April 27 1998
last rev. June 20 1998

Carlo Penco

Brandom's solution of Kripke's puzzle

talk given at CMU (Pittsburgh) in
seminar on dynamic semantics held by Horacio Arlo-Costa and Tim Fernando
8 and 15 april 1995


Brandom's "solution" of Kripke's puzzle in Making it Explicit [573-583] is to be read on the background of four main ideas, plus his general concern on inferential role semantics. I will give some hints about these basic presuppositions, because, once they have been accepted, Kripke's puzzle seems to have no more appeal (at least from Brandom's point of view). If already acquainted with Brandom's general ideas, you may skip part I and go directly to part II.

0. Inferential role (normative) semantics
1. deixis presupposes anaphora
2. indirect speech as relation between tokens
3. rigidity is an anaphoric phenomenon (generalizing anaphora)
4. proper names' senses are anaphoric chains

5. a reconstruction of Brandom's analysis of Kripke's puzzle
6. generalization of the analysis
7. some critical comments

0. Inferential role semantic. One of the main tenets of Making it Explicit is an inferential approach to meaning: the meaning or the "conceptual content" of an expression is its inferential role. The original idea is traced back to Frege's Begriffsshrift: here Frege suggests that two sentences have the same conceptual content if they may be substituted one another preserving the goodness of the inferences in which they appear, or preserving the consequences we may trace from them (the original example of Frege is a pair of sentences in active and passive form).

  • Commitments and entitlements The original contribution of Brandom to the inferential approach to meaning is an highly normative definition of inferential role in term of entitlements and commitments. The definition of "commitment" is much stronger that the definition given, for instance, in Levi; for Brandom a commitment is not just an expression of what you are committed when you recognize a set of beliefs and update it, but it represents the claims you have to avow because they follow from what you claim, even if you are not aware of them: in making a claim you are committed to the consequences recognized by the society as valid consequences of that claim.
    Given these assumptions a general definition of inferential rolein terms of commitments and entitlements has an intuitive appeal. Brandom makes some reference to Dummett's view of meaning as given by circumstances and consequences of application (or as introduction and elimination rules). Entitlements are to be interpreted as the circumstances of applications or premises which entitle you to make a claim; commitments are the consequences you are bound to accept, given the claim you have made.
  • Conceptual content of noninferential perceptual reports Brandom's inferentialism is intended in a broad way, not strictly relying on logical premisses and consequences, allowing also nonlinguistic circumstances of application to figure as entitlements. The coherence of this step with an inferentialist view of meaning is given via a connection with an idea of Sellars' Empiricism and Philosophy of Mind : any utterance can be understood properly only as a move in the game of giving and asking for reasons; in order to utter a "meaningful" perceptual report you need not only to have a reliable responsive disposition, but you need to understand the commitments held in uttering the report. Therefore a perceptual report, like "it is red", is a proper meaningful utterance only in a game where the speaker is committed to the consequences of the utterance, for instance: "it is not blue" or "it has a color". A parrot who has been trained to utter the same sounds when facing a red thing does not give to the sounds any conceptual content.
    A problem remains: which kind of inferential role may be given to subsentential parts? It is easy to define an inferential role for predicates (let us think, for instance, to a semantic network given via meaning postualtes); but it is really hard to define an infernetial role for indexicals and proper names. One of the most original aspect of Brandom's analysis is to enclose also these kinds of subsemntential expressions in an inferentialist framework, via the concept of anaphora.
    [back to top]

  • 1. deixis presupposes anaphora: the role of substitution [459-464] In recent times a great stress has been posed on indexicals (Kaplan, Perry, Lewis), as establishing the basic nature of word-world relation. Then it has been considered that everything can be used indexically (to fix the reference). Therefore some special place has been given to tokenings instead of types. Eventually it has been insisted on the priority of unrepeatable tokenings (indexicals & demonstratives) over repeatable ones (common nouns, def./indef.descriptions). No attention, in this concern, has been given to anaphora for it appears to deal with intralinguistic matters. The aim of Brandom is to show that anaphora, far from being just an intralinguistic device, is the expression of the commitments of speaker which permit deixis to work. This general attitude is embedded is the strongly normative inferential role semantics (see before), where a great importance is given to the substitutional role of linguistic items.

  • Inferential content of terms is given via substitution. To know the inferential content of a term is to have the right substitution-inferential commitment, that is to be committed to substitute the term to preserve inferences. The basic idea comes from Frege: in order to speak of an object you have to be able to recognize it again as the same; you need to have some criteria for identity of objects. To be able to manage identity statement you must have the ability to make the appropriate substitutions. This may be intended in two main relevant ways:
    a) to be able to use the same term inside different inferential structures such as:
    bachelor (Todd)
    not married (Todd)

    b) to be able to substitute the term with another one referring to it.

    Todd = the smartest speaker of H-T seminar
    Tired (Todd)
    Tired (the smartest speaker of HTS)

    Brandom gives much importance to the symmetric substitution-inference structure of singular terms; the preceding inference is derivable and can be derived from:

    Tired (the smartest speaker of HTS)
    Tired (Todd)

    The symmetric structure characterizes singular terms, while the asymnmetric susbtitution-inference structure characterizes predicates: from: [Tired (a man) -therefore - Tired (an animal)] you cannot obviously derive: [* Tired (an animal)- therefore - Tired (a man)
  • Main step in the argument The main passage in the argument is perhaps the following: if we do not have substitutional role we do not have inferential role; therefore if an expression lacks substitutional role, it is meaningless; we do not know what to do with it.
  • Demonstratives Demonstratives are "unrepeatable"; they may occur only once in a certain context of utterance. Their function is to pick up an object; but to pick up an object you need to refer to the same object again: you need identity criteria. From that and from the preceding "main step of the argument" it follows that, in order to work as meaningful pieces of language, demonstratives have to be used in substitutions. But they cannot because they are not repeatable. But how to pick up the same object referred by a context dependent demonstrative? Human languages are endowed with anaphoric devices which serve the purpose. Demonstratives may be appear in some substitution structure as anaphoric initiators: "that pig is grunting, so it must be happy"

    "...the capacity of pronouns to pick up a reference from an anaphoric antecedent is an essential condition of the capacity of other tokens (which can serve as such antecedents) to have references determined. Deixis presupposes anaphora. No tokens can have the significance of demonstratives unless others have the significance of anaphoric dependents; to use an expression as a demonstrative is to use it as a special kind of anaphoric initiator" (462)

    The anaphoric dependent "it" is a repeatable expression which may give rise to substitutions like:

    "It is Wilbur" or " The pig he pointed to saying 'that pig' is Wilburg, the big one"

    Anaphoric dependence is not substitution, but "a more primitive sort of commitment" which builds up the link from unrepeatable tokenings to repeatable ones (or between unrepeatable events and repeatable contents - p.465): "substitutional commitments governs the use of repeatable expressions. Anaphora is required to generate repeatable from unrepeatable tokenings, paradigmatically deictic ones." (467).
    [back to top]

  • 2. indirect speech as relation between tokens [534-539] This part of the background relies heavily on a generalization of Davidson's analysis of indirect speech, without the awkward paratactic aspect of the theory (1). The three main feature of Davidson's theory are:
    1. it focuses on tokens rather than on types
    2. it displays a sentence token related to the token of the original speaker
    3. it centers on the problem of the relation between reporting and reported tokens.
    The relation between reported and reporting token is that they share the same content = what is preserved by good translation. The two tokenings are therefore called "samesaying": the reporting token specifies the content of the reported token by being a samesaying with it. Brandom accepts the basic ideas in 1 and 2, while rejecting the paratactic aspect and keeping the idea that you need to make it explicit the relation between reporting and reported speech.

  • samesaying and commitments The work done by the notion of samesaying is given in Brandom by the notion of identity of commitments undertaken by the speaker and the reporter: a relevant distinction has to be clearly traced between the case in which a reporter undertakes the same commitment of the original speaker (deferring the responsability of justifications to the speaker) and the case in which a reporter does not undertake the same commitment.
  • Contents and contexts Contents of indexical & anaphoric expressions is to be evaluated at the context of the ascriber's utterance, not at the context of the ascribed utterance. This is an application of the point 3 of Davidson: indexicals are to be evaluated in direct and indirect speech according, respectively, to the context of reported and reporting tokens; we need to distinguish between reporting with quotation and proper indirect speech:
    Tim says "I am confused": /I am confused/ is evaluated at the context of the reported token
    Tim says that he is confused: /he is confused/ is evaluated at the context of the reporting token
    [back to top]

  • 3. rigidity as anaphoric phenomenon (a generalization of anaphora) [468-473] A typical aspect of demonstratives is that they pick out the same object in all possible worlds; that is a feature for which Kripke introduced the term "rigid designator" and Kaplan invented the operator "dthat" to stipulate that, while a definite description (!xDx) is not a rigid designator, dthat (!xDx) is to be considered as fixing the reference. This analysis is considered as a particular application of a more gerenal feature of language: anaphora.

  • Rigidity, anaphora and proper names As Chastain 1975 suggested, rigid designation seems to be a case of a more general feature of language: anaphoric chains. The proper work of anaphora is to make a direct link to the anaphoric initiator, rigidifying it. A general suggestion is given (to be developed later) that tokenings of proper names can be understood as elements in an anaphoric chain that is anchored in some name-introducing tokening.
  • Causal chains, anaphoric chains Brandom develops Chastain's insight claiming that causal theories of proper names appear as "dark ways of talking about the sort of anaphoric chains that link tokenings of proper names into recurrent structures" (470) His idea is that anaphora reveals a general primitive recurrence structure which is exploited by many kinds of terms (proper names, mass terms, and so on).
  • levels of commitments Brandom considers three levels of commitments: assertional, substitutional and anaphoric ones, related respectively to the level of sentences, subsentential elements (singular terms and predicates) and unrepeatable tokenings (paradigmatically deictic uses of singular terms): "the articulation characteristic of specifically discursive commitments is to be understood most broadly in term of inference, the details of which require attention to substitution, the details of which in turn require attention to anaphora."
    [back to top]

  • 4. proper names' senses are anaphoric chains [572] Two competing views are normally discussed about proper names: the Frege-Russell descriptive view, where for each name we have a cluster of definite descriptions which constitute the sense of the name (Searle) and the Mill-Kripke causal theory for which names have no senses but refer directly to their referents. Does it exist a third way between these two main divides? Brandom tries to develop a third way, exploiting some ideas developed by Evans and McDowell.

  • Singular thoughts and de re senses Brandom gives a general definition of "singular thoughts": Object dependent thoughts,...are those that can be entertained only if the singular terms occurring in the sentence tokenings that express them do succeed in picking out objects. Prime among them, the suggestion is, are those propositional contents, whose expression requires the use of demonstratives or indexicals." (586)
    This definition avoids the idea of "singular proposition", as an entity whose constituents are considered to be "real" entities, the referents themselves. The singular proposition "London is pretty" is an ordered pair consisting of the town and the property of being pretty: (London, Being Pretty).
    An escape from singular propositions is suggested by Frege when he says that the content of a thought belongs to the realm of "senses". The problem is that nobody ever agreed about what senses are, especially senses of proper names. Are they (clusters of) definite descriptions? McDowell suggested that, analyzing Frege's texts, we have another possibility: de re senses are something very peculiar, not descriptions, but constraints given by the objects themselves, which figures in the thoughts not as constituents but as the way in which we build up our cognitive relation with the world. He sustains in a well known and debated paper that both causal theories and descriptive theories share a basic metaphysical dualistic assumption: the existing of a divide between language (mind) and world. The "third way" gives a completely different picture, where the problem of the relation between language and world disappears (2). The attempt made by Brandom has the great merit to try to give some substance to this idea of "de re senses" in a way which can be described and grasped formally.
  • Conceptual access to objects: senses as anaphoric chains A way to give substance of the idea of "de re" senses of name-tokens is to treat de-re senses in analogy with anaphoric chains; in a discourse a chain of name-tokens can be considered analogous to an anaphoric chain; this chain goes back to the anaphoric initiator (the first time the name occurs in the discourse of the speaker). The (structure of the) anaphoric chain can be considered analogous to the Fregean sense of a proper name: "Anaphoric chains contribute to both the theoretical task for which Frege postulated senses: they are the way in which objects can be given to us, and they determine the reference of the expressions occurring in them. Anaphoric chains of tokenings - explained in terms of inheritance of substitutional commitments - provide a model for object-involving, de re senses." (572) Given that conceptual content is given in term of inferential power, we may speak of the conceptual content of proper names, without being committed to senses as definite descriptions). This content is a constellations of tokens given by the relevant anaphoric chain.
  • Transparency principle In the most common interpretations of Frege senses are "transparent": "one cannot grasp two different senses without realizing they are different" (571). If we accept the general idea of singular thoughts, we have two choices: (i) retain transparency principle and give away senses: these thoughts are determined by objects alone without intervention of senses of proper names or indexicals (direct referent theory); (ii) abandon transparency principle and retain the idea of a de-re sense for proper names as stated before. Anaphoric chains cannot guarantee the transparency: one can grasp two different anaphoric chains without realizing whether they are linked to the same or to different objects. Therefore substitutional commitments made by a speaker using a chain may differ from that of somebody else who knows more.
  • Normative aspects of reported and reporting speech Tokens of different types (Todd is tired, he has worked hard, I'll ask him...) form classes in anmaphoric chains. These classes are subject of substitutional (and then also - indirectly - inferential) commitments. A basic tool for the analysis of indirect speech is: always distinguish different attitudes toward reported speech (attributing vs. undertaking). To distinguish between reported and reporting speech we need to make it clear which kinds of attitude are taken in a report; in particular De re ascription of belief such as "s believes of Tim the he is tired" is understood in terms of:
    1) attributing a commitment to the speaker
    2) undertaking an existential commitment (568-569)
    [back to top]

  • 5. Brandom's analysis of Kripke's puzzle (573-581) Kripke's puzzle has been analyzed in different ways. Linking proper names to anaphoric chains gives a way to analyze Kripke's puzzles as the behavior of a speaker who does not realize that - two different tokens of different types are about the same individual (London-Londres) - two different tokens of the same type are about the same individual (Paderewski case) The reason of the speakers' failure lies in his linking the tokens to two different anaphoric chains where the anaphoric initiators are supposed to be different. Giving this analysis is not immediately equivalent to explain away the puzzle; in what follows we try to show the steps Brandom takes in order to dismantle Kripke's puzzle, and to claim that the so called puzzle leads to some substantive conclusions about the logical behavior of proper names.

  • Weak and strong form of Kripke's puzzle Brandom individuates two different forms of the puzzle, to show that the real difficulty is in the "strong" form (the one actually clearly defined by Kripke in his paper); the "weak" form is easy to be dismissed, as we shall see soon.
    weak - attributing inconsistency to the believer :
    Bel (Peter, Pa & not Pa)

    strong - attributing inconsistency to the reporter :
    Bel (Peter, Pa) & not Bel (Peter, Pa)

    In the case of the weak interpretation we may easily resolve the puzzle, abandoning the form of de-dicto reports which attribute contradictory beliefs to the believer and skip to a de-re form which avoids it. Example: from: "Peter believes that Paderewski has musical talent and that Paderewski has not musical talent" to: "Peter believes of Paderewski, who has musical talent, that he has no musical talent"

  • Strong puzzle and the limitations on the disquotational principle The puzzle is linked to two main principles, the translation and the disquotational principles; Brandom suggests that the translation principle is not so central, given the fact that the puzzle can be given also in the pure form without need for translation (the Paderewski case is such an example). Therefore Brandom gives the utmost attention to the Paderewski case and to the disquotational principle:

    " if a speaker sincerely assents to 'p' then he believes that p"

    This principle is analyzed as constituted by two sub-principles, concerning respectively:

    (a) linguistic avowals of beliefvs.attributions (reports) of belief
    (b) expression used by the believer vs. expression used by the reporter.

    Brandom accepts (a) and rejects (b): an assertion with a certain content of belief is an evidence for attributing a belief, but it is not true that the very same words used to avow the belief are to be used in reporting the belief. This follows from the reconstruction of the davidsonian analysis of indirect speech given above: specifying the relation between reported and reporting tokenings involves "subtelties which the disquotational principle simply ignores." (577)
  • Brandom's strategy Kripke himself acknowledges three categories of exceptions to the disquotation principle: ambiguity, indexicals and pronouns. Brandom's point is not to reject disquotational principle tout-court, but to reject its general applicability to proper names. This kind of answer parallels answer of the kind Sosa 1996 has given, which relies essentially on the problem of ambiguity. Brandom's strategy is to show that also proper names are locutions which do not fall under the disquotational principle, exactly under some circumstances similar to the ones aknowledged by Kripke. I will give here just two examples:
    (i) As far as ambiguity is concerned, Kripke makes the case of a person using the same name-kind (Cicero) to refer to the famous orator and to a spy in the II World War; this is a case of clear ambiguity and cannot give rise to any puzzle. Brandom points out that Kripke gives no evidence for making us accepting a real difference between the cases fo Cicero and of Paderewski; in fact the difference can be understood only with facts not accessible to the speakers entertaining the relative beliefs.
    (ii) Distinct "anaphoric" chains of tokenings of "it" may be anchored in antecedents picking out either different objects or the same object. Both of these structures are the one characterizing the use of proper names (cicero/cicero or paderewski/paderewski). Given the similarity of the way to pick up the referent via a chain (call it anaphoric or causal), the behavior of proper names can be considered analogous to the behavior of indexicals or pronouns, and therefore automatically excluded by the disquotational principle.
    Brandom quotes Kripe's idea that differences in the beliefs of speakers (different descriptions given by speakers of a language) do not change the reference of a name so long as the speaker "determines that he will use the name with the reference current in the community": Brandom suggests that this idea is like a rough account of what it is to use a pronoun with a certain antecedent. In both cases (proper names and pronouns) to take a token as continuing a chain is to be committed to take it as inheriting its substitutiona-inferential role from the anteceding tokenings and to be committed to determine the referent by tracing the chain back to the "anaphoric" initiator (580-1)
    [back to top]

  • 6. Generalization of the analysis Brandom's analysis of Kripke's puzzle aims to a dissolution of the puzzle based on ideas very similar to Kripke's (on proper names as chain of tokenings linked to an original fixation of the reference). Given the general framework, his aim is to provide a very general alternative to the causal reference theories, while retaining some general patterns of their explanation in a different setting, more similar - but not identical - to a Fregean perspective.

  • General results The analysis of Kripke's puzzle gives some general results, not only about proper names, but about a general restriction of the use of the disquotational principle and a general restriction about the possibility of detecting contradictions "by pure logic and semantic introspection". (i) Proper names can be used in such a way that the disquotation principle does not apply to them. This conclusion is motivated by an approach to proper-names' reference in term of chains of tokenings, which is shared by Kripke himself. The result is that there are good reason to treat proper names on an anaphoric model, which explains away the puzzle. (ii) "as you cannot tell "by pure logic and semantic introspection" whether two chains ...are anchored in one object or in two for ordinary anaphoric dependents, so one cannot for the anaphoric chains that govern the use of proper names" [581] (iii) you cannot tell in general from the lexical type of an expression whether it is used in conformity with the disquotational principle (e.g. only attributive uses and not referential uses of definite descriptions have to follow the disquotationl principle).
  • Mill vs Frege p.582 Both the "invisibility" of the anaphoric analysis of proper names and the idea that the puzzle has to do with belief come from a Millian picture of the working of proper names. Bur even Kripke's theory is not easy to be given in pure millian terms: demonstratives are not directly referential because they require at least a sortal to pick out the referent; pronouns convey also gender information or implicit personal sortals. On the other hand it is difficult to think of a fregean theory as defining a sense of a name as a set of properties (as Kripke suggests); properties belong to the realm of reference and not to the realm of sense. To properly understand the counterposition between Mill and Frege we need to make Frege less "descriptivist" than it is supposed to be, and to give some space to the possibility of "de re senses". In that case, however, the criticism of the causal theory collapses.
  • Tactile rather than visual model of understanding? Just a quotation [583] "Constellations of tokenings given by anaphoric chains (extending this structure to proper names) are a general feature of language. Such chains anchor our thought and talk in particular objects that it is about. One can grasp an anaphoric chain as one grasps a stick; direct contact is achieved only with one end of it. Conceptual contents are therefore best thought of on a tactile rather than a visual model. As something to be grasped, rather than something to be represented.
    [back to top

  • 7. Comments I would give some hints for a discussion, isolating some points where the attention should be posed:

  • Idiolect and common senses Brandom did not say how he would have expressed Pierre's beliefs. What can we deduce from his general assessment of Kripke's puzzle? In the Paderewski case we should say that Pierre believes are linked to two different anaphoric chains where the anaphoric initiators (be what it may be, names or demonstratives) are understood by Pierre to refer to different individuals. But this is just a way to recast the puzzle, not to solve it. Maybe the suggestion could be expressed in the following "Fregean" way: the reference in indirect speech (in belief contexts) is an indirect reference, that is the sense in normal speech. In this case Pierre is reported as referring not to Paderewski himself, but to the two different senses (anaphoric chains) he attaches to the name. We could say something like: " Pierre believes that Paderewski, as the individual linked to the anaphoric initiator x in the context C, has musical talent, and Pierre believes also that Paderewski, as the individual linked to the anaphoric initiator x1 in context C1, has no musical talent".
    If this is the intended suggestion we are back to the main problem posed by Kripke to a "moderate" Fregean solution to the puzzle (note 29 of Puzzle about Belief): we are in this case stuck in a not perspicuous set of individual idiolects, where everybody attach his own senses to a name, and it is not easy "to obtain a requisite socialized notion of sense" which could be grasped by most of the individuals belonging to a linguistic community (3).
    We could try an attempt like that:
    weak puzzle: Pierre believes of Paderewski, who almost everybody knows to be a musical talent, the he is not a musical talent.
    strong puzzle " Pierre believes that Paderewski, as the individual linked to the anaphoric initiator x in the context C, has musical talent, and Pierre believes also that Paderewski, as the individual linked to the anaphoric initiator x1 in context C1, has no musical talent AND Paderewski is used in both of his chains to refer to an individual which is individuated by most of the chains used in the community as the same individual which is typically referred to as a musical talent".
    This last formulation imposes on Pierre a change of the senses he attributes to his tokens of Paderewski: if earlier he was committed to forbid substitutions of the token /Paderewski/ in the two different chains, when offered a clarification, Pierre has to permit new substitutions, therefore to change the inferential meaning attached to the two chains. He has to use the two different chains as derivation from some former hypothetical chain which begins whit a unique anaphoric initiator. He has then to recognize that the two chains are two variants of a unique one: he has changed the sense attached to the his tokens of Paderewski.
  • Transparency and Cognitive value Brandom says that anaphoric chains perform the main role Fregean senses are supposed to perform: they are "the way in which objects can be given to us, and they determine the reference of the expressions occurring in them". But there is another role Frege wanted the notion of sense to perform: to explain the cognitive value of sentences . It is difficult to define clearly how this cognitive aspect is maintained by treating senses as anaphoric chains. This is, however, a point which relies widely upon McDowell's analysis of de re senses and the possibility of an analysis of the concept of sense which is neither descriptivist nor direct-referential. Certainly, with all their difficulties, the two alternatives are more clearly stated and easier to accept, than a vision where it is difficult to distinguish clearly how the conceptual aspect is expressed.
    The attempt given by Brandom makes a clarification: if we accept something like de re senses, we have to give up the transparency condition for senses given by Frege. This is a very delicate point, because the transparency condition is a central condition for which Frege needed the concept of sense: "one cannot grasp two different senses without realizing they are different". Transparency of thoughts is retained by direct reference theories abandoning fregean senses and accepting the idea of singular propositions made by ordered pairs of individuals and properties. But Frege required transparency for thoughts, without saying what it is for a proper name to have the same sense; he just stressed the difference between two thoughts with names (Hesperus and Phosphorus) with different senses. Therefore to interpret anaphoric chains as not epistemically transparent is rather vague, because two anaphoric chains are always regarded as different chains (it is a unavoidable syntactic matter); we cannot have guaranty that we understand when two different chains pick up the same individual or different ones. But even Frege cannot guarantee thet we know of two different senses that they pick uo the same individual. We have to make a distinction between two different cases:
    a) a case of deep ambiguity: we do not realize which sense we are using; we may distinguish a de re sense in a veridical case and an allucination; given the definition of "de re" sense as depending on the object, we have two different senses when we speak of an allucinated object without realizing it, and when we speak of a real object. We might be unaware of the kind of sense we are using, therefore we cannot tell which sense we are using(4)
    b) a case of superficial ambiguity: we do not realize that we are using different senses (differnt chains) to refer to the same individual. We may leave aside the problem of deep ambiguity: any chain we use purport to refer to some individual; if it is a mock individual our thought has a differnt status that the one we entertain when speaking of a real individual. [But I will not go so far as to say that we have "mock thoughts" for reasons given by Bell]. We are always dealing with different chains which, being sintactically different one of the other, are always recognizable as different. The only problem is the classical fregean problem of not being aware the two senses denote the same individual.
    We might then define a weak transparency principle: two different anaphoric chains are always considered different (picking up different objects) by default, unless new information is provided to link the two chains. Certainly every chain is different for every speacker; we just need to know when they converge to (an ideally unique chain with) the same anaphoric initiator. Chains are changing structures: they develop and they merge; still we may recognize a chain as the same if we maintain the same substitutional commitments in picking up the same intended referent.
  • Conceptual confusion What is the content of a belief expressed by a sentence with a proper name? The two main answers are: a singular proposition, which we may express with an ordered pair consisting of a property and an object (think of - Londond, the property of being pretty - ) and a thought, whose component are senses (the sense of a predicate and the sense of a proper name). (*) Brandom gives an enriched version of the second answer, but, given his analysis, he has to face an unexpected difficulty: if he wants some kind of conceptual (=inferential) content to be attached to proper names, he has to credit Pierre with conceptual confusion.
    Speaking of chains is for Brandom speaking of conceptual contents as the substitutions a speaker is committed to when he uses some tokens in a chain; but here two problems arise:
    (i) the definition of what kind of confusion is the confusion which arise of thinking of two different chains as arising from two different individuals where they are connected with the same individual.
    (ii) the difference between the substitutions a speaker is committed to by the knowledge shared in his linguistic community and the substitutions a speaker aknowledges, maybe wrongly as in the case of Pierre.
    The second problem is the problem of how to treat belief revision and the first is the problem on conceptuial confusion, treated by Kripke only about predicates (you cannot attribute belief in case of conceptual confusion). I will develop here only the first problem because it seems to me that it is a point which naturally comes out from Brandom's perspective.
    A chain is a conceptual introduction to an individual: it picks up the individual sortally (if London were not a town but a person or an animal it would not be London) and binds you to certain commitments given by the lexicon you use and the information shared in the society. Not realizing that two chains are linked to the same individual object (in this case to the same town) can be considered, under this perspective, a case of conceptual confusion. Therefore we cannot attribute a proper and coherent belief to the speaker because "we have ground to suspect conceptual or linguistic confusion" (Kripke note 23).
    Apparently for Kripke Pierre is not in a case of conceptual confusion, because there are no senses involved; but for Brandom Pierre does not recognize that his two different conceptual contents (two chains) are considered as the same content (as a unique chain) by the society. (5)
    We should say then that, being in some kind of conceptual confusion, Pierre does not really believe, but "claims to believe" (I use here Marcus' terminology, not Brandom's; here he seems to be compelled to a position similar to Barcan Marcus').
    However our notion of belief, being different from the notion of knowledge, should account for false beliefs, and therefore for contradictory beliefs. In fact it is hard to imagine somebody entertaining a false believe which does not clash in its consequences at some time with a true one. Attributing beliefs only when consistency is retained seems to be too much of a requirement; explicit aknowledge inconsistency seems to be required to refuse really entertaining of beliefs (and rationality) to a speaker; to permit belief revision when new relevant information is givent to the believer, we need to have wrong and contradictory beliefs come to light as such.
    Therefore it seems reasonable that Brandom tried to confer proper beliefs to the speaker (to Pierre); but he wants to attribute Pierre a proper belief, and not just a "claiming to believe", he seem obliged to abandon the kripkean assumption for which we should not attribute a belief in case of conceptual confusion. And also this choice seems not easy to accept.
    [back to top]


    (1)Davidson ["On saying that " 1969]exploits the English peculiarity of "that", expression used both for a demonstrative and for introducing an indirect clause (this does not happen normally in other languages). In Davidson's theory the token occurring within the "that" clause is not strictly part of the ascription; the ascription is just something like "S says that", with "that" used demostratively; then a token follows to display the referent of "that", without punctuation - and for that the theory is called "paratactic".
    (2)See the detailed analysis of McDowell in De re senses, Singular Thought and Imnner Space and Language and World
    (3) Individuals and properties This attempted possibility to obtain some socialized notion of sense through anaphoric chains (or through causal chains translated into anaphoric reading) find a great difficulty in more complex judgments; where a property is not attributed as a "typical" property used by the community to refer to the individual; it is the case of towns. Is London pretty or is it not pretty? In an encyclopedia you certainly will find information about the musical talent of Paderewski as a typical feature of the man, but you will not find information about the beauty of London as a typical feature of the town. It really depend on taste and on parts of London under consideration (as it actually happened to Pierre). Therefore it is difficult to assert Pierre believes of London, which is not pretty, that it is pretty. Even if most of the people agreed that London is not pretty, this is not a typical property of the town; people may know or may not know whether Paderewski has or has not musical talent; but the beauty of London is not question of knowing, but question of aesthetic judgment. Therefore even in the weak case we have difficulty even in Brandom's formulation.
    (4) We might say that sometimes we do not realize which kind of thought we are entertaining, that is we do not realize which substitutions we may make in a chain we picked up from somebody else; somebody is saying "he will come and bring some cakes" and we continue "Let's hope he will bring many of them" whithout the least idea of whom we are speaking of, but that we are speaking of the individual the speaker is referring to, be it a real individual or just a mock one (in the case the speaker was joking)
    (5) We may express the same point in a more standard language: a rigid designator is a function from possible words to individuals; but (assuming that the function is computable) we may have different procedures which compute the function, different procedures attached to the function: not recognizing that they are computations of the same function can be understood as a conceptual confusion in the use of our intellectual tools (as if we could not realise that 2+2 gives the same result than 4-2 because we are not able to manage the algebraic structure of integers, but only the one of natural numbers).
    (*) [[If we accept that anaphoric chains are an to be identified with something like conceptual contents of names, Brandom can be interpreted as suggesting some kind of cluster theory revised, where instead of clusters of definite descriptions, we have clusters of anaphoric chains. Different anaphoric chains are to be considered convergent or correspondent (a term taken from McDowell 290 de re) if they pick up the same individual.]]