By Sari Pietikäinen, Alexandra Jaffe, Helen Kelly-Holmes, Nik Coupland
Sociolinguistics from the Periphery "presents a fascinating book about change: shifting political, economic and cultural conditions; ephemeral, sometimes even seasonal, multilingualism; and altered imaginaries for minority and indigenous languages and their users"
Review of New Essays on Semantic Externalism and Self-Knowledge
Date: Thu, 16 Sep 2004 15:00:13 +0200 From: Anne Reboul Subject: New Essays on Semantic Externalism and Self-Knowledge
EDITOR: Nuccetelli, Susana TITLE: New Essays on Semantic Externalism and Self-Knowledge SERIES: Bradford Book PUBLISHER: The MIT Press YEAR: 2003
Anne Reboul, Institute for Cognitive Sciences, CNRS, Lyon, France
DESCRIPTION This book could be seen as a companion to the collection of papers recently published on Tyler Burge (see the review on the Linguist list: issue 15.1691) and his own version of externalism. A short introduction on externalism for non-philosophers may be warranted. Semantic externalism is the position, first advocated by Putnam (1975), according to which meanings are at least in part determined by states of affairs in the world and thus are not purely mental and private. This seems to lead to the conclusion that a subject could not have a privileged access to his/her own thought, which contradicts a central tenet in philosophy of mind. The book is dedicated to the examination of this conflict between semantic externalism and privileged access. It should be clear that this question has repercussion in linguistics, notably in Gricean type pragmatics, where meaning is defined in psychological terms. The discussion is thus of quite real importance for all linguists who adopt a mentalistic view of language use and meaning.
SUMMARY The book opens with an Introduction by the editor, Nuccetelli. She notes that ''the attempt to hold externalism concurrently with privileged self-knowledge might face a reductio, since it might appear that substantial propositions about one's environment could be known by simple deduction from non-empirical premises'' (1). Nuccetelli presents the opposition between internalism (all mental properties including those with content are local in the sense that they preserve across internal replicas) and externalism (some mental properties -- e.g. those with content -- are not local in this sense). This can be reduced to claims about supervenience (roughly, co-variation) either on internal or external conditions. Externalism seems to be supported by theories of direct reference. There is, however, a compatibility problem between externalism and privileged and authoritative access, which can be seen from two perspectives: either externalism or privileged and authoritative access should be abandoned. The compatibility problem stems from a potential reductio (first proposed by McKinsey) whose formulation could be: (1) I am thinking that water is wet; (2) if I am thinking that water is wet, then some empirical condition obtains (from externalism, which can be argued for from thought experiments, i.e., a priori); (3) therefore, some empirical condition obtains. In other words, if externalism and privileged access are correct, a priori empirical knowledge should be possible. Nuccetelli then presents the contents of the chapters.
The first chapter, by Martin Davies, deals with ''The problem of armchair knowledge''. He begins by recalling the McKinsey argument showing the incompatibility between externalism and privileged access under its reductio form. He notes that if deduction from a priori known premises to an empirical conclusion goes through, then indeed inconsistency threatens. Thus ''externalist philosophical theory, when taken together with a plausible claim about self-knowledge, gives rise to an instance of what [Davies] call[s] the problem of armchair knowledge'' (25). The problem lies in the transmission of evidential support from the premises to the conclusion. Building on propositions by Wright, Davies argues that epistemic warrant from premises to conclusion is limited to those cases where the truth of the conclusion is not a precondition of the warrantability of one premise. Such problems, according to him, are relatively widespread and are related to the issue of begging the question in that both hinge on the question of background assumptions.
The second chapter, by Crispin Wright, is devoted to ''Some reflections on the acquisition of warrant by inference''. Wright begins with a distinction between transmission of warrant and closure of warrant (across entailment). Closure is weaker, depending merely on the fact that the existence of warrant for premises ensures the existence of warrant for the conclusion. Transmission, on the other hand, depends on warrants for the premises giving warrant for the conclusion. Transmission fails -- though closure does not -- in cases where warrant for the premises depends on (antecedent) warrant for the conclusion (begging the question). Non transmissible warrants may be connected with the holism of empirical confirmation: in such cases, there is information-dependence of warrant. This may occur for inferential warrants. Another type of cases has to do with non-inferential warrants acting as preconditions for cognitive activities such as perception, memory or self- knowledge. This seems to be the kind of failure of transmission, which occurs in McKinsey cases. However, operational necessity, in such cases, entitles the subject to believe a priori that the relevant defeaters do not obtain, thus allowing both semantic externalism and authoritative self-knowledge.
Brian McLaughlin writes the next chapter, on ''McKinsey's challenge, warrant transmission, and skepticism''. He begins by noting that given the closure principle, the issue is ''whether it is possible to know a priori a contingent environmental proposition'' (80), where a contingent environmental proposition a priori entails the existence of the external world. McKinsey's incompatibilism would go through if there was any ''reason to believe that we can't know a priori that the external world exists'' (80). By contrast, some compatibilists claim that privileged self- knowledge and a priori mental externalism can refute some kinds of skeptical hypotheses. McLaughlin then notes that neither Wright's nor Davies' argument (see above) shows ''that no compatibilist argument could be one in which warrant transmits'' (88). McLaughlin distinguishes two notions of a priority: a weak notion in which a belief is a priori if it warranted but not on the basis of empirical evidence; a strong notion in which a belief is a priori if it is warranted but not on the basis of empirical evidence and if it is empirically indefeasible in the sense that it cannot be rebutted by empirical evidence. The same distinction applies to privileged access to mental states, yielding four possibilities for a given mental state: weakly privileged, weakly a priori external; weakly privileged, strongly a priori external: strongly privileged, weakly a priori external, strongly privileged, strongly a priori external. The three first possibilities would involve a failure of warrant transmission, though the fourth would not. However, such a case is considered as implausible by McLaughlin.
The fourth paper is by Michael McKinsey and deals with ''Transmission of warrant and closure of apriority''. He begins by rehearsing his incompatibilist argument. This leads him to the conclusion that either semantic externalism or privileged access is false. However, it seems that semantic externalism is true given ''semantic facts about cognitive predicates that contain small-scope proper names and indexical pronouns'' (99). Thus, privileged access should be restricted to logically narrow properties. McKinsey then turns to the problem of transmission of warrant. He agrees that there is a failure of transmission in his argument but insists that closure is sufficient to the validity of the argument and that warrant is not necessary. He notes that closure will apply similarly in strong and weak apriority arguments. However, he admits that there is a logical possibility that some form of semantic externalism might be consistent with privileged access. Yet, ''most forms of externalism imply that certain cognitive premises have very strong empirical consequences'' (112) and they are not subject to the closure objection. He concludes that ''[privileged access] is false, and so there is simply no point in looking for forms of externalism that are consistent with it'' (113).
''The reductio argument and transmission of warrant'', by Jessica Brown, rejects failure of transmission of warrant in the reductio argument on the grounds that Wright uses an internalist epistemology which is incompatible with externalism. ''Thus, the concern that drives Wright to his conditional analysis need not be shared by epistemological externalists, for whom epistemic entitlement is partly a matter of the subject's relations to her environment'' (121). What is more, Wright's solution is as problematic as the reductio itself. ''It seems then that we need to look for an answer to the reductio argument elsewhere'' (130).
Dretske devotes the sixth chapter to ''Externalism and self- knowledge''. He begins by noting that metaphysical and epistemological doctrines in philosophy are often difficult to reconcile and that in such cases, ''there is a temptation to trash the metaphysics in order to salvage the epistemology'' (131). In the specific case of the compatibility between externalism and privileged access, however, he would suggest ''to reexamine (...) a suspect epistemology'' (131). Dretske begins by reminding the reader that his externalism is not restricted to mental content, but generalizes to all forms of representation. He then turns to the sort of epistemology embodied in the current view of privileged access, noting that it makes a wrong assumption: ''it assumes that knowledge of what you think -- for instance, that there is water -- is (or requires) knowledge that you think'' (133). Thus, ''if we think, we can know -- in a privileged and authoritative way -- what we think. What we can't know, at least not in the same authoritative way, is that we are thinking it'' (136). Introspection is like perception in that it gives us the content of our cognitive states, but not the fact that it is content. Thus, ''perception is not an answer to the skeptic's problem about other minds'' (138). Additionally, ''carrying information (...) is not closed under (known) logical implication'' (140). Knowing that we are thinking is not done through introspection, neither is it subject to privileged and authoritative knowledge. Rather, it is learned by human children through their parents and language plays a major role. ''Nonhuman animals never learn these things. This is exactly what one would expect on an externalist theory of the mind'' (141).
Gary Ebbs then turns to ''A puzzle about doubt''. He begins by noting that the standard view of the relevant thought experiments supposes subjectively equivalent worlds such that a subject might never know, without empirical investigation, which world she is in. However, Ebbs argues, this is wrong because this analysis ''conflicts with the truism that to express a thought, one must have some idea of what that thought is'', which entails, on an externalist view, that one must know in which world one is. On Ebbs' view, ''minimal self-knowledge is a practical aspect of ordinary competence in the use of language, not a kind of second-order propositional knowledge as many philosophers assume'' (147). Additionally, ''most philosophers assume that for a person to entertain the thought that she is actually in one of her subjectively equivalent worlds, she need only picture herself in it'' (150) in the sense of adding to her present subjective experience a caption referring to that world. Adopting this standard picture leads to a puzzle as to what words refer to. To dissolve this puzzle, Ebbs proposes that in fact we can know without empirical investigation which world we are in. His first assumption is to the effect that ''it is epistemically possible that p for a given person only if she can make sense of its actually being the case that p'' (154). But for that, it is not enough that she pictures herself in such a world: she must be able to express that possibility and to do that she would have to suspend all her substantive beliefs, which would deprive her language of its semantics, preventing her from expressing the possibility. ''Statements that we can't make sense of doubting in the sense described above are among the statements that we properly take ourselves to know without empirical investigation'' (162).
The editor of the book, Susana Nuccetelli signs the eighth chapter, ''Knowing that one knows what one is talking about''. She wants to ''argue (...) that externalist claims about the dependence of content on environmental factors presuppose certain theses about the semantics of natural- kind terms that, if sound, would make those claims eligible for empirical justification instead'' (169). She uses the example of singular propositions, where a causal chain is involved linking the (proper) name and its referent. This excludes Fregean senses. As for natural-kind terms, though they do have senses, these fall short of being Fregean. Just as singular terms may fail to have referents, so do natural-kind terms (e.g. phlogiston). Thus, if semantic externalism is correct, then the justification of belief about substantial entailments from words and thoughts to the environment rests (at least in part), on empirical investigation. This leads to an answer to incompatibilist arguments: self-ascriptive beliefs about propositional attitude contents could be a priori justified under the weak notion of apriority (see above).
Anthony Brueckner deals in chapter 9 with ''Two transcendental arguments concerning self-knowledge''. He begins by noting that ''what is needed for a fully adequate reply to the skeptic about self-knowledge is an explanation of how it is that I know that I am thinking a water thought'' (186). He then turns to two transcendental arguments, one by Moran and one by Bilgrami. They both reject infallibility and transparency of self-knowledge, as well as the notion of an introspective faculty. They both agree on the non-inferential and not evidentially based character of self-knowledge. Bilgrami tries to show that the notion that believing to be in a given mental state is enough to guarantee that one is indeed in that mental state is a condition of the possibility of responsible agency. Brueckner introduces here Shoemaker's notion of a self- blind person, incapable of second order beliefs and question whether such a person would be incapable of responsible agency (entailing freedom, responsibility and appropriateness of reactions). This leads him to the conclusion that Bilgrami's argument fails. He then turns to Moran's. Moran distinguishes ''between two points of view upon one's own intentional mental states: the theoretical/empirical point of view and the transcendent point of view'' (194). The first one is evidential (and thus subject to a Moore's paradox), while the second is not, but is based on actual states of affairs in the world (and thus not subject to a Moore's paradox). A rational agent should adopt the transcendental point of view, which is necessary for responsible action. This is however false. Hence, again there could be a self-blind but nonetheless responsible agent.
In chapter 10, Joseph Owens turns to ''Externalism, Davidson, and knowledge of comparative content''. He examines Davidson's views on externalism and self- knowledge, arguing that Davidson misunderstands the incompatibilist arguments. Owens begins by a distinction between physical externalism of the Putnam kind (P- externalism) and social externalism of the Burge kind (B- externalism). One key issue here is whether one should not only think that ''the twins merit different psychological characterizations'' (202), but also ''assign different psychological states to the twins'' (202). Davidson would accept the first possibility, but not the second, thus qualifying as a linguistic but not a metaphysical realist. He also rejects B-externalism as incompatible with self- knowledge, while he accepts P-externalism as compatible with self-knowledge, rejecting incompatibilist arguments on the grounds that they rely on an introspective view of self-knowledge as perception of inner mental objects. This, however, is mistaken in Owens'view in that the mistake is not about the mechanisms but about the nature of self- knowledge. The nature of self-knowledge is (wrongly) taken to be that to know that I'm thinking a water thought, I must be able to reject the possibility that I'm thinking a twater thought. Davidson's rejection of B-externalism is based on the fact that, given that, on B-externalism, public conventions determine content, ''this makes content independent from intention'' (212), opening the possibility that one would know what one is thinking or saying. This conclusion is not, however, correct.
Kevin Falvey devotes the eleventh chapter to ''Memory and knowledge of content'', turning to the second kind of incompatibility argument. He begins by noting that our knowledge of our past mental states is not as authoritative as knowledge of our present mental states. It may also be more objective, though less direct. Thus, ''first-person authority extends primarily over one's present-tense attributions of intentional mental states'' (221). However, a more pressing question is to whether one can still be authoritative about one's own thoughts if one is radically out of touch with one's past mental states, through, for instance, world switching. This is Boghossian's argument, ''directed against Burge's idea that a central part of our notion of first-person authority involves the self- verifying character of first person attributions'' (221). This self-verifying property stems from the fact that one cannot entertain a second-order judgment about a first order thought without thinking that first-order thought. Boghossian's argument rests on an equation between remembering and having once known and not forgotten, an equation which Falvey contests. To show this, Falvey introduces the notion of disjunctive kinds of stuffs, which takes care of world switching. Thus, the right to rely on preservative memory is not itself entirely free of empirical presuppositions.
In chapter 12, Sanford C. Goldberg asks ''What do you know when you know your own thoughts?''. He wants both to acknowledge that there are valid incompatibilist arguments and that semantic externalism is well motivated. His view is that one should distinguish between two kinds of self- knowledge. The first kind are first-person present-tense judgments, or basic self-knowledge, and self-verifying. Goldberg then turns to a variation on Boghossian's argument, noting that it relies on the notion that world switching is accompanied by concept shifting in the sense of a replacement. This however is unconvincing and indeed implausible. Nevertheless, an adaptation of Boghossian's argument does go through. However, this argument does not threaten compatibilism between externalism and basic self- knowledge as this is not the kind of self-knowledge involved in the Boghossian style argument. What is more, such a distinction between two types of self-knowledge is independently motivated.
Chapter 13, by Richard Fumerton, deals with ''Introspection and internalism''. Fumerton begins by pointing out critics of externalism use arguments which rely on half of Leibniz's law in a way strongly reminiscent of the Cartesian dualist argument. After having outlined various internalist and externalist commitments, Fumerton revisits the incompatibilist argument. He supposes ''that the externalist will concede that we know through introspection what we are thinking of in a way in which we cannot know through introspection truths about the causal origin of our internal states'' (270), a concession that, according to him, is fatal to some externalist commitments.
Mathias Steup examines, in chapter 14, ''Two forms of antiskepticism''. He goes back to the McKinsey's argument, noting that Sawyer and Warfield have argued that there is no contradiction in spite of appearances between semantic externalism and privileged and authoritative access, which they take to be a basis for an antiskeptical argument. Steup, though sympathetic with the attempts, is however doubtful about the soundness of the arguments. According to him, ''there is reason to think that, while it might be possible to gain a priori knowledge of some very general propositions about the external world, the premises of Warfield's argument are, in point of fact, not knowable a priori'' (282). One point is that Warfield's argument falls under failure of transmission of warrant, thus begging the question against the skeptic. However, Steup thinks that an appeal to background knowledge might do the trick: I know that I have a body and that there is an external world because I can know that the skeptical alternative is false through my background knowledge, which defeats it. Thus, skeptical alternatives ''are defeated in the strong sense that their negations are beyond a reasonable doubt'' (291).
CRITICAL EVALUATION This is an excellent book in which most papers are of very high quality. Its relevance to linguists should be obvious. All the linguists who think that, whatever ''the function'' of language is, language can be used to describe reality presumably think that words have extensions as well as intensions and may also think that, where words refer to categories of objects in the world, e.g. natural kinds, the extension of these words play a major role in their semantics. In this, they are semantic externalists. They may also think however that meaning should be described in psychological terms and that speakers have privileged access to what they mean by a given utterance. Hence the resolution, if any, of the conflict between semantic externalism and privileged access is of special interest to at least some linguists. Though I don't think that any specific paper in this collection provides a complete and completely satisfying solution to this problem, most of them make an interesting contribution. Of special interest for linguistics are, in my opinion, Dretske's, Ebbs', Nuccetelli's, Owens' and Steup's papers. Each of them contributes a more or less important piece to what could be a solution to the puzzle. They also happen to be perhaps more accessible to a non-philosopher audience than are other papers in the collection, which though highly sophisticated and fascinating for philosophers, tend to be more technical.
REFERENCES Putnam, H. (1975) ''The meaning of 'meaning''', in Mind, Language and Reality. Philosophical Papers, vol. 2, Cambridge, Cambridge University Press, 215-271.
Hahn, M. & Ramberg, B. (2003),Reflections and replies. Essays on the philosophy of Tyler Burge, Cambridge, MA.,The MIT Press.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Anne Reboul is a First Class Research Fellow at the French Center for Scientific Research (CNRS) in France. She has a Ph.D. in Linguistics (EHESS, Paris) and a Ph.D. in philosophy (University of Geneva, Switzerland). She has written some books, among which an Encyclopaedic Dictionary of Pragmatics and quite a few papers in French and English, on pragmatics and/or philosophical subjects.