Occasion-Sensitivity - Charles Travis
Descripción
The Philosophical Quarterly Vol. No. ISSN –
January doi: ./j.-.._.x
expressed can one (even the subject ...) retrospectively determine what must have been intended’ [by the action one has performed] ... we can only know what we intended to do after we have actually acted’ (p. ). No The doubt these things are often Scots Philosophical Association and theBut University St Andrews true – it would be a clot who denied these possibilities. to ofmy mind, the possibilities do not impugn the standard case wherein agents form intentions and act them out knowingly. Right now I am finishing my review of Pippin’s book and have full first-person knowledge of this as I type. So it is quite wrong of Pippin to use the opacities of social dependence and the time-ordering within which a clear knowledge of one’s intention is sometimes achieved to challenge the causal model. Donald Davidson himself was clear that someone who is capable of first-person knowledge may get things wrong in a particular case. First-person knowledge of one’s intentions is not first-person infallibility. Hegel insists that the moral subject has ‘The right of intention [which] is that the universal quality of the action shall have being not only in itself, but shall be known by the agent and thus have been present all along in his subjective will’ (PR §). ‘Been present all along ...’, Hegel writes. This does not read as the conclusion of one who believes that intentions emerge only when actions are finally completed. Evidently these reflections are to be carried on further. In this task one could not have a more stimulating companion than Robert Pippin’s splendid book. University of Glasgow
D K
Occasion-Sensitivity. B C T. (Oxford UP, . Pp. vii + . Price £..) This is a book of essays in philosophy of language dating from to . They treat some familiar philosophical issues, and painstakingly develop objections to fundamental assumptions of traditional logically orientated analytical philosophy (LOAP). I have space for two examples, Travis’ approach to a kind of sceptical argument widely discussed around the middle of the twentieth century, and his objections to LOAP approaches to natural-language semantics. These objections bear on assumptions which are central to several of his other targets, including Grice’s distinction between standard meaning and implicature, bivalence according to which natural-language declarative sentences must be either true or false, some Quinean ideas about the opacity of belief ascriptions, and the Fregean account of belief as a state which relates the believer to the sense of a sentence. I found Travis’ sensible ideas about language more compelling than the methods he uses to develop them and some of the concerns that drive his work. The latter derive from ‘ordinary-language philosophy’, LOAP’s main analytical-philosophical competitor. Like John Austin, Travis grounds philosophical claims on intuitions about what competent native speakers of English would or should say in hypothetical situations of his own devising. My methodological reservations hark back to an exchange between Benson Mates, ‘On the Verification of Statements about Ordinary Language’, and Stanley Cavell, ‘Must We Mean What We Say?’, both in Inquiry, (), pp. –, –. Mates argued that native speakers’ intuitions are no substitute for empirical © The Author The Philosophical Quarterly © The Editors of The Philosophical Quarterly Published by Blackwell Publishing, Garsington Road, Oxford , UK, and Main Street, Malden, , USA
BOOK REVIEWS
investigations of language use. Even the most careful philosophers tend to be unreliable reporters of their own linguistic behaviour, let alone of that of others. Furthermore, even accurate reports of linguistic behaviour may not tell us much about meaning or inferential structure. A question hotly debated at the time was whether the term ‘voluntary’ meaningfully applies to unexceptionable everyday behaviour. What if people refrain from saying such things as ‘He tied his shoes voluntarily’ not because of any philosophically interesting features of the meaning of the term ‘voluntary’, but just because they do not like to spend time talking about voluntariness in connection with things they do not disapprove of ? Cavell replied that if noses must be counted to decide how words are properly used, only competent native speakers (including ordinary-language philosophers) can count them reliably. Because good evidence about language use comes from the behaviour of native speakers, Cavell (pp. –) thought that if ordinary-language philosophers cannot rely on their own linguistic intuitions ‘there would be no noses to count’. This reflects the then prevailing attitude that empirical investigations cannot contribute much to the evaluation of philosophical ideas. I do not think many contemporary philosophers have read Mates, but a substantial number of them now side with him in advancing considerations other than the proper use of colloquial English in their philosophizing; one example is the reliance on findings from neuroscience and cognitive psychology in recent philosophical literature on mind and mental states. So-called experimental philosophers do use claims about what people say, but they try to support them by sophisticated empirical research. In view of this, I was impressed by how well Travis’ methodology agrees with Cavell’s side of the debate. For example, he argues that Grice’s ideas about conversational maxims cannot ‘serve as the core of a predictive theory of what is said or implicated’ (p. ), without so much as mentioning any anthropological or socio-linguistic research. In order to apply the resources of formal logic to utterances in natural language conversations one must identify well formed sentences in the natural language of interest Lnat, and associate them with well formed formulae in an ideal logical notation Lideal like Russell’s and Whitehead’s. LOAP philosophers, including Carnap and Davidson, assumed that they could disambiguate English sentences by means of the differences between the Lideal counterparts. They thought they could rely on inferential relations among Lideal sentences to systematize, evaluate and account for what speakers take to be inferential relations among the Lnat counterparts. Just as Russell believed his theory of definite descriptions could dissolve philosophical problems with regard to talking and thinking about non-existents, LOAP philosophers believed that one can resolve a number of other problems by studying the logic of Lideal counterparts to puzzling Lnat expressions. This programme cannot get off the ground unless grammatically well formed natural-language declarative sentences have the same truth-conditions and underlying logical structure as the well formed formulae in the Lideal used to study them. It must also be assumed that whenever a competent native speaker utters a grammatically well formed Lnat sentence, the conventional meanings of its component words and the identities of the things the speaker uses them to talk about are sufficient to © The Author The Philosophical Quarterly © The Editors of The Philosophical Quarterly
BOOK REVIEWS
determine the truth-conditions of what is said. Otherwise there would be no way of deciding which ideal-language sentences to use in analysing any given naturallanguage utterance. Travis objects quite rightly that the truth-conditions of natural-language sentences depend upon and vary with the circumstances in which they are uttered. In the context of some conversations, the sentence ‘this leaf is green’ could be uttered to say what is true of an autumn leaf painted green. But in other settings where what is at issue is the leaf’s natural colour, the same sentence could be uttered to say what is false about that same leaf (p. ). This is one of many examples of what Travis calls ‘occasion-sensitivity’: conventional meanings and the identities of items spoken about typically constrain without completely determining the truthconditions expressed by uttering a natural-language sentence on any particular occasion. Given conventional meanings and objects referred to, what fixes truthconditions are features of the occasion of utterance. This is why the truth-conditions of Lnat sentences vary over occasions of utterance. Isolated from these, naturallanguage declaratives typically have no definite truth-conditions. It follows that standard meanings, identities of referents and grammatical rules of vernacular English seldom if ever suffice for mapping Lideal formulae onto English sentences. Suppose an Lideal sentence p has the same truth-conditions as an Lnat sentence uttered on one or more particular occasions. Even if the two sentences range over and make reference to the very same things, they may have different truthconditions relative to other situations. If so, Davidson’s and other typical LOAP natural-language semantics programmes are unworkable. But perhaps the only reason why the sentence ‘this leaf is green’ can have different truth-conditions relative to different occasions of utterance is that it is ambiguous. A Davidsonian might say that although an ambiguous sentence can be used to say different things, the conventional meanings of its words determine that it must always be used to say one or other of them. So although conventional meanings and the identities of things referred to cannot single out a unique ideal-language counterpart for an ambiguous natural-language sentence, they can narrow candidate counterparts down to a small, easily managed selection. Travis does not think the phrase ‘is green’ is ambiguous between one sense for painted leaves and another for unpainted ones. But even if it is, he says, there are ‘indefinitely many’ different occasions on which either disambiguation could be uttered to express different truth-conditions. The verbal characterization of a disambiguation such as the sentence ‘the leaves are painted green’ is itself subject to different understandings relative to different occasions of utterance, and so on ad infinitum (pp. –). If so, a Davidsonian analysis of a single natural-language utterance would require the identification and application of infinitely many different rules. Mates would object. If Travis’ claim about indefinitely or infinitely many different understandings is supposed to be an a priori truth, it is not obvious how to establish it. If it is a posteriori, and the best evidence for it comes from Travis’ own linguistic intuitions, how reliable are they? If it purports to describe real-world linguistic or cognitive behaviour, Travis’ claim needs empirical support. Travis © The Author
The Philosophical Quarterly © The Editors of The Philosophical Quarterly
BOOK REVIEWS
rejects views which exhibit ‘poor method in linguistics, a bad fit of theory to data’ (p. ), and says that ‘simple observation of linguistic phenomena’ is a source of evidence for philosophical claims (p. ). It is a peculiarity of this book that it makes no use of empirical linguistic data. Even so, it is certainly plausible that at least some natural-language sentences can be used to say enough different things to raise difficulties for the application of formal logic to questions about what their logical implications are. For example, if ‘p’, ‘p ⊃ q’ and ‘q’ are respectively the Lideal counterparts of the English sentences () ‘the leaves are green’, () ‘if the leaves are green, then it is not autumn’, and () ‘it is not presently autumn’, then the validity of ‘(p & p ⊃ q) ⊃ q’ (modus ponens) might certify the inference of () from () and (). But the conventional meanings of ‘leaves’, ‘green’, etc., and the identity of the leaves () is used to talk about on any given occasion do not tell us whether () has the same truth-conditions as the antecedent of (). Nor do they tell us whether the horseshoe is the Lideal counterpart of any given utterance of ‘if ... then ...’. It all depends upon contextual facts which the identities of referents and the rules and conventions of Lideal and English leave open. Ch. takes up what Travis calls ‘the argument from illusion’. Suppose N is a putative knower, and p is a sentence which N uses on some particular occasion to make a contingent factual claim. What N says is true, and N is convinced somehow that it is true. The argument from illusion assumes that p is false in a possible situation (Travis calls it a ‘ringer’) which resembles N ’s actual situation so closely that N has no way of distinguishing the one from the other. Then for all N can tell, he might be mistaken, and so does not really know that p is true, no matter what reasons he has to believe it is true (p. ). According to this argument we cannot attain knowledge of contingent truths which measures up to a standard Travis ascribes to Cook-Wilson: N knows that p only if N ‘could not have the grounds he has for taking it that p while p was not so’ (p. ; cf. pp. ff.). It is not just that one cannot know what is not the case. There is to be ‘no ... possibility, not even a very remote or outlandish one’ that N might be mistaken, given whatever convinces N that p is true. Drawing on some Austinian ideas, Travis argues that the argument from illusion fails because sentences describing what someone might be mistaken about are every bit as occasion-sensitive as sentences like ‘this leaf is green’. If the sentence ‘for all N knows, p might be false’ is occasion-sensitive, it too has no definite truth-conditions or truth-value apart from occasions of utterance, and it can have different truthconditions relative to different occasions. Suppose N is convinced there are sheep nearby, having heard them bleating. If bleating goats sound too much like bleating sheep for N to tell the difference, then for all N knows, this might be a sheep-free goat-bleating ringer situation. But if there were no goats to hear, N ’s inability to distinguish goat- from sheep-bleats would not make it true to say that N might be wrong. Similarly for other ringers which the sceptic might invoke to show that N might be wrong (pp. , ). Thus the sceptic is required to show how under the circumstances a putative knower might be mistaken. In some cases the putative knower might be wrong, but the argument from illusion fails to show that putative knowers might always be wrong. © The Author The Philosophical Quarterly © The Editors of The Philosophical Quarterly
BOOK REVIEWS
I agree that one need not doubt a knowledge claim unless one has some idea of how the putative knower might be wrong. I agree that there are often ways to rule out specific error possibilities. But no matter how many challenges to knowledge claims we can rebut, the sceptic can argue that on at least some occasions there are ringer possibilities which cannot be eliminated merely because no one is aware of them. Alternatively, sceptics can hypothesize error sources like Descartes’ deceitful demon which cannot be ruled out even by people who know a lot about them. Some sceptics believe it might be possible for a neuroscientist to manipulate brains in vats to make them believe what is not the case. Some people (the neuro-scientist, for one) can tell whether N is a brain in a vat, but N cannot. It takes more than occasionsensitivity to refute such sceptics. If all that the argument from illusion can hope to show is that Cook-Wilson knowledge is unattainable, does it deserve the attention Travis devotes to it? Travis disapproves of ‘the philosopher’s airy lack of interest in practical affairs’ (p. ), but the question whether Cook-Wilson’s standard can be met is usually, if not always, irrelevant to real-world practical affairs. Motorists try to ask directions only from people who know the relevant navigational facts. It would be good to find an informant who could not possibly be mistaken, but wondering whether informants know the way typically amounts to no more than wondering whether they can deliver true descriptions of the route. Furthermore, typical motorists do not refrain from deciding that an informant knows the way because they do not know anything that makes it impossible for them to be mistaken about what their informant knows. A theory of knowledge concerning itself with practical affairs should apply to the epistemics of scientific as well as everyday non-scientific practices. But scientific journals are full of claims accepted on evidence which does not rule out the possibility of error. Sometimes the probability of error is high enough to warrant caution, but some scientific claims are based on a large and diverse body of evidence which makes mistakes highly improbable. For Travis, this makes no difference: no matter how persuasive it is, ‘mere evidence never yields knowledge’ (p. ). As I understand it, this is because no matter how low the probability may be, conditional on all the available evidence, all it takes to fail the Cook-Wilson test is a ‘slight possibility ... no matter how outlandish or far-fetched’ that the relevant claim is false. Epistemologists who want to understand, evaluate or make contributions to real-world epistemic practices need not worry about whether Cook-Wilson knowledge is possible. I think this makes it reasonable for epistemologists to turn away from the sceptical arguments which exercised Austin and his contemporaries, and pursue questions about the truth-conduciveness of specific epistemic practices. A number of them have done just that. Turning away from speculations about what people would say in imaginary situations, they use resources from the natural, social and behavioural sciences (including cognitive psychologists), and from statistics, probability theory, machine learning, etc., to investigate detailed, realistic questions about methods for producing and using evidence to develop, evaluate and apply empirical and theoretical claims and explanations in real-world situations. Readers who believe that philosophers should concern themselves with scientific and mundane practical affairs should welcome this trend. © The Author
The Philosophical Quarterly © The Editors of The Philosophical Quarterly
BOOK REVIEWS
However, readers of various philosophical persuasions should welcome Travis’ carefully crafted essays as illuminating illustrations of the strengths and weaknesses of twentieth-century analytical philosophy. University of Pittsburgh
© The Author The Philosophical Quarterly © The Editors of The Philosophical Quarterly
J B
Lihat lebih banyak...
Comentarios
