Tommaso Piazza A Priori Knowledge Toward a Phenomenological Explanation
PHENOMENOLOGY & MIND
Herausgegeben von / Edited by Arkadiusz Chrudzimski x Wolfgang Huemer Band 10 / Volume 10
A Priori Knowledge
Toward a Phenomenological Explanation
Frankfurt I Paris I Ebikon I Lancaster I New Brunswick
Ire.vrin. Malta. Portugal by Gazelle Books Services Limited White Cross Mills Hightown LANCASTER.co.uk
Livraison pour la France et la Belgique: Librairie Philosophique J. mechanical. +33 (0)1 43 54 03 47 . LA1 4XS sales@gazellebooks. F-75005 PARIS Tel. microfilming. place de la Sorbonne . Turkey. D-63133 Heusenstamm nr. recording or otherwise without written permission from the Publisher.com
United Kingdom. Fax +33 (0)1 43 54 48 18 www. NJ 08854-8042 trans@transactionpub. photocopying.O. detailed bibliographic data is available in the Internet at http://dnb.Vrin 6. Frankfurt www. Iceland.Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliographie.de
North and South America by Transaction Books Rutgers University Piscataway.ddb.com ISBN 10: 3-937202-92-7 ISBN 13: 978-3-937202-92-1 2007
No part of this book may be reproduced. with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system.fr
¤2007 ontos verlag P. Box 15 41. for exclusive use of the purchaser of the work Printed on acid-free paper ISO-Norm 970-6 This hardcover binding meets the International Library standard Printed in Germany by buch bücher dd ag
.ontosverlag. stored in retrieval systems or transmitted in any form or by any means. electronic.
Logical Truth.Table of Contents
1. and the A Priori Reductive and Non-Reductive Conceptions of Analyticity Implicit Definition. Analyticity. Realism about Logic Introduction Logical Principles. Justification and Epistemic Relativity Objective Truth Resnik’s Attack Wittgenstein on the necessity of ‘1 inch = 2.54 cm’ and logical inference Dummett’s Objection Rule Following considerations and the adoption of a convention Summarising Remarks Wright’s Attack Conclusion 57 60 64 65 75 79 84 87 89 109
. A Priori. Analyticity. and Implicit Definition Empiricism. and the Recalcitrant A Priori Problems with Implicit Definition BonJour’s Objection Fodor and Lepore’s Objection Horwich’s Objection Hale and Wright’s defence of the traditional connection Logic and Convention Coda 1 3 5 8 10 13 23 31 46 52
and Eidetic Seeing Is Holism a Possibility for the Empiricist? Intuition of Essences and the Analytic/Synthetic Distinction Husserl’s Conception of the Analytic/Synthetic Distinction Eidetic Variation Passive synthesis and Concept Constitution Knowledge of Reality and Conceptual Truth Absolute vs Relative Objectivity Are Conceptual Truths True? 138 145 150 156 157 164 168 174 177 179
. Phenomenology and Rational Insight Naturalism and Justification Phenomenology. the A priori and Rational Insight Boghossian’s Argument against Relativism Epistemological Realism about Justification Conclusion 111 115 117 119 122 123 123 126 131 132 134 135
TABLE OF CONTENTS
3. Justification. Objective Knowledge Introduction What the Tortoise Said to Boghossian What Boghossian would say to the Tortoise Rule-circular Arguments The Side-Argument Rejecting the Side-Argument First Horn: Simple Internalism and Rational Insight Second Horn: Epistemic Responsibility and the Lack of Epistemic Irresponsibility Realism.
The first one concerns the very nature of the statements allegedly known a priori. Addressing the general question about whether we do know anything a priori involves addressing two distinct sub-questions. analytic statements are truth-apt (indeed. So. that is to say concerns the question whether these statements do express the kind of things that can in principle be known. and to suggesting a plausible way out of the difficulties my analysis will hopefully bring to the fore. Accordingly. if these statements do express the kind of things that can in principle be known. therefore a priori.Introduction
Do we know anything a priori. unlike synthetic statements. analytic statements do not have cognitive content. is empirical and inductive in nature. Therefore. to inquiring into existing answers to these questions. they do not say anything about reality. is entirely due to the meanings of their constituting expressions. they can. like mathematical and logical knowledge. Contemporary empiricism has delivered a unified answer to both questions with the notion of analyticity. and indeed are (at least sometimes and locally). that is to say independently of experience? Do we know a priori anything objective concerning reality? Do we know logic? Do we know it a priori? This book is devoted to critically analysing these problems. the distinctive nature of analytic truth yields a satisfactory account of our knowledge of it. However. This feature of analytic statements is due to the fact that their truth. The second subquestion deals with the way. a priori knowledge is not any kind of knowledge of reality. then its truth can be known simply by understanding the statement. As synthetic statements. in fact it avoids the intuitive drawbacks of Mill’s solution according to which alleged pieces of a priori knowledge. For a natural suggestion is that if the truth of a statement is entirely determined by the meaning of its constituting expressions. on the empiricist reading of the notion of analyticity. Mill’s solution is consistent with the empiricist prin-
. true) statements. and the question concerning what kind of knowledge one acquires when knowing them. a priori statements are the kinds of things that can be known. This empiricist solution has exerted a considerable influence among empirically minded philosophers. In accordance with this traditional answer. However. the truth of an analytic statement can be known merely by understanding it. all a priori statements are analytic statements. known a priori.
Kant held that an analytic statement is one characterized by the fact that its predicate term expresses a concept which is contained by the concept expressed by its subject term.vi
ciple according to which every piece of knowledge of reality stems from and is justified on the basis of experience. It is too narrow. However. because its predicate concept is not contained in its subject concept. for this reason he also claimed that the a priori epistemological status of analytic statement is not problematic: given the containment theory. and a fortiori not analytic. while synthetic statements are. it is too wide. The viability of this empiricist solution clearly depends on the very notion of analyticity called into question. by inductive and empirical means. Kant notoriously also held that analytic statements are not ampliative with respect to our knowledge of reality. in fact. However. Notoriously. to admit that our mathematical or logical knowledge is a priori is not to admit that our knowledge of reality has sources other than our experience of it. the empiricist notion of analyticity is not Kant’s notion. to a degree sufficient for knowledge. this solution has the unpalatable feature of imposing a conception of mathematic and logic as just contingently true. yet the concept [professional philosopher] expressed by the predicate term is actually contained by the concept [daughter of a professional philosopher] expressed by the subject term. If mathematic and logic say nothing about reality. to know an analytic statement it is sufficient to possess
. However. for it applies just to statements of the subject-predicate form: statements like (K1) Everything is spatio-temporal or is not spatio-temporal is not analytic according to Kant’s definition. The solution based on analyticity allows the empiricist to defend her epistemological principle without loosing the necessity of mathematical or logical statements. Secondly. admitting that Kant’s notion constitutes an improvement on such notion is probably not that far from the truth. Kant’s analyticity has been found wanting for two reasons. for a statement like (K2) Every daughter of a professional philosopher is a professional philosopher is false. for just beliefs in contingent statements can be justified.
it reduces to knowledge of logical truths. and if a competent speaker of a natural language is indeed in a position to answer such a question whenever it arises concerning two distinct expressions. the epistemology of analytic statement is not anymore Kant’s. Unless we are told how logic is known. and to be acquainted with the principle of non-contradiction. So long as they do.INTRODUCTION
both its predicate. we cannot stay content with the contention that a priori knowledge of analytic statement is unproblematic because. that logical principles are simply self-evident – it is clearly vital for a sound empiricist theory of the a priori to provide such an account. understanding what they mean coincides with (because it requires) appreciating that they are
. at bottom. because according to this notion analytic statements do not say anything about reality. The basic idea is that the meaning of certain expressions – in the case at issue the meaning of the logical constants – is determined by constraining those expressions to have whatever meaning is required for the truth or the correctness of certain basic contexts that contain them. it is Frege’s. Rather.and the subject-concept. A simple suggestion is that logical principles are known either because they are implicit definitions of the logical constants they contain. or because they are deducible from such principles. we might say. Given the wider nature of the empiricist conception with respect to its applicability just to statements of the subject-predicate form. Though Frege considered the latter epistemological task as completely unproblematic – convinced. Frege held that a statement is analytic when it is either a substitutional instance of a logical principle – much in the way “it rains o it rains” is a substitutional instance of the principle “p o p” – or can be reduced to a substitutional instance of a logical principle with the aid of definitions – much in the way “every bachelor is an unmarried man” can be reduced to “every unmarried man is unmarried”. If. pace Quine. however. if they perform the role of implicit definitions. Both features are preserved within the empiricist notion of analyticity. as he was. the question about how analytic statements can and are known reduces to the question about how predicate and propositional logic is known. the question whether two expressions of a natural language are synonymous is not intractably unintelligible. This is the contemporary proposal I shall consider at the beginning of the first chapter. It follows that no one understanding such context (sentences or rules of inference) can fail to appreciate that they are true (valid).
nor it is reducible to the instantiation of a logical principle if synonymous are substituted by synonymous. More than this. the theory seems to make good sense of the traditional suggestion according to which logical principles are selfevident. The notion of implicit definition seemingly makes it available to the empiricist a plausible way out. J. the admission of a priori knowledge of synthetic statements entails that experience is not the only source of knowledge and justification. The problem is that synthetic a priori knowledge is not consistent with the empiricist epistemological principle. So. it follows that no one fully understanding what they say can fail to appreciate that what they say is true. in fact. Since accepting such principles is constitutive of the capability of understanding what they say. Accordingly. An empiricist should welcome this account for a third important reason. at the same time and under the same respect. is not the same as the meaning of the (conjunctive) predicate “not green and not blue and not gray …”. and against the idea that the meaning of certain expressions is implicitly defined by the resolve to accept as true given contexts featuring these very expressions. Fodor and H.viii
true (valid). advanced by L. the meaning of any colour term “c” could not be grasped in the first place if it were equivalent to the infinite conjunction of the negative predicates constructed out of each colour expression other than “c”. As it makes available a notion of logical truth according to which logical principles are just definitions of a devised sort of the logical constants they contain. In contrast with analytic statements. Given the potential infinity of colour discriminating expressions. statements like (H1) are to be counted as synthetic under the standard (Fregean) empiricist notion of analyticity. (H) is not the instantiation of a logical principle. Take the following statement: (H) If something is entirely coloured of red it is not. Lepore. Boghossian’s idea that meaning coincides with conceptual role. The meaning of “red”. against P. synthetic statements are about reality. it seemingly makes available the view that statements like (H1) are just definitions of a devised sort of the color predicates they contain. The first chapter of this book will be devoted to taking into consideration several criticisms. therefore acceptable from an empiricist epistemological point of view. entirely coloured of green. In accordance with a
. BonJour. and by P. Horwich.
every transition accepted within a language constitutes a conceptual role.
. If the identification is unqualified. It follows that conceptual role semantics itself must be rejected. both criticisms will be considered as striving against one and the same epistemological suggestion concerning logical knowledge and knowledge of statements like (H)1. However. this objection
My suggestion as to the equivalence of the implicit definitional approach and the conceptual role semantics is apparently taken for granted in the literature. this means that the meaning of “brown cow” is not a function of the meaning of “brown” and of the meaning of “cow”. If meaning is identified with conceptual role. does the transition from “x is a brown cow” to “x is dangerous”. So. The natural alternative is to qualify the identification of meaning and conceptual role by narrowing the scope of meaning-constituting inferences to analytic inferences. 133). However. An example is seen in Horwich 1998: “For simplicity I am focusing on the case in which implicit definition proceeds by regarding a sentence […] as true.INTRODUCTION
presupposition shared within the contemporary debate. constituting a conceptual role for the expression “brown cow”. nor the transition from “x is brown” to “x is dangerous” are legitimate. Fodor and Lepore have objected to Boghossian’s proposal that the conceptual role semantic is the hostage of a dilemma between two unpalatable options. So long as it is conceded that we should allow for the advertised correspondence between rules of inference and axiom schemata. the analytic/synthetic distinction has been convincingly rejected by Quine. However this conceptual role is not a function of the conceptual role of “brown” and “cow”. because neither the transition from “x is a cow” to “x is dangerous”. according to Fodor and Lepore. for instance. such arguments do not have any immediate bearing on the suggestion that the meaning of the logical constants is constituted by the conceptual roles specified by their introduction and elimination rules. Accordingly. this second alternative also proves unviable. Against Fodor and Lepore’s suggestion I shall point out that Quine’s arguments are correctly understood as being directed against the notion of synonymy. Quine’s argument then has a direct bearing just on the suggestion that those analytic statements which are reducible to logical principles by means of definitions can be know a priori. So. But this discussion carries over in an obvious way to the case of implicit definition in which certain rules of inference are regarded as valid” (p. The unqualified identification of meaning and conceptual role entails the violation of the compositionality of meaning.
Horwich’s objection is directed against the notion of implicit definition. This is why the existence problem matters: the defined expression. I will end the first chapter by critically assessing Horwich’s argument. that our resolve to accept as true the implicit definer determines that the expression comes to possess that meaning. However. that is to say not to entail fresh existential commitments whose fulfilment could be ascertained just by additional epistemic work. According to Horwich. besides our resolve. may suffer from reference failure. the main problem of the traditional connection – the thesis that implicit definition explains a priori knowledge – is that it is undermined by the Wittgensteinean conception of meaning needed to make good sense of the model of implicit definition. Since a meaning is constituted by the regularity of use centred on the resolve to accept as true the implicit definition. that it is unique. Hale and Wright grant the point when they require an implicit definition not to be arrogant.x
can be resisted also in its qualified variety. they also suggest that the wittgensteinean conception doesn’t impede establishing the traditional connection. A major problem concerns the relation between meaning and reference fixing. and Hale and Wright’s replies to it. it is arguably by constraining the identity of its reference to be such that it makes the implicit definition true. Hale and Wright have recently rehearsed the point that a wittgensteinean conception of meaning is indeed needed for making sense of the model of implicit definition. Horwich contends that the model actually explains how certain expressions receive a meaning only provided that we know that the required meaning exists. and may
. and that we are able to explain why it is so. it constitutes a possibility that. To do so I will rehearse Boghossian’s characterization of Quine’s attack against synonymy – both in its error theoretic and non factualist varieties– and will show how it can be resisted in both its varieties. if reality doesn’t cooperate. If an implicit definition performs the role of determining a meaning for an uninterpreted expression. Borrowing from the early debate issuing from Prior’s provocative proposal of “tonk” as a legitimate logical constant. yet at the cost of undercutting the connection with (a priori) knowledge. I will not dwell upon the question whether the wittgensteinean conception of meaning is indeed needed to make sense of implicit definition. The identification of meaning with use helps coping with the four problems. Hale and Wright assert that an implicit definition (or a set of rules) is not arrogant. the content thereby constituted turns out to be false.
The most natural proposal. Tennant 2005. and this amounts to a non-conservative extension of the language2. 628-629. It describes. the speaker must not be able to get away with less than is required when one is giving the listener entitlement to make certain inferences from what has been asserted”. however. the relevant question is about whether an implicit definition of that sort can be safeguarded from reference failure by some other condition. Accordingly.
. these mutual obligations and entitlements have to be in balance. only if it is conservative and in harmony: that is to say. The elimination rule of an expression is not in harmony with the introduction rule if it licenses from a context containing the expression the inference of more (or less) than is required to introduce such context. however. then F# is true.
“The introduction rule states the conditions under which a conclusion with that operator dominant can be inferred. harmony cannot constraint implicit definitions of the first kind. Clearly. the corresponding Carnap-conditional is If xFx. if the addition to a language of the new expression (together with the linguistic apparatus which defines it) does not allow one to derive something which is underivable within the same language without the new expression. The corresponding elimination rule states what the listener is entitled to infer from the speaker’s assertion. As it is clear. p. If #F is an implicit definition of “#”. if you like.INTRODUCTION
be accepted as true (respectively. valid) a priori. Likewise. The listener must not be allowed to infer more than is required to have been “put in” to the justification of the assertion. is that such implicit definitions be transformed in corresponding so-called Carnap-conditionals. whose meaning constituting role is safeguarded from reference failure by their conditional formulation. and to implicit definitions provided by pairs of rules. In this case then it is possible to prove statements not containing that expression which it was not possible to prove before. The first thing to notice is that Hale and Wright’s proposal is meant to apply both to implicit definitions provided by simple sentences. the obligations that have to be met by the speaker in order to be justified in asserting the conclusion in question.
and logical truth is a by-product of such rules. The first one is to endorse conventionalism. Since no new knowledge has been introduced. If the conventional rules of inference just play the function of facilitating the exchange of information. However. that their consequents are known a priori.xii
As Horwich remarks. It seems to follow that no new knowledge is made available once it is introduced by means of the new expression through an implicit definition of it. in the light of the function they are to perform. In particular. This result does seem to stand in clear opposition to the idea that implicit definition plays a primary role in the epistemology of the a priori. as intended by the theory. First of all. or predicates. There remain to be analyzed implicit definitions framed in terms of introduction and elimination rules. for the adoption of certain conventions. advocating the epistemological role of implicit definitions of the conditional form does nothing to show. it is easy to see how non-conservative extension and harmony suffice to vindicate the view that implicit definition conveys a priori knowledge. There remains the case of the logical constants. the Carnap-conditional is uninteresting from an epistemological point of view. only against the backdrop of our accepting the validity/truth of other rules/conditionals whose status cannot be in turn explained by implicit definition. but rather as natural desiderata. If a pair of harmonious introduction and elimination rules implicitly defines a logical constant. then no fresh existential commitment has been introduced by their adoption. In particular. it might be suggested that all implicit definitions are to take this form if they are both not to be arrogant and to convey substantial a priori knowledge. This move would in fact prevent the foregoing problem from arising. Two distinct reactions to this circularity are available. framed in terms of introduction and rules. If the rules are in harmony. it means that whatever existential commitments are ratified within a language after the introduction of the new expression must have been ratified before. harmony and conservative extension guarantee the validity/truth of these rules/conditionals. are we thereby justified in believing that these rules are valid (that the corresponding conditionals are true)? Unfortunately. the same knowledge must have been available before. take in consideration implicit definitions of individual terms.
. and justify the adoption of the conservative extension and harmony constraint not in the light of their conductivity to validity/truth.
according to which realism and objective truth should be rejected in the case of logic. is that it is a priori knowledge just of conventional truths. I shall leave open the issue about whether such a justification is enough to ensure objective a priori knowledge of the validity of such rules. Boghossian. Objective truth can be characterized as a property that sentences (or propositions) possess independently of our holding them to be true. most importantly the objectivity of normative epistemological principles. negative answer to this question. evidence transcendence should be seen just as a sufficient. and whether. it can be known along the circular pattern described by Dummett. I will firstly pause to stress the connection between the idea that logical truth is objective and several pressing questions in general epistemology. that is to say necessarily to employ the very rules of inference whose validity is under dispute. so conceived. and recently developed by Boghossian. The argument concludes with the rejection of objec-
. but not as a necessary condition for objectivity. BonJour. to the effect that an anti-objectivist construal of logic entails strongly unpalatable forms of epistemic relativism and scepticism. According to a recent proposal of Wright. on the one side. M. The idea that logical truth is objective has been attacked in both senses of the notion. I will take up the issue in the third chapter. when discussing a recent proposal of Boghossian. Since Dummett’s paper on the Justification of Deduction. such justification is alleged to be necessarily circular. The second chapter will be devoted at assessing the questions whether logical truth can coherently be conceived as objective truth. however. Before taking into consideration an influential. and argue that it does not stand in the way of justifying our rules of inference. The second alternative is to welcome the circularity at issue as unavoidable. objective truth has been identified with a non-epistemic property. According to Wright even an evidentially constrained truth-predicate may be shown to deserve a realist – objectivist – interpretation if (a) it satisfies Cognitive Command. however. Resnik. and much in the same vain by P.INTRODUCTION
The shortcoming. Along with Dummett’s characterisation. (b) it allows for a Socratic resolution of the Euthyphro Contrast and (c) the facts allegedly reported by true sentences within the area perform a Wide Cosmological Role. To show this I will rehearse a very general argument put forward by L. has put forward an argument (echoing Benacerraf’s argument against mathematical Platonism).