of metalinguistic information is pre-semantic. Hence, if Perry wants to explain cognitive significance in terms of meaning, he cannot appeal to this kind of metalinguistic information to do so. This information is simply outside the realm of semantics. In fact, this metalinguistic strategy is very similar to Salmterion of adequacy for semantics.
As I said earlier, in his (1988) paper Perry was trying to Filipe Martone that paper is not entirely correct. He interpreted Wettstein the proposition that Cicero is self-identical , there is no difference whatsoever between their cognitive significance, and hence there is nothing for direct reference theorists to worry this challenge. After all, there is an important difference becould possibly be explained was claiming in that paper. He was claiming that it is not business to explain cognitive significance, not that there is no phenomena of cognitive significance at all. And this challenge is not successfully met by Perry (1988), since his solution is not based on any semantic properties of proper names. Unfortunately, the solution he offered in his (1997) is not able to meet it either. Let us see why.
In Perry (1997), his theory of reflexive content is a little bit different. He claims that, for an utterance g of the senis something like R: There is a person x and a convention C such that (i) C is exploited by g; (ii) C permits one to designate x (iii) x uses LISP 9 .

79
A convention C for Perry is the convention that is esperson or thing is assigned a name, a permissive convention is established: that name may be used to designate that perexploiting these conventions, and it is in virtue of them that we refer to the objects we do. Essentially, then, conventions C are things lik -historical chains, which determine the referent of a given token of a name, and they R. Furthermore, R is known by every competent speaker: even if still grasp R because we know how language works and how names refer.
can be informative (Perry does not explicitly talks about these cases in that paper): these names exploit different conventions, C and , and this fact is expressed by the resame convention is being exploited twice, and this information is encoded in its reflexive content. Since all competent speakers grasp reflexive contents effortlessly, this is not something too esoteric. Contents R like the above are easily ance is produced.
However, Perry believes that these conventions C are not in any sense an aspect of the meaning of the names they introduce. They are not like the characters of indexicals that, in a context, determine the referent. The determination of reference is done before any semantic evaluation. Conventions C, then, are pre-semantic. I quote: The role of context in resolving the issues of which naming conventions are being exploited is quite different from its role with indexicals. In the case of indexicals, the meaning of a given expression determines that certain specific contextual relationships to the utterance and utterer who is speaking, or to whom, or when determine designation. Different facts are relevant for different indexicals, and the meaning of the indexical determines which. Names is not that they are tied, by their meanings, to different relationships to the utterance or utterer. The role of context is simply to help us narrow down the possibilities for the permissive conventions that are being exploited (PERRY, 1997, p. 7, italics mine).
As we can see, Perry believes that conventions C are not in any sense encoded at any level of the meaning of a proper name. They are external relations that fix its reference. Figuring out which relation is being exploited and thus which thing is being referred to is a matter of contextual ingenuity for sure. But this is not something that is determined by meanings, or even by applied meanings, as happens with indexical expressions. Discovering which name and which permissive convention is at play is like figuring us which meaning is being intended in this situation. If this is correct, then remains unscathed.
In fact, I am not sure that this proposal is able to account for cognitive significance at all famous Paderewski case. Imagine that Peter is a huge fan meet him and to get his autograph. Imagine also that Peter same person. It can be informative for him (and certainly the same name, since they exploit exactly the same permissive convention: Paderewski was baptized only once twice. The convention C. If this is correct, then the utterance ought to be trivial, but it is not. content is implausible qua theory of content. I am just is considerably strengthened, for one of the most promising erence theorists is inadequate. Cognitive significance seems to depend on things external to official or reflexive contents, like contexts, background kn tions, and so on. In order to find a plausible solution, it would not hurt to look more attentively to pragmatics and not so much to semantics. tendia originalmente.