In his three Critiques, Immanuel Kant provides a system of philosophy that encompasses ethics, aesthetics, metaphysics,...
34 downloads
1004 Views
3MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
In his three Critiques, Immanuel Kant provides a system of philosophy that encompasses ethics, aesthetics, metaphysics, and epistemology. As Kant's is a seemingly complete system, one may reasonably infer that it contains an account of the nature of truth. However, Kant's elliptical remarks on the subject make it difficult to specify the precise nature of his account. This book considers explanations by a number of authors concerning Kant's account of truth, and proposes an alternative to these views.
Lori J. Underwood is Assistant Professor of philosophy at Christopher Newport University. She received her Ph.D. in philosophy from the University of Missouri.
Kant's Correspondence Theory of Truth
amencan university studies
Series V Philosophy Vol. 195
PETER LANG New York • Washington, D.C./Baltimore • Bern Frankfurt am Main • Berlin • Brussels • Vienna • Oxford
Lori J. Underwood
Kant's Correspondence Theory of Truth An Analysis and Critique of Anglo-American Alternatives
PETER LANG New York • Washington, D.C./Baltimore • Bern Frankfurt am Main • Berlin • Brussels • Vienna • Oxford
Library of Congress Cataloging-in-Publication Data Underwood, Lori J . Kant's correspondence theory of truth: a n analysis and critique of Anglo-American alternatives/Lori J . Underwood. p. cm. — (American university studies. 5, Philosophy; v. 195) Includes bibliographical references (p. ). 1. Kant, Immanuel, 1724-1804—Contributions in concept of truth. 2. Truth. I. Title. II. American university studies. Series V, Philosophy; v. 195. B2799.T8U53 121'.092—dc21 2002011018 ISBN 0-8204-6240-3 ISSN 0739-6392
338526 Die Deutsche Bfbllothek-CEP-Einheitsaufnanme Underwood, Lori J.: Kant's correspondence theory of truth : a n analysis and critique of Anglo-American alternatives / Lori J . Underwood. - New York; Washington, DC/Baltimore; Bern; Frankfurt am Main; Berlin; Brussels; Vienna; Oxford: Lang. (American university studies: Ser. 5., Philosophy; Vol. 195) ISBN 0-8204-6240-3 B2799.T8 U53 2003
METO LIBRARY
The paper in this book meets the guidelines for permanence and durability of the Committee on Production Guidelines for Book Longevity of the Council of Library Resources.
m> © 2003 Peter Lang Publishing, Inc., New York 275 Seventh Avenue, 28th Floor, New York, NY 10001 " www.peterlangusa.com All rights reserved. Reprint or reproduction, even partially, in all forms s u c h a s microfilm, xerography, microfiche, microcard, and offset strictly prohibited. Printed in the United States of America
for my parents, Linda and C. L. Murray and my husband David
SC TABLE OF CONTENTS List of Figures
ix
Acknowledgments
xi
Chapter One: The Problem of Truth in Transcendental Idealism
1
Chapter Two: An Incomplete Account of Truth
17
Chapter Three: The Verificationist View
31
Chapter Four: The Constructivist View
57
Chapter Five: The Coherence View
73
Chapter Six: The Hybrid View
115
Chapter Seven: An Idealist Correspondence Theory
126
Bibliography
149
SC FIGURES Figure One
119
Figure Two
119
Figure Three
120
Figure Four
131
§g ACKNOWLEDGMENTS Henry Allison, Kant's Transcendental Idealism: An Interpretation and Defense. New Haven and London: Yale University Press, © 1983. Reprinted by permission of the publisher. All rights reserved. Henry Allison, Idealism and Freedom: Essays on Kant's Theoretical and Practical Philosophy. New York: Cambridge University Press, © 1996. Reprinted by permission of the publisher. All rights reserved. Austin, J.L., "Truth" in Proceedings of the Aristotelian Society supplementary issue, 1950. Reprinted by courtesy of the Editor of the Aristotelian Society: © 1950. All rights reserved. Lewis White Beck, Proceedings of the third International Kant Congress. Dordrecht: Kluwer Academic Publishers, © 1972. Reprinted by permission of the publisher. All rights reserved. Jaakko Hintikka, Knowledge and the Known: Historical Perspectives in Epistemology. Dordrecht: Kluwer Academic Publishers, © 1974. Reprinted by permission of the publisher. All rights reserved. Immanuel Kant, The Critique of Pure Reason. New York: Copyright © 10/69 by Bedford/St Martin's From: Kant's Critique of Pure Reason by Smith. Reprinted with permission of Bedford/St. Martin's. All rights reserved. Carl J. Posy, Kant on Causality, Freedom and Objectivity. Minneapolis: University of Minnesota Press, © 1984. Reprinted by permission of the publisher. All rights reserved. Hilary Putnam, Philosophical Papers, Volume 3: Realism and Reason. Cambridge: Cambridge University Press, © 1983. Reprinted by permission of the publisher. All rights reserved. Hilary Putnam, Reason, Truth, and History. Cambridge: Cambridge University Press, © 1981. Reprinted by permission of the publisher. All rights reserved.
xii Kant's Correspondence Theory of Truth Robinson, Hoke, "Two Perspectives on Kant's Appearances and Things in Themselves," in Journal of the History of Philosophy volume 32:3 John Hopkins University Press, Baltimore: © July 1994. Reprinted by permission of the publisher. All rights reserved. Russell, Bertrand, Human Knowledge: Its Scope and Limits. Routledge, New York: © 1948. Reprinted by permission of the publisher. All rights reserved. Frederick Schmitt, Truth: A Primer. San Francisco: Perseus Books/ Westview Press, © 1995. Reprinted by permission of the publisher. All rights reserved. Norman Kemp Smith, A Commentary to Kant's Critique of Pure Reason. Boston: Brill Academic Publishers, © Koninklyke Brill nv, Leiden, The Netherlands. 1923. /Palgrave Macmillan New York. Reprinted by permission of the publishers. All rights reserved. Strawson, P.F., The Bounds of Sense: An Essay on Kant's Critique of Pure Reason. Routledge, London and New York: © 1966. Reprinted by permission of the publisher. All rights reserved. Ralph Walker, The Coherence Theory of Truth: Realism, AntiRealism, Idealism. Pages 2, 61-69, 75,77. London and New York: Routledge, © 1989. Reprinted by permission of the publisher. All rights reserved.
Chapter One The Problem of Truth in Transcendental Idealism Introduction This domain is an island, enclosed by nature itself within unalterable limits. It is the land of truth—enchanting name!—surrounded by a wide and stormy ocean, the native home of illusion, where many a fog bank and many a swiftly melting iceberg give the deceptive appearance of farther shores, deluding the adventurous seafarer ever anew with empty hopes, and engaging him in enterprises which he can never abandon and yet is unable to carry to completion. (A235/B295)
Given this passage, we can conclude that a central project of the Critique of Pure Reason is to establish the limits within which truth can be obtained and distinguished from illusion by finite rational subjects. The nature of Kant's conception of truth, which it seems is at the heart of this project, will be my subject for this work. My aim is two-fold; first I wish to consider alternative interpretations of the nature of truth in Kant's theoretical philosophy by analyzing current Anglo-American literature on the subject and second, I wish to explain what the requirements are for a plausible theory of truth given an interpretation of transcendental idealism that is consistent with the apparent spirit of Kant's espoused principles. As for the former goal, Kant's references to truth in the First Critique are elliptical at best. Consequently, opinions as to the latter vary greatly among scholars. In constructing my view, I will draw largely on Kant's remarks in the Critique of Pure Reason, and minimally on some of his other primary texts. In the course of this inquiry, I will focus on a direct interpretation of Kant's philosophy and will argue for three theses: (1) in my interpretation of transcendental idealism, three kinds of utterances will be capable of being truth bearers. All truth bearers for Kant are judgments; the three kinds being analytic, synthetic a posteriori, and synthetic a priori judgments. I will give an account of transcendental idealism that considers Kant's central theses in light of some 20th century developments in metaphysics and epistemology. I acknowledge that Kant may not have
2
Kant's Correspondence Theory of Truth
subscribed to these principles in their current form, but I think this strategy is consistent with an interpretation of transcendental idealism as a viable alternative to current epistemological theories. In other words, I will provide an account of transcendental idealism as I think Kant would construe it today. (2) I will be arguing that there is an account of truth implicit in Kant's transcendental idealism. Specifically, I will argue that there is a definition of truth implicit in Kant's account of how the various types of judgment combine to make knowledge possible. Further, I will argue that the nature of Kant's conception of truth has yet to be fully captured by any of the considered interpretations drawn from Anglo-American literature on Kant's theoretical philosophy. (3) Finally, I will argue that the nature of the truth maker and the relationship between judgment and truth maker are different for each of the previously mentioned kinds of truth bearing judgments. The combination of these three conceptions of truth will constitute a tri-level theory of truth, which is faithful to the text of the Critique of Pure Reason. I believe this interpretation also augments a well-formed and reasonable construal of Kant's transcendental idealism. The varied and sometimes vague manner in which Kant addresses the issue of truth has led to a debate over what the nature of truth might be in his transcendental philosophy. Interpretation of his intent on this issue range from limited correspondence, to coherence, to constructivism. By considering and evaluating some of the various prongs of this debate among Anglo-American authors, I hope to develop a clearer understanding of some central current views on the matter. I also hope to clarify the nature and source of some of the major obstacles to isolating and identifying the concept of truth in Kant's transcendental philosophy. A review of recent Anglo-American literature on Kant's theory of knowledge reveals a number of different interpretations of his conception of truth. I will begin by examining some of these interpretations according to their compatibility with the requirements of the text and with a well-formed, interpretation that coheres with what I take to be the intent of transcendental idealism. In so doing, I will be able to narrow my field of inquiry so as to focus in on an accurate circumscription of the nature of truth in the Kantian system.
The Problem of Truth in Transcendental Idealism
3
Terms and Definitions Before beginning an examination of the literature, it is important to introduce and define some of the fundamental terms involved. I draw these definitions from a number of sources. While I acknowledge that the specific definitions of the terms to be discussed may vary in the literature and that there is no consensus on the meaning of many of these terms, what I intend to provide here are, simply, some fundamental definitions, which will serve as a basis for further development. Empirical Realism Kant is an empirical realist. This means that, for Kant, the objects of experience constitute the actual. The world as we experience it is public and objective, and the truth or falsity of our judgments of experience is determined by their relationship to the objects of experience. Kant writes, "empirical realism is beyond question." (A375) Insofar as the empirical employment of the understanding is concerned, the set of the objects of possible experience is co-extensive with what is real. To this extent, Kantian metaphysics and epistemology are quite similar to scientific realist accounts. However, the similarity dissolves with Kant's postulation of the transcendental. Transcendental "I entitle transcendental all knowledge which is occupied not so much with objects as with the mode of our knowledge of objects in so far as this mode of knowledge is to be possible a priori. A system of such concepts might be entitled transcendental philosophy." (A 12 B25) Transcendental knowledge is a function of transcendental concepts, pure intuitions, and their synthesis in transcendental judgments correlated with the transcendental object. Whereas purely logical judgments are devoid of all content, transcendental judgments are devoid only of all empirical content. They are the formal conditions for the possibility of experience. They are termed transcendental because as necessary conditions for the empirical, they cannot themselves contain anything empirical. Through transcendental judgments, we determine possible objects according to the pure concepts and pure intuitions of the understanding. The formers insure that our ex-
4
Kant 's Correspondence Theory of Truth
perience will operate according to certain principles, and the latter individuate appearances according to space and time. A transcendental judgment is the application of these pure concepts and pure intuitions to the form of an object in general. Transcendental judgments function in the original unity of apperception. It is in the original unity of apperception that the synthetic unity of appearance is determined. In other words, through the original unity of apperception, a coherent, communicable world of appearances is developed by finite rational minds. The object of the original unity of apperception is the transcendental object. Although there is disagreement, I will define the transcendental object as the form of an object in general. Through the original unity of apperception, we create the parameters of experience. The objective unity of consciousness that is created in this way is an objective condition of all knowledge. It "is a condition under which every intuition must stand in order to become an object for me" (B138) Kant describes apperception as the T in my experiences. In order for a finite mind to have an experience, he must orient himself within that experience. In other words, before a potential object of experience can be experienced by a given subject, that subject must initiate a process by which that potential object can be conditioned so as to be an object for him. This process is the original unity of apperception, and it must occur transcendentally because it is causally prior to the empirical. The term 'transcendental' then, refers to the principles, intuitions, judgments and object that are both dependent on finite subjects and necessary as conditions for the possibility of any experience by those subjects. Kant entitles his view 'transcendental idealism' precisely because he argues that the formal conditions of experience, which are supplied by the finite subject, are real in so far as they apply to the world of experience, but ideal in that they are not qualities of things as they are in themselves. Essentially, transcendental reflection leads us to realize that the world, as it appears to us, is empirically real but transcendentally ideal. Transcendent Kant distinguishes the transcendental from the transcendent. Whereas transcendental principle set the limits of the legitimate employment of the understanding, transcendent principles go beyond the
The Problem of Truth in Transcendental Idealism
5
limits set by the transcendental principles. Transcendent principles purport to apply beyond the limits of possible experience. In other words, a transcendent judgment is a judgment that purports to have as its object a thing as it is in itself, a noumenal object. (A296/B353) Examples of transcendent principles include "There is life after death", and "Lying violates the Moral Law." As these examples show, transcendent judgments (principles) have truth conditions that may only be verified beyond the limits of possible experience. Hence, transcendent judgments cannot be knowledge claims for us because they extend beyond the legitimate employment of the understanding. In addition to transcendent judgments, there is a transcendent object (or objects). The transcendent object is the thing in itself. It is the transcendent object that is given in sensibility, and I argue that it is in this capacity that the transcendent object serves as the material condition for the possibility of experience. Judgments: Kantian Truth Bearers Anything that can reasonably be said to possess the property of being either true or false is a truth bearer. There is disagreement over precisely what sorts of entities are candidates for being truth bearers. Definitions range from very narrow conceptions (only sentence tokens are capable of being true or false) to the very broad conception espoused by Kirkham (even teddy bears could be bearers of truth as long as the supply was plentiful enough to have a teddy bear for each way in which an individual might want to express himself (Kirkham 1995, 61-62). For Kant, all truth bearers are judgments, although not all judgments are truth bearers. Judgments that entail knowledge of an object are truth bearers. Objects of all other judgments are merely objects of thought, and need only be consistent. That is not to say that there are no truth bearing judgments about ideas, mental states etc. Kant allows for such abstract objects as objects of truth bearing judgments. Merely regulative or problematic judgments, however, cannot be substantive truth bearers. Their truth can be assumed for the purpose of argument or motivation, but as they have no satisfiable truth conditions, they cannot be bearers of truth. Kantian judgments should be understood as sentence types rather than sentence tokens. A sentence type is an abstract entity that expresses
6
Kant's Correspondence Theory of Truth
some relationship. A sentence token is an expression of a sentence type. Ink markings on paper, certain combinations of sounds and certain movements of the hands are all examples of sentence tokens. It must be the case that Kantian judgments are sentence types and not sentence tokens because he refers to concepts as predicates of possible judgments. A possible judgment is one that as yet has no sentence token. Therefore, if there are established predicates for possible judgments, judgments must be abstract types rather than concrete tokens. The concept of body means something, for instance, metal, which can be known by means ofthat concept. It is therefore a concept solely in virtue of its comprehending other representations, by means of which it can relate to objects. It is therefore the predicate of a possible judgment. (A69/B94)
Kant holds that there are three modes of judgment. These are problematic, apodeictic, and assertoric. (A74/B100) Problematic judgments express logical (non-objective) possibility. They are judgments that may be assumed for the moment, but their acceptance is in no way required by the understanding. Such judgments may be assumed for the purpose of argument, and in being so assumed, may be useful in indicating the path to truth. (A75/B101) Another mode of judgment is apodeictic judgment. Apodeictic judgments include tautologies and contradictions; either their truth or falsity is necessary. They are, in a purely logical sense, bearers of truth. (A75/B101) The final mode of judgment is assertoric judgment. Assertoric judgments are purported statements of fact (propositions). Such judgments deal with reality and as such are bearers of truth or falsity (depending on their accuracy relative to what actually obtains). True assertoric judgments have both objective validity and objective reality, in that they reflect both possibility and actuality. According to Kant, everything that is incorporated into the understanding is incorporated through these modes of judgment. "We first judge something problematically, then maintain its truth assertorically, and finally affirm it as inseparably united with the understanding, that is as necessary and apodeictic." (A76/ B101) For Kant, finite rational minds must think 'about' their objects. In other words, to think (at least
The Problem of Truth in Transcendental Idealism
1
purposeful thoughts) one must be engaged in an act of judging. Thus when referring to Kantian truth bearers (or potential truth bearers) I will use the term 'judgment'. Objective Validity To be a truth bearer, according to Kant, is to have objective validity. Objective validity is real possibility. It is the possibility of being true or false. Judgments have objective validity in virtue of their relation to what could obtain in reality. Kant distinguishes between two kinds of possibility, logical possibility and real possibility. The former only requires that there is no contradiction, i.e. that the law of non-contradiction is not violated. The latter, real possibility, requires more than accordance with the law of non-contradiction. A judgment is objectively valid just in case it has, or could have, as its object an object of possible experience. (Bxxvii/) The objective validity of a judgment entails the possibility of the empirical reality of its object. (A35/B52) Objective Reality Objective reality is the relation of a judgment to a given object. It is given to empirical judgments by the function of transcendental judgment. For this reason, Kant argues that transcendental truth is the foundation of all empirical truth. Without the relation of a judgment to its given object, empirical judgments are devoid of content and cannot be modes of substantive knowledge. Whereas objective validity refers to the real possibility of a judgment (given the synthetic unity of experience), objective reality refers to the correlation between a judgment of experience and a real object of experience. A judgment has objective reality in virtue of its relation to an appearance, which is determined by the formal and material conditions of the possibility of experience. (A156-157/B195-196) Knowledge Kant's goal in the Critique is to provide an accurate account of the nature and limits of knowledge. All knowledge involves judgment and all judgment involves subsuming something under a concept. To be knowledge, a judgment must have both objective validity and objective reality. That is to say, it must be both possible and actual. In other words,
8
Kant's Correspondence Theory of Truth
it must be about something (an object) and the relation depicted in the judgment must be true of the object it describes. That this is the nature of knowledge is affirmed by Kant's explanation of truth as "the agreement of knowledge with its object" (A58, A191). This definition holds for empirical, analytic, and transcendental judgments. To understand what Kant takes knowledge to be, one must first understand his distinction between levels of judgment. Kant assesses the structure of knowledge in terms of concepts joined to intuitions in a judgment. There are three levels of judgment holding according to Kant, and only one of them is a dependable path to truth. These three levels are opinion, belief, and knowledge, and the key to understanding the truth conditions for empirical judgments is to understand the notions of subjective and objective validity. The Critique seems clear in its indication that Kant thinks knowledge is justified true belief. The first level of judgment holding is opining. Suppose I have an opinion about the temperature on Mercury. I think that Mercury is further from the sun than earth, so it must be cold there. I do not have a strong conviction concerning this belief. That is, I think it is probably the case, but since I lack any evidence about relevant conditions, I would be easily dissuaded from my opinion. Thus the judgment "Mercury is colder than Earth" has a low degree of subjective validity (in that I do not hold it to be something that is true, I just hold it as something that is probably true). It has no degree of objective validity, since Mercury is in fact warmer than earth. Some opinions, of course, will turn out to be true, but since they are not judgments that are or need be supported strongly, they are not proper paths for truth seekers. The remaining two levels of judgments have, as a minimum condition, a high degree of subjective validity, i.e., the holding of a thing to be true. Yet, subjective validity is a necessary, but not sufficient condition for having a true empirical judgment. To know something (to be true) one must first hold the relevant judgment as a belief, that is, to hold that it is true for me, i.e. subjectively valid. This is the second level of judging—belief. To have subjective validity to the degree sufficient for belief, justification is required. That is the essential difference between opinion and
The Problem of Truth in Transcendental Idealism
9
belief. I can hold any opinion and can change my opinions without cause. However, to claim something as my belief, I must have some reason to believe it. Kant addresses the issue in terms of degrees of confidence. Confidence in a judgment, it seems, only makes sense in reference to justification. Therefore, given this talk of confidence, it seems reasonable to conclude that Kant holds that justification is a necessary condition for knowledge. If it makes sense to speak of Kant as endorsing a view of justification for empirical judgments, it makes sense only if one considers him to be an externalist for justification. The third level of judging is knowing. When a judgment is known, it must be believed (subjective validity), and that belief must match up with a fact/state of affairs in the world (i.e. be true). That is to say, the belief is objectively valid, it is true for everyone. In other words, what is believed must actually be the case. For Kant, although the world is in some sense constructed by the finite subject, it is also outside of him. We know this because we perceive the world spatially, and according to Kant, space is out outer sense. (A 26/B42) Thus you are justified in believing snow is white just in case phenomenally, snow is white. If our holding of the judgment be only subjectively sufficient, and is at the same time taken as being objectively insufficient, we have what is termed believing. Lastly, when the holding of a thing to be true is sufficient both subjectively and objectively, it is knowledge. The subjective sufficiency is termed conviction (for myself), the objective sufficiency is termed certainly (for everyone). (A822/B850-1)
Truth Specific definitions of the nature of truth vary from theory to theory. Without assuming any particular theory, by truth I will mean what prima facie we mean in our ordinary talk—"telling it like it is," the consonance between a judgment (or set of judgments) and the way things actually are. When using the phrase "is true" I will intend the ascription above characterization of truth to the judgment in question. For example, "It is true that cats hate water" means that the judgment "Cats hate water" describes the world accurately. Since I refrain from defining 'world' at this time, the definition provided will allow for both realist and idealist con-
10 Kant's Correspondence Theory of Truth ceptions of truth. Redundancy theorists claim that the predicate 'is true' is trivial and adds nothing new to the judgment it modifies. According to this account, "p is true" means exactly the same thing as "p." Kant's view could not accurately be classified as a redundancy view. For Kant 'is true' is an analyzable predicate that means "the agreement of our knowledge with its object." (A237/B296) A simple judgment consists in the claim that there is a certain relationship between a concept and an object. An example of this is "Kittens are small cats." In itself, that judgment simply asserts that there is a certain relationship between the concept 'kitten' and the predicates 'small' and 'cat.' The judgment "It is true that kittens are small cats" says something more. It says that the relationship between concept and predicates described in the simple judgment holds in reality. Specifically, the predicate 'is true' indicates that the judgment in question accurately maps the relationship between its object(s) and their properties. This being the case, Kant cannot be a redundancy theorist. The predicate 'is true' has a specific meaning when it is applied. For each judgment, the predicate 'is true', if applied to it, would mean that the truth conditions for that judgment have been satisfied and the predicate 'is true' would fail to appropriately apply if those conditions were not met. Truth Conditions By the term, "truth condition(s)" I will mean simply the thing or set of things that must be the case if the judgment in question is true. Such conditions may include but are not restricted to states of affairs, fit with a set of established beliefs, and/or logical relationships. The truth conditions for the empirical judgment "My cat chewed a hole in my shoe" are that there is a set of states of affairs, namely, (1)1 have a cat, (2) I have a shoe, (3) there is a hole in my shoe and (4) the hole was put there by my cat. In addition to these states of affairs obtaining individually, they must also relate to one another in the appropriate way. (This last qualification is significant as it serves to eliminate as candidates for truth judgments that result from fallacious reasoning such as confusing cause and effect, ignoring a common cause etc.)
The Problem of Truth in Transcendental Idealism
11
Truth for analytic judgments is a function of the connection between the subject/predicate relation in the judgment and the semantic and syntactic structure of our language as we have determined to consistently employ it and as it accords with the principle of non-contradiction. The truth condition for the judgment "all bachelors are unmarried men" is that it accurately depicts the way in which our language connects the concepts of bachelor and unmarried man. Other sorts of judgments, like those with truth functional connectives, have a different set of truth conditions. Take the judgment, "If my cat chewed a hole in my shoe, then she is in trouble", as an example. The truth conditions for that judgment are that it is not the case that the antecedent is true and the consequent false. There is no one state of affairs to which the true judgment must correspond. Such judgments, in fact, have the same truth conditions when they refer to no state of affairs whatsoever (If p, then q.). Complex judgments, like those with truth functional connectives, may be broken down into their simple parts, and these simple parts are either matters of definition, or matters of fact, which purport to conform to states of affairs. The truth conditions of the simple parts involve an analysis of the accuracy of the definition, or an analysis of the relationship of the judgment to the relevant state(s) of affairs. In short, what would have to be the case for the complex sentence to be true is a function of what would have to be the case for the constituent sentences to be true. Let us return to our example. The truth functionally connected judgment is "If my cat chewed a hole in my shoe, then she is in trouble". The truth conditions for that judgment are; again, that it is true so long as the antecedent does not obtain while the consequent does not. How do we know the truth conditions for this particular judgment? We come to know by breaking the complex judgment down into its constituent parts and analyzing what their relationship to the relevant state(s) of affairs would have to be for them to be true. Thus, the truth conditions for the judgment, "If my cat chewed a hole in my shoe, then she is in trouble" are that it is not the case that there is a set of states of affairs such that (1) my cat chewed a hole in my shoe, and (2) it is not the case that she is in trouble.
12 Kant's Correspondence Theory of Truth Truth Maker For any judgment to be meaningful, it must have both sense and reference. In using the term 'truth maker', I will be referring to the referent of a given judgment. The referent is that individual whose relationship to the relevant judgment determines whether or not the truth conditions for that judgment are satisfied. Again, a truth maker may be a state of affairs, a set of logical rules, the content of a set of beliefs, or something else entirely. Truth makers may be internal to the judging subject, external to the subject, or purely linguistic. For correspondence theories of truth, the truth maker is generally something external to the judging subject. This external truth maker may be an object or a state of affairs. An example of an object as truth maker is the truth maker for the judgment; "The book on the table is open." That judgment is true just in case there is an object, (namely a book) which is in a certain place, (on a table) and is in a certain position (open). We can assess whether the truth conditions for a given judgment have been satisfied only if there is an object to which the judgment either conforms or fails to conform. The object that the judgment is about is the truth maker for that judgment. It is that which determines whether the truth conditions have been satisfied. I said that the truth maker in correspondence theories is generally external to the subject. However, there are exceptions to this general rule. A correspondence theorist is almost certainly going to allow that we can make true reports about our ideas, feelings, beliefs, etc. The objects of such judgments would not be external to the subject, but the truth condition for the judgments would still be that they correspond to their objects. Thus, my report "I believe Elvis is still alive" would only be true if there was a correspondence between my report and an actual belief I possessed concerning the existential status of the Elvis Presley. For some theories of truth, including certain coherence theories, the truth maker is always internal to the thinking subject. The truth maker is a set of beliefs, and the judgment is determined to be true or false in virtue of its degree of "fit" with that set. Essentially, a belief is true just in case it does not contradict any of the beliefs in the established set. (The better a given belief meshes with the web of established beliefs, the more
The Problem of Truth in Transcendental Idealism
13
confidence one ought to have in its truth.) Imagine a very simple subject, Tom, who has only five beliefs in his set. These beliefs are: (1)1 have only one sister, and her name is Rose. (2) No sister of mine would marry a lunatic. (3) All musicians are lunatics. (4) My sister is married to Paul. (5) All people who play instruments are musicians.
Tom has a judgment that is a candidate for a new belief; "Paul plays the flute." The truth conditions for that judgment are that it is true just in case it fits with (i.e. does not contradict) Tom's established set of beliefs. In this case, it is Tom's set of established beliefs that is the truth maker for his new judgment. It is their inter-connection and content that determines whether or not the new judgment will fit. (In the example in question, the new judgment's failure to fit with the subject's set of beliefs entails the judgment's falsity.) Thus it is the set that is the truth maker qua its role of satisfying or failing to satisfy the truth conditions for the judgment in question. For other theories of truth, truth conditions are purely linguistic. Take for example judgments such as "All squares are non-circles." The truth conditions for this judgment are that it is true just in case what we mean by 'square' can never be consistent with what we mean by 'circle'. In this case, the truth maker is the structure and content of our language. It is this structure and content that determines whether the truth conditions for the judgment in question have been satisfied. Certain judgments fail to have truth makers, and thus, except in an abstract semantic sense, fail to have meaning for us. Take the judgment "In 1999, the present king of France is bald" as an example. In a correspondence theory of truth, this judgment is meaningless for us because there is no present king of France in 1999. The judgment has no referent. The truth conditions for the judgment are that the judgment conforms to the object described in the judgment. Since there is no object (namely a present French king), there can be no conformity or lack of conformity between the judgment and its object. It is impossible to determine if the truth conditions are satisfied because there is no accessible truth maker to satisfy them.
14 Kant 's Correspondence Theory of Truth Such judgments would be termed 'problematic' by Kant, because although their truth can be assumed, it can never be demonstrated. Such judgments are problematic precisely because they are unverifiable (and also unfalsifiable). That is to say, one cannot determine if their truth conditions have been satisfied. As such, one cannot accurately claim to have knowledge from purely problematic judgments. Meaning For a judgment to be a truth bearer, it must first be meaningful. To have a meaning, according to Kant is to have "an ascribed significance." (A84/B117) (Note that all truth bearing judgments are meaningful, although not all meaningful judgments are truth bearers. Kant maintains that there are meaningful judgments that are purely regulative.) A judgment's having ascribed significance is a function of its form and matter. What Kant refers to as form and matter is roughly what we mean by sense and reference. Matter and Form—These two concepts underlie all other reflection, so inseparable are they bound up with all employment of the understanding. The one [matter] signifies the determinable in general, and the other [form] its determination—both in the transcendental sense, abstraction being made from all differences in that which is given and from the mode in which it is determined. (A266/B322)
Kant explains the meaningfulness of judgments in terms of their form and matter. An empirical judgment is meaningful for us only if its form is one determined a priori by our understanding, for anything to be a possible object of knowledge for us, it must be in a form that is comprehensible to us, given our limitations as finite rational minds. Among these limitations is the need to think discursively; we can't directly intuit (intellectual intuition) we must think about our objects. As we have discursive intellects, all knowing for us is conceptualizing. In other words, we judge by subsuming an object under a concept. This is the form of all judgments. The matter of empirical judgments is their object in a stronger sense; it is that which they are about. The matter of empirical judgments is constituted by appearances. Kant's account of knowledge is unusual, how-
The Problem of Truth in Transcendental Idealism
15
ever, in that the matter of empirical judgments must have a certain form as well. Just as the form of a judgment in general is determined by the structure of our understanding, so is the form of appearances. For Kant, there are conditions for the possibility of experience. For any appearance to be a possible object for us, its form must be determined according to the pure principles of the understanding and the pure intuitions of space and time. Both pure principles and pure intuitions are dependent on the judging subject. Therefore, the objects, which constitute the matters of our empirical judgment, must themselves be of a certain form. That which in appearance corresponds to sensation I term its matter, but that which so determines the manifold of appearance that it allows of being ordered in certain relations, I term the form of appearance. That in which alone the sensations can be posited and ordered in a certain form, cannot itself be sensation; and therefore, while the matter of all appearance is given to us a posteriori only, its form must lie ready for the senses a priori in the mind. (A20/B34)
Since, for Kant, transcendental and analytic judgments are meaningful for us; they too must have form and matter. Transcendental and analytic judgments have the same form as empirical judgments, namely, the form of a discursive judgment. The specific nature of the referent for transcendental judgments is debatable. For this reason, I will allow an account of reference for transcendental judgments to develop over the course of my inquiry. The matter for analytic judgment is significantly less problematic. The matter of analytic judgments, is again, that which they are about. For purely logical analytic judgments, their matter is their form. "All P's are P's" is true because the sentence is a logical tautology. It has a form such that any instantiation of it is necessarily true. Another way of understanding this point is to say that purely logical judgments are selfreferential. The matter for material analytic judgments is their subject concept. For example, the judgment "All bachelors are unmarried men" is not analytically true because of its form. That is to say, it is not selfreferential in the same sense that purely logical analytic judgments are. "All bachelors are unmarried men" is an analytic truth because the con-
16 Kant 's Correspondence Theory of Truth cept of the nature of a bachelor includes the predicates of unmarried and man. Nevertheless, it is an analytic truth because one need have no immediate recourse to particular bachelors to know its truth. The matter of material analytic judgments is the nature of the thing indicated by the object concept. Thus for Kant, all cognitive judgments must be meaningful. All meaningful judgments must have matter and form; sense and reference. The general form of all judgments is the same; it is the form of a conceptualizing. The nature of the matter for a given judgment will be dictated by the kind of judgment it is. This is Kant's basic account of meaning. Summary The essence of Kant's conception of theoretical truth is determined by the elements of his account of knowledge and the nature of his conception of certain components of truth theory (truth bearers, truth conditions, and truth makers). This being the case, I will evaluate several authors' proposed conceptions of truth for Kant's theoretical philosophy based on how the authors assess and apply these factors. I will then consider how these factors ought to be construed and applied according to textual evidence in the Critique of Pure Reason, and how they must be understood in accordance with certain necessary tenets of transcendental idealism.
Chapter Two An Incomplete Account of Truth Henry Allison In Kant's Transcendental Idealism: An Interpretation and Defense, Henry Allison provides an interpretation of transcendental idealism which purports to entail an in-depth explication of "Kant's views on the nature, conditions and limits of human knowledge." (Allison 1983, 3) However, this explication does not contain a complete account of truth as it functions within transcendental idealism. Allison's omission of an interpretation of truth is problematic for a number of reasons. First, an account of transcendental idealism that does not explicate the nature of truth is incomplete. This is not, of course, a tremendous problem for Allison in and of itself, since a good general interpretation of transcendental idealism need not address all aspects of the Critique. However, Allison's failure to provide an account of truth does ultimately leave his interpretation with some problematic implications. In addition, because of these implications, the incompleteness of his account detracts from the veracity of his interpretation. Drawing from Allison's interpretation in the aforementioned book, as well as his subsequent modifications and defenses of this interpretation, I will provide what I take to be an accurate explanation of his concept of an epistemic condition and the tenets of his interpretation which arise from that concept. I will then consider some difficulties that arise from his failure to explicate truth in relation to such epistemic conditions. Allison on the Nature of Epistemic Conditions Allison does not directly consider truth conditions or truth makers. Instead, he explicates what he calls 'epistemic conditions.' If he provides any substantive account of truth in his analysis of transcendental idealism, it is embedded in his account of epistemic conditions. However, as I will argue, the conception of truth inherent in Allison's account is incomplete and, as a result, his overall interpretation of transcendental idealism is flawed. Allison argues that transcendental idealism is essentially an epistemic, rather than a metaphysical, thesis. To show this, he introduces the
18 Kant 's Correspondence Theory of Truth notion of an 'epistemic condition', by which he means that without which our representations would not relate to objects (i.e. possess objective reality). He maintains that understanding the nature of these conditions is fundamental to a correct understanding of transcendental idealism. Epistemic conditions are, according to Allison, distinct from both psychological and ontological conditions. Although both epistemic and psychological conditions are subjective (as they reflect the structure of the finite mind), psychological conditions do not objectivate (i.e. serve as necessary conditions for the representation of an object) as epistemic conditions do. Epistemic conditions also have shared qualities with ontological conditions in that both are objective (in the sense of objectivating). However, epistemic conditions apply to representations whereas ontological conditions apply to things in themselves. (Allison 1996, 4-5) Allison's definition of ontological conditions as applicable only to things in themselves is indicative of the problems that arise from the incompleteness of the account of truth in his interpretation of transcendental idealism. Because of his failure to account for the nature of the truth makers for the different kinds of judgment in transcendental idealism, Allison undermines the strength of Kant's empirical realism. For Kant, the world (the world of appearances) has an ontological status independent of finite subjects. If it did not, Kant's view would not be substantively different from Berkeley's empirical idealism. External objects would continue to exist even if there were no finite subjects. They simply would not have any properties that are knowable for us. It is not that empirical objects are dependent on finite subjects for their ontological status. Rather it is that the conditions for the possibility of our experience of those objects are dependent on finite subjects. If it were not the case that there were ontological conditions for empirical objects, then Kant would not be an empirical realist. Allison asserts that a commitment to this sort of epistemic condition implies a commitment to a kind of idealism, which is distinct from both phenomenalist and metaphysical idealisms historically attributed to Kant. For Kant's idealism, "the basic point is that the transcendental method requires one to consider objects in terms of the conditions of their representation, and this means that an object is understood idealistically as the
An Incomplete Account of Truth
19
correlate of a certain mode of representation." (Allison 1996, 5) In short, Allison is arguing that what makes objective knowledge possible for us is the fulfillment of certain epistemic conditions. These conditions are determined by the structure of our understandings and are satisfied by the conformity of representations to that structure. It is certainly true that, for Kant, knowledge requires the satisfaction of the conditions of the possibility of experience. However, Allison's definition of the role of the conditions of the possibility of experience demonstrates yet another way in which his failure to explicate a truth maker for judgments is problematic for his interpretation. The difficulties for his interpretation arise from his limited conception of meaning for Kant, and this limited conception arises from Allison's overly narrow focus on epistemic conditions. He interprets the phrase "possibility of experience" as qua the possibility of what can be experienced, i.e. what can be empirically known. He focuses solely on the experienced to the exclusion of the experiencing. "Possibility of experience" can also mean the possibility of the nature of experience qua possibilities in terms of which is comprised what is, what can be, and what cannot be empirically known and is itself not empirically known. Moreover, to say that an experience and/or an experiencing conforms to the nature of experience is not equivalent to saying that the experienced conforms to the nature of experiencing. This is a possibility that Kant explicitly denies. According to Kant, views like Allison's commit a particular sort of error. Kant explicates this error in the following passage. The error, which quite obviously is the cause of this mistaken venture, and which indeed excuses though it does not justify it, lies in employing the understanding, contrary to its vocation, and in making object, that is, possible intuitions, conform to concepts, not concepts to possible intuitions, on which alone their objective validity rests. (A289/B345)
Allison asserts that there are two different operations implied in the discussion of epistemic conditions found in the introduction to the transcendental logic. He argues, however, that the two operations are characteristic of the same activity, namely synthesis. In one case a particular is unified under a concept and in the second undetermined empirical intuitions are determined under the categories, space, and time through the
20 Kant's Correspondence
Theory of Truth
original unity of apperception. Although the contrary is commonly assumed to be the case, the central claim of this paragraph [A79/B104-5] is the identity of the understanding and its activity (function) as considered in general and transcendental logic. In short, the two disciplines are concerned with one understanding, possessing a single characteristic activity, which they analyze at different levels... In fact, it is precisely because general and transcendental logic deal with the same activity at different levels that Kant thinks it possible to move from the determination of the forms or functions of the former to those of the latter. (Allison 1983, 123)
In the way he emphasizes that transcendental idealism is an epistemic rather than metaphysical thesis Allison is, I think, providing an account of the meaning of judgments which focuses on sense and ignores the importance of specifying the nature of reference. This omission is evidenced by Allison's failure to specify a truth maker for various kinds of judgments. Judgments about representations qua what is represented are meaningful for us just in case the representations are objectivated by the satisfaction of epistemic conditions. If we follow Allison's lead, we need not concern ourselves with an ontologically grounded truth maker at all, because to do so would be to be concerned with things in themselves. (Remember, for Allison, ontological conditions apply only to things in themselves.) Instead, our focus must be on the relationship between our subjective representations qua our representing of empirical objects, and the relationship between them and objective representations qua what is represented subjectively, which have been objectivated through synthesis in which objectivating epistemic conditions are satisfied. Since he denies the application of ontological conditions to anything other than things in themselves, Allison cannot subscribe to a two-world view. If only noumenal objects have ontological conditions, then it cannot be the case that there are two ontologically independent worlds, one noumenal and one phenomenal. Thus, Allison is led to a "two-aspect" view instead of the two-world view. The two-aspect view is a view according to which there is one set of things that exist, and for each thing there is a way that it seems to us and a way that it is in itself. The satisfaction of epistemic conditions, i.e. the process of having representations be meaningful for us, takes place entirely wimin the realm of the former;
An Incomplete Account of Truth 21 objects as they appear to us. I think an examination of Hoke Robinson's criticism of Allison's 'two-aspect' view, as well as Allison's replies to those criticisms, will bring to light a number of problems inherent in Allison's interpretation of transcendental idealism in general. Using these criticisms as a background, I will argue that at least some of the problems that Robinson points out in Allison's interpretation of transcendental idealism arise from the incompleteness ofthat interpretation's account of truth. Robinson and Allison: Criticisms and Replies Robinson and Allison have entered into a debate over the correct interpretation of the phenomenal/noumena distinction. I will argue that the well-founded objections that Robinson makes against Allison's interpretation are implications of Allison's incomplete account of the Kantian conception of truth. Allison's failure to give an explicit account of truth that explains the nature of truth conditions and truth makers leads him to a problematic assessment of the phenomena/noumena distinction. Therefore, in explaining how his reading of this distinction is flawed, we will see why his failure to explicate fully Kant's conception of truth is a problem for his interpretation. There has been, to say the least, much said about the issue, but for the present purpose, an abridged introduction to the distinction will suffice. The basic notion of transcendental idealism is that we, as subjects, shape our reality to some extent. Thus, certain features of our experience (space and time, as well as certain categorical regularities) are empirically real, but transcendentally ideal. That is to say, they should not be taken as applicable beyond the realm of our experience because they are ontologically dependent on finite subjects like ourselves. This brings us to a distinction between two classifications of 'stuff; stuff about which we are warranted in making substantive knowledge claims (phenomena), and stuff about which we are not warranted in making such claims (noumena). Kant uses the word Sinnenwesen (sensible entities) to describe phenomena. He distinguishes these sensible entities, objects as appearance, from noumena. Under the heading of noumena, we place things as they are independently of how we intuit them,
22 Kant's Correspondence Theory of Truth as well as mere objects of thought. These are Verstandeswesen, intelligible entities. With the phenomena/noumena distinction, Kant is demonstrating what he takes to be the limits of theoretical knowledge. How one interprets the metaphysical view arising from this distinction sets the stage for determining not only the simplicity, but the viability of transcendental idealism as an epistemic and metaphysical theory. An examination of some competing theories will, I think, demonstrate that there is an interpretation of this distinction, which is both faithful to the spirit of transcendental idealism as Kant conceives it, and a viable candidate for a rational worldview. In order to demonstrate what is at issue in the Allison/Robinson debate concerning the phenomena/noumena distinction, I will use Robinson's definitions for the relevant terms. (I think his definitions are, for the most part, correct.) (Robinson 1994, 413) The terms to be defined are representation, appearance, and thing in itself. A representation is an idea in the Lockean sense. A thing in itself is an entity that is entirely independent of our consciousness, or ability to be conscious, of it. Finally, there is the appearance. Unproblematically, we can say that Kantian objects are appearances, however, that is where the agreement ends. The definition of this term is disputable, and, as Robinson points out, it is the difference in defining the term appearance that demarcates between the various interpretations. Traditionally commentators have argued that Kant's phenomena/noumena distinction implies that there are two worlds, a phenomenal world and a completely separate and experientially inaccessible noumenal world. The standard 'two-world', or 'two-object' view, conflates appearances and representations. It maintains that there are two sets of material objects; objects as they appear to us (representations), and objects as they are independently of us. Although there is textual support for this reading, this interpretation makes transcendental idealism untenable. There are a number of problematic implications that arise from the two-world view. Two problems that Robinson points out are the problem of affection and the problem of phenomenality. The problem of affection has been cited by many commentators, including Jacobi and Allison, as being the fundamental difficulty with the
An Incomplete Account of Truth 23 two-world view. The problem results from the conflation of representations and appearances. According to Kant, representations develop as a result of the application of our cognitive faculties to a manifold of sensibility given to the understanding. The manifold of sensibility develops "as the result of [affection] of the mind by the object." However, if the object is itself an appearance, then it cannot produce the sensory manifold out of which it was formed. If the appearance cannot be the source of affection, then the source must be things in themselves, a possibility that Kant denies. Since under the two-world view all possible explanations of affection are untenable, the two-world view must be rejected. (Robinson 1994, 415) A second problem with the two-world view is the problem of phenomenality. The problem is that if appearances, the objects of our knowledge, are themselves mere mental representations, then Kant's transcendental idealism is simply Berkeleyan empirical idealism without Berkeley's sustaining God, (Robinson 1994, 416) and with the apparently ad hoc postulation of things in themselves. Because of these problems, among others, which arise from the two-world view, it is a view that is inconsistent with both the spirit of transcendental idealism and major tenets of the Critique of Pure Reason. Henry Allison suggests an alternative to the two-world view that is more sympathetic to the project of transcendental idealism. Whereas the two-world view is primarily a metaphysical account, the two-aspect view is an epistemological account. It is basically the view that there is one set of entities, each of which has two aspects, a noumenal one and a phenomenal one. An aspect is a set of properties exhibited by the object. According to this interpretation, the distinction between phenomena and noumena represents a difference in how we consider objects at the meta-level. Objects may be considered subjectively, i.e. as they appear to us, or they may be considered objectively, i.e. as they exist independently of the subjective conditions of our understanding. Transcendental idealism, according to the two-aspect view, is the doctrine that our theoretical knowledge is limited to things considered subjectively. Under the two-aspect view, the appearance/thing in itself distinction is not a distinction between two kinds of things, nor is it a distinction between a thing considered as appearance and that thing considered as a
24 Kant's Correspondence Theory of Truth thing in itself. "It is rather between a consideration of a thing as it appears and a consideration of the same thing as it is in itself." When considering things as appearance, we are considering them as they are in their relation to the conditions under which they are given to our sensibility. When considering those same things as they are in themselves, we consider them in abstraction of those sensible conditions. (Allison 1983, 240-241) Allison's Conflation of Two Senses of 'Object' Textual disagreements aside, Robinson objects to what he sees as Allison's conflation of two distinct senses of the term 'object' (Robinson 1994, 422-423). Allison describes epistemic conditions as applying only to representations. A subjective representation represents an object for us once it has met the epistemic conditions. An object in this sense, i.e. in the sense of being dependent on the judging subject, could not have an in-itself aspect. Robinson concludes that since Allison's two-aspect view entails that a single object has both an as-appearance and an in-itself aspect, and Robinson claims that epistemic conditions apply only to representations, Allison must be sliding between two necessarily distinct notions of object. As a consequence of this ambiguity, Allison's two-aspect view fails to specify adequately the subject/object relationship. But an object that exists only in the sense that we hold a representation to refer to it depends upon our representation, and thus can scarcely have an in-itself aspect. Allison slides from this notion of object to one in which the object . . . is "what we intuit (the phenomenal object) [rather t h a n ] . . . our intuition of it.," it is the "something that appears," the "thing itself." The independence thus acquired is supposed to allow the object to have both an as-appearance aspect. . . and an in-itself aspect as well. But it is difficult to reconcile the apparent independence of an in-itself aspect with the apparent subject-dependence of an asappearance aspect. (Robinson 1994, 423)
Robinson argues that the aforementioned conflation leads Allison to a problematic 'filtration model' of transcendental idealism. This model is, according to Robinson, incompatible with an accurate and consistent interpretation of transcendental idealism. Robinson analyzes what he takes to be all the reasonable interpretations of Allison's sense of 'object,' and argues that all are problematic.
An Incomplete Account of Truth 25 Allison's 'Filtration Model' According to Robinson, there are three possible ways to sort out the conflated sense of 'object' in Allison's two-aspect view. The first way to sort out the possible dependence relationship between subjective representations and their objects may be referred to as 'illusion'. This is the Berkeleyan approach in which representations are in no way dependent upon independent objects. Since this interpretation does not allow for the in itself aspect advocated by Allison, this cannot be the relationship he has in mind. (Robinson 1994, 423) The second possible interpretation of the relationship between subjective representation and object is the idealism interpretation. According to this reading, objects exist by virtue of their representation. In this view, the ontological conditions of an object are its epistemic conditions. This view, according to Robinson, is unacceptable to Allison as well because, again, objects would be prohibited from possessing the in itself aspect which Allison prescribes. (Robinson 1994, 423-424) Robinson argues that the remaining possible interpretation of the relationship between subject and object is the only one acceptable for Allison. This view implies what Robinson calls the filtration model. According to this model, there is a range of things in themselves that exist completely independent of the knowing subject. They exist by virtue of meeting some ontological condition. Among things in themselves, there are two kinds; those that can satisfy our epistemic conditions, and those that cannot. Our epistemic conditions work as a kind of filter. The epistemic conditions allow things in themselves, which can satisfy them to become appearances for us, while disallowing things in themselves, which cannot. Moreover, those things in themselves, which cannot fit into a unified world with others, will also be disallowed. Those things that survive both filtrations become the objects of possible experience. Thus, for each appearance which we have (because it made it through the filter), there is a concurring way that that thing is in itself. (Robinson 1994, 424) The filtration model is problematic for a number of reasons. First, it is not textually supported. Kant describes what Allison calls our epistemic conditions as determining our objects of possible experience, not filtering out the objects that are not candidates for possible experience.
26 Kant's Correspondence Theory of Truth He describes the process as one of synthesis, not filtration. Textual difficulties aside, the filtration model, as Robinson points out, implies a level of isomorphism between phenomenal and noumena which transcendental idealism must prohibit. It is this implied isomorphism that is at the heart of Robinson's criticisms of the two-aspect view. The Problematic Isomorphism of Allison's 'Two-Aspect' View In what I think is the force of his objections to Allison's interpretation, Robinson addresses the isomorphism between phenomena and noumena, which the filtration model implies. (Robinson 1994, 425) If the two-aspect view is correct, and there is a single set of entities each of which has a phenomenal and a noumenal aspect, then there is necessarily a structural isomorphism between things considered as noumena and things considered as phenomena. According to Robinson, the two-aspect doctrine, because of the filtration model that Allison must utilize in employing it, denies one of Kant's fundamental assumptions, the non-spatio-temporality of things in themselves. It is in the original unity of apperception that our understandings individuate objects out of the undetermined manifold of sensibility according to space and time. If all objects were already individuated noumenally, then they would have to be individuated spatiotemporally in the same way, a possibility that Kant explicitly denies. There are difficulties, however, with spatiality, in particular what may be called the isomorphism problem. This problem, according to Allison, is due to the assumption that the two aspect view requires an in-itself aspect for every appearing object (though not every object with an in-itself aspect must be capable of appearing). But this would mean that, since objects as appearances are individuated by their spatiotemporal characteristics, so would be the underlying thing in its in-itself aspect, contrary to the Kantian doctrine of the non-spatiality of things in themselves. (Robinson 1994, 425)
Moreover, even if the individuation problem is put aside and Allison does not attribute spatiotemporal and numerical designation to the thing as it is in itself, the two-aspect view would have to imply that one underlying thing could possess contradictory aspects. Its phenomenal aspect would be causal, spatial, temporal etc., and its noumenal aspect
An Incomplete Account of Truth 27 would not. It is difficult to see how one could maintain coherence while failing to deny that a single thing could exhibit contradictory qualities. Allison's Response to Robinson Allison claims that, for Robinson, a dual-aspect theory, if it is coherent, must be one "which assigns distinct properties to distinct aspects of some underlying thing" (Allison 1996, 15) Although he admits that this characterization is accurate for most dual-aspect theories, he thinks that this sort of dual-aspectism is not what Kant has in mind. (Allison 1996, 15) Allison argues that such charges are misdirected given the nature of his view, because they imply a more metaphysical than epistemic reading of the doctrine. His dual-aspect theory refers to modes of representing, not modes of being. Transcendental idealism, on a double-aspect reading, does not require that phenomenal and noumenal properties be assigned to the same things, as is suggested by the metaphysical version of dual-aspectism and Robinson's . . . example... Since (on Kant's view) being representable in these disparate ways is a function of different modes of representing rather than the nature of the things represented, it is not (except in a trivial sense) a property of things in themselves. (Allison 1996, 15)
In response to the isomorphism objection, Allison provides an explanation of sameness, which he thinks is what is really in question in such objections. He concedes that even if a metaphysical isomorphism is denied (as in the previous response), his dual aspect view does imply "that it is in some sense the same thing that is to be considered from two points of view. And this raises the question of how this sameness is to be understood." (Allison 1996, 16) He points out that the problem is not one of direct contradiction, because it is not the case that an appearance is also considered a nonappearance. The problem is that there seem to be two conflicting ways of representing the same underlying thing. An empirical dual-aspectism, according to Allison, would assert that there are simply two ways of representing the thing in itself. However, says Allison, "the thing in itself is one of the ways of considering the thing." (Allison 1996, 16) By making the issue epistemic rather than metaphysical, Allison believes the objec-
28 Kant's Correspondence Theory of Truth tion in question can be circumvented and coherence can be maintained. The only way to maintain coherence is to argue that the thing underlying the two conflicting modes of representing is the "transcendental object=x", and not the thing in itself conceived as thing in itself. This allows us to distinguish the thing as we can represent it to ourselves problematically (although independent of our epistemic conditions) from the way it would have to be considered as completely inaccessibly to any finite discursive intellect. This distinction, argues Allison, is plausible because it is close to the transcendental object/noumenon distinction which Kant draws in the Critique. Reply to the Two-aspect Debate First, I want to emphasize that I think there is a great deal of merit in Allison's focus on the epistemic nature of transcendental idealism. Viewing transcendental idealism as a fundamentally metaphysical thesis can lead to a number of inconsistencies including the traditional 'twoworld view' of the phenomena/ noumena distinction which is at the foundation of Strawsonian criticisms and characterizations of transcendental idealism as a phenomenalism. Moreover, I think that that the text of the Critique of Pure Reason supports an epistemic reading of the appearance/thing in itself distinction. However, I think Allison's interpretation over-corrects for this problem, and in doing so creates an entirely new set of problems for transcendental idealism. At the heart of this new set of problems is, I think, Allison's account of significance that emphasizes sense to the exclusion of secured reference. By a secured reference, I mean an established truth maker beyond agreement with the rules of logical syntax or mere coherence, which Kant calls a merely negative condition of truth. Truth is a merely negative condition of truth for Kant because, although truth would not be possible if experience were not coherent, coherence is not the essence of truth. I will fully explicate the role of coherence in truth in chapter seven. What I have in mind is an ontologically grounded referent. I think that if Allison were to include an account of the secured reference for judgments, then many of the problems proposed by Robinson would be solved. However, Allison's dual-aspect view would also be significantly altered.
An Incomplete Account of Truth 29 Secured Reference and Allison's Equivocation of 'Object'. If Allison were to allow for an ontologically grounded referent for judgments which would, in the appropriate cases, serve as the ultimate truth maker for those judgments, then he would be able to distinguish the two senses of object which he is accused of conflating. Instead of conflating appearance and thing in itself (or, as he later suggests, appearance and transcendental object), Allison could posit that there is a set of entities upon which our appearances are ultimately based. There is a way that set of things is in itself, and there is a way that the set is after it has been determined by our epistemic conditions in the original unity of apperception. Positing this sort of secured reference does not require Allison to slide between two notions of object to assert two modes of representing. There is a way we represent the object objectively to ourselves, and there is a way we represent the object to ourselves subjectively. There is also a further mode of representing. It is the way in which the object is (or would be) represented in a non-discursive intellect, namely as it is in itself (as creation/ Werke). This tri-aspect view, in which the third aspect is the ultimate ground from which the second, and consequently the first aspects are formed, would allow Allison to interpret transcendental idealism as a primarily epistemic thesis. This is the case because Allison could focus on the relationship between our appearances and their reliance on our epistemic conditions without allowing direct access to things in themselves by making them simply other aspects of things as they objectively appear to us. The Grounding Model versus the Filtration Model. If Allison were to amend his dual-aspect view to a tri-aspect view as I have suggested, he would also evade Robinson's attribution of the filtration model. Instead of postulating a model in which our epistemic conditions serve to filter a set of things in themselves, some of which are candidates for appearances, and others which are not, Allison could postulate that there is a set of things in themselves which serve as the ground out of which our epistemic conditions construct appearances. Thus, there is no sub-set of things in themselves which already satisfies the conditions for the possi-
3 0 Kant 's Correspondence Theory of Truth bility of experience (again, a possibility which Kant denies). Finally, by postulating this third aspect, or ontological ground for appearances and subsequently representations, Allison can avoid Robinson's charges of unwarranted isomorphism. Since appearances and things in themselves are distinguished under this view, there would be no reason to infer that contradictory qualities such a spatiality and nonspatiality applied to the same individuals. Moreover, there would be no reason to infer that things in themselves are individuated in the same way as appearances. In both cases, the result would be more compatible with the text of the Critique and with transcendental idealism in its most plausible form. Summary I have argued that some of the problems for Allison's interpretation of transcendental idealism arise from his omission of an explicit account of truth. Because he does not fully explicate what truth is within transcendental idealism, Allison fails to account fully for truth conditions and the possibility of ontologically grounded truth makers. His failure to do so leads him to an interpretation of the phenomena/noumena distinction that is inconsistent with some basic assumptions of transcendental idealism. In other words, his failure to explicate fully what truth is for Kant leads him to a failure to distinguish and account for both the sense and ultimate reference of our judgments. By demonstrating the problems that arise from such an omission, I have shown that it is reasonable to conclude that interpretations, like Allison's, that fail to explicitly account for truth in transcendental idealism, are prone to systematic difficulties. Consequently, I think it is also reasonable to conclude that such views are incomplete in a way that makes them inadequate as interpretations of Kant's transcendental idealism.
Chapter Three The Verificationist View An Introduction to Verificationism Another of the major Anglo-American traditions in Kant scholarship is exemplified by the work of P.F. Strawson. At the heart of this tradition is the claim that Kant's system of theoretical philosophy is only viable when transcendental idealism is understood as a kind of complex empiricism. Strawson denies the viability of transcendental idealism at all, and concludes that Kant must either admit a transcendental realism, or be relegated to Berkeleyan idealism. According to Strawson's verificationist interpretation of transcendental idealism, a judgment is known to be true just in case it is verified by experience. A verificationist conception of justification of this sort entails a correspondence theory of truth. The truth maker for judgments that are not purely logical may be either objects in a transcendentally real world, or perceptual objects in an empirically ideal world like Berkeley's. In any case, Strawson maintains that the conception of truth to which Kant must be committed is inconsistent with a true transcendental idealism. Others have taken Strawson's verificationism and adapted it such that it fits with the basic suppositions of Kant's transcendental idealism. Among these is Carl Posy. Posy, like Strawson, gives an account of the theory of justification implicit in Kant's view, rather than an account of his conception of truth. However, there is an account of truth implicit in both accounts of justification. Posy focuses his interpretation of what Kant would require for one to be justified in asserting a judgment. He ultimately concludes that the only viable account of truth in transcendental idealism has to be one implicit in what he calls "assertabilism". P.F. Strawson In The Bounds of Sense, Strawson sets forth a comprehensive interpretation of the first Critique that embraces this tradition, and included in his interpretation is an account of what Strawson thinks Kant's conception of truth must be. He argues that Kant can have no coherent account of the
32 Kant's Correspondence Theory of Truth truth of transcendental judgments. Strawson argues that a well-founded theory of truth for Kant's system would necessarily be verificationist. He lends support to this claim by arguing the project of transcendental idealism fails. Specifically, he thinks Kant fails to establish the necessity of the categories as conditions for the possibility of any experience. And the necessity of application of such concepts will be traceable simply to the necessity of bringing objects under concepts, i.e. to pure understanding alone. So much for the general form of the argument and the assumption on which it rests. Serious doubts arise when we look a little more closely. (Strawson 1966, 78)
Assuming Kant's failure to demonstrate the necessity of the categories, Strawson claims that the only meaningful conceptions of truth are empirical truth and formal logical truth (because there is nothing necessary beyond the empirical and the formal). For Kant, a judgment that is true beyond the purely logical sense must be one that corresponds to the empirical object or state(s) of affairs it is about. The first Critique clearly supports Strawson's verificationist conception of what it means for an empirical judgment to be true. In many places, Kant refers to truth as "the agreement of knowledge with its object."(A58/B82) I will not take issue with this portion of his interpretation. However, I do take issue with his claim that the conceptions of truth for empirical judgments and truth for purely logical judgments are the only legitimate conceptions of truth to be found in the first Critique. If justified, Strawson's claim not only undermines the richness of Kant's conception of truth, it undermines his central project. Therefore, in dealing with Strawson's interpretation, I will try to show that his basis for dismissing the possibility of non-empirical and yet informative truths is unjustified. In doing so, I will demonstrate that there is room in a sound interpretation of the Critique for a richer account of truth. Strawson's objections to Kant's transcendental idealism are numerous and their exposition can, and does, fill volumes. However, for the sake of both brevity and relevance to the issue at hand, I will focus on two main points which I think are the essential ground of Strawson's purely empiricist interpretation of Kant's conception of truth; I will refer
The Verificationist View 33 to these as (1) the 'principle of significance' interpretation and (2) the Deduction of Categories objection. It is my aim to demonstrate the inaccuracies in these interpretations and objections and in doing so to show that Strawson's denial of the significance of the categories, and hence the foundation for his interpretation of Kant's conception of truth, is unjustified. The Principle of Significance Much of Strawson's empiricist interpretation of the Critique is based on a thesis which he attributes to Kant and terms 'the principle of significance' (hereafter, PS). PS: This is the principle that there can be no legitimate, or even meaningful employment of ideas or concepts which does not relate them to empirical or experiential conditions of their application. If we wish to use a concept in a certain way, but are unable to specify the kind of experience-situation to which the concept, used in that way, would apply, then we are not really envisaging any legitimate use ofthat concept at all. (Strawson 1966, 16)
The importance of PS to Strawson's interpretation of Kant is widely acknowledged. Wenceslao Gonzalez, among others, acknowledges that PS is one of the major principles on which Strawson builds his empiricist reading of Kant. (Hahn 1998, 336) Strawson's reading of Kant as a verificationist is, I think, motivated by his understanding of PS. However, as I will show, Strawson's definition of PS is problematic, and it is, in large part, this problematic definition which leads him to his unfavorable interpretation of transcendental idealism. Strawson maintains that the doctrine of PS is a formal expression of one of Kant's most fundamental claims, namely "that we can have no non-empirical knowledge, and hence no knowledge at all, of anything which is not an object of possible sensible experience." (Strawson 1966, 241) Before analyzing Strawson's interpretation of Kant's principle of significance, it is necessary to explain one of the terms to be discussed. Strawson's definition of PS involves a distinction between schematized and unschematized categories. Schematism is an activity of the imagination in which the pure concepts of the understanding are applied to the
34 Kant's Correspondence Theory of Truth manifold of sensible intuition. Imagination contains elements of both sensibility and understanding and is able, through the commonality of pure time, to subsume representations under pure concepts. In short, a schematized category is one that has been applied to the objects of possible experience. (A140-145/B179-184) As I understand it, Strawson's reading of PS leads him to criticize transcendental idealism along the following lines. Al (pi) According to PS, we cannot legitimately employ concepts without relating them to their experiential conditions. (p2) Due to the nature of the original unity of apperception, the synthetic a priori judgments in which the categories are employed occur prior to any possible experience (as they are conditions for it) and thus, unschematized, cannot have experiential conditions. (CI) Therefore, given PS, judgments involving unschematized categories are not legitimate. (C2) Moreover, if we reject this illegitimate employment of the categories, there remains little to bar us from understanding Kant's system as either a transcendental realism or an empirical idealism.
The line of reasoning in the above argument is warranted only if Strawson's interpretation of PS is correct. However, I do not think his interpretation of PS is correct, and in demonstrating that it is erroneous, I will demonstrate that the argument predicated on it cannot provide a sound foundation for concluding that Kant's only legitimate notion of truth is empirical verificationism. Meaning, Truth and Knowledge There are two passages from the Critique of Pure Reason which Strawson cites in favor of PS, and his interpretation of PS leads him to claim that there is credence in the charge that transcendental idealism is really nothing more than phenomenalism (and thus that truth for Kant is nothing more than correspondence to phenomenal objects). Even space and time, however free their concepts are from everything empirical, and however certain it is that they are represented in the mind completely a
The Verificationist View 35 priori, yet would be without objective validity, senseless and meaningless, if their necessary application to the objects of experience were not established. . . Apart from these objects of experience, they would be devoid of meaning. And so it is with concepts of every kind. (B195) For all categories through which we can attempt to form a concept of such an object [a divine object] allow only of empirical employment, and have no meaning whatsoever when not applied to objects of possible experience, that is, to the world of sense. Outside this field they are merely titles of concepts, which we may admit, but through which (in and by themselves) we can understand nothing. (B724)
While prima facie it seems true that Kant thinks synthetic judgments have no meaning for us without their connection to objects of possible experience, and that therefore only empirical and analytic judgments are capable of truth bearing. However, I think that transcendental idealism in its best form would have to modify that claim. Kant cannot (and I think that given a broad overview of transcendental idealism he does not) mean there is no other legitimate employment of concepts. Recall that, for Kant, meaning is "an ascribed significance." In other words something has meaning only if we ascribe significance to it given the rest of our worldview. (B117). A judgment is meaningful just in case we know what it would be for that judgment to describe accurately what is actually the case. In other words, a judgment is meaningful for us just in case we know its truth conditions. However, knowing the truth conditions for a judgment does not entail immediate knowledge of its truth maker. Moreover, Kant does not require that the relation to the object be immediate. It is possible to know the truth conditions for a judgment without having those conditions involve a direct relationship between the judgment and its object. Those pure synthetic judgments therefore relate, though only mediately, to possible experience, or rather to the possibility of experience; and upon that alone is founded the objective validity of their synthesis. (A157/B196)
The fundamental point of disagreement between Strawson's understanding of the principle of significance and mine is the definition of the phrase 'objects of possible experience'. For Strawson's empiricist read-
36 Kant's Correspondence Theory of Truth ing to be correct an object of possible experience would have to be an object that we could possibly experience directly (as a phenomenon). There are passages in the Critique that seem to indicate that this is what Kant has in mind. "The pure concepts can find no object, and so can acquire no meaning which might yield a concept of some object." (A147/B186) However, there are other passages that imply a different understanding of what Kant would have counted as an object of possible experience. The possibility of experience is, then, what gives objective reality to all our a priori modes of knowledge. Experience, however, rests on the synthetic unity of appearances, that is on a synthesis according to concepts of an object of appearances in general... Experience depends, therefore, upon a priori principles of its form, that is, upon universal rules of synthesis of appearances. Their objective reality, as necessary conditions of experience, and indeed of its very possibility, can always be shown in experience. (A157/B196) Pure a priori concepts . . . cannot indeed contain anything empirical; yet, none the less, they can serve solely as a priori conditions of a possible experience. Upon this ground alone can their objective reality rest. (A95)
I believe that a broader interpretation of what it is to be an object of possible experience better preserves the spirit of transcendental idealism. On this broader interpretation, an object of possible experience can be (1) a phenomenal object, state of affairs or established linguistic/logical structure or (2) the transcendental objects which serves as the ground and ultimately the source of objective reality for the former. If an object of possible experience were restricted to those in the first set, it would make no sense to regard transcendental truth as ontologically prior to empirical truth. If to be true, (that is to have objective validity and objective reality and the appropriate relation between judgment and object), is necessary prior to experience, then it makes no sense to require that the phrase 'object of possible experience' refers only to an object that we could conceivably experience directly, i.e. a phenomenal object. In short, for transcendental idealism, as Kant construes it, to make sense, objects of possible experience must include both objects that it is possible for us to experience and the objects that are necessary as the
The Verificationist View 37 grounds of our possible experience. This is all that we were called upon to establish in the transcendental deduction of the categories, namely, to render comprehensible this relation of understanding to sensibility, and, by means of sensibility, to all objects of experience. The objective validity of the pure a priori concepts is thereby made intelligible, and their origin and truth determined. (A128)
Although it makes sense to speak of a judgment that is meaningful but not capable of being a truth bearer, it does not make sense to claim that a judgment is a truth bearer but is not meaningful. How could a judgment bear truth for us if it were nonsensical? Passages such as this one make no sense if to have meaning qua a relation to an object of possible experience is to be immediately related to the phenomenal. The broader interpretation of an object of possible experience explains the necessity of the categories, and in fact the necessity of transcendental idealism. To construct the most reasonable interpretation of Kant's view, it is necessary to take his broader definition of an object of possible experience as the correct one. For this reason, I argue that the principle of significance that Kant employs does not, and if transcendental idealism is tenable cannot, preclude legitimate, truth bearing judgments that are not immediately connected to empirical objects. For these reasons, I suggest a modification of PS: PS': This is the principle that there can be no meaningful employment of ideas or concepts, nor even any substantive understanding of such ideas or concepts, which does not relate them to empirical or experiential conditions of their application.
PS' is, I think, more faithful to Kant's remarks and intentions in the Critique of Pure Reason. Moreover, it fails to support Strawson's inference that Kant's postulation of the transcendental employment of the categories and space and time fails on Kant's own terms. The apparent failure resulted from Strawson's interpretation of the principle of significance, not the nature of the principle as Kant intended. Given this failure, Strawson falls short in his attempt to show that on the basis of Kant's principle of significance, the only truth conditions which could be appli-
3 8 Kant 's Correspondence Theory of Truth cable to a judgment are those associated with analytic truths and empirical verificationism. The Deduction of the Categories Objection Another of Strawson's objections to the possibility of non-empirical and yet informative truth is his objection to Kant's deduction of the categories from the Table of Judgments. To set the stage for this objection, I will briefly explain how Kant arrives at his Table of Judgments, and, subsequently, the Table of Categories. For Kant, the understanding is "a non-sensible faculty of knowledge." (A68/B92) We know that the only modes of knowledge are intuitions and concepts, and we know that intuition requires sensibility. Therefore, the human understanding must yield knowledge by means of concepts. Concepts rest on functions, a function being "the unity of the act of bringing various representations under one common representation." (A68/B93) Concepts are only useful for the understanding given this function which can only take place by means of judgment. Thus, all acts of the understanding are judgments, and the understanding may therefore be understood as simply the faculty of judgment. As such, the functions of the understanding can be discovered by giving an exhaustive account of the functions of unity in judgments. (B94) Abstracting all content from judgments and considering only the form of understanding can reveal the exhaustive account of the functions of judgment, also called the Table of Judgments. In doing this "we find that the function of thought in judgment can be brought under four heads, each of which contains three moments." (A70/B95) The Table of Judgments, according to Kant, "appears to depart in some, though not in any essential respects, from the technical distinctions ordinarily recognized by logicians." (B96) Kant's Table of Judgments is basically a representation of the connective concepts of formal, what Kant calls general, logic. These are the general forms of intellectual synthesis with all content abstracted. The Table of Judgments can serve as a clue to the discovery of the Table of Categories if it can be shown that these formal connective concepts can have objective validity and objective reality in addition to their formal content. The study of this is transcendental logic which does not
The Verificationist View 39 abstract all content from its concepts—only empirical content. Kant claims that an analysis of the formalized table which reflects the primary ways in which we judge (and thus the way we think about the world) allows us to infer what the pure concepts of the understanding must be in order to serve as the necessary conditions of the possibility of experience. The same understanding, through the same operation by which in concepts, by means of analytical unity, it produced the logical form of judgment, also introduces a transcendental content into its representations, by means of the synthetic unity of the manifold in intuition in general. (B105)
In other words, we are able to deduce the pure concepts of the understanding that are used in our transcendental judgments. The result of this deduction is an exhaustive list of the pure concepts of the understanding, the categories. These concepts, when employed transcendentally, determine the undetermined manifold of sensibility in such a way that experience is made possible for us. The deduction is made possible because the Table of Judgments represents an "exhaustive inventory" (B105) of the powers of the understanding. Because the list of functions in the Table of Judgments is exhaustive, Kant concludes that the necessity of the categories has been demonstrated. Both tables are developed systematically from the principle of the faculty of judgment. Otherwise, the list would be haphazard and inductive and therefore not necessary. (A81) Strawson examines Kant's deduction of the Table of Categories (i.e. the primitive, or non-derivative, a priori concepts of pure understanding) from the table of logical judgments. He performs this examination from a position enlightened by more recent work in logic. Strawson claims that since Kant claims to have found the list of primitive concepts (which he calls the categories), and relegates all other pure concepts derived from the categories to the position of mere 'predicables' his list should be looked at with serious skepticism. (Strawson 1966, 80) Kant claims to have found the one set of determinative concepts that are, (with space and time and the undetermined manifold of empirical sensibility), the conditions for the possibility of any experience. Since he arrives at this list without the advantage of more recent and well-
40
Kant's Correspondence Theory of Truth
developed systems of logic, Strawson claims that there are good reasons to view his list skeptically. However, he focuses his criticisms on Kant's transcendental deduction itself and not on the flaws in the Table of Judgments. It would be pointless to dwell on imperfections of detail in Kant's Table of Judgments. He has made, at least, a list of features for which it is claimed that they are all fundamental to the classification of logical forms of proposition, and there are no others which are. (Strawson 1966, 79)
The strength of Kant's claim requires not only that the category be a possible logical form or feature of experience. Rather, it must be an essential form or feature. In claiming that the categories are an essential form of experience, Kant is claiming that without the categories, one of the necessary resources for experience would be missing. In other words, without the categories there could be no experience whatsoever. [This feature must be] one which exhibits, as no other can, some part of those fundamental and indispensable resources themselves. Only under this condition can we safely even try to derive a category . . . from a logical form or feature of propositions. (Strawson 1966, 80)
Strawson argues that Kant in no way proves that he has derived his categories under these conditions, and thus, his entire Table of Categories is suspect. Not only does Strawson contend that Kant's transcendental deduction of the categories does not work, he also claims that the categories themselves are an unnecessary proposal in the first place. It is true that knowledge (in any sense, at least, in which we are concerned with it) is essentially expressible in propositions. But it does not follow that every form, or every formal feature, of propositions which the logician thinks it worthwhile to distinguish is absolutely necessary in the expression of knowledge. (Strawson 1966, 79)
Strawson points out a number of problems with Kant's transcendental deduction of the categories. First, since Kant claims that the list in the Table of Judgments and the Table of Categories is exhaustive, he has precluded any justified inference from formal logic to transcendental
The Verificationist View 41 logic. A logician can take a "certain indispensable minimum equipment of notions" and "distinguish indefinitely many forms of propositions, all belonging to formal logic." (Strawson 1966, 79) That being the case, if there is a category for every form of proposition, there must be indefinitely many categories. Therefore, Kant cannot have provided an exhaustive list of categories in his Table of Categories. Second, argues Strawson, a logician chooses a set of primitive concepts. Therefore it is not be the case that what is represented systematically is non-arbitrary, as Kant suggests (A81). Given that the Table of Judgments is arbitrary (and therefore not necessary), the same can be said for the derived Table of Categories. Strawson thinks Kant's view is an improvement over Berkeley's precisely because he does attempt to account for the objective unity of our experience. However, sense can easily be made of the unity of experience without an appeal to the categories, i.e. by an appeal to transcendental realism. Our experience is unified because it is the experience of an objectively set and ordered world. Because Kant shows neither the completeness nor the necessity of his Table of Categories, he fails to demonstrate the general necessity of transcendental idealism. If there is no reason to suppose that there is transcendental logic, then there is no need to suppose that there are transcendental judgments that are truth bearers. The results of the appeal to formal logic are not merely meager. Their meagerness is such as to render almost pointless any critical consideration of the detail of Kant's derivation of the categories from the Table of Judgments. (Strawson 1966, 82)
In fact, Kant's deduction of the categories is only useful in so far as it points us toward progress to the extent that we are prepared to fill out the questions by assuming "all those theses of the Transcendental Aesthetic which can be detached from the doctrine of transcendental idealism." (Strawson 1966, 82) These criticisms are the basis of Strawson's view that, for Kant, the only legitimate truth bearers are analytic judgments and empirical judgments. The truth conditions for the latter are that they appropriately correspond to their objects, and the truth conditions for the former are that they are consistent with the rules of language (formal and informal). IF
42 Kant's Correspondence Theory of Truth Kant fails to establish the necessity of the Table of Categories as conditions for the possibility of any experience, thus the postulation of truth conditions for transcendental judgments is pointless and meaningless. Reply to the Deduction of the Categories Objection I will provide what I take to be good reasons for accepting Kant's deduction of the categories. Moreover, I will show that even if Strawson's objection to the derivation of the categories is granted, it is not therefore necessary to concede the project of transcendental idealism as a whole. Most importantly, if this is not conceded, then neither is the grounding of synthetic a priori judgments, and if this grounding is not relinquished, there is no longer good reason to deny the possibility of non-empirical truths. So long as Kant can establish this possibility, there is no strong support for the claim that his view is for all real intents and purposes committed to either a transcendental realism or an empirical idealism. Strawson's main evidence for the prudence of skepticism concerning Kant's Table of Categories is that for any logician, the choice of a set of primary concepts is arbitrary. Never the less, Strawson admits that Kant "has made, at least, a list of features for which it is claimed that they are all fundamental to the classification of logical forms of proposition, and there are no others which are." (Strawson 1966, 79) This is, I think, sufficient to ground Kant's deduction. Kant claims to have isolated the basic forms of judgment. Since to employ the understanding is to judge, then to isolate the basic forms of judgment is also to isolate the way in which the understanding is employed. Once we discern how the understanding is employed in abstraction of all content, it is possible to deduce how it would have to be employed in abstraction of empirical content alone. So long as Kant has established the fundamental forms of judgment, and there are no others that are, he has provided an adequate basis for his deduction. It is only the fundamental forms of which there are not others that Kant wants to argue are necessarily constitutive of experience. The fact that there may be an infinite list of derivative categories is irrelevant. The derivative categories are not necessary conditions for the possibility of experience in general.
The Verificationist View 43 Even if we grant Strawson's argument against the deduction of the categories, we need not grant his inference to the implied implausibility of transcendental idealism. Even if the particulars of Kant's account are incorrect (which, given the consistency of the ways in which we judge and thereby understand seems unlikely), it need not be the case that his structure is flawed as well. The most important part of Kant's claim is that the world as we know it must conform to the nature of our understanding. This can be granted even if it is not granted that the exact nature of the understanding has been precisely determined. Let us imagine that I know of a certain building with a certain sized door and no windows. I also know that the building was completed with nothing inside it. I can derive from these premises that there is no unassembled, non-squishy object inside the building that is larger than the doorway. I may err in my assessment of what those objects inside are, but this does not invalidate my claim as to the structure of the objects in general. Similarly, even if Kant errs in his assessment of what the particular categories are, that does not thereby invalidate the basic claim of transcendental idealism in general. Furthermore, contrary to what Strawson suggests, the importance of the transcendental idealism with which we are left need in no way be seen as a "meager result we might have attained directly from the original distinction between intuitions and concepts, sensibility and understanding." (Strawson 1966, 82) On the contrary, the fundamentally important implications of transcendental idealism stand independently of Kant's assessment of the contents of the categories. Even if the minimal conclusion that categories exist and shape our reality is granted, then Kant provides us with justification for rejecting arguments for the possibility of doing special metaphysics. Given the knowledge that the structure of our understanding shapes the objects of our experience, there is good reason to refrain from making inferences about the way things are in themselves on the basis of the way they seem to us. The seemingly uninteresting conclusion that we must shape our experience according to categories has tremendous import. If granted, then there is good reason to demand agnosticism in reference to judgments concerning things as they are outside the parameters of our experi-
44 Kant's Correspondence Theory of Truth ence, and thus good reason to reject scientific realist accounts like Strawson's. A reconstruction of Strawson's wrong categories argument given these considerations would have to be considerably different. Although (1) it seems that there are a great number, if not infinitely many, forms of judgment which a logician can imagine, formal logic demonstrates that there is a set of primary judgments which hold for all rational individuals. (2) Although there could not be infinitely many conditions for the possibility of any experience (otherwise we would never get to possible experience), there can be a set of primary conditions which, with their derivatives, could be satisfied as the minimal conditions for the possibility of any experience. (3) Therefore, a well-constructed table of judgments may serve as a sufficient basis for deriving the categories. I think I have demonstrated that there is good reason to be suspicious of Strawson's objections to Kant's transcendental ideality thesis. And given this suspicion, it is reasonable to leave open a possibility which Strawson's objections were designed to close. The possibility to which I am referring is the possibility of non-empirical truth. Without establishing that the transcendental idealism/empirical realism dichotomy fails, Strawson cannot prohibit the possibility of truth conditions for synthetic a priori judgments. Strawson is not alone in rejecting Kant's postulation of synthetic a priori judgments as truth bearers. Carl Posy interprets Kant as having a merely empiricist conception of synthetic truth. He argues for an assertabilist conception of truth in which what it is for a judgment to be true is just that the subject be justified in asserting it. Posy sees this view as a further development of Strawson's view, but as one that better preserves what he takes to be the spirit of transcendental idealism. Carl Posy In "Transcendental Idealism and Causality: An Interpretation of Kant's Argument in the Second Analogy" (Posy, 1984, 20-41), Posy presents what is basically a modified Strawsonian interpretation of Kant's conception of truth. Whereas Strawson contends that the transcendental ideality thesis fails, Posy identifies transcendental idealism with a "special sort of verificationism." (Posy 1984, 21) Posy contrasts his special verificationism with classical verification of the variety espoused by Straw-
The Verificationist View 45 son. Realizing that verificationism is strictly a theory of justification rather than truth, Posy clarifies his definition of the standard view by adding "what I am calling standard verificationism combines a testability criterion of significance with an old-fashioned correspondence theory of truth." (Posy 1984, 22) Transcendental Idealism as Radical Verificationism The first modification Posy suggests for the standard view is to focus directly on truth rather than truth as an aspect of meaning. According to this version of verificationism, transcendental idealism is a thesis that simply identifies truth and verifiability. "That is: a judgment will be true if and only if there is sufficient verifying evidence to assert it (or there will demonstrably be such evidence). Similarly, it will be false if its denial is thus verifiable." (Posy 1984, 23) Posy calls this view 'radical verificationism.' Given radical verificationism, truth is always relative to an evidential situation, and those judgments verified by all possible evidential situations are said to be true a priori. There are, however, certain problems inherent in identifying transcendental idealism and radical verificationism. If transcendental idealism were in fact identical with radical verificationism, "it turns out that the criterion of truth for an objective judgment must invoke not only the actual state of evidence for that judgment, but also (at least implicitly) a whole host of potential evidential situations that go beyond the actual epistemic state." (Posy 1984, 30) For example, if it is objectively true that Alex is a black cat, then it must be the case that any number of imaginable epistemic situations, varying in time and perspective, would continue to confirm that the subject 'Alex' belongs to the property class 'black things'. Otherwise the claim will only really be, "Alex has as yet failed to appear non-blackly." Rather than being an objective claim, this would be merely a subjective claim about perceptions. However, given Kant's causal argument in the Second Analogy, radical verificationism has an advantage over classical verificationism. In radical verificationism, it is possible to maintain that any judgment about events "already contains all the facts about the past," and given the na-
46
Kant's Correspondence Theory of Truth
ture of determinism, contains facts about the future as well. (Posy 1984, 35) Remember that for Kant, the unity of our experience is, in part, a function of the determinism implied by the necessary application of the category of causation to all possible experience. Because we know that determinism is true for the world of experience, we can know that for any event, that event was caused by a past event and will cause a future one. Moreover, because determinism is true, we can know that certain events will always occur subsequently and in a causal relationship. This knowledge forms the basis of our objective scientific knowledge. For example, the judgment "A causes B" is a fact, i.e. is objectively true, so long as, given determinism, A always has caused B, and always will. It is the necessary deterministic property of our unified experience which gives science its predictive success. This advantage, however, does not come without a price. First, truth must be ultimately identified with phenomenal verification. Thus, any judgment that cannot be phenomenally verified cannot have a truth condition or be meaningful. In other words, Kant would be inextricably bound to phenomenalism. As Posy points out, "Kant certainly never intended to declare such judgments devoid of meaning." (Posy 1984, 37) This leads Posy to introduce a further modification of classical verificationism. He suggests the need to identify transcendental idealism not with radical phenomenal verificationism, but with semantic verificationism. (Posy 1984, 37) Transcendental Idealism as Semantic Verificationism Posy claims that transcendental idealism may be identified with semantic verificationism "without entailing phenomenalism and without denying significance to ethical, metaphysical or theological discourse." (Posy 1984, 37) For the purpose of distinguishing this assertabilism from the two previously discussed versions of verificationism, Posy explicates the distinction I addressed earlier concerning truth conditions and meaning. Posy treats meaning and truth conditions, as "two distinct components of the 'meaning' of a judgment." These are: "(a) the sense in which the judgment purports to describe the world; and (b) the recognition of
The Verificationist View Al what would count as a verification of the judgment" (37) He terms component (a) truth conditions, and component (b) 'verifiability' or 'assertability' conditions. Posy gives no unique interpretation of truth conditions, and defines assertability conditions as those conditions under which the understanding outlines for us the state(s) of affairs under which an individual would be justified in asserting the judgment as knowledge. (Posy 1984, 38) Posy maintains that to understand any judgment, one must understand both its truth conditions and assertability conditions. For empirical judgments, transcendental idealism is the general claim that verification conditions are the "dominant semantic factor in assessing our judgments", but not the only factor. Whereas, for the realist, semantic success is identical with correspondence to the world, for the idealist semantic success is identical with warranted assertability. This passage, I think, encapsulates Posy's assertabilist construal of transcendental idealism, and in so doing, captures what he takes to be Kant's theories of empirical and transcendental truth. The idealist perfectly well admits that such judgments may be true in the standard (correspondence) way, but he or she simply holds that fact to be irrelevant to the ultimate semantic success or correctness of the judgments. The realist, in turn, never denies that part of what we understand about these judgments is their associated verification conditions; but for him these conditions have no bearing on the semantic success of the judgments. That success simply amounts to correspondence with the world. Since "truth" is almost universally associated with the realistic notion of correctness, it would be better to use a phrase like "semantic success" as a neutral term to cover truth for the realist and warranted assertability for the idealist. (Posy 1984, 38)
Posy distinguishes truth as correctness (the traditional notion of truth) from truth as warranted assertability. For the realist, truths as correctness and truth as warranted assertability are identical. One is warranted in asserting the judgment "snow is white" just in case a certain state of affairs obtains in the transcendentally real world, namely, that the object snow has the property of whiteness. The realist would likely concede that the judgment "snow is white" would not be meaningful for one for whom the verification conditions were inaccessible (as they might be for someone blind since birth).
48 Kant's Correspondence Theory of Truth Moreover, such a judgments may still have no meaning for the person who can access the verification condition, yet does not have the tools to interpret what he has accessed. Imagine that the same individual, blind since birth, undergoes an operation that successfully gives him sight. Initially, before he learns to interpret visual data, and attempt to verify that snow is white would remain meaningless for him. However, the truth of the judgment, both as correctness and as warranted assertability, is independent of these conditions of verifiability. For the realist, truth and verifiability are distinct. Idealists too hold that truth as correctness and truth as warranted assertability are identical. Posy maintains that an idealist may admit that in a trivial sense, truth is a matter of correspondence, i.e. that "snow is white" is true just in case snow is white. However, for idealists, truth in the objective sense of truth as correctness is ultimately identical with warranted assertability. A judgment is verifiable just in case the conditions of its warranted assertability are not known. For any judgment for which the warranted assertability conditions are not satisfiable, either in principle or in practice, that judgment cannot be true. Posy concludes that, as an idealist, Kant is committed to the identification of transcendental idealism and assertabilism. This view entails that 'truth' has a different meaning for empirical and transcendental judgments. Empirically, there are two senses in which a judgment can be true. In the trivial sense, a judgment is true just in case it corresponds to its object. Empirical judgments are ultimately true, however, only in so far as we are warranted in asserting them. That is, they are true just in case they are verifiable for us. For transcendental judgments, there is no truth in the trivial sense. If these judgments are true, then they must be necessarily true and their truth must be a condition for the possibility of experience. The relation of transcendental objects to their objects is one of determination rather than correspondence. True just means "in satisfaction of assertability conditions." Thus, we know that the principle of causality is necessarily 'true', because we could not make semantically successful judgments about successions and events if we did not assume it to be true. (Posy 1984, 39) Transcendental judgments are true in virtue of the fact that they make it possible for us to meet the warranted assertability conditions of empirical
The Verificatiortist View 49 judgments. One of the advantages of assertabilism, according to Posy, is that it leaves itself open to several possibilities concerning the relationship of our judgments to a mind-independent reality. He points out that his view in no way requires that the object-classes of our judgments conform to external reality, but he does not deny that conformity either. He writes, I did not deny that our judgments purport to be true in virtue of their correspondence to a "transcendentally real" world; nor did I deny that recognizing the conditions for that truth is part of understanding these judgments. That is why I was careful not to say that our singular concepts or terms refer to the mind-bound object classes. For our singular terms purport to refer to something else all together, and they might conceivably succeed in that reference... As to the existence of mind-independent objects, it [assertabilism] is simply agnosticism, or better, silence. (Posy 1984, 39)
Posy's point here is that the transcendental idealist intends for her judgments to correspond to things in themselves. Our particular (i.e. empirical) judgments are not intended to refer to phenomenal objects whose characteristics are in anyway bound to the minds of finite subjects. We think we are describing the world as it is in itself, we just accept the fact that we may be wrong and remain agnostic as to whether we are successful in describing things as they actually are. Posy then indicates that there is another class of judgments for which we may have truth qua semantic success; namely the class of thing-initself type judgments. This class of judgments includes ethical and theological judgments, and the class is such that we are incapable of verification. For these kinds of judgments, Posy claims that the idealist would agree with the realist that evidence is irrelevant to semantic success, because semantic success here just means "old-fashioned truth" (presumably correspondence to a mind-independent world). (Posy 1984, 39) I think what Posy has in mind is this, when making judgments about things in themselves, the individual who is judging is, in effect, assuming realism. Not only is realism assumed, but it seems transcendental realism is assumed. Once realism is assumed, we return to the conception in which the truth or falsity of a judgment (and thus its conditions of warranted assertability) is independent of the judgment's verifiability for us. Since Kant, according to Posy, "never intended to declare such judg-
50 Kant's Correspondence Theory of Truth ments devoid of meaning," (Posy 1984, 37) he must be willing to allow a realist account of truth for judgments about things in themselves. Since, according to Posy's transcendental idealism as assertabilism view, there is more to meaning than knowledge, "it is perfectly possible for such judgments to be meaningful and true." (Posy 1984, 39) In other words, we know what the truth conditions are for judgments about things in themselves, we simply realize that we are unable to verify whether or not those conditions have been met. Take the judgment "God is eternal" as an example. According to Posy's interpretation, that judgment is meaningful for Kant. It is meaningful because we know its truth conditions, what it would be for the judgment to be correct. "God is eternal" is true just in case the transcendentally real entity, God, possesses the quality of being eternal. The awareness of these truth conditions makes the judgment meaningful for us, even though we realize that we are unable to verify whether those conditions are met. Posy argues that assertabilism has a number of advantages. First, unlike Strawson's verificationism, it does not leave Kant committed to phenomenalism because there is truth, of a sort, which applies to nonempirical judgments. Second, it explains the necessity of our synthetic a priori judgments (they are necessary for our semantic success). Finally, identifying transcendental idealism with assertabilism allows Kant the latitude to declare judgments about things in themselves meaningful. There are, however, some difficulties that undermine and ultimately overpower the view's apparent advantages. Textual Difficulties with Assertabilism Foremost among the difficulties with assertabilism, is its complete lack of support from and indeed contradiction with the text of the first Critique. First, as Posy admits, Kant in no way addresses linguistic criteria for truth or meaning. (Posy 1984, 40) There simply is no evidence in either edition of the Critique, or in Kant's other essays on theoretical philosophy that demonstrates that he thought there was any connection between our semantic success and truth. Posy indicates that since truth is often associated with correctness, the idealist is better served by the notion of semantic success. Kant, however, does speak of truth in terms of correctness.
The Verificationist View 51 Kant's distinction between knowledge, belief, and opinion specifically goes against Posy's interpretation. Kant asserts that there are judgments which have meaning for us (which I presume requires semantic success), yet are none the less not true. These are false opinions and false beliefs. Suppose I believe that it is aerodynamically impossible for bumblebees to fly. This judgment could grant me semantic success, as it is coherent with prior theories of aerodynamics. Nevertheless, I do not have knowledge according to Kant. I have a false belief. I know my belief is false because when I look out into the world of common experience I see that bumblebees do in fact manage to lug their chubby behinds from flower to flower despite their small wing size. It is this incorrect judgment that leads me to a new piece of knowledge, namely that my theory of aerodynamics is false because it makes false predictions. For Kant, we only have knowledge when we have true beliefs, and to have a true belief entails, for Kant, the correctness of our judgments. (A822-3/B850-51) Perhaps Posy means that the idealist is better served by limiting nonempirical judgments to semantic success rather than correctness. However, this too is textually problematic. For both empirical and transcendental judgments, Kant describes objective validity as a necessary but not sufficient condition for the truth of a judgment. (A125, A788/B816) Specifically, objective validity is the relation of a judgment to its object in such a way that the judgment is capable of being either true or false. To have objective validity is to be a truth bearer. So, objective validity, plus the satisfaction of the relevant truth conditions (appropriate relationship etc.), is the definition of truth. Kant characterizes objective validity as "real possibility". (Bxxvii) Real possibility is, I think, far more indicative of a criterion of correctness than a criterion of semantic success. Second, Posy's characterization of truth for idealists versus realists is suspect given Kant's explication of transcendental idealism. Posy's interpretation of the realist / idealist distinction wrongly assumes that because Kant posits idealism, he must also posit a purely epistemic conception of truth. In other words, because Kant argues for idealism, his conception of truth must make the relevant condition for truth reliant on the epistemic situation of the subject. This position is contrary to Kant's conception of truth as he describes it in the Critique.
52 Kant's Correspondence Theory of Truth Kant's principle of significance entails that, for a judgment to be meaningful for us, it must have both sense and reference. We must understand the judgment's truth conditions, i.e. what would be for what is assumed in the judgment to be the case. This is the judgment's sense. The judgment must also have a reference, and contrary to what Posy suggests, our empirical judgments do not purport to refer to transcendentally real objects. That would be impossible since, according to Kant, transcendentally real objects have no properties that are knowable by finite subjects. "The true correlate of sensibility, the thing in itself, is not known, and cannot be known, through these representations; and in experience no question is ever asked in regard to it." (A29) Truth in Assertabilism and Synthetic a priori Judgments Even if the textual conflicts are ignored or explained, there are substantive problems for reconciling Kant's intended project and assertabilism. At the foundation of transcendental idealism is the idea that the understanding employs functions transcendentally, and the function of the understanding in this way is a necessary condition for the possibility of any experience. As necessary conditions for experience, transcendental judgments must be ontologically prior to experience. There can be no experience until that experience has been unified by the transcendental function of the understanding. Moreover, if there were no finite subjects to employ the understanding in this way, there would be no objects with knowable properties. Consequently, any interpretation of transcendental idealism that could not account for the fact that the application of the categories to the undetermined manifold of sensibility precedes experience would be less than optimal, if not flawed. Semantic success, i.e. the fulfillment of assertability conditions, entails, of course, the employment of language. The use of language is an empirical employment of the understanding. This is evidenced by the fact that language is taught and learned over time. Prior to an adequate amount of experience, we are unable to employ language. (I am distinguishing language from other forms of communication. It is certainly the case that newborns communicate their needs etc, but they do not do so
The Verificationist View 53 through language in any meaningful sense.) Since language use is empirical, and since nothing empirical could be said to be prior to any possible experience, then language use qua fulfillment of assertability conditions could not apply to transcendental judgments. Kant not only maintains that synthetic a priori judgments are true, he maintains that they are necessarily true. So, a view like assertabilism, which in essence isolates such judgments from being true, is certainly inadequate. Perhaps, however, Posy is claiming that the very employment of concepts is a linguistic activity. If concepts are linguistic entities, then their employment is surely a linguistic activity. Given this definition of concepts, characterizing truth as semantic success for transcendental judgments seems to make sense. Unfortunately for assertabilism, Kantian concepts are not linguistic. Kant speaks of concepts as things that may or may not be capable of being captured by language. Despite the great wealth of our languages, the thinker often finds himself at a loss for the expression which exactly fits his concept, and for want of which he is unable to be really intelligible to others . . . . It is advisable to look about in a dead language, to see whether the concept and its appropriate expression are not already there provided. (B369)
Such passages demonstrate that, for Kant, concepts exist apart from their linguistic expression. So, although Kant may think language is necessary for the communication and connection of our judgments, he does not think that employing concepts is, in itself, linguistic. Therefore, given that synthetic a priori judgments necessarily precede experience, and judgment are not themselves linguistic, assertabilism could not accurately describe what Kant intends as truth for synthetic a priori judgments. The Meaningfulness of Judgments Concerning Things In Themselves Posy's view also contradicts Kant's conception of transcendental idealism in regard to the meaningfulness of judgments about things in themselves. In particular, Posy cites judgments about ethical and theological entities as being meaningful. Unless we equivocate on the term
54 Kant's Correspondence Theory of Truth meaning, this suggestion is directly contradicted by Kant's remarks in the first and second Critiques. Let us consider ethical judgments. A distinction needs to be made here between necessary moral truths, and the truth of particular moral judgments. Particular moral judgments such as, "killing that mime would be wrong", are judgments whose objects are appearances. As such, they are not judgments about things as they are in themselves, and their meaningfulness to us does not support Posy's contention about the possibility of our having meaningful judgments about things in themselves. As for possible judgments whose objects are things in themselves, Posy also fails to show that these have any meaning for us beyond problematic meaning. Recall that a problematic judgment is only meaningful for us to the extent that we could assume its truth for the moment. Posy seems to want to attribute assertoric meaning to propositions about things in themselves, and this is, for Kant, impossible. Concerning the ethical judgments, as imperatives, they are commands, not declarative judgments. If there is meaning in the semantic success of a command, it is not the same sort of meaning which attaches to declarative theoretical judgments. The fulfillment of assertability conditions for a command (i.e. being justified in uttering a command) seems in no way substantive or equivalent to truth analogous to the truth of empirical theoretical judgments. Posy's claim concerning the meaningfulness of theological judgments whose objects are things in themselves is completely unsubstantiated by the text of the first Critique. Kant addresses this issue directly. If. . . the question be whether this being [the ground of our appearances] is substance, of the greatest reality, necessary, etc., we reply that this question is entirely without meaning. For all categories through which we can attempt to form a concept of such an object allow only of empirical employment, and have no meaning whatsoever when not applied to objects of possible experience. (A696/B724)
Assertabilism and Agnosticism Finally, Posy maintains that it is important to neither assert nor deny the correspondence of our empirical judgments to things as they are in themselves. We must leave open this possibility. (Posy 1984, 39) Al-
The Verificationist View 55 though Kant does advocate agnosticism concerning things in themselves, this is precisely not the sort of agnosticism he has in mind. It is true that Kant does not deny the existence of things in themselves. In fact, he affirms that they do exist as the "true correlate of sensibility" (A29). To this extent, we must remain agnostic about the relationship between our reality and things as they are in themselves. That is not to say, however, that the objects to which our empirical judgments correspond may be things in themselves. This, in fact, is precisely what Kant wants to deny. The objects of our empirical judgments are and must be appearances. (Bxvi-xvii) To allow otherwise would be to allow special metaphysics. Thus the agnosticism advocated by Posy is not consistent with Kantian transcendental idealism. I think I have demonstrated that Posy's transcendental idealism is not a viable interpretation of truth for Kant's theoretical philosophy. Assertabilism lacks textual support, as well as consistency with the major assumptions of transcendental idealism as Kant construes it. In short, although it may be possible to construct a version of transcendental idealism for which truth was equivalent to assertabilism, there is no evidence that it would resemble Kantian transcendental idealism in any way. Summary Neither Strawson's phenomenal verification view, nor Posy's assertabilist view succeeds in capturing truth as Kant intends it for his theoretical philosophy. Strawson would concede this and maintain that this incongruity is acceptable, given that the central project of transcendental idealism fails. I have shown, however, his arguments for why transcendental idealism fails are unconvincing. Posy, on the other hand, thinks his interpretation is sympathetic to transcendental idealism. Posy cites the advantages of assertabilism for consistency with Kant's argument in the Second Analogy, and for its ability to account for the 'truth' of synthetic a priori judgments. As I have shown, however, the view is far too inconsistent with several of Kant's most basic assumptions. Moreover, Posy's view lacks adequate textual support.
56 Kant's Correspondence Theory of Truth Both authors present viable interpretations of truth. These interpretations are not, however, in any way suitable for Kantian transcendental idealism. It is the aim of this paper to find the best interpretation of truth for Kant, and the best interpretation would have to be both reasonable and consistent with the basic tenets of transcendental idealism. It may be reasonably concluded, then, that the best interpretation of truth for Kant's theoretical philosophy is not to be found in the various varieties of verificationism considered in this chapter.
Chapter Four The Constructivist View An Introduction to Constructivism A second set of interpretations of truth for the Critique is somewhat more compatible with transcendental idealism. These interpretations do not deny the consistency of transcendental idealism as a theory, but they do interpret the theory empirically. Such views are neither coherence nor correspondence views in the traditional sense. The type of view in question is constructivism, and a version of it is (at least implicitly) attributed to Kant by several authors, including Hillary Putnam and Jaakko Hintikka. They have suggested that Kant's view of truth is a relativistic constructivism. A relativistic constructivist view is one that combines a constructivist account of meaning with intuitionism. In a constructivist account of meaning, the meaning of a judgment consists in the stipulation of what is to count as proof for it. In other words, to understand a judgment is to know, for any construction, if that construction could serve as proof of the truth of that judgment. (Dummett 1978, 153) There are varieties of constructivism that are simply varieties of standard coherence theories. Such theories hold that the essence of truth is the possession of a proof. I call Putnam and Hintikka's view relativistic constructivism because it combines a variety of constructivism with a variety of intuitionism. Intuitionism is the claim that entities exist "when and only when they are proved to exist and that . . . states of affairs obtain when, only when, and because they are proved to obtain." (Kirkham 1995, 187) When combined with Kant's account of the dependence of the knowable properties of empirical objects on the finite subject, the relativism of the view is made clear. For Kant, we construct the objects of experience by determining the manifold of undetermined empirical sensibility according to the categories and the pure intuitions of space and time. There are no knowable objects until we have constructed them for ourselves. In a concrete sense, we stipulate what will serve as the truth-makers for our empirical judgments. The truth-makers for our empirical judgments are empirical objects, and we stipulate them by determining their knowable formal prop-
58 Kant's Correspondence Theory of Truth erties in the original unity of apperception. In other words, our intuition of undetermined empirical intuitions, subsumed under the pure concepts of the understanding, creates the objects that will demonstrate the proof of our empirical judgments. For analytic judgments, we also construct the truth-maker when we stipulate what will count as proof for our judgments. Since the truthmaker for those judgments is the structure of our formal and informal languages. Again, these are structures that do not exist until we create them, and in constructing them, we construct that which will serve as the truth-maker for our judgments. The status of transcendental judgments is somewhat problematic for this view, and the authors do not agree on precisely what that status is. For these reasons, I will address the exposition of this status in the course of reconstructing the authors' views. It is important to note that both constructivism and intuitionism are usually taken to be theories about mathematics. I am using them here because I think they are closest to what is intended by both Putnam and Hintikka. In the case of Putnam, he calls his account of truth for Kant an 'internal realist' account. However, by internal realism, he means that reality is internal to the finite subject, i.e. that he constructs the objects that are the truth-makers for his empirical and analytic judgments. Putnam According to Putnam, all truth for Kant is completely relative to finite minds. This dependence is complete because, as an extreme version of constructivism implies, the world is completely our construction. There is no degree of objectivity in truth to the extent that there is a dependence on anything outside the mind. Putnam interprets Kant as an idealist in a strong sense, because he thinks Kant must deny that things in themselves have any substantive role whatever in theoretical knowledge. This assumption is predicated on Kant's claim that there can be no knowledge beyond the realm of human experience. We can know nothing outside of possible experience, and things as they are in themselves are beyond the parameters of that possible experience. Therefore, our knowledge cannot be dependent in any way on things as they are in themselves. Putnam writes,
The Constructivist View 59 As I have said, the only answer that one can extract from Kant's writings is this: a piece of knowledge (i.e. a 'true statement') is a statement that a rational being would accept on sufficient experience of the kind that it is actually possible for beings with our nature to have. 'Truth' in any other sense is inaccessible to us and inconceivable by us. (Putnam 1981, 64)
In this passage, Putnam establishes a number of points relevant to his interpretation of the Kantian conception of truth. First, he concedes that he intends his view to be an interpretation and not a reconstruction of Kant's remarks in the Critique. (As he correctly points out, one must extract a substantive conception of truth from Kant's writings.) However, he does argue that the conception of truth he describes is a necessary implication of transcendental idealism as Kant construes it. Second, Putnam establishes that he thinks Kant is committed to a variety of what I call relativistic constructivism. A statement is only true if it can be verified by our possible experience. Thus for there to be truth, there must be objects of possible experience. For there to be objects of possible experience that can serve as the truth-makers for our judgments, we must construct those objects for ourselves through the transcendental employment of the understanding. Therefore, the truth of our empirical judgments is a function of their 'goodness of fit' with the appearances we have constructed. In other words, the essence of truth is a variety of coherence. The variety of coherence Putnam attributes to Kant is a coherencebased version of assertabilism. Putnam maintains that we should read Kant as having rejected the correspondence theory of truth "and as saying that the only sort of truth we can have an idea of, or use for, is assertability (by creatures with our rational natures) under optimal conditions (as determined by our sensible natures). Truth becomes a radically epistemic notion." (Putnam 1983, 210) A First Approximation to the Correct Interpretation of Kant's Conception of Truth To understand Kant's positive account of truth, Putnam recommends that we first keep in mind his negative account. Kant explicitly rejects Berkeley's idealism and the similitude theory of reference that it implies. According to a similitude theory of reference, our ideas of objects are
60 Kant's Correspondence Theory of Truth copies of the mind-independent things that cause them. (This mindsindependence is relative of course, because for Berkeley what is independent from our minds is present in the mind of God.) (Putnam 1981, 60) Putnam suggests that a "first approximation" of Kant's conception of truth may be grasped by assuming that he believes that what Locke says about secondary qualities is true of all qualities. From the assumption that all qualities are secondary, it follows that whatever we can meaningfully say about an object is said about that object as it appears to us—not about that object as it is in itself. It also follows from this reading of Kant that there can be no 'similitude' "between our idea of an object and whatever mind-independent reality may be ultimately responsible for our experience ofthat object." (Putnam 1981, 61) In other words, for Kant, there can be no similarity between things as we perceive them and things as they are in themselves. There can be no such similitude, because there can be no isomorphism between appearances and things in themselves. As Putnam correctly asserts, judgments about ordinary empirical objects are not judgments about things in themselves. They are judgments about things as they appear to us. This holds for objects of inner sense (which Putnam calls sensations) and external objects. Putnam stresses that it is important to realize that our sensations (red, hot, sour, etc.) are no more transcendentally real than spatially determined objects of experience. Sensations are, on Putnam's interpretation, "no more and no less directly knowable than so-called 'external' objects." (Putnam 1981, 62-63) To recapitulate what has been covered, according to a first approximation of Kant's conception of truth, there is no distinction between primary and secondary qualities. Because there is no such distinction, there is no reason to distinguish between sensations and external objects. It is the view that "All properties are secondary (i.e. all properties are Powers)." (Putnam 1981, 63) That is to say, both the material and formal properties of appearances are dependent on intuiting subjects. Both kinds of empirical objects serve as transcendentally ideal referents for our judgments, and in both cases the intuiting subject constructs the referent for his judgments.
The Constructivist View 61 Putnam argues that this first approximation of Kant's conception of truth needs to be modified because it suggests that different properties are being attributed to the same noumenal object. On such a view, to say that Hugo is a black cat is to suggest that there is a noumenal object with the power (disposition) to appear to me both blackly and catly. This implies that for each empirical object, there is a single phenomenal object with the power to produce that empirical object in me. Kant (as was argued in chapter two) denies the possibility of this sort of isomorphism. Putnam takes Kant's rejection of isomorphism between phenomenal and noumenal objects as "the point at which he all but says he is giving up the correspondence theory of truth." In other words, Putnam argues that isomorphism is a necessary condition for a correspondence theory of truth. The Correct Interpretation of Kant's Conception of Truth Putnam argues that the first approximation of Kant's conception of truth must be modified to transfer the power to produce empirical objects for us from individuated noumenal objects to the noumenal world as a whole. But the Power is ascribed to the whole noumenal world; you must not think that because there are chairs and horses and sensations in our representation, that there are correspondingly noumenal chairs and noumenal horses and noumenal sensations. There is not even a one-to-one correspondence between things-forus and things in themselves. (Putnam 1981, 63)
Putnam reiterates that Kant's denial of isomorphism "means that there is no correspondence theory of truth in his philosophy." "Truth is ultimate goodness of fit." (Putnam 1981, 64) Putnam argues that neither the sensible or intellectual properties of experience are caused by noumena. Putnam argues that Kant's denial of our access to noumenal objects is tantamount to a denial of the correspondence theory of truth. According to Putnam, if we are to specify a truth-making correspondence between our judgments and their objects, then we must either have direct access to noumenal objects, or there must be an isomorphism between appearances and things in themselves (noumenal objects). Basically, Putnam argues that truth in any non-trivial way requires structural
62 Kant's Correspondence Theory of Truth isomorphism between our beliefs and a mind-independent reality. He rejects both possible modes of specifying correspondence as inconsistent with transcendental idealism and concludes that Kant's conception of empirical truth must not be a correspondence conception. Putnam's rejection of the first mode of specifying correspondence is predicated on Kant's denial of our access to things as they are in themselves. To specify (1) a set of judgments (2) a set of states of affairs that obtain (3) and a truth-making correspondence relationship between the two, one may maintain that we have direct access to things as they are in themselves. (Putnam 1983, 206-207) A useful theory of truth will make the truth conditions of our judgments verifiable. For truth conditions to be verifiable, we must have direct access to the truth-maker for our judgments. According to Putnam, the necessary access would be access to things as they are in themselves. Having this access would sufficiently specify the truth of our judgments because there would be one set of judgments (true judgments), which contains relationships that mirror relationships that obtain in the world as it is in itself. (Putnam 1983, 207) Noumenally, there is one set of things and relationships which obtain, and thus, there could only be one set of judgments which corresponds to the nature of those things and their relationships. However, as Putnam points out, we do not have direct access to noumenal objects. Therefore, if his analysis of the necessary nature of a correspondence theory of truth is accurate, then the first possibility for a correspondence account of truth for Kant has been exhausted. In short, if a correspondence theory requires direct access to things in themselves, then Kant could not subscribe to a correspondence conception of truth. If Putnam is right, then the only other tenable interpretation of a correspondence theory of truth would entail an isomorphism between things as intuited by us (appearances) and things in themselves. One must "postulate a built-in structure of the world, a set of essences, and say that this structure itself singles out one correspondence between signs and their objects." (Putnam 1983, 225) In other words, there can only be essences built into the empirical world if there are isomorphic essences that obtain in the noumenal world. In addition, there must be a one to one correspondence between appear-
The Constructivist View 63 ances and noumena if there is a noumenal essence that is isomorphic with each empirical essence. Since Kant denies that there is a one to one correspondence between things and themselves and appearances (because appearances are individuated according to our spatial and temporal determinations), he also denies this second possibility for a correspondence theory of truth. With both possibilities excluded, we must conclude that the Kantian conception of truth is something other than correspondence. In standard form, I take Putnam's argument to be as follows. (PI) If Kant subscribes to a correspondence conception of truth then he maintains that either we have immediate access to things as they are in themselves through some perceptive faculty or we have mediate access to things as they are in themselves through the isomorphism between appearances and things in themselves. (P2) According to Kant, it is not the case that we have any perceptual or intellectual immediate access to the nature of things as they are in themselves. (CI) Therefore, if Kant subscribes to a correspondence conception of truth, then he must maintain that we have mediate access to things as they are in themselves through the isomorphism between appearances and things in themselves. (CI) If Kant subscribes to a correspondence conception of truth, then he must maintain that we have mediate access to things as they are in themselves through the isomorphism between appearances and things in themselves. (P3) If there is an isomorphism between appearances and things as they are in themselves, then there must be a one to one correspondence between appearances and things in themselves (because there must be one noumenal essence which causes each phenomenal essence.) (P4) According to Kant, it is not the case that there can be a one to one correspondence between appearances and things in themselves (because things in themselves cannot be individuated in the same way as appearances are). (C2) Therefore, there cannot be mediate access to things in themselves by the isomorphism between things in themselves and appearances. (C3) Therefore, given PI, CI & C2, Kant cannot subscribe to a correspondence conception of truth.
Since our judgments cannot be said to correspond to things in themselves either directly, or indirectly as the correspondence to appearances
64 Kant's Correspondence Theory of Truth that have a one to one correspondence with things in themselves, then "there is no correspondence theory of truth in [Kant's] philosophy." (Putnam 1981, 64) Therefore, truth can only be a kind of epistemic assertabilism. According to this view, a true statement will be a statement "that a rational being would accept on sufficient evidence of the kind that it is actually possible for beings with our nature to have." (Putnam 1981, 64) The kind of evidence we can have, is the kind we construct qua appearances. Putnam infers from his denial of direct access to and isomorphism with the noumenal world that sensation must also be a purely phenomenal property. (Putnam 1981, 63) If sensible properties are no more transcendentally real than formal properties like space and time, then they too are entirely dependent on finite subjects. If I understand Putnam correctly, he thinks that to claim that there is something, i.e. sensation, that is part of an appearance, yet which is not constructed by a finite subject, amounts to a claim that there is a one to one correspondence between appearances and things in themselves. This one to one correspondence is necessary because, if sensation is not constructed, there must be a thing in itself for each appearance such that that thing in itself can be the power that causes the sensation in the appearance. Putnam reasons that since Kant denies this one to one correspondence, he must also affirm that empirical sensation is a construction. What Putnam seems to have in mind for his account of Kantian sensation is an adverbial account. There is not red 'out there' in a transcendentally real sense, which causes a sensation in me. Rather, certain appearances 'appear redly to me.' The red is in the intuiting, it does not exist in the world independently. It is because of the radical nature of Putnam's account of the dependence of objects of experience on finite subjects, which is illustrated by his account of sensation, that I classify his view as a radical (relativistic) constructivism. A Reply to Putnam In Putnam's view, the conception of truth as correspondence described in the Critique must be merely the correspondence between empirical judgments and the constructed empirical objects that serve to establish their truth or falsity. To deny this kind of correspondence would
The Constructivist View 65 be to undermine Kant's empirical realism. To preserve that realism it must be possible, in some sense at least, for Kant to claim that "Hugo is a black cat" is true just in case Hugo is a black cat. Putnam does not deny this trivial notion of correspondence, he simply calls it "goodness of fit." The issue is not what is required for this trivial notion of correspondence. Two points must be considered in analyzing Putnam's account of the Kantian conception of truth. First, I will argue that Putnam's interpretation does not provide adequate truth conditions for all of the kinds of judgments that can be truth-bearers according to Kant. Second, I will argue that a fundamental premise of Putnam's argument against a Kantian correspondence conception of truth is not well founded. The Problem of Non-empirical Truth. One of the fundamental problems with Putnam's interpretation is that it does not leave room for analytic or synthetic a priori truths. Putnam defines a true judgment as a judgment "one would accept on sufficient experience of the kind that is actually possible for beings with our nature to have." (Putnam 1981, 64) It makes no sense to say of a priori judgments that they are true just in case they would be verified by experience. Since Putnam's view does not account for the truth conditions of all the kinds of judgments that Kant claims are truth-bearers, the view is problematically incomplete. It is possible that Putnam simply omits his account of a priori truth, but would nevertheless allow for such an account. I do not think so, however. Putnam's interpretation, like many empiricist interpretations, underestimates the importance that Kant places on analytic and transcendental truth. Isomorphism and the Rejection of Correspondence. Putnam reasons that for Kant to subscribe to a correspondence conception of truth, he must maintain that either we have immediate access to things as they are in themselves through some perceptive or intellectual faculty or that we have mediate access to things as they are in themselves through the isomorphism between appearances and things in themselves. He rejects the possibility of a correspondence conception of truth for Kant because Kant expressly denies both the former and the one to one correspondence between things in themselves and phenomena implied by the latter.
66 Kant's Correspondence Theory of Truth It is certainly true that if a correspondence view of truth required us to have substantive knowledge of things in themselves, immediately or mediately, then Kant would have to reject such a view. However, I do not think that a relationship of either type is necessary to secure one specified correspondence between our true judgments and their objects. As I will argue, neither relationship (as Putnam construes them) is necessary, and therefore, Putnam does not provide sufficient reason to reject the possibility of a correspondence conception of empirical truth for Kant. It is not the case that the transcendental idealist needs to secure access to knowledge of things as they are in themselves in order to specify a one to one correspondence between our words and thoughts and their objects. Moreover, it is not necessary for there to be access to knowledge of things in themselves for the one to one correspondence between an empirical judgment and its object to confer truth on the requisite judgment. All that is needed to secure correspondence is the existence of a publicly accessible, objective world. Kant argues that there is one objective, publicly accessible reality that we all experience. This reality is determined by the parameters of the synthetic unity of experience. It is determined for all according to the matter of experience (sensibility) and the form of experience (transcendental conditions), and may be extended beyond an individual's particular experiences by means of inference from the transcendental laws governing experience. This is equivalent to saying that, although the sensible world has no absolute magnitude, the empirical regress (through which alone it can be given on the side of its conditions) has its own rule, namely, that it must always advance from every member of the series, as conditioned, to one still more remote; doing so by means either of our own experience, or of the guiding thread of history, or of the chain of effects and causes. (A522/B550)
In this passage Kant affirms that, out of what we are all given in sensation, we form an objective world which operates according to certain laws. The application of these laws insures that there will be a single, unified objective world of appearances, which hold even for the relationships between appearances that have as yet not been intuited by any
The Constructivist View 67 subject. According to Kant, the empirical world is set and objective. There will be one series of sets of affairs that have, do, will obtain. This is insured by the fact that we all have the same material given to our sensibility, and we all determine it according to the same a priori principles. Thus, there is one real world, and therefore one set of true judgments which correspond to that world. There are not infinitely many sets of entities to which true judgments might correspond. There is only one, and it is accessible to finite subjects. Given these considerations, we must reject Putnam's claim that a correspondence theory of truth requires some immediate or mediate knowledge of things as they are in themselves. Consequently, we must also reject his subsequent rejection of the possibility of a correspondence conception of truth for Kant. Rejecting Putnam's argument also allows us to reject a problematic implication of his conception of transcendental idealism, namely, his constructivist account of sensation. Sensation and Intuition. Putnam's constructivist account of sensation, which arises from his denial of any relationship between phenomenal and noumenal essences, directly contradicts Kant's account of sensation. Kant's account of sensation supports an affirmation of a correspondence conception of truth (though of course not in the transcendental realist sense) without thereby implying either immediate knowledge of things as they are in themselves or a one to one correspondence between things in themselves and appearances. Kant distinguishes intuitions from sensations because the latter "do not of themselves yield knowledge of any object." (A28/B44) This distinction must be made because of Kant's account of sensation and his denial that we may have any substantive knowledge of things in themselves. All of our knowledge of things in themselves is negative knowledge. We know what properties do not belong to things in themselves. (Kant posits a slightly different view concerning our moral knowledge, but for our purposes we need deal only with theoretical knowledge.) However, he does posit a strong connection between the sensual aspects of an appearance and things in themselves. He describes the thing in itself as the correlate of sensibility. (A29)
68 Kant's Correspondence Theory of Truth To explain how a sensation could have a thing in itself as its correlate, Kant distinguishes between two aspects of an appearance. He describes an appearance as having both form and matter. Sensations are the matter of appearance, and they are only accessible to us a posteriori because our minds have a set form ready for them a priori. The matter of an appearance then, is the correlate of what is given in a transcendent object, but is accessible to us only after it has been given form by the understanding. What Kant means by this is that we cannot perceive anything given to us in sensation without perceiving it in a certain form. That form includes spatial and temporal individuation and the empirical laws imposed by the pure principles of the understanding. This is the ground of Kant's claim that sensation is not itself intuition, and that the former cannot itself yield knowledge. Sensation can only yield knowledge when it has been determined according to the predetermined form imposed by the understanding. That is to say, we can only receive the matter (sensation) in an appearance when its form has been determined. Because of the strong connection Kant postulates between things in themselves and sensation, and his distinction between the matter and form of an appearance, it is unreasonable to suggest that he subscribes to an intuitionist constructivist account of sensation. Kant makes these distinctions precisely because he does not think that the form and matter of an appearance are equally transcendentally real. (Putnam 1981, 64) This wrong-minded interpretation of sensation, along with the overstatement of the requirements for correspondence discussed previously, together give sufficient reason to reject Putnam's denial of a correspondence conception of truth for Kant. Since his argument for his constructivist assertability view seem to be that it is the only default given the impossibility of correspondence, his positive account must also be rejected for lack of evidence. Putnam's interpretation of Kant's conception of truth is incomplete, Furthermore, his rejection of correspondence does not follow either from the text or from the necessary rejection of one-to-one isomorphism between noumenal and phenomenal objects. Given these problems, it is reasonable to conclude that his view is at best incomplete and at worst inconsistent with transcendental idealism as Kant construes it. For these
The Constructivist View 69 reasons, Putnam's constructivist account is not adequate as an interpretation of Kant's conception of truth. Jaakko Hintikka I would like to begin by saying that Hintikka's comments concerning the status of non-mathematical truth for Kant are sparse. My remarks are therefore intended as an interpretation, not reconstruction, of what I think is his implicit account of theoretical truth for Kant. Hintikka argues that according to the most reasonable understanding of transcendental idealism, Kant's mathematical constructivism can be translated to apply to all objects of possible experience. He argues this based on what he thinks is the appropriate connection between mathematics and individuation. According to Kant, the only adequate explanation for why our mathematical arguments apply to experience must assume that we put the properties and relations dealt with in these arguments into the objects ourselves. These properties and relations are the properties and relations of individual objects. The only way we can know individual objects is through the senses. Therefore, the only way to put the requisite mathematical properties and relations into all individual objects is in the act of sensation. "It follows, Kant seems to have thought, that all our knowledge of individual objects which applies to them universally and necessarily is really about properties and relations which we have ourselves put into objects in the act of sensation." (Hintikka 1974, 131-132) Hintikka argues that Kant is inconsistent in relying on the "mere registration of sensations" to establish the application of mathematical arguments to individual objects. He maintains that Kant's theory of mathematics fails because of his "unfaithfulness to his own principles." Even though Kant generally emphasizes the "active and constructive aspects of our knowledge;" "this is just what he neglected to do here." (Hintikka 1974, 132) In short, Hintikka thinks that Kant's exposition of his theory of mathematics is inadequate because it fails to assert that properties and relations of our mathematical arguments are placed in objects by us when we create those objects for ourselves. He claims that a better exposition of a theory of mathematics for transcendental idealist would focus on the
70 Kant's Correspondence Theory of Truth importance of constructivism for Kant. However, even in the course of explaining the role of mathematics in the individuation of objects, as Kant construes it, the nature of those objects as constructions is revealed. To understand that nature, we must consider the relationship between mathematics and sensibility. Mathematics and Sensibility Hintikka's reconstruction of Kant's argument for the connection between mathematics and appearances is roughly the following: (1) Both mathematical reasoning and the conditions for the intuition of empirical objects are concerned with the existence of individuals. (2) Since for Kant we can only know a priori of an object what we put into it, then all of our a priori judgments must arise entirely from us. (3) Both the structure of mathematical reasoning and the structure of experience arise from the structure of our apparatus of perception. (4) Therefore, since we see that individuals as mathematical objects are entirely constructed by us, we ought to conclude that individuals as empirical objects in general are entirely constructed by us as well. (Constructivism is true for the world of appearances.) (Hintikka 1974, 178-180)
Again, because it undermines the importance of our construction of object by focusing too much on the givenness of intuitions, this argument is an inadequate exposition of the relationship between mathematical arguments and the individuation of argument. However, the argument fails to provide us with a clue as to the nature of truth for judgments about empirical objects (individual objects as Hintikka describes them). All the knowable formal properties of individual objects are put into them as they are constructed. Therefore, the truth of judgments consists in their correspondence to a reality that is dependent on and constructed by finite subjects. Reply to Hintikka Incompleteness Revisited. One of the problems in Putnam's interpretation of Kant's conception of truth as constructivist is also found in Hin-
The Constructivist View 71 tikka's account. That problem is the problem of incompleteness. Hintikka does not provide a substantive account of the truth of transcendental judgments. He considers such judgments only as they are employed in the construction of individual objects. The Fundamental Role of Sensibility in Experience. Hintikka also provides reading of transcendental idealism that misconstrues the role of noumenal objects in the determination of our objects of experience. His complaint about Kant's theory of mathematics is that his arguments for it focus on sensibility in a way that undermines the importance of our construction of our individual objects. It seems to me that Kant ought to have focused on these activities [the activities which we in fact perform in order to gain knowledge about individual objects] rather than on the mere registration of sensations. It is instructive to note that Kant actually spoke of objects as "given to us in sensation." I would like to submit that only very rarely can we passively wait until the objects we are interested in are given to us. usually we have to go out and look for them ourselves. (Hintikka 1974, 132)
This passage demonstrates that Hintikka's interpretation of the nature and role of sensibility in knowledge is problematic. First, for Kant, particulars are necessarily given to our sensation. If they were not, there would be nothing out of which we could construct our object. To determine an individual to have certain knowable formal properties, there must be an individual to determine. Moreover, it makes no sense, for Kant, to speak of objects we are interested in prior to their intuitions being given to us in sensibility. To have an interest in something, it must be an object for you. Nothing can be an object for you that is not a part of the synthetic unity of experience, and there is no synthetic unity of experience unless the understanding is given the manifold of undetermined empirical intuitions out of which to construct this synthetic unity. Therefore, it would be inconsistent for Kant to suggest that we could be interested in an object, the intuition of which was not first given to our sensibility. I think Hintikka's remarks here are indicative of a misunderstanding of what Kant means by sensibility. For Kant, the sensibility is our mode
72 Kant's Correspondence Theory of Truth of receptivity. It is not, and indeed could not be limited to empirical sensation. This is the case because, for the reasons given in the preceding paragraph, there must be intuitions that are given to the sensibility prior to any empirical entity. If the constructivist account of truth implicit in Hintikka's view requires that we emphasize our construction of objects in such a way that the importance of what is given to us in the sensibility must be compromised, then the account is inconsistent with Kant's basic assertion that knowledge requires both intuition and concept. Given this inconsistency, Hintikka's constructivist account could not adequately capture what Kant intends in his conception of truth. Summary I have shown that constructivist views like those of Putnam and Hintikka fail to capture a conception of truth as Kant intends it. Each has problematic implications of its own, but both deny the significance which Kant attributes to things in themselves as distinguishing transcendental idealism from Berkeleyan sorts of idealism. Finally, such accounts contradict Kant's textual remarks concerning the nature of an external object. More importantly, constructivist views introduce these difficulties without in turn introducing an interpretation that makes transcendental idealism more feasible as a rational worldview. For these reasons, the best account of truth for transcendental idealism cannot be a purely constructivist account.
Chapter Five The Coherence View An Introduction to Coherence Views Many scholars have suggested that truth in transcendental idealism is essentially a matter of coherence between our judgments. An extreme version of a coherence theory of truth is attributed to Kant by constructivists like Putnam. Not all those who think Kant subscribes to a coherence theory of truth think his view must be so radical. Ralph Walker and Norman Kemp-Smith provide interpretations of Kant's conception of truth that do not commit transcendental idealism to the sort of phenomenalism Kant wants to avoid. Walker provides an account of truth for each of Kant's three kinds of truth bearing propositions, analytic judgments, empirical judgments and transcendental judgments. He argues that the truth of analytic judgments is trivial for Kant, that truth for empirical and transcendental judgments is an impure coherence theory, and that there is a fourth set of meaningful judgments that nevertheless correspond to things in themselves. Smith argues that Kant was the first coherence theorist. He claims that since all acts of judgment are simultaneously applications of empirical and transcendental judgment, truth consists simply in the coherence of these judgments. Smith maintains that although there are things in themselves, truth is essentially the coherence between our judgments. Ralph Walker A coherence theory of truth, as Walker construes it, is one in which the truth of a judgment is its coherence with a certain set of beliefs. Substantively, truth does not mean the correspondence between judgments and a mind independent reality. Correspondence is only a negative criterion of truth to the extent that "is true" means "corresponds to a fact." However, the set of facts is determined by coherence. "Truth is not a matter of correspondence with some reality that obtains independently of our beliefs about it." (Walker 1989, 62) Walker argues that Kant has an impure coherence conception of truth, because Kant does not hold that truth is always a matter of coherence. (Walker 1989, 61) Walker believes that Kant's coherence concep-
74 Kant's Correspondence Theory of Truth tion of truth must be impure, because Kant maintains the importance of a transcendent object that serves as the given in our judgments. This transcendent object is, by its nature, independent of any beliefs we might have about it. I will explicate Walker's account of truth for analytic, empirical, transcendental and noumenal judgments. I will then argue that, although Walker correctly describes Kant's structure of experience, his conclusion that a coherence account of truth is necessary given this structure is unwarranted. Analytic Judgments Apart from the impure coherence conception of truth that Walker ascribes to empirical and transcendental judgments, he gives an account of analytic truth. He argues that Kant's conception of analytic truth is unproblematic because Kant thought "it was somehow merely verbal and empty-too trivial really to be worth calling knowledge." (Walker 1989, 61) I think that this explanation is indicative of a lack in Walker's interpretation. Specifically, he does not seem to acknowledge the difference between the Kantian version of the analytic/synthetic distinction and the version of that distinction in formal logic. I say this because given Kant's take on this distinction, truth for analytic judgments may be purely verbal, but it is certainly neither empty nor too trivial to be considered knowledge. Walker's apparent failure to grasp this distinction is important not only as it relates to his account of analytic truth in Kant, but also as it relates to the nature of synthetic truth. Therefore I think a brief consideration of it is necessary before undertaking an explanation of Walker's interpretation. The distinction between analytic and synthetic judgments in formal logic is laid out in terms of the nature of the relationship between subject and predicate concepts. If the predicate concept is contained within the subject concept, then the judgment is analytic. If the predicate concept is outside the subject concepts, then the judgment is synthetic. What in formal logic is construed as a two-term relationship, Kant defines as a four-term relationship. Every judgment involves an understanding subject, an object (or objects), a subject concept and a predicate concept. For Kant analytic judgments include not only those judgments in which the predicate concept is contained within the subject concept,
The Coherence View 75 but also those in which the predicate concept is a necessary consequence of the subject concept. (A6-7/B10-11) The distinction boils down to one simple principle; if the connection between the subject concept and the predicate concept is based solely on the principle of contradiction, no matter how mediated that connection is, the judgment is analytic. Otherwise, the judgment is synthetic. Kant's own examples serve to illustrate the nature of his distinction. The judgment, "All bodies are extended" is an analytic judgment, because the predicate can be derived from the subject by conceptual analysis alone. The judgment, "All bodies are heavy" is synthetic because you need experience in addition to conceptual analysis to derive the predicate 'heavy' from the concept of a body. (Bll) It is important to note that, as Allison points out in The KantEberhard Controversy, Kant is unclear on this issue in the first Critique. In the section from which the earlier examples were drawn, he describes the distinction as a distinction concerning whether or not the predicate is "thought through identity" with the subject. In judgments in which predicates are thought through identity with the subject, the judgment is analytic. Otherwise it is synthetic. (A7 / B l l ) It seems possible that this is still a description between two kinds of logical relationships. However, Kant is quite clear when he addresses this issue in another of his works, On Discovery. The example he uses, the judgment "Bodies are divisible," illustrates his point. Since divisibility is a logical consequence of being a body, even though it is not strictly contained in the subject concept, the judgment is described as an analytic one. Given the complexity of Kant's analytic/synthetic distinction, it hardly seems to be the case that the truth of analytic judgments is merely trivial. Furthermore, Kant would certainly not hold that analytic judgments, in his sense, fail to constitute knowledge for us. We judge the truth of a principle, at least in part, by how many true consequences it has. Kant explicitly affirms this. "There is truth [of a concept] in respect of its consequences. "The greater the number of true consequences that follow from a given concept, the more criteria there are of its objective reality." (B114) Since for Kant, anything that is thought through identity with the subject is an analytic judgment, and
76 Kant 's Correspondence Theory of Truth consequences are predicted on the basis of an in-depth analysis of the subject term, then all predictions of consequences are thus analytic judgments. This point can best be demonstrated by means of an example from the natural sciences. My interpretation of Galileo's argument against the geocentric universe is predicated on an analytic judgment. If ours is a geocentric universe, then Venus will not show phases. Venus shows phases. .•. Ours is not a geocentric universe.
The consequent 'Venus will not show phases' is thought through identity with the subject 'our geocentric universe.' It is a prediction that is based on an analysis, an a priori conceptual analysis, of what it would mean for ours to be a geocentric universe. Based on this analytic judgment, we are able to make predictions that will either confirm or disconfirm its truth. The second premise, of course, is observational and thus not analytic. However, it is the analytic judgment in the first premise that makes the subsequent observational premise useful to us. This example, and many which are analogous, show that analytic judgments certainly can count as non-trivial knowledge for us. It is on the basis of such conceptual analyses that much of the predictive success of our natural sciences stands. Thus, we must conclude that analytic judgments are not true in a merely trivial sense. They are significant in adding to our knowledge. Furthermore, we must conclude that synthetic judgments extend knowledge beyond where it could be extended by any degree of pure conceptual analysis. An account of the truth of synthetic judgments must therefore be able to account for truth in a way that does not rely purely on formal relations between judgments. I think it is this consideration that motivates Walker to argue that Kant's coherence conception of truth must be an impure one. Empirical Judgments Synthetic a posteriori, or empirical judgments are fairly simple to analyze. An empirical judgment is true just in case it is coherent with the formal and material conditions of truth. The formal conditions of truth are the pure concepts of the understanding and the pure intuitions of
The Coherence View 11 space and time. The material conditions of truth are what are given in sensibility. The Kantian conception of truth for empirical judgments is not, however, a pure coherence conception. As I have said, for Kant, a synthetic judgment must extend beyond the formal relationships between our judgments. For Kant, there must be a given, and the importance of this given is especially apparent in a certain class of empirical judgments. This class of empirical judgments, namely reports of what is given in sensation, is one for which truth is not essentially a matter of coherence. "That seems blue to me." "That feels hot." "That tastes sour." All of these are reports of sensation and, as such, their truth is fundamentally a matter of correspondence. There is a class of statements, reports of what is given in perception, for which truth does not consist in coherence but requires the correspondence with independent facts. For all other statements about the world as we can know it through experience, truth consists in coherence, coherence with the a priori principles on the one hand and with the report of what is given on the other. (Walker 1989,63)
To understand why such reports must be reports of what is given, it is important to understand that Kant has an adjectival rather than an adverbial account of sensation. I am not "appeared redly to", rather, what appears to me appears to me as something red. Red then is not a way I perceive. It is a way what I perceive is. What Walker has in mind seems to be this. Our empirical judgments are true in virtue of coherence insofar as they must conform with principles of the understanding. Reports of what is given in perception, however, are matters of correspondence to independent facts. By independent facts, given his definitions of dependence and independence, Walker can only mean direct correspondence to noumena. Walker describes a pure coherence theory of truth as one in which truth is entirely a matter of coherence. Correspondence to mind independent reality place no role in the establishment or satisfaction of truth conditions. The coherence theorist holds that for a proposition to be true is for it to cohere with a certain system of beliefs. It is not just that it is true if and only if it coheres with that system; it is that the coherence, and nothing else, is what truth
78 Kant's Correspondence Theory of Truth consists in. In particular, truth does not consist in the holding of some correspondence between the proposition and some reality which obtains independent of anything that may be believed about it. (Walker 1989, 2)
Given this explanation of a pure coherence theory of truth, we can infer the notions of dependence and independence Walker is using. With this inference, we can better understand what he means when he describes Kant's coherence conception as impure. According to Walker's description, one may preserve a pure coherence theory of truth so long as the facts to which our judgments correspond are facts that are shaped by our beliefs about the world. This is what Walker calls a trivial notion of correspondence. An impure coherence theory of truth, then, must be one in which there is some degree of non-trivial correspondence. That is to say, there must be some way in which certain judgments correspond to a reality that is independent of our beliefs about it. Any fact, that is, anything to which a judgment might correspond is either dependent or independent. A fact is dependent if it is constituted, in any part, by our beliefs about it. Correspondence to such facts is, again, only a trivial correspondence and one that does not compromise a substantive coherence theory of truth. The 'fact' to which a judgment corresponds is only independent if it is in no way related to our beliefs about it. Correspondence of our judgments to these types of facts is a non-trivial correspondence. Reports of what is given in sensation correspond in this non-trivial way, and thus make Kant's coherence conception of truth an impure one. Our judgments about appearances must correspond either to the form or matter of experience. The form of appearances is dependent on our beliefs and as such any correspondence to the form of a judgment is only trivial correspondence. Judgments like reports of sensation, however, correspond to the material conditions of experience alone. As it is given to us, the material conditions of experience cannot have their source in us. Their source must be something independent of us, in Walker's sense. The independent source in this sense is the ground of our appearances, the transcendent object—the thing in itself. According to this view, the judgment "That red bird flew away" is a judgment that is true in virtue of its coherence with the formal conditions of the understanding. More precisely, the judgment is true just in case it
The Coherence View 79 fits in the synthetic unity of appearances that is determined in accordance with those formal conditions. However, the judgment "That seems red to me" is not true in virtue of coherence. It is merely a report of the material conditions of experience and as such is true in virtue of its correspondence to the source of the sensation. As I have shown (and Walker does not seem to deny), this source must be things as they are in themselves. I have argued that trivial versus non-trivial correspondence in Walker's view is determined by the nature of the fact to which the judgment in question corresponds. In turn, the nature of facts is determined by their relation to the formal and material conditions of experience. Given this structure, to understand the impure coherence theory that Walker attributes to Kant, we must first fully understand the distinction between the formal and material conditions of experience. The material conditions of experience, as previously described, are what are given in sensation. As givens, they are not dependent on us. The formal conditions of experience are dependent on finite subjects, and as such "hold in the world of appearances in virtue of the fact that we believe them—they are elements of the coherent system that determines the truth." The formal conditions of the understanding are constitutive of the empirical world and "hold within it simply because they are fundamental to the coherent system of our beliefs, and not because of any correspondence to a reality that obtains independently of those beliefs."(Walker 1989, 62-63) Because the formal conditions of empirical truth apply to the world precisely because we believe them, and because they do not themselves correspond to anything outside the mind of the subject, truth for empirical judgments is, at least to this extent, dependent on coherent belief. Consider again the judgment "that red bird flew away." Except for the raw perceptual data, the 'fact' to which this judgment corresponds is dependent on the understanding of the judging subject. First, 'that red bird' indicates that the individual making the judgment, I will call her Linda, is picking out a particular thing in space. Linda is indicating that bird as opposed to any other individual that might be subsumed under the same concept. Second, she is picking out an individual in time. She is making a judgment about a bird that was nearby and a certain time (t-1), but that was then absent at a subsequent time (t2). Space and time are
80 Kant's Correspondence Theory of Truth pure intuitions that are dependent on judging subjects as formal conditions of experience. In addition to space and time, the fact to which Linda's judgment corresponds is dependent on other formal conditions of experience. Linda's judgment assumes that the bird is actual. It assumes that there was a unity of experience of the bird, namely that it was the same bird from tl to t2 - tn. Linda's judgment also assumes that there was a causal relationship between the flapping of the bird's wings and the bird lifting from its perch and moving away. Existence, Unity, and Causality are all categories. They are all formal conditions of experience placed in the 'fact' by the judging subject. Except for the perceived redness, that about which Linda was judging was entirely a function of the formal conditions of the understanding. This structure, according to Walker, requires that Kant espouse a coherence conception of truth, albeit an impure one. Another distinctive fact about Kant's coherence conception of empirical truth, as Walker sees it, is that the whole truth is something we can never attain. For Kant, the phenomenal world is not the world as anyone actually believes it to be, but the world as we would believe it to be under ideal circumstances: circumstances which we cannot in fact realize. (Walker 1989, 65) Walker takes the phenomenal world to be the totality which would arise from a subject collecting and processing all of the perceptual data which could be attained from the noumenal world, plus the knowledge that this totality was complete. Of course, no finite subject could actually do that, so although truth is a matter of coherence, the whole truth can never, in principle, be attained (Walker 1989, 66) That no subject can know the totality of experience is affirmed explicitly by Kant. The sum of all of the possible objects of our knowledge appears to us to be a plane, with an apparent horizon—namely, that which in its sweep comprehends it all, and which has been entitled by us the idea of unconditioned totality. To reach this concept empirically is impossible, and all attempts to determine it a priori in advance have proved in vain. (A759/B787)
One might well ask, however, why this is significant to Walker's interpretation of the Kantian conception of truth. It is significant because the 'impure' elements of his interpretation, namely, the affirmation of the
The Coherence View 81 role of the transcendent object as the ground of appearances and this affirmation of the impossibility of knowing the totality of objects of possible experience, serves to distance his account from constructivist accounts like Hintikka 's and Putnam's. As I argued in the last chapter, constructivist views, although internally coherent, are fundamentally inconsistent with transcendental idealism as Kant construes it. Thus, by allowing that Kant's coherence theory is impure in the two aforementioned ways, Walker avoids the problems inherent in constructivist views and presents an interpretation of truth for Kant that is more consistent with the fundamental tenets of transcendental idealism than are pure coherence interpretations. Since Kant does not regard the whole truth about the phenomenal world as something practically attainable, Walker claims that Kant rejects the principle of bivalence. The principle of bivalence holds that there are no statements that are neither true nor false. Walker argues that the coherence theorist is not entitled to claim that the principle of bivalence is true unless he either constructs his theory in such a way that the coherent system of beliefs can determine the truth or falsity of every statement (by, for example identifying the set of coherent beliefs with God's beliefs) or defines what is to count as a statement to those things capable of having a truth value (as certain verificationists do). Walker argues that there are three possible positions for a coherence theorist. (1) He may subscribe to the principle of bivalence in a way similar to those discussed in the previous paragraph. Under these sorts of conditions, for every statement 'p' either 'p' or 'not p' is a part of the coherent set of beliefs. (2) He may deny bivalence by maintaining that there is a determinate totality of truths, and for any statement, it is either part ofthat totality or it is not. Since under these conditions it is possible that neither 'p' nor 'not p' is part of the determinate totality of truths, such statements will be neither true nor false. (3) The coherence theorist may also hold that the coherent set of beliefs does not form a determinate totality. In such a view, there may be certain values of p which are determinately neither true nor false. Walker attributes the third position to Kant's transcendental idealism. What he has in mind are the antinomies, where both a judgment and its negation are false. Both the judgment and its negation are false for the
82 Kant's Correspondence Theory of Truth antinomies since they rest on a false assumption (that there are truths about the world as a totality). This is, according to Walker, effectively a rejection of bivalence; it just reflects a decision to use the word false for statements such that neither they nor their negations meet the requirements of coherence. (Walker 1989, 66) Finally, to explicate the nature of Kant's impure coherence conception of empirical truth, Walker distinguishes three senses of the term 'objective'. In the first and weakest sense, to say something is objective is to say that one may be mistaken about it. Objectivity is a necessary element of Kant's coherence theory. By maintaining that there must be objectivity, that is the standard of coherence, a coherence view may avoid the charge that to be true is nothing more than to be believed. To be true means not only to be believed, but also to be coherent with the coherent set of beliefs. (Walker 1989, 67) To establish that Kant's view allows for objectivity in this weakest sense distances Kant from solipsism. In requiring what Walker calls 'the standard of coherence', we eliminate the possibility that for Kant, reality is whatever we make of it. It is possible to imagine constructivist views even more extreme than Putnam's in which reality is simply what our minds make it to be. Interpretations of this sort completely eliminate the possibility of empirical realism in any substantive sense, and thus are incompatible with transcendental idealism as Kant intends it. That there must be objectivity in the minimal sense of having judging subjects be susceptible to error is a requirement for Kant. Moreover, if we consider Kant's claim that coherence is a negative condition of all truth, it seems reasonable to agree with Walker concerning the role of the standard of coherence in establishing objectivity in this minimal sense. In the second sense, it is to say that things exist independently of our perception of them. With Kant's definition of an appearance as an undetermined object of empirical intuition he is providing a "graphic way of expressing his coherence theory." (Walker 1989, 68) For Kant, appearances exist independently of our representing them to ourselves. To illustrate his second definition of objectivity, Walker cites the following passage: "Representation in itself does not produce its object in so far as existence is concerned, for we are not here speaking of its causality by means of the will." (A92/B125)
The Coherence View 83 This passage refers to Kant's appearance/representation distinction. The representation is that which we represent to ourselves subjectively. The appearance is that which is represented. It is that which has been determined transcendentally, but not yet empirically. By determined transcendentally, I mean that the appearance has had all of the formal conditions of experience imposed on it. It is the result when material conditions of experience have added to them all of the forms that make knowledge possible for us. Walker argues that the representation/appearance distinction is indicative of Kant's having a coherence conception of truth because for Kant the laws, which hold of appearances, do not hold in virtue of their correspondence to anything independent. They hold simply because the laws constitute a coherent system of principles which serve as the truthmaker for all empirical judgments. (Walker 1989, 69) Again, the laws Walker has in mind are the formal conditions of experience, the categories and the pure manifold of space and time. It is certainly the case that, for Kant, an appearance is an object of possible experience prior to any empirical determination. It is also the case that appearances are determined according to the formal conditions of experience and that these conditions must be internally coherent. What Walker must prove, however, is that it is this internal coherence that is the essence of what it means to be a fact. If to be coherent with principles is the essence of being a fact, then a judgment corresponds to a fact, i.e. a judgment is true just in case the belief expressed by the judgment fits with the coherent set of beliefs constituted by appearances. In my analysis of his view, I will consider whether he proves that coherence is the essence of what it means to be a fact. In the third sense, objective means public, i.e., common to everyone. (Walker, 67-68) Objectivity in this sense means that the standard of truth, or truth-maker, for our judgments is available to everyone. This objectivity holds for everything in the world of appearances. To put it simply, objectivity in this sense means that we move about in the same world. According to Walker, this third sense of objectivity lies beyond the scope of argument because it involves an appeal to things as they are in themselves. The only reasonable explanation for why there is a common
84 Kant's Correspondence Theory of Truth world of appearances is that they are somehow grounded in things in themselves. (Walker 1989, 77) This point returns again to the distinction between the formal and material conditions of experience. The reason that Walker holds that reports of sensation are true in virtue of their correspondence to something independent of our beliefs about it is that every judgment requires a given. Just as there must be something given to us in representation, namely appearances, such that we might make empirical judgments, there must also be something given to us prior to the determination of appearances. There must be something given in sensibility upon which the formal conditions of experience are imposed. Since Kant is an empirical realist, the given in appearances cannot come from us. As I have argued, that which is independent of us is, according to Walker, the thing in itself. Therefore, what is given in sensation is the transcendent object, the thing in itself. I think Walker provides an accurate account of what Kant intends by the material conditions of experience. Because they must be prior to any beliefs/determinations we make about the world (be these determinations transcendental or empirical), the material conditions must have their foundation in something entirely outside us, namely, in the transcendent object. This is one of the aspects of Kant's view which Walker believes commits him to an impure, rather than a pure coherence conception of truth. Specifically, that because Kant maintains that we build the world out of something, (.e. that which is given in the undetermined manifold of empirical intuition), then he cannot have a pure coherence theory of truth. This given must be, according to Walker, a thing in itself as it is not conditioned by the formal conditions of experience. (Walker 1989, 63) For Walker to prove his theory concerning Kant's conception of truth, he must prove that this reliance on the material conditions of experience proves only that Kant's basic coherence theory of truth must be modified with minimal elements of correspondence. In other words, he must prove that this reliance is not indicative of the fact that for Kant, truth is essentially a matter of correspondence with coherence being only a minimal condition. In my later analysis of his view, I will consider whether he sufficiently grounds this claim.
1
The Coherence View 85 Transcendental Judgments As has been implied in the discussion of truth for empirical judgments, Walker argues that the truth of transcendental judgments is entirely a matter of their coherence. The truth of transcendental judgments cannot be determined by experience. Nevertheless, there must be synthetic a priori knowledge because we rely on concepts like cause and object that cannot be derived from experience. Since appealing to a divinely granted harmony between our appearances and the way the world really is (Leibniz's move) is problematic, because of its circularity, Kant concludes that there is only one possible explanation for what makes synthetic a priori judgments true. "Since it is not 'the object alone' that 'makes the representation possible', 'the representation alone must make the object possible' (A92/B124)" (Walker 1989, 61-62) In other words, transcendental judgments are true in virtue of the fact that they are "fundamental to the coherent system of our beliefs." They create the framework for this coherent system by determining the undetermined manifold of empirical intuition (what is given in sensation) and forming it into the synthetic unity of appearance. This is done by imposing the formal conditions of experience on what is given in sensibility. As the concepts transcendental judgments employ are merely formal conditions, it seems that their truth must be a matter of coherence. The internal coherence of these principles translates into the coherence of the phenomenal world. Truth for transcendental judgments is strictly a matter of their coherence, and this is demonstrated by their role in establishing the coherence of our world of appearances. "For he [Kant] takes the view the material given in sensation can mean something to us only when it has been synthesized and made the subject of a kind of judgment." (Walker 1989, 63) This explication of coherence as the essence of transcendental truth is fundamental to Walker's impure coherence reading of Kant's conception of truth. Transcendental judgments establish the 'facts' about the empirical world. By determining the form of appearances and the structure within which appearances relate to one another, transcendental judgments create the unity of our experience. It is to this unity, to these facts about the empirical world, which our empirical judgments purport to correspond. If these facts are facts in virtue of being coherent with a
86 Kant's Correspondence Theory of Truth certain set of beliefs (the formal conditions of experience), the correspondence of our empirical judgments to facts is correspondence in a merely trivial sense. The essence of what it means for an empirical judgment to be true would be for it to be coherent with the set of coherent transcendental judgments. Transcendental judgments, then, are true because the set of concepts and intuitions they employ are internally coherent. They are necessarily true because they constitute the set of judgments that must be made if the world is to be ordered in such a way that we can understand and therefore experience it. If they are purely formal, it seems prima facie, that their truth must be purely a matter of coherence. Consider again the judgment, "that red bird flew away." That judgment is true just in case it corresponds to a certain kind of fact about the world, namely, a state of affairs. Thus, the judgment "that red bird flew away" is true just in case there is a set of causally related states of affairs in which there is a particular red bird which was present at tl, but moved away at t2 and was gone at t3. As I have argued, the obtaining of this set of states of affairs is predicated on the imposition of the formal conditions of the understanding on the material conditions of experience. In other words, the fact is a fact because it was determined to be so by coherent transcendental judging. To some extent, the fact in question is a fact because it coheres with these formal conditions. The crucial question, then, is to what extent is the factness of a fact dependent on coherence? If Walker establishes that a fact is primarily a fact because of this coherence, he establishes that Kant's correspondence conception of truth is trivial and secondary to his coherence conception. Noumenal Judgments Walker also admits of a second way in which Kant's coherence conception of truth is impure. Coherence only applies to the world about which we can form beliefs, the phenomenal world. "Kant's coherence theory is impure in another way also. It is restricted to the world of appearances." (Walker 1989, 63) Nevertheless, as Walker rightly points out, the world of appearances is not all that there is. The truth of judgments about things as they are in themselves (noumenally) consists in their correspondence to that independent reality. It is simply a corre-
The Coherence View 87 sponding which we are unable to verify or falsify. Specifically, Walker has in mind judgments which are meaningful for us, but which nonetheless purport to correspond to things as they are in themselves. The examples he gives are ethical judgments that cannot be empirical because they incite us to act contrary to our empirical nature. (Walker 1989, 64) Walker must explain the apparent gap between agents as they are in their merely empirical natures, and agents with moral responsibility. In other words, he must explain how individuals who, as empirical beings, are ruled by their natures could be required by reason to act contrary to those natures. By our empirical natures, we are disposed to taking any actions that might bring us to fulfill our desires. This urge to act on the basis of desire ought to make us individuals who act only in our own interest. Our ethical judgments, however, reveal to us that there are cases in which we ought not act only in our own interest. Imagine that your spouse has given you money to buy him a ticket to the biggest football game of the year. Instead of buying the tickets as you promised, you want to spend the money on new philosophy books. However, you know, based on your awareness of certain ethical truths that you should not break a promise. This knowledge motivates you to act contrary to your own desires and keep your promise. Since there is a motivation to act contrary to the nature of your empirical self, certain judgments, like ethical judgments, must be about noumena and not phenomena. Therefore, there are judgments, which are meaningful for us which nevertheless correspond to things in themselves. Summary Walker concludes that given Kant's account of transcendental idealism, truth for empirical and transcendental judgments is ultimately a matter of coherence. The coherence of transcendental judgments establishes the coherent set of beliefs, and thus their truth is determined by their internal consistency. Empirical judgments are true just in case they cohere with that coherent set. Other than the material condition of experience qua what is given in sensibility, correspondence as an element of truth only applies outside our realm of possible argument and possible knowledge. Therefore, as far as Kant's account of knowledge is con-
88 Kant's Correspondence Theory of Truth cerned, truth is coherence. However, because we must acknowledge these non-verifiable instances of correspondence, Kant's conception must be characterized as an impure coherence conception. Reply to Walker I agree with much of what Walker has to say about the structure of experience in Kant's transcendental idealism. I think that, unlike previously considered interpretations, his recognizes the importance of the transcendent object for Kant's epistemology and metaphysics. Specifically, he recognizes that something transcendent must serve as the material condition for the possibility of experience. He also recognizes that there are meaningful judgments whose truth conditions can never be analyzed by finite subject. Although I agree with Walker's analysis of the structure of experience in transcendental idealism, I think his interpretation of the necessary account of truth accordant with this structure is problematic. To prove his impure coherence theory Walker must prove that empirical and transcendental truth are substantively a matter of coherence. He must also show that, with the exception of the role of the transcendent object in supplying the material conditions of experience and in serving as the truthmaker for our noumenal judgments, the correspondence which is relevant to making true judgments is correspondence in a trivial sense only. As I will argue, Walker's reading grants too much to substantive correspondence on certain accounts, and too much to substantive coherence on others. Substantive and Trivial Notions of Correspondence and Coherence Much reference has been made to substantive versus trivial notions of correspondence and coherence. Before proceeding with a critique of Walker's interpretation, I wish to stipulate precise meanings for the terms, as they are key to both my critiques of various views, and to my interpretation of truth in transcendental idealism. Correspondence in the trivial sense is simply the notion that a true judgment will correspond to a fact about the world. As we have seen, the truth of such facts could be entirely a matter of internal coherence. Therefore, one can admit correspondence in this sense and still advocate
The Coherence View 89 a substantive coherence theory of truth. A substantive correspondence account of truth would be like the account Walker gives for the truth of noumenal judgments. What it means for these judgments to be true is essentially for them to correspond to the way things are. Walker maintains that this notion of correspondence cannot apply to empirical or transcendental judgments because this notion of correspondence requires that those entities to which our judgments correspond exist independently of how we perceive or believe about them. Coherence in the trivial sense is what Kant refers to as a negative condition of all truth. To be true requires coherence in this trivial sense because no set of true beliefs could be inconsistent. If there is a supposedly true belief that fails to cohere with the rest of our true beliefs, then the truth of the inconsistent belief is immediately called into question. Coherence as a criterion of truth in this sense is consistent with a substantive correspondence account of truth. It could simply be the case that a beliefs failure to cohere with the established set of beliefs is evidence of its falsity (i.e. evidence that it does not in fact correspond to the way things actually are). This trivial notion of coherence is not what Walker has in mind. Walker is arguing for a substantive coherence account of truth for Kant. Given this account, the predicate 'is true' means 'coheres with the established system of beliefs'. Failure to cohere with the set is not merely evidence of a beliefs falsity; it constitutes the falsity ofthat belief. The significance of this distinction becomes apparent when considering the types of Kantian judgments Walker thinks are subject to Descartes' evil demon. Truth and Noumenal Confirmation To account for the apparent gap between finite subjects as parts of nature and finite subjects as moral agents, Walker maintains that there are judgments that are meaningful for us even though we can never confirm their truth. His example is the claim that we must be free in a sense stronger than the sense the compatibilists allow. We must believe this claim because if it were not true then there would be no justification for holding individuals morally responsible. We cannot confirm their truth because they are true in virtue of their correspondence to things in them-
90
Kant's Correspondence Theory of Truth
selves. (Walker 1989, 64) In the case of such judgments, Walker argues that truth is, substantively, a matter of correspondence. The coherence of these judgments is only a minimal condition of their truth. In this case, however, Walker extends substantive correspondence beyond where Kant can allow it to goKant seems to deny that substantive judgments about things in themselves are meaningful for us. He claims that we can make analogous judgments, which are only about the world of appearances, but judgments that would only be confirmed by their direct correspondence to things in themselves are not meaningful for us. How much of this [moral] character is ascribable to the pure effect of freedom, how much to mere nature, that is, to faults of temperament for which there is no responsibility, or to its happy constitution can never be determined; and upon it therefore no perfectly just judgments can be passed. (A551n)
Walker's suggestion that assertoric judgments that could be confirmed only by corresponding directly to noumena is in conflict with Kant's remarks on the subject. Such judgments are in fact always problematic because we can only construct them based on an analogy to how things operate in the world of appearances. It is not our judgments whose truth would be determined by direct correspondence to noumena, it is only the analogous judgment we would make if we could know about noumena that would be so confirmed. Therefore, even though it is the case that Kant's conception of truth involves some degree of substantive correspondence, as I think it does, it cannot be substantive correspondence in this sense. The Matter of Appearances: Putnam's Objection Reconsidered Walker's definition of substantive correspondence seems to imply that for a judgment to be true in virtue of its correspondence to an object, that object must be entirely independent of any beliefs we have about it. (This is similar to Putnam's objection to Kantian correspondence.) If this is in fact what substantive correspondence must mean, then Walker is correct in arguing that empirical propositions cannot be true in virtue of their correspondence to their object. However, I see no good reason for
The Coherence View 91 limiting substantive correspondence in this way. Because the 'facts' to which empirical judgments purport to correspond are made up of both form and matter, they have sufficient objectivity to serve as the truthmakers for our empirical judgments. Even though it is true that our transcendental judgments impose upon appearances the necessary formal conditions for the possibility of experience, the truth of our judgments about those appearances is as much a function of matter as of form. What is given in the appearance as sensibility is what distinguishes that appearance from others, which have been determined according to the same formal conditions. In other words, the truth of the content of our empirical judgments can only be explained in virtue of the matter that is given to us, not imposed by us, in the appearance. As I have argued previously, the Kantian account of the phenomenal world is sufficient to ground a substantive correspondence conception of truth. Because it is the matter of a given judgment that distinguishes it from judgments about other similarly determined objects of possible experience, what it means for an empirical judgment to be true is for it to correspond to the appearance it purports to describe. The form which is imposed on appearances is a significant element of truth, it is not, however, the primary factor which determines the truth of this judgment over its contrary. Individual truths, as they are predicated on difference in matter rather than form, are true primarily in virtue of correspondence. Walker admits that 'reports of sensation' are true in virtue of correspondence. What he does not admit, however, is that what is given in sensibility is what distinguishes appearances from one another, and thus it is the primary factor in determining empirical truth. The substance of truth is, in part, a function of the imposition of the formal conditions of experience, and to this extent, coherence is relevant to truth. Coherence, is however, secondary to correspondence, and its role therefore makes for an impure correspondence, rather than an impure coherence conception of truth. Remember Kant's analytic/synthetic distinction. For a judgment to be synthetic its predicate cannot be obtained by any amount of conceptual analysis of the subject. If truth were essentially a matter of coherence between judgments and the formal conditions of experience, then the truth of any judgment could be determined by analyzing the content
92
Kant's Correspondence Theory of Truth
of the pure concepts of the understanding and the concepts of space and time. However, if this were possible, then the judgments in question would not be synthetic because they would not take us beyond an analysis of the concepts of their form. Therefore, Kant could not have intended a primary coherence conception of truth. Primary and Secondary Aspects of Truth To explain the difference between what I call the primary and secondary aspects of Kant's conception of truth, I must first distinguish differentiation from individuation and reiterate the distinction between objective validity and objective reality. I am claiming that it is the matter of appearances, which differentiates between appearances and thereby determines which judgments are true of their objects. I need to specify, however, that I am not rejecting Kant's claim that it is the pure intuitions of space and time that individuate appearances. Space and time separate appearance from one another in the structure of our experience. The contents of our experiences are distinct, however, because of what is given to us in sensation. The distinction between individuation and differentiation is grounded in the distinction between objective validity and objective reality. A judgment is objectively valid so long as it reflects real possibility. A judgment reflects real possibility so long as it does not contradict the necessary conditions for the possibility of experience. Objective validity is a purely formal consideration and it consists in the coherence of our judgments with the pure principles and intuitions of the understanding. Objective reality is the relation of a judgment to something actual, that is, to a differentiated appearance. As such, objective reality is not purely formal. It connects our judgments to both the form and matter of experience. It is an assessment of the relation of an objectively real judgment to its object that determines if that judgment is true. If knowledge is to have objective reality, that is, to relate to an object, and is to acquire meaning and significance in respect to it, the object must be capable of being in some manner given. Otherwise concepts are empty; through them we have indeed thought, but in this thinking we have really known nothing. (A155/B194-195)
The Coherence View 93 If the knowledge corresponds to its object, then the judgment is true. Since (1) the truth of a judgment is ultimately determined by an assessment of its objective reality, (2) objective reality requires differentiated appearances, and (3) appearances are differentiated based on their matter, not their form, then truth must ultimately be a matter of correspondence. In other words, correspondence is the primary aspect of Kantian truth. I argue that Walker is incorrect in asserting that coherence is the substantive conception of truth for Kantian empirical judgments because what sorts our true judgments from false ones is the objective reality of their objects. In other words, it is that they correspond to a differentiated appearance in the world as we experience it. The coherence of our judgments with the transcendental conditions of experience insures only objective validity. As such, it is only a secondary aspect of the Kantian conception of truth. For Kant, even objectively valid judgments can be false. Therefore, the substance of truth must be something beyond coherence. It must be correspondence. Evidence of the primary status of correspondence in distinguishing truth from falsity is found in Kant's account of the distinction between truth and dreams. Both truth and dreams must conform to the requirements of the understanding. If truth were merely a matter of this form, there would then be no way to distinguish truth from dreams. Since Kant thinks there is a way to make such a distinction, then his substantive account of truth must not be an account that focuses on the form of our judgments. That which distinguishes truth from falsity in our empirical judgments must be the matter of our appearances. The coherence of appearances and dreams with the conditions for the possibility of experience is a negative condition of all empirical truth. The correspondence to the appearance, which is made up of both matter and form, is the substantive, or positive, criterion of all truth. The empirical truth of appearances in space and time is, however, sufficiently secured; it is adequately distinguished from dreams, if both dreams and genuine appearances cohere truly and completely in one experience, in accordance with empirical laws. (A492/B521)
If Walker acknowledges the significance of the matter of appearance in grounding the objectivity of the phenomenal world and differentiating
94 Kant's Correspondence Theory of Truth individual appearances, then his view is not really an impure coherence view in any substantive sense. Once the objective reality of the phenomenal world is granted, and the suggestion that correspondence require correspondence to things in themselves is rejected (I argued for this rejection in the previous chapter), then one is hard pressed to explain Kant's definition of truth as "the agreement of knowledge with its object" as something other than substantive correspondence. Kant does not claim to reject the correspondence theory of truth for empirical propositions and he does not need to reject it. If Walker denies the significance of the matter of appearance in determining the truth of our judgments, then he commits himself to an account of sensation like Putnam's, which is inconsistent with the text of the Critique, and that fails to account for the difference between objective validity and objective reality. Since Kant thinks both objective validity and objective reality are necessary conditions for the possibility of being true, any view that conflates them must show good cause. One possible motivation for this move would be Kant's purported solution to the problem of the evil demon. Kant and the Evil Demon Walker's claim that there is a substantive correspondence view of truth attached to our noumenal judgments is motivated by a need to explain our dual status as empirical and moral beings. There are also motivations for suggesting that empirical and transcendental truths are, for Kant, essentially a matter of coherence. One must be able to explain how Kant could respond to Descartes' evil demon hypothesis without resorting to Humean skepticism. Kant sees a progression from an erroneous to an accurate understanding of metaphysics as a science. This progression takes the form of a three-step movement from dogmatism to skepticism to transcendental idealism. These three steps are encapsulated in the views of Descartes, Hume and Kant respectively. (A759/B787-A769/B797) Descartes takes the dogmatic view that things must exist independently of me as the cause of my intuitions. However, the existence of these external entities is not immediately perceived, it is only inferred from the fact that I have a certain set of perceptions.
The Coherence View 95 That the existence of external, or outer, objects can only be inferred from perception leaves them with a merely doubtful existence. (A3 67) There are three possible responses to this situation; dogmatism, skepticism, or transcendental idealism. Since the first two are problematic, Kant argues that the correct solution to the paralogism is to understand that outer objects as we perceive them are empirically real but transcendentally ideal (A368-A380) As a solution to the paralogism of ideality in regard to outer sense, one may take Descartes' suggestion that we can be certain of the existence of the external objects which serve as the cause of our perceptions because there is a God who insures the harmony between my perceptions and external objects, and He would not deceive me as to the nature of those objects. This solution to the paralogism leaves one open to the possibility of a deceiving evil demon. That to which I think my judgments correspond could be not what actually is, but what some evil power wants me to think actually is. The possibility that there is such an evil deceiver leads one to the second step in analyzing metaphysics as a science, namely, skepticism. Hume argues that with the exception of analytic relations of ideas, whatever can be known can be known through experience alone. Therefore, our knowledge cannot go beyond our perceptions. The possibility of an evil demon is irrelevant, because we do not suppose that we can have any knowledge, immediate or mediate, of what might be the source of those perceptions. The skeptical solution, however, is problematic. Although we need not be concerned with an evil demon, the restriction on the limit of our knowledge seems arbitrary. Hume was in error, according to Kant, because he inferred the contingency of laws like causality from the contingency of our determining in accordance with such laws. The passing beyond the concept of a thing to possible experience reveals that such laws are necessary for the a priori synthesis that makes experience possible. The skeptical solution to the paralogism of ideality, then, is inadequate because it restricts knowledge without critically analyzing the limits of knowledge. (A766/B794-A767/B765) The solution to the paralogism of the ideality of outer objects, then, must lie in transcendental idealism. One must understand that the objects
96
Kant's Correspondence Theory of Truth
of outer sense which serve as the cause of certain perceptions, or representations, we have are real. They are real because they are part of what is actual. Upon transcendental reflection, however, we realize that the form of outer objects, both their spatiality and their accordance with certain laws like causality, are dependent on the form of such objects which is supplied by us as finite subjects. Therefore, we need not be concerned with the possibility of an evil demon because the form of the appearances, which we represent in perception, has its foundation in the pure principles of the understanding and the pure intuitions of space and time. Both the form of our appearances and the form of our representations have their foundations in us. All of the forms of experience, and in fact all forms of thought, however, do depend on the laws of logic. Descartes points out that an evil demon could even deceive us as to the veracity of the laws of logic. Since, according to Walker, Kant thinks the laws of logic apply to both things in themselves and appearances; it seems both are susceptible to evil demon problems. According to Walker, this is a problem for correspondence conceptions of truth, but not for coherence conceptions. Walker argues that because noumenal judgments are true in virtue of their correspondence to things in themselves, because of their reliance on the laws of logic, those judgments could be erroneous if an evil demon were at work. However, because the truth of phenomenal judgments is essentially a matter of coherence, they are not susceptible to evil demon problems in virtue of their reliance on the laws of logic. Whereas in noumenal judgments the laws of logic serve to bridge the logical gaps between belief and reality, in phenomenal judgments the laws of logic are just among the principles that must determine the formal conditions of experience. Because they need no outside justification, the laws of logic in the application to phenomenal judgments do not imply the possibility of an evil demon. (Walker 1989, 75) If Walker is correct in his definition of substantive correspondence and in his analysis of Kant's solution to the evil demon problem, then it seems Kant must be committed to some version of a substantive coherence account of truth. However, as I will argue, his definition of substantive correspondence is problematic. Furthermore, the coherence of our beliefs can be seen as the solution of the evil demon problem in a
The Coherence View 97 way that does not imply a substantive coherence conception of truth. As such, Walker cannot substantiate his claim that truth for Kant is essentially coherence with minimal elements of trivial and substantive correspondence. Walker would argue that in making the truth of empirical judgments dependent on correspondence, we have thereby subjected transcendental idealism to the evil demon problem. I do not, however, think that this is accurate. If there is an evil demon, then he misleads us as to the form of our experience, the matter of our experience, or both. As I will argue, Kant has a solution to the evil demon problem given all three possibilities. More importantly, it is a solution that does not commit him to the rejection of a correspondence conception of truth. As I argued, the form of appearances is significant to the truth of our empirical judgments, it simply is not primary. By allowing that the form of an appearance is both significant for truth and dependent on us, we avoid concerns that an evil demon may be misleading us as to the form of our experience. We are not deceived as to form because we provide it, and what is given in the undetermined manifold of sensibility has not form and thus its form cannot be distorted. It may still serve to differentiate between true and false judgments about appearances, and because we determine its formal conditions, we could not be misled as to its form. In the way I have explained, correspondence here is still primary, but coherence plays an important secondary role. Suppose that the evil demon deceived us as to the matter of our experience. If committed to a correspondence account of truth for empirical judgments, would Kant thereby be giving up his solution to the problem posed by such a deceiver? The answer is no. Kant would only be giving up his solution if he maintained that the matter of our appearances grants us mediate access to knowledge of the nature of things as they are in themselves. If this were the case, then there would be a gap between our judgments and what made them true, namely, the nature of things in themselves. It would be a gap that the evil demon could fill. However, this is not Kant's view. I have argued that it is the matter of our judgments that differentiates between appearances with different sensual contents. That does not imply that our intuition of sensual properties is an intuition of the nature of
98
Kant 's Correspondence Theory of Truth
things in themselves (about which we could be wrong if there were an interfering evil demon). Remember that for Kant, sensation is not equivalent to intuition. A sensual quality can only be intuited after it has been determined according to the formal conditions of experience. It follows that, regardless of what we are given as matter (be it the actual transcendent object or a maliciously manipulated version of it) we can establish an objectively real world that can serve to specify a one to one correspondence between the states of affairs that have, do or will obtain and our judgments about them. Kant's solution to the evil demon is not to deny empirical correspondence. Rather, his solution is to redefine it. Our empirical judgments are true in virtue of their correspondence to an objectively real phenomenal world. Whether this world in any way reflects the structure of its ground is irrelevant. The evil demon cannot corrupt our knowledge of things in themselves because we cannot have knowledge of things in themselves. The sensation that has its correlate in things in themselves, although necessary for the differentiation of appearances, cannot be conceived without being determined according to the necessary conditions for the possibility of experience. That our perception of sensation may not accurately relate the nature of its ground is not important. In fact, for it to do so would be impossible. That there is a ground, however, is not irrelevant. It is the common ground of our appearances in the given that insures the objective reality of the phenomenal world and thus insures the viability of a correspondence theory of empirical truth. In short, an evil demon could not deceive us as to the form of our appearances, and a deception as to the matter of appearances would be irrelevant in securing the correspondence of our empirical judgments to their objects. As such, Kant's solution to the problem of the evil demon entails neither an affirmation of a substantive coherence theory of truth for empirical judgments nor a rejection of a correspondence theory of truth for empirical judgments and the appearances to which they correspond. As I have shown, Walker does not provide sufficient reason for thinking that Kant has an impure coherence theory as opposed to an impure correspondence theory (as Kant himself seems to suggest). Moreover, as I have shown, Kant can account for both the possibility of an evil
The Coherence View 99 demon and the fourth paralogism without claiming that the truth of transcendental and empirical propositions is essentially a matter of coherence. One must now consider whether Walker's coherence view fits better with the rest of Kant's theoretical framework. I think that there is a reading of transcendental judgment that makes correspondence the primary factor in its truth. However, since Walker primarily focuses on rejecting correspondence as a substantive conception of empirical truth, I will reserve the discussion of transcendental truth as correspondence for chapter seven. For the moment, however, let us grant that transcendental judgments are, in a sense, true purely by virtue of the fact that they establish the coherent framework for the synthetic unity of appearances. Even so, for the reasons I have given previously Walker cannot establish that Kant subscribes to a primarily coherence conception of truth that is tainted by correspondence. When the role of the transcendent object is thus conceded, and subsequently the objectivity of the world of appearances is conceded, there remains no good reason for contradicting Kant's claim that coherence is a negative condition of empirical truth. Walker must maintain that our empirical judgments are about appearances. If appearances are set by not only the formal, but also the material conditions of experience, and thereby an objective phenomenal world is fixed, one is hard pressed to conclude that judgments about that world are not true primarily in virtue of their correspondence to those phenomena that they purport to describe. The purely logical criterion of truth, namely, the agreement of knowledge with the general and formal laws of the understanding and reason, is a conditio sine qua non, and is therefore the negative condition of all truth. (A59/B84)
Given Kant's description of the role of coherence in truth, my interpretation of it as secondary, rather than primary, is a better fit. Although a negative (what I call secondary) condition of truth, coherence is nevertheless vital to any conception of truth in transcendental idealism. Truth as coherence may be truth about form only, but it is precisely because the forms of appearances are coherent with the pure principle of the understanding and the pure intuitions of space and time that we can avoid the possibility of an evil demon. Even this negative condition is an
100 Kant 's Correspondence Theory of Truth indispensable part of what it means for a judgment to be true. Norman Kemp-Smith Kant is the real founder of the coherence theory of truth. He never himself employs the term coherence, and he constantly adopts positions which are more in line with a Correspondence view of the nature and conditions of knowledge. But all that is most vital in his teaching, has proved really fruitful in its after history, would seem to be in line with the [coherence] positions which have since been more explicitly developed. (Smith 1923, 36)
Norman-Kemp Smith has long been acknowledged as the first person to 'recognize' that Kant makes the groundbreaking move from a correspondence to a coherence theory of truth. This interpretation, or at least variously modified versions of it, has become the generally accepted background view for contemporary Kantians. The motivating idea behind such view is, I think, the assumption that full-fledged correspondence theories of truth are fundamentally inconsistent with a true idealism. In other words, the contemporary wisdom is that to attribute a correspondence theory of truth to transcendental idealism is to deny that it is a true idealism at all. Reality, Categories and Facts Smith asserts this necessary connection between transcendental idealism and coherence. He bases this claim on Kant's postulation of the pure principles of the understanding and the fundamental role of judgment. Concerning the pure principles of the understanding, Smith focuses on the necessity of their role in establishing true judgments (facts). For Smith, an 'asserted fact' is "our description of a given appearance"— a judgment, (xxxvi) A fact simpliciter, then, is a judgment for which the truth conditions are met. Ultimately, of course, everything is real, including what we title appearance; but in the articulation of human experience such distinctions are indispensable, and the criteria that define them are prescribed by the context in which they are being employed. Thus facts cannot be established apart from principles, nor principles apart from facts. (Smith 1923, xxxvii)
The Coherence View
101
There is much to be gleaned from this dense passage which, thoroughly understood, reveals the basic motivation for connecting to coherence. Due to this density and importance, I think a solid understanding of Smith's view is best facilitated by a line by line analysis of the passage. "Ultimately, of Course, Everything Is Real, Including What We Title Appearance. " This claim is, and must be, at the foundation of any nonskeptical interpretation of transcendental idealism. Smith takes it to be a metaphysical claim. We know it is a metaphysical claim because for Kant, (transcendental) logic is essentially a metaphysical, not an epistemic project. (Smith 1923, xxxviii) There are, however, at least two readings of this claim as a metaphysical thesis. (There are, of course, those, like Allison, who read this as a purely epistemic claim.) One metaphysical reading of Kant's claim that everything is real interprets it as both a metaphysical and an epistemic claim. In regard to appearances, they are real in the sense of being actual. To be actual is to have been determined according to the category of actuality. A category, of course, cannot reasonably be said to apply to things in themselves, because that application of the categories is a function the finite understanding performs in the determination of appearances. Therefore, things in themselves cannot be real in the literal sense of being actual. We realize, however, that there must be something that is not an appearance, yet is the ground of our appearances. In other words, there must be things in themselves. The claim that things in themselves are real, however, must be understood as a problematic claim. We only know that they are real by a metaphorical extension of the actuality of our appearances to non-appearances. We can not know that things in themselves exist in the literal sense of having the category apply to them. We must, nevertheless, assume that they exist in an analogous way to the way that appearances exist. This is one metaphysical reading of Kant's claim that everything is real. However, I do not think it is Smith's reading. Smith, I think, takes Kant's claim that both appearances and things in themselves exist quite literally. It seems that he thinks there are two sets of actual things. One of these sets is the phenomenal world, and the other is the noumenal world. There is a real world as appearance and a real world in itself. This is the
102 Kant 's Correspondence Theory of Truth view to which I will refer in my analysis of Smith's interpretation of Kant. Regardless of how one reads Kant's claim that everything is real, there are two significant points. First, we must keep in mind the importance that Kant places on affirming the existence of things in themselves. Second, we must also remember the strong sense in which empirical reality is reality for Kant. Kant maintains that what appears to be the case to us is real in the same sense for a transcendental idealist as it is for a transcendental realist. However, upon reflection, the transcendental idealist realizes that all of the knowable properties of our objects are dependent on finite subjects. "For the world is a sum of appearances; and there must therefore be some transcendental ground of the appearances. . . " (A696/B724) Again, although Kant denies that we may have substantive knowledge of things as they are in themselves, he consistently maintains that there is something which is independent of appearances and serves as their ground. If we take a thing in itself in its loosest sense, then a thing in itself is simply something that exists independently of a transcendental ground. Since Kant thinks that there is something that actually serves as such a ground for appearances, it is reasonable to infer that, at least in this limited sense; Kant maintains the existence of things in themselves. Moreover, Kant says that we must think of things in themselves as real, in a sense. Though this sense is qualified relative to modern realists, it is a meaningful sense nevertheless. "We are enabled to prove the truth of this first estimate of our a priori knowledge of reason, namely, that such knowledge has only to do with appearances, and must leave the thing in itself as real [wirklich] per se, but as not known by us." (Bxx) In other words, although we can only make problematic judgments about things in themselves, we nonetheless are aware of their necessity. There must be something transcendent that serves as the raw material for our determination of appearances. In short, for Kant, appearances must be appearances of something. As for the reality of appearances, Kant is far more explicit. "All outer perception, therefore, yields immediate proof of something real in space, or rather, is the real itself. In this sense, empirical realism is beyond question." (A375) Since actuality is a category, all appearances are
The Coherence View 103 necessarily real, i.e. actual. Kant claims that there is an objective reality about which we make judgments precisely because the phenomenal world is real. "But in the Articulation of Human Experience Such Distinctions [Between Appearances and Things in Themselves] are Indispensable, and the Criteria that Define Them are Prescribed by the Context in Which They are Being Employed. " Smith emphasizes, quite correctly, that, although both appearances and things in themselves are real, the distinction between the two is, in essence, what Kant's theoretical philosophy is about. It is precisely the failure to recognize this distinction that is responsible for what Kant calls "special metaphysics." According to Smith, "'transcendental' is primarily employed by Kant as a name for a certain kind of knowledge. Transcendental knowledge is knowledge not of objects, but of the nature and conditions of our a priori cognition of them." (Smith 1923, 74) Nevertheless, Smith wants to deemphasize the strictness of the transcendental/empirical distinction. "The distinction between the empirical and the a priori must not be taken (as Kant himself takes it in his earlier, and occasionally even in his later utterances) as marking a distinction between two kinds of knowledge. They are elements inseparably involved in all knowledge." (Smith 1923, 36-37) "Thus Facts Cannot be Established Apart from Principles, Nor Principles Apart from Facts. " The set of facts is, according to Smith, the set of true judgments. Once we recognize that what we know it is, at least in part, a product of the structure of our understandings, then we realize that knowledge is impossible without appeal to the principles employed by the understanding. Remember, that for Kant, all knowing is judging. As such, knowledge requires both intuitions and concepts. Intuitions are given to, and determined by the understanding according to its principles (principles being the application of concepts). Pure Concepts, Judgment and Coherence What motivates Smith's assertion of the passage in question is obvious. He is simply encapsulating the basic tenets of transcendental ideal-
104 Kant 's Correspondence Theory of Truth ism. What is less obvious, however, is what motivates him to connect these tenets to a coherence theory of truth. I think the motivation lies in the fact that we need to employ such principles for experience, and our experience is thus, to a degree, dependent on us. (This is the same constructivist idea that, in part, motivates Putnam's verificationism.) It is not, however, the necessity of the pure concepts alone which motivates Smith's coherence view. Smith cites the fundamental role which judgment plays in the Kantian account of knowledge as evidence for his assertion that Kant is committed to a coherence conception of truth. "The fundamental thesis of the coherence theory finds explicit formulation in Kant's doctrine of judgment: the doctrine that awareness is identical with the act of judging." (Smith 1923, xxxvii) "The distinction between the empirical and the a priori is not . . . a distinction between two kinds of synthesis or judgment, but between two elements inseparably involved in every judgment." (Smith 1923, 40) Smith is, in essence, denying that the understanding functions in two separate capacities, one transcendental and the other empirical. Thus we are constantly determining the reality about which we are judging. Every synthetic judgment has both an empirical and a transcendental aspect. Whenever we judge we are simultaneously making a truth bearing judgment and creating the truth-maker for that judgment. Under these conditions, our judgments are true so long as we do not contradict ourselves when we judge. Following Smith's line of reasoning, it becomes obvious that it is his denial of the significance of the empirical/pure (transcendental) distinction that leads him to a coherence interpretation of Kant's conception of truth. He wants to undermine their separation as forms of knowledge and as forms of judgment. By claiming that the empirical and the transcendental are merely two aspects, which simultaneously function in every judgment, Smith eliminates the possibility for a determined unified reality to which empirical judgments might correspond. If the transcendental and the empirical employment of the understanding are seen as separate, a coherence interpretation is less credible. When the transcendental and empirical employment of the understanding are seen as distinct, the picture looks far different from the one suggested
The Coherence View 105 by Smith. According to that picture, the understanding in its transcendental employment determines our unified world of appearance when it determines the undetermined manifold of empirical sensibility in the original unity of apperception. After this activity, experience is possible. We then make truth bearing synthetic judgments (in addition to our analytic ones) which are determined to be true or false based on their relationship to the objective reality, which was set by the transcendental judgments. To this extent, Smith's strong coherence interpretation can be seen as one predicated on his conjunction of empirical and transcendental judgment. Essentially, Smith is claiming that since Kant believes that our faculty of judgment (the understanding) is simultaneously responsible for making truth bearing judgments and for making the truth making judgments which will satisfy the truth conditions for those judgments, it is only reasonable to conclude that Kant is subscribing to a coherence conception of truth. Smith thinks that Kant's view requires us to concede that our knowledge is in no way dependent on independent sense data, and that therefore, the only standard of truth must be coherence. Reply to Smith Smith claims that Kant cannot actually have a correspondence conception of truth, even though it is this type of conception that is suggested by most of Kant's remarks in the Critique. According to Smith, the fact that Kant could not subscribe to a correspondence theory of truth is made obvious by an examination of his accounts of knowledge and experience. It is his position that Kant must have a coherence theory of truth because a correspondence theory is not possible within Kant's theoretical framework. I wish to address some of the assumptions that lead Smith to this conclusion. First, I want to address Smith's inference from the necessity of the pure concepts of the understanding and of judgments which employ them to the attribution to Kant of a coherence conception of truth. It is true that there can be no facts without concepts, because facts are knowledge claims and all knowledge for us requires the application of a concept. It is also true that for Kant, knowledge, and even awareness, requires judgment. These two facts do not, however, make the coherence theory
106 Kant 's Correspondence Theory of Truth of truth a necessity for Kant. While it is true that some theories of truth hold that judgments (or propositions as such views usually call them) exist independently of subjects, this is by no means universal. It is perfectly reasonable to hold that truth is dependent on finite subjects (because they are making the truth assertions which they must confirm or disconfirm) while still maintaining that this confirmation takes place by means of correspondence to reality. This is especially true if the definition of truth-bearers is restricted to propositions or the like. Thus, it is possible to have a correspondence theory of truth even if truth-bearers are necessarily subject dependent. The next question is, is it possible to have a correspondence theory when the truth-makers are not entirely independent of finite subjects? Again, I point out that correspondence theories need not require complete independence from the perceiving subject. (I will clarify this in my discussion of the transcendental and transcendent object.) The connection between principles and coherence is not, however, Smith's strongest support for his coherence interpretation. He maintains that since the grounding of judgments and the judgments themselves happen simultaneously, and both arise from us, in making a true judgment, we are merely maintaining coherence. To show that Kant is not mistaken in his description of truth as correspondence, it is necessary to underscore the importance of the empirical/transcendental distinction. The Importance of the Empirical/Transcendental Distinction To defend Kant's textual description of his conception of truth as a correspondence conception, it is necessary to defend a strict interpretation of the empirical/transcendental distinction because it is the lax interpretation of this distinction that obscures the line between appearances and representations When this line is obscured, describing Kant as having a coherence conception of truth is far more reasonable. Therefore, by weakening the justification for blurring the empirical/transcendental distinction, I will also provide doubt as to the reasonableness of interpreting Kant as a coherence theorist. As Kant is not always consistent in his use of the terms representation and appearance, I will specify what I intend when using them. By
The Coherence View 107 representation, I mean the purely subjective reproduction of an object of thought or experience in the mind. By appearance, I mean an empirical object as it is independently of our representing it. The appearance is the represented; it is "the undetermined object of an empirical intuition" (A20/B34) Appearances are those things which serve as the correlates of our judgments and whose nature and situation either confirm or disconfirm those judgments. (By judgments here, I of course am referring to only truth bearing judgments.) Appearances are the possible objects determined for us in the original unity of apperception. As I mentioned previously, at its crux, this debate is a disagreement over the fundamental way in which transcendental idealism, as a thesis, must be understood. It can be taken as a strictly metaphysical, as a strictly epistemological thesis, or as a thesis that cuts across this distinction. Epistemic interpretations, like Allison's, entail some version of a two-aspect view (each object has an in-itself aspect and an as-appearance aspect). This reading is problematic for reasons already discussed. The alternative is some sort of metaphysical reading. Recall the two types of metaphysical readings discussed earlier. The first, which I take to be Smith's, interprets actuality as applying to things in themselves. The second, a version of which is argued for by Robinson, cuts across this distinction and sees actuality as applying to things in themselves only in an analogous sense. Again, according to Smith's metaphysical reading, transcendental idealism entails a strict two-world view, according to which there are two sets of ontologically distinct objects; phenomenal objects and noumenal objects. Smith argues that since transcendental and empirical judgment function together in every case, and since transcendental idealism is primarily a metaphysical doctrine for which there are two distinct and noninteracting worlds, the apparent correspondence between our judgments and their objects is not really correspondence but coherence. However, given careful consideration of the text and of the spirit of transcendental idealism, both of the premises of this argument must be rejected. First, if transcendental and empirical judgment functioned together in every case, then the notion of an original unity of apperception that creates a unified reality of objects prior to any experience would make no sense. If whenever we make an empirical judgment, we are simultane-
108 Kant 's Correspondence Theory of Truth ously making its truth making judgment, then we are creating the empirical world as we go. Given that Kant thinks his is an empirical realism in this strictest sense, this almost solipsistic view is unacceptable. Furthermore, Kant makes reference to the importance of unschematized categories as per their role in creating this unity of consciousness, i.e. in transcendental apperception. If transcendental and empirical judgment were always employed in concert, it would be impossible for there to be an important employment of unschematized categories. For Kant, however, the employment of unschematized categories is a necessary condition for the possibility of experience. "For only by means of these fundamental concepts can appearances belong to knowledge or even to our consciousness, and so to ourselves" (A125) In other words, if there were no empirically independent transcendental employment of the understanding, then there would be no explanation for the unity of consciousness. The categories, as applied to the concept of an object in general, establish the unity of consciousness into which appearances will fit. This unity is established in the absence of any empirical employment of the understanding. In fact, the unity of consciousness must be independent of and priori to any empirical employment of the understanding. Transcendental apperception is the "unity of consciousness which precedes all data of intuitions, and by relation to which representation of objects is alone possible." (A 107) It is in this way that the finite subject determines the manifold of appearances according to the pure intuitions of space and time and the categories. The categories are the concepts that serve as the rules to construct this unity of consciousness. These transcendental judgments synthesize what is given in pure sensibility, as determined by space and time, according to the rules provided by the pure concepts of the understanding. [E]ven the purest objective unity, namely, that of the a priori concepts (space and time), is only possible through relation of the intuitions to such unity of consciousness. The numerical unity of this apperception is thus the a priori ground of all concepts, just as the manifoldness of space and time is the a priori ground of the intuitions of sensibility. (A 107) This transcendental unity of apperception forms out of all possible appearances, which can stand alongside one another in one experience, a connection of all these representations according to laws. (A 107)
The Coherence View
109
To deny that there is a transcendental employment of the understanding which is entirely distinct from the empirical employment is thus to contradict one of Kant's fundamental assumptions concerning the nature of knowledge and experience. Correlatively, Smith's metaphysical reading of transcendental idealism is committed to a two-world account of the phenomena/noumena distinction and is thus untenable from the start. Just as obscuring the empirical/transcendental distinction undermines the importance of transcendental judgment, any interpretation that is committed to a two-world view negates the significance of the basic thesis of transcendental ideality, namely, the distinction between representation and appearance. As I will show, transcendental idealism is doomed in the absence of this distinction. The Transcendental Object In addition to undermining the separateness of the transcendental and empirical functions of the understanding, Smith's view disregards the transcendental deduction of the categories. Kant needs to show that the categories are necessary as conditions for the possibility of experience because there is a fact not explained by the metaphysical deduction—the continuity of experience across subjects. (A90-91/B123-124) If reality is dependent on particular subjects, then there is no reason to think that reality would not differ from subject to subject. Individuals differ in many ways—why not in this one? The answer is that all finite subjects necessarily apply the pure principles of the understanding. The nature of finite, discursive understanding requires it. So, experience is dependent on a finite subject qua finite subject, it is not finite subject qua individual. This distinction at first seems to be nothing more than philosophical hair-splitting. After all, the continuity of experience can easily be explained if all individual subjects put the world together in the same way. However, even if all finite subjects did put the world together in the same way, this would not explain the commonality of our experience. There is an important question to which Smith does not provide a satisfactory answer. That question is, what are we putting together? In other words, what is the raw material for the transcendental employment of the under-
110 Kant 's Correspondence Theory of Truth standing? Smith infers that because we determine the nature of reality (by putting together the synthetic unity of appearances), truth must be a matter of coherence. The error which leads Smith to draw this inference is his assumption that knowledge does not begin with sense data. According to Kant, it does begin with sense data—sense data in the form of givens. I am referring to the undetermined manifold of intuition that is the 'object' of the original unity of apperception. Over the course of the Critique, Kant purports to establish that we all order reality according to the same basic blueprint (the categories and space and time). Nevertheless, he needs more to explain the uniformity of reality. It seems that, for the most part, we all move around in the same world. Imagine that someone asks you for directions to the movie theater. You provide her with a list of instructions to follow, based on a seemingly unproblematic assumption, namely, that experience is communicable because you are experiencing the same world. To say that we all structure the world in the same way, that we have the same 'blueprint' for reality, is not enough to explain the consistency of the empirical world across subjects. After all, a contractor can use the same blueprint on two different houses (and follow it precisely) and have the houses turn out to be completely different. It is possible, for instance, that the builder built the first out of brick and the second out of straw. It is unlikely that if he said, "go and turn at the house just like this one", that his directions would be very helpful to anyone. However, when we communicate about our experiences, this is exactly what we do. We assume that the other person is able to follow what we say because their 'house' is just like ours. We can assume this because not only do we have the same blueprint, we start with the same raw material. Throughout the Critique, Kant makes reference to 'transcendental object(s)'. There is, however, no lack of controversy over exactly what Kant means by 'transcendental object.' Allison claims that it is unclear if Kant always means to pick out the same sorts of things with the term. (Allison 1983, 242-246) There are passages in which it seems that by transcendental object Kant means noumena or things in themselves. (A3 66) In other passages, Kant describes the transcendental object as giving objective validity to the categories as the correlate of the original
The Coherence View 111 unity of apperception. (A 109) Allison argues that Kant cannot attribute all of these qualities to the same entity, otherwise he would be stuck with an undesired transcendental realism. (Allison 1983, 244) It is possible, however, that Allison's concerns about inconsistency are unwarranted. There is a way in which the definitions of transcendental object as noumenon and transcendental object as a necessary (but not sufficient) condition for the objective validity of the categories are compatible without implying the transcendental realism that Kant wishes to avoid. First, it is important to remember that Kant in no way denies the existence of things in themselves. To deny this would relegate Kant to an idealism like Berkeley's which he vehemently denies in the Refutation of Idealism. In fact, Kant refers to things in themselves as things which exist apart from their transcendental ground, and which can remain standing while we seek an outside cause of their existence. We can, and must, act as if they are real. (This works even if we concede that we can only say that noumena exist problematically, in a way that is analogous to the way that phenomena exist.) I think Kant's remarks in the first Critique imply a distinction in mind between two kinds of objects. Even if this implication is not present, it is at least a distinction that serves as a reasonable reading of transcendental idealism. These two kinds of objects are a transcendental object, and a transcendent object. They differ drastically in kind, of course, but both serve as necessary components for the possibility of experience. Just for this reason, the categories represent no special object, given to the understanding alone, but only serve to determine the transcendental object, which is the concept of something in general, through that which is given in sensibility, in order thereby to know appearances empirically under concepts of objects. (A251)
The transcendental object is the concept of an object in general, and as such, it is the object of the original unity of apperception. In addition, synthetic a priori judgments need something given in sensibility. What is given in sensibility is the transcendent object. It is noumena serving as the transcendent ground of our appearances. To put it briefly, the transcendent object is the raw material out of which our appearances are con-
112 Kant 's Correspondence Theory of Truth structed. It is the given that, in conjunction with the concept of an object in general and the pure concepts of the understanding, combines to produce for us a world of possible experience. This structure does not commit Kant to the transcendental realism that he explicitly rejects. Space, time and the categories (the parameters of possible experience) do not apply to things in themselves. As such, nothing noumenal can be an object of experience for us. However, the noumenal insofar as it serves as the undetermined manifold of sensibility, can serve as a correlate for the transcendental judgments that make experience possible. Instead of committing Kant to transcendental realism, this interpretation has Kant committed to an empirical realism, transcendental idealism, and a transcendent realism. These three theses are not only acceptable for Kant, they are accepted by him. That he accepts transcendental idealism and empirical realism is obvious. Likewise, in arguing that things in themselves must, in a sense, exist, Kant is affirming transcendent realism. This does not imply that the noumenal world is one that we are capable of knowing, it merely confirms that the noumenal world is not empty. My interpretation of the transcendental object succeeds where Smith's interpretation fails, because it preserves the importance of both the transcendent and transcendental objects to Kant's view. Moreover, my interpretation fits better with Kant's remarks about both truth and the unity of experience. In preserving the importance of these two kinds of objects, I also weaken the case for reading Kant as subscribing to a coherence theory of truth. The case is weakened because, given that there is an objective ground out of which our appearances are constructed, it no longer seems to be the case that our truth-makers are entirely a product of our own understandings. There is something to the appearances that serve to confirm or disconfirm our judgments that is not a product of our faculty of judgment. That something is that which is given in the original unity of apperception, it is the transcendent object. Hence, since our truth-makers are not entirely a product of our own faculty of judgment, true must mean something more than the coherence between our judgments. I have shown that there are good reasons to doubt Smith's claim that Kant is committed to a coherence conception of truth. First, he overstates
The Coherence View 113 the requirements for correspondence. Second, he mistakenly deemphasizes the importance of the empirical/transcendental distinction for knowledge and judging. Finally, Smith ignores the significance of the need for a transcendental deduction and the importance of the transcendental and transcendent objects. Given these considerations, a pure coherence theory of truth no longer seems viable for transcendental idealism. Summary Walker fails to establish that his interpretation captures what Kant intends by his conception of truth. There is a great deal of textual evidence for a correspondence reading of truth for transcendental idealism. Because of this evidence, any view that suggests a contrary reading must show that the alternate view is somehow more in the spirit of transcendental idealism than the textual view. As I have shown, his view does more to bring out than to suppress conflicts within transcendental idealism. His reading is indicative of a misconstrual of Kant's analytic/synthetic distinction, and it ignores the pervasive importance that what is given in sensibility plays in the determination of the truth or falsity of our empirical judgments. Furthermore, one need not adopt Walker's reading to avoid evil demon problems for transcendental idealism. Smith's pure coherence reading is problematic because it undermines the transcendental/empirical distinction. As a result, Smith commits himself to a two-world view of the phenomena/noumena distinction which is inconsistent with transcendental idealism in its own right, and which undermines the importance of both the transcendental and transcendent object. Moreover, the two-world view undermines transcendental idealism as a viable alternative to contemporary epistemological theories. Thus, I do not think Smith's view serves as the best possible interpretation of Kant's view. For these reasons, neither Walker's impure coherence theory nor Smith's pure coherence theory is adequate as a conception of truth for transcendental idealism. Walker's reading does show, however, that there must be some relationship between correspondence and coherence that serves to fully determine the truth of our truth bearing judgments.
114 Kant 's Correspondence Theory of Truth Possible interpretations of this relationship will be the subject of the next two chapters.
Chapter Six The Hybrid View An Introduction to Hybrid Views There is a set of interpretations of truth for transcendental idealism, similar to Walker's impure coherence view that does not posit a single standard for satisfying truth conditions. The members of this set I will call hybrid views, because they claim that transcendental idealism requires a hybrid of correspondence and coherence to make sense of the truth of all of its kinds of judgments. There is not consensus among these views as to what the exact combination of coherence and correspondence ought to be. Heimo Hofmeister and Paul Copan explicate Kant's conception of truth in this way and I believe their two views represent a good sample of the various ways in which the hybrid view is conceived by contemporary authors. Hofmeister's View Hofmeister argues that Kant must be understood as having a hybrid conception of truth that combines empirical correspondence and coherence as nominal conditions of truth with an account of transcendental truth as the agreement of knowledge with its object. His view is a hybrid that emphasizes a kind of transcendental correspondence account of substantive truth. He argues that for Kant, neither empirical correspondence nor coherence captures what Kant means by truth. According to Hofmeister's interpretation, both empirical coherence and correspondence are merely negative conditions for truth in transcendental idealism. His criticisms are grounded in his belief in the primacy of transcendental truth. The Inadequacy of Coherence According to Kant, a coherence theory of truth defines truth as agreement with the formal procedures of the understanding and reason, that is, as the agreement of propositions with each other. (Logic, 51) Hofmeister claims, and I agree, that a coherence theory can never establish truth beyond these formal conditions. As such, it is too restrictive and is only a negative condition of all truth because it cannot account for errors in content as it can for errors in form. "There the criterion of truth
116 Kant 's Correspondence Theory of Truth refers to the formal rules of thought only. Therein for Kant lies its univocity." (Hofmeister 1972, 317) The Inadequacy of Correspondence Hofmeister argues that, just as coherence is only a negative condition for truth, so it is with correspondence. He cites Kant's remarks about correspondence as a general criterion of truth in the Transcendental Logic as evidence that, for Kant, correspondence is only a negative criterion for truth. The problem is that a general criterion of truth would have to abstract particular content from all objects in order that it is universally applicable to all judgments. However, to be useful in picking out true judgments from false ones, a criterion of truth must be able to differentiate among judgments based on particular content. Therefore, the notion of a general criterion of truth as correspondence that is at the same time useful in establishing the truth of particular judgments is absurd. Therefore, correspondence, the agreement of a judgment with its object, can only be a negative criterion for truth. (A58/B83) Its [truth's] definition as the agreement of knowledge with its object is valid only as a nominal definition; it is still insufficient, however, because it is based upon the simple verbal reference to experience and cannot count as a theory of truth for Kant if it is to relate, in the sense of naive empiricism, to an unproblematic given object. (Hofmeister 1972,318)
Hofmeister's claim that the correspondence theory of truth is inadequate for Kant seems to be based on his assumption that such a theory of truth implies a simple empiricism. In other words, to hold that a true judgment is just one which agrees with its object suggests that all true judgments, and thus ultimately all knowledge, must be derived solely from experience. This is not Kant's account of knowledge, one of the central aims of the Critique of Pure Reason is to demonstrate that we have synthetic a priori knowledge. Thus, if Hofmeister is correct about what a correspondence theory of truth implies, then this surely cannot be what Kant has in mind as a sufficient account of truth. Otherwise, transcendental idealism would be a doomed theory.
The Hybrid View 117 Transcendental Truth Hofmeister argues that to give an account of truth for empirical propositions, one must first give an account of truth for transcendental propositions. Transcendental logic can provide the criterion for establishing the truth of empirical judgments because transcendental logic recognizes the understanding as a system of principle without which no object can be thought. Only in transcendental logic can we have an account of truth that does not lose all content in its definition. (Hofmeister 1972,318) Upon considering transcendental logic, I realize that one thing is common to all of my thoughts, namely that they are mine. I recognize myself as an I, and I recognize that it is the I that makes judgments, and it is only for the I that such judgments are valid. In other words, I cannot think of an object without thinking of it as a part of my consciousness. On this basis, I realize that by a spontaneous act of the understanding, I create the world of which I am conscious by combining the formal conditions of experience (the pure principles of the understanding) with the material conditions of experience (the a posteriori manifold of sensibility). "It is in the combination of a priori rules with the a posteriori manifold that experience generates the truth of its judgments, by virtue of generating its object." (Hofmeister 1972, 319) Should I engage in transcendental reflection of the process just described, Hofmeister maintains that I must conclude that the truth of empirical judgments is not an agreement between knowledge and its object. Rather, empirical truth is the result of the agreement of knowledge with its object. (Hofmeister 1972, 319) It seems that according to Hofmeister, the two are mutually exclusive. Empirical truth must either be the agreement of knowledge with its object or the result of the agreement of knowledge with its object. Hofmeister bases his interpretation of Kant's conception of truth on the fact that transcendental truth is a necessary condition for the possibility of empirical truth. Empirical truth may be prima facie a combination of correspondence to objects of experience and coherence between the principles of judgment. In the full sense, however, empirical truth is itself nominal. Truth, in its substantial sense, is the agreement of knowledge with its object, which is determined in the original unity of apper-
118 Kant 's Correspondence Theory of Truth ception. This sense of truth, transcendental truth, alone makes all other truth possible. "Judgments of experience are therefore true and in agreement with their object if they rest upon the transcendentality of truth. As mere judgments of experience they are, however, only hypothetical, conditional, and not absolutely true." (Hofmeister 1972, 320) Reply to Hofmeister I do not disagree with Hofmeister concerning the primacy of transcendental truth. It is true that, for Kant, empirical truth would be impossible were it not for the fact that our transcendental judgments determine the objects of possible empirical experience. I do disagree, however, with Hofmeister's inference from the primacy of transcendental truth to the conclusion that empirical truth is not substantive truth. Although there are necessary conditions for the possibility of empirical truth, this does not necessarily imply that empirical truth is itself only truth in a nominal sense. Kant's Empirical Realism Hofmeister's conclusion that empirical truth is only truth in a trivial sense undermines one of Kant's fundamental assumptions, namely, empirical realism. Kant maintains that we must understand appearances as real. They in fact constitute the real. Only by ascribing to them reality in the strongest sense can we avoid the conclusion that everything is illusion. (B70) Thus, when we make an empirical judgment, we are making a judgment about something real. Concerning assertoric judgments, which are necessarily empirical because we can only speak of the non-empirical problematically, Kant claims that their affirmation or negation "is viewed as real (true)." (A75/B100) Because appearances are real for Kant, and because the affirmation or negation of assertoric judgments is a matter of establishing reality, or truth, it is unreasonable to suggest that empirical truth is, for Kant, merely trivial. Knowledge as a Four-Term Relationship I think that Hofmeister's description of empirical truth as trivial is predicated on a misconstrual of the relationship between finite subject
The Hybrid View 119 and empirical object for Kant. The essence of Kant's "Copernican revolution" is a redefinition of the subject object relationship. The traditional, i.e. transcendental realist, conception of the subject/object relationship is one in which the representations of the subject conform to the structure of the object as it is in itself. Given this picture, an empirical judgment is true just in case our judgment about an object as we represent it corresponds to the object as it is in itself.
s<
o
Figure One
Many scholars (including, I think, Hofmeister) have taken Kant's Copernican revolution to be a simple reversal of this picture. According to this view, the nature of appearances (empirical objects) is entirely determined by the epistemic conditions of the finite subject. Given this structure, if there were no finite subjects, there would be no objects. This view is problematic for Kant because it suggests empirical idealism.
s
>o
Figure Two
In think that this is the view that Hofmeister has in mind, and it is for this reason that he thinks empirical truth is only truth in a nominal sense. Since, however, this view does not support empirical realism to the degree Kant wants to establish it; it is inconsistent with the intended scope of Kant's transcendental idealism. The correct understanding of the subject/object relationship in Kant is one which construes it as a four-term relationship between the finite subject, the object as appearance, the transcendental principles which constitute the epistemic conditions for the subject, and things in themselves which must exist as the ultimate ground of appearances. In this view, our judgments about appearances are true in virtue of correspondence to those appearances. Those appearances, however, only have knowable properties for us in virtue of the fact that the undetermined manifold of empirical sensibility—the transcendent object (thing in itself) is determined according to the formal conditions of experience.
120 Kant 's Correspondence Theory of Truth These conditions are, again, the pure intuitions of space and time and the categories. Those formal conditions, as functions of the understanding, are dependent on the finite subject. The illustration was gathered from the Kant lectures of Alexander Von Schönborn at the University of Missouri, Columbia. According to this view, an empirical judgment is true, is a substantive way, just in case it corresponds to its object. Formal Conditions + Manifold
/
! p.
S
O
Figure Three
Moreover, this picture preserves the importance of our epistemic conditions in making objects knowable for us. This view preserves both Kant's empirical realism and his transcendental idealism. Summary Because Hofmeister's view conflicts with Kant's remarks concerning the substantive nature of empirical truth, and because it suggests a reading of the subject/object relationship that, could be interpreted as an empirical idealism, Hofmeister's interpretation of Kant's conception of truth is inadequate. He wrongly assumes that because correspondence as a general criterion for truth is only a negative condition, that correspondence itself is only a negative condition for truth. What Kant has in mind in the transcendental logic is not to establish that empirical truth as correspondence is trivial. Rather it is his intention to show that empirical truth is grounded, at least in part, in the structure of our understanding. In addition to truth as correspondence, we must also add the knowledge that the knowable properties of the objects to which our empirical judgments correspond are dependent on our understandings. This, with correspondence, is a sufficient set of criteria for empirical truth. Hofmeister's hybrid view fails because of implications that are inconsistent with a reasonably favorable reading of transcendental ideal-
The Hybrid View 121 ism. His is not, however, the only interpretation of Kant's conception of truth as a hybrid of correspondence and coherence. Paul Copan also suggests a version of this view. Copan's View Paul Copan begins his analysis of the Kantian conception of truth by acknowledging that it is not necessary for an idealist philosopher to ascribe to a coherence theory of truth. (Copan 1998, 11) He also distinguishes Kant's transcendental idealism from subjective idealisms like Berkeley's. Transcendental idealism is different from subjective idealism because "Kant believes that there are things in themselves, not simply appearances." (Copan 1998, 12) He concludes that, given the nature of Kant's transcendental idealism, truth for Kant must be a 'partial synthesis' of correspondence and coherence conceptions of truth. Phenomena andNoumena Revisited Copan correctly points out that for Kant, the phenomena/noumena distinction is not a distinction between two ontologically distinct sets of objects. Moreover, the noumenal is necessary for the phenomenal because "the things in themselves furnish the basis for any appearance. Without the noumenal there could be no phenomenal, for how can there be appearances of things if there are no actual things?" (Copan 1997, 12) According to Copan, this thesis is argued for explicitly by Kant. He cites the following passage as evidence for his claim. "For the world is a sum of appearances; and there must therefore be some transcendental ground of the appearances, that is, a ground which is thinkable only by the pure understanding" (A696/B724) From this, Copan concludes that object is to be understood in a twofold sense. For every object, there is a way that we perceive it, and a way that it is in itself. To this extent, his is a version of what Allison calls the two-aspect view. Copan's version of the two-aspect view entails that things in themselves and things as they appear to us are simply "two possible ways of apprehending one set of objects." (Copan 1998, 12) I take Copan to mean that there is one set of things that exists. For each thing there is a way we apprehend it (as appearance) and a way God would
122 Kant 's Correspondence Theory of Truth apprehend it (as thing in itself)Correspondence as Kant's Primary Conception of Truth Given this information, Copan thinks we ought to infer that Kant subscribes to a correspondence conception of truth, with a few coherence-based modifications. Since their grounding in things in themselves gives them objectivity, (appearances are just the way we understand things as they are in themselves) appearances may serve as the truthmakers for our empirical judgments. Thus, the judgment "Grass is green" will be true just in case in the phenomenal world the object 'grass' has the property 'green'. In other words, the judgment will be true just in case it corresponds to what obtains in the phenomenal world. Copan argues against authors like Smith and Walker who claim that coherence is the primary conception of truth for Kant. He allows that coherence is important because we must have consistency between our concepts and our intuitions, but "a mere coherence theory will not work. It must be supplemented by something mind-independent. . . . noumena—the objective grounding of all appearances."(Copan 1997, 15) A coherence account of truth would not be sufficient for Kant because "Kant assumes that the human mind does not create appearances; they are not merely private sensations. Rather, they are based on something external to us." This external ground cannot be accounted for by a pure coherence view because a pure coherence theory is one "in which knowledge and the object of knowledge are conflated into identicality." (Copan 1997, 15) Kant's Partial Synthesis Despite his conclusion that a pure coherence account of truth is insufficient for Kant, Copan does not infer that Kant must have a pure correspondence view. Kant's conception of truth must be seen as a synthesis of correspondence and coherence because although Kant maintains that there is an independent ground for our appearances, he "introduces an inter-subjective element into the discussion as well." (Copan 1997, 15) The inter-subjectivity to which Copan is referring is Kant's postulation of the necessity of transcendental judgment. To Copan, this means that we can only have appearances that cohere with the structure of our
The Hybrid View
123
understandings. There must be agreement between our cognitions and their objects. (Copan 1997, 15) While it is certainly true that for Kant, all appearances must cohere with the pure concepts of the understanding, whether this implies the need to have a substantial and positive criterion of Kant's theory of truth be grounded in coherence remains an open question. Summary Copan argues that each object has two-aspects, an as appearance aspect and an in-itself aspect. Truth must include coherence because the as appearance aspects must cohere with the structure of our understandings. However, truth is primarily a matter of correspondence. Our empirical judgments are ultimately true in virtue of their correspondence to the external object that grounds our appearances. It is not the case that appearances are copies of things in themselves. They are just the aspects of things in themselves that cohere with the pure principles of the understanding. the fact that an agreement of our cognitions with their objects is possible does not require that these objects be originals and that our cognitions be their "copies." When we are being appeared to, the objects we receive through intuitions have an existence by themselves, independent of our constructions of them. They appear to us as having certain properties—properties which [we] have not constructed—even though these properties cannot be well-defined and are limited by our own "distorting filters" of intuitions and concepts. (Copan 1997,16)
Copan concludes that since what appears to us is not of our own construction, although coherence with the pure principles of the understanding is a negative condition for all truth, the primary sense in which our judgments are true is in virtue of their correspondence to appearances as they are grounded in things in themselves. Response to Copan The Two-Aspect Problem Revisited Copan subscribes to a two-aspect view of the phenomena/noumena distinction. As such, his view is subject to the same criticisms leveled
124 Kant 's Correspondence Theory of Truth against Allison's view. These criticisms were addressed in detail in Chapter Five, so I will only summarize briefly here. Foremost among the problems with the two-aspect view is the implied filtration model. Copan admits to this filtration view by claiming that our appearances are clouded by the "distorting filters" of intuitions and concepts. The problem with the filtration model is that it implies a structural isomorphism between appearances and things in themselves. This, again, is problematic because appearances are individuated according to our spatial and temporal intuitions. For there to be a one to one correspondence between our appearances and things in themselves, things in themselves would have to be similarly individuated. Since Kant thinks space and time are intuitions that we add to our appearances, he would not allow these designations to apply to things in themselves. This is not substantive knowledge of things in themselves. We do no what relationships are present noumenally, but space and time as we understand them are not present. Knowledge of Things in Themselves Correlative to the isomorphism problem presented by Copan's filtration model is his assertion that we can know things in themselves. He suggests that in intuition we are presented with properties of the object as it is in itself which were not constructed by us, but which have been well defined and limited by our intuitions and concepts. (Copan 1997, 16) Kant explicitly prohibits the possibility of our having direct, substantive knowledge of things in themselves, even in a limited and well-organized way. Kant maintains that we can only know things in themselves problematically. "The true correlate of sensibility, the thing in itself, is not known, and cannot be known through these representations; and in experience no question is ever asked in regard to it." (A30/B45) Because he does not incorporate this prohibition into his view, Copan's interpretation comes dangerously close to making Kant a transcendental realist. Finally, Copan gives no substantive account of transcendental truth. One may infer that they are true in virtue of their coherence with one another, or one may infer that they function as rules and as such have no truth-value. It is unclear if Copan has one of these or some other account
The Hybrid View 125 of transcendental truth in mind. This omission is problematic because Kant explicitly states that there are transcendental judgments and that there is transcendental truth. A view, like Copan's, which does not address the nature of truth for all kinds of judgments, is problematically incomplete. Summary Because Copan's view is committed to the problematic two-aspect view of the phenomena/noumena distinction, and because it allows that we have knowledge of things in themselves, it is inconsistent with transcendental idealism as Kant construes it. In addition to being inconsistent with some of the basic theses of the theory for which it purports to provide a theory of truth, Copan's view is problematically incomplete. Therefore, his is not a reasonable interpretation of what Kant means by 'truth'. Conclusion Hofmeister and Copan argue that neither a coherence nor a correspondence view alone will suffice for transcendental idealism because Kant demands that appearances cohere with the pure principles of the understanding while maintaining that things in themselves exist. Neither Hofmeister nor Copan, however, succeeds in providing a conception of truth for Kant, which allows for both of these factors without compromising with Kant's empirical realism or his transcendental idealism. Given that they fail to preserve the basic structure of Kant's system, hybrid views, at least of the sort proposed by Hofmeister and Copan, cannot fit into an interpretation of transcendental idealism that is consistent with Kant's explained intent and scope.
Chapter Seven An Idealist Correspondence Theory A Brief Review In the preceding chapters, I have suggested that all of the standard views fall short of capturing a conception of truth in Kant's theoretical philosophy that is (1) textually supported, (2) consistent with a reasonable reading of transcendental idealism, and (3) tenable as an account of truth in its own right. Having given a negative analysis of how the standard views fail to meet these desiderata, a positive delineation on my part is in order. In the course of presenting my account, I will explain exactly what I mean by correspondence, and what I take the nature of the relevant objects for each kind of truth to be. I will begin by explaining what I take to be the basic tenets of a correspondence theory of truth. Then, as I am advocating a global account of truth that is made up of local accounts, I will break my interpretation into three parts that ultimately interconnect to form the whole. These are: (1) analytic truth, (2) empirical truth, and (3) transcendental truth. Correspondence Theories of Truth in General In general, correspondence theories of truth are founded on the intuition that truth is, in some way, a relation to reality. The classical version of the theory describes this relationship as follows; something is true if and only if it corresponds to a fact about the world, and it is false if and only if it does not correspond to such a fact. Given this basic intuition, it is unsurprising that most correspondence theories of truth accompany realist accounts of metaphysics rather than idealist ones. However, although the correlation between correspondence theories of truth and realism is prevalent, it is neither universal nor necessary. Note that this definition, which I think captures the essence of a correspondence theory of truth, makes reference to facts about the world, not facts about things as they are independently of any intuition of them. Although tradition certainly suggests otherwise, I see no good reason for thinking more is required of a correspondence theory of truth other than a demand that true judgments accurately map an objective reality. While
An Idealist Correspondence Theory
127
the objectivity of Kant's empirical reality requires a role for things in themselves. The empirical reality itself is in large part dependent on the finite rational mind. Given the Kantian distinction between representations and appearances, Kant satisfies this demand for the relationship between judgments and reality. The synthetic unity of appearances, as determined by the original unity of apperception, has sufficient objective reality to serve as the truth-maker for our empirical judgments. I will address this point in more detail in my analysis of empirical truth. In its most basic version, a correspondence theory of truth holds to the following definition: X is a true judgment = x is a judgment, and there is some fact y such that x corresponds to y; X is a false judgment = x is a judgment, and there is no facty such that x corresponds to y.
What is a Fact? The challenge of assessing correspondence theories of truth is to discern more precisely what facts are, and in what way judgments do or do not correspond to them. G.E. Moore addresses this problem in Some Main Problems in Philosophy. Moore suggests that to serve adequately their role in the correspondence theory of truth, facts must exist. They do not exist in the way ordinary physical objects do, but they are objectively physical objects. (Moore 1953, 267) In claiming that facts are 'objectively physical objects', I take Moore to be claiming that facts obtain in the world. He argues that for each true belief, there is exactly one fact to which that belief refers. For false beliefs, there are no facts to which they refer. A fact then must be, in one sense, something present in the world. However, facts are not ordinary physical objects. For example, it is a fact in the world that fire is hot. Fire itself is not a fact, but that it has certain constant properties is a fact. Moore describes facts in another way as well. He describes the term fact as interchangeable with the term 'truth' under certain conditions. The term truth applies "not only to the class I mean but also to the forms of words by which we express them." (Moore 1953, 306) Thus facts for Moore have a dual nature. They are both what obtains in the world, and
128 Kant's Correspondence Theory of Truth expressions of what obtains in the world. A belief is a fact then, just in case it corresponds to the appropriate fact. A problem with Moore's suggestion, and with most views that assert the dual nature of facts as both language/mental acts and features of the world, is espoused by Lawrence E. Johnson. He points out that for the correspondence theory of truth to work; it seems that facts are necessarily tied to cognition, yet necessarily distinct from it. Facts must be things that are stated (as sentences, judgments, etc.), and they must also be hard features of the world. This appears to be an enormous obstacle for some supporters of the correspondence theory of truth, and has prompted many proponents to advocate an analysis of correspondence to reality in terms of states of affairs rather than in terms of facts. This is useful in that it is easier to justify a skeptical response to the assertion that facts are both physical and linguistic than it is to justify a skeptical response to the assertion that there are states of affairs which obtain in the world. If those truth-makers which obtain in the world are referred to as states of affairs, and the term 'fact' is reserved for the truth-bearers which purport to correspond to the relevant state(s) of affairs, then 'facts' no longer have a dual nature. Facts are not 'objectively physical objects' instead, they are assertions, or possible assertions, which contain relationships that match relationships between states of affairs in the world. This solution to the problem of the dual nature of facts works because of the vagueness of the term 'states of affairs.' A state of affairs could be an event, the spatial and/or temporal presence of an object with certain properties, an obtaining mental state, or a number of other similar kinds of things. Moreover, because it does not require a one to one relationship between true beliefs and their facts, and no relationship at all between false beliefs and facts, this interpretation allows for partial truths. A purported fact could accurately map some aspects of a state of affairs, and thus be a partial truth. This is not a possibility in Moore's view, but it does regularly occur in ordinary language use. Since correspondence as correspondence to a state of affairs is a somewhat weaker claim than correspondence as correspondence to a fact about the world, it better captures our use of language and avoids certain problems. As I will argue, for Kant, the truth condition for empirical
An Idealist Correspondence Theory
129
judgments is that they correspond to the relevant state of affairs in the way just described. Nevertheless, more needs to be said about the nature of correspondence because correspondence can occur in a number of ways. What Does Correspondence Mean? Now we must explain exactly what it means for one thing to correspond to another. Schmitt argues that there are three kinds of correspondence theories of truth, each of which conceives of the nature of correspondence in a different way. These are correspondence as logical relation, correspondence as correlation and correspondence as congruence. I will explicate each of these kinds of correspondence and argue that a variation of each constitutes one aspect of the Kantian conception of truth. Correspondence as a Logical Relation. Moore's view maintains that a belief is true just in case it corresponds to a fact. In this view, truth is essentially relational. What it is to have a true belief (a fact, in one sense) 'p' is for there to be a fact in the world 'p', the content of which is identical with the content of the belief. G.E. Moore's account of truth posits just this sort of correspondence. A true belief will have the same content as the fact to which it refers. As I have argued, there are a number of problems with a correspondence as logical relation account of correspondence, at least as Moore construes it. These problems center on the dual nature of facts in Moore's account, and the identity relationship between truth bearer and truth-maker, which this dual status of facts implies. Frederick Schmitt captures the spirit of Moore's view: The belief that p corresponds only to the fact that p. The belief that p is false just in case there is not fact to which it corresponds.. . .This theory . . . makes correspondence a logical relationship between judgments and facts: a belief is true when it corresponds to a fact in the sense that it has the same content as some fact. (Schmitt 1995, 170-171)
There are a number of difficulties with Moore's view. The most pressing of these concerns the claim that there is (and indeed must be) an
130 Kant 's Correspondence Theory of Truth identity relationship between a true belief and the fact to which it corresponds. How is it possible, in any real sense, for a non-physical entity, like a belief, to be identical in content with a fact that is a part of the physical world? How can the idea of snow have the same content as the object snow? Although Moore's correspondence as logical relation is problematic because it suggests an identity of content between things that can have no such identity, I will show that a version of this will be useful in explicating one aspect of the Kantian conception of truth. This single interpretation of correspondence will not however, be sufficient to explicate the entirety of Kant's conception. For this reason, other interpretations of the nature of correspondence must be considered. Correspondence as Congruence. Correspondence can also be interpreted as congruence. Correspondence as congruence theories maintain that there must be a structural isomorphism between judgments and facts/states of affairs. According to proponents of this type of theory, there is simply a natural relationship between our beliefs about the world and the world as it in fact is. (Schmitt 1995, 170-171) Relationships between our judgments and their corresponding states of affairs are not merely a matter of historical convention; it is a natural relationship that developed as it did necessarily. Just as two correctly matched puzzle pieces fit one another perfectly; so a true judgment perfectly matches the state of affairs to which it corresponds. Again, a state of affairs denotes anything that obtains in reality. One well-known theory that analyzes correspondence as congruence is Bertrand Russell's account of truth. Like most correspondence theorists, Russell claims that truth consists in a proper relationship between a belief and a fact. He further maintains that every belief is a four-way relationship involving a subject, two objects, and an object relation. The subject is the individual holding the belief. One of the object terms is something like the subject term in a sentence. It is the thing that the believer believes is somehow affecting something else. The something else that is being affected is the other object; this object is related to the object term in a sentence. The second object is that thing which the believer believes is being affected by the first object.
An Idealist Correspondence Theory
131
Finally, there is the object relation, which is very much like the verb in a sentence. It describes how the first object relates to the second. In the judgment "A change in humidity precedes rain" the object in the first sense is 'a change in humidity'; the object in the second sense is 'rain'; the object relation is 'preceding' and the subject is the believer, asserter, or thinker of the judgment. According to Russell, "everything that there is in the world I call a 'fact'" (Russell 1948, 143) Facts, for Russell, are also necessarily objective. "I mean by a 'fact' something which is there, whether anyone thinks so or not." (Russell 1948, 143) Facts and judgments are not identical. A fact is something that holds in reality, while a true judgment is an assertion of a fact. Essentially, what Russell calls a fact is what I have referred to as a state of affairs. For his definition of fact to work, however, Russell must include in his account of facts something that I have included in my account of states of affairs. He must include the possibility that reports about beliefs, feeling etc. can be true or false. Such factual relationships are certainly not present whether anyone thinks so or not. All empirical facts for Kant are the result of the conditioning of the undetermined manifold of sensibility by the finite rational consciousness. Nevertheless, there is an objective truth as to whether one's beliefs etc. have been accurately reported. Therefore, an adequate correspondence as congruence account of truth will include certain beliefs in the set of facts. For facts, there is a three-term relationship. There is an object, which affects another object according to an object relation. Essentially, the fact relationship is the belief relationship minus the subject. (Kirkham 1995, 120) A belief is true in this view just in case the relationship between objects in the belief matches the relationship between the corresponding objects in reality. This point is best explained by an example. David i Cosette 4-
=
Cosette 4-
breaking 4-
=
breaking 4-
vase 4-
=
vase 4-
Figure Four
132 Kant 's Correspondence Theory of Truth Suppose that David believes that Cosette broke the vase. In this case, David is the subject; Cosette is the first object, and the vase is the second object. The object relation is breaking. When presented with such an example, how is one to assess its truth? According to Russell, truth is analyzed by looking at the direction in the relationship within the belief and comparing it to the direction within the relationship in the fact. David's belief is true if and only if the relationship described in Figure four holds. David's belief that Cosette broke the vase is true just in case there is a fact that has a three term relationship including the objects Cosette and a vase, and an object relationship of breaking. Furthermore, the relationship must be in the same direction as David's belief. In other words, Cosette must be the object that is affecting the vase through a relationship of breaking. If no matching fact (three termed relationship) for David's belief exists, or if it exists in a different direction, then David's belief is false. This account of correspondence is correspondence as congruence, because it posits that facts must be structurally isomorphic, or congruent, with beliefs. No true belief could have a structure that fails to match up with any fact in the world. In addition, in asserting that facts are a threeterm relationship in the world, Russell provides an answer to Johnson's concerns about the odd dual nature of facts. For Russell, facts are just relationships in the world and statements people make are declarations of belief, not facts. Moreover, the correspondence that is necessary for truth just consists in relational isomorphism between our beliefs and the requisite facts. Thus, there is no problem of mental properties like beliefs having literally the same content as facts in the world. I will argue that part of the Kantian conception of truth is a correspondence as congruence account. However, correspondence as logical relation and correspondence as congruence are not sufficient to fully explicate all aspects of truth necessary for transcendental idealism. There is a third notion of correspondence, which must also be utilized. This notion of correspondence is correspondence as correlation. Correspondence as Correlation. Theories that define correspondence as correlation hold that each truth bearing entity (judgments in this case) is
An Idealist Correspondence Theory
133
directly correlated with a state of affairs. The state of affairs and the judgment need not be in any way structurally isomorphic, it merely needs to be the case that the judgment taken as a whole directly corresponds to the state of affairs taken as a whole. (Kirkham 1995, 119-120) Most philosophers who contend that correspondence should be understood as correlation also maintain that there is no necessary relationship whatsoever between our judgments (beliefs, judgments etc.) and facts about the world. Some, like J.L. Austin refer to the relationship as one of pure historical convention. J.L. Austin advocates a correspondence as correlation theory of truth. Although he agrees that truth must entail a correspondence to reality, he contends that truth-bearers and states of affairs need in no way be structurally isomorphic. "There is no need whatsoever for the words used in making a true statement to 'mirror' in any way, however indirect, any feature whatsoever of the situation or event." (Austin, 1970, 125) Austin maintains that the correspondence between truth-bearers and states of affairs, which is a necessary condition for the ascription of truth, is nothing more than convention. (Austin 1970, 121-122) Austin objects to the correspondence as congruence view because he contends that they cannot account for one-word utterances which none the less have truth-values. (Austin 1970, 129) To replace the two sets of relations, which Russell suggests are structurally isomorphic, Austin advocates a single four-term relationship. The first term is a statement, which according to Austin is the information conveyed by a declarative sentence. The sentence is the second term, and as explained, it is the medium through which a statement can be conveyed. The third term is states of affairs and the fourth is types of states of affairs. Statements may be conventionally correlated to types of states of affairs, or to a particular historical state of affairs. In either case, a statement is true just in case it accurately describes the state of affairs to which it is correlated by convention, and a statement is false if it does not. (Austin 1970, 129) The correlation between statements and states of affairs that determine truth serves to help us learn the language. The relationship is not one of necessity. A good way to wrap one's mind around the distinction between correspondence as congruence and correspondence as correlation is to con-
13 4
Kant 's Correspondence Theory of Truth
sider one of the dichotomies between the philosophies of Plato and Aristotle. One school of thought on Plato's Theory of Forms holds that for Plato, the world of the Forms must mirror the world of appearances. The form of cat must mirror, or resemble, a cat. The form of chair must resemble a chair. If we translate this concept concerning forms to ideas, beliefs, judgments, etc., we can get an idea of what correspondence as congruence proponents have in mind (no pun intended). It seems to be the case that if the correct conception of correspondence is as congruence, then our concepts, if not our language, could not have developed other than they did. This necessity arises from the fact that our concepts, and thereby our judgments, must be constructed so as to mirror the world to which they correspond. This being the case, and assuming that the world operates causally, since there is only one set of things which will obtain in the world, there is only one congruent structure which mirrors that set. Therefore, our concepts necessarily developed as they did. Opposed to this sort of congruence view of forms is Aristotle's account of forms. According to him, the form of each thing is within that thing. Our concepts and judgments are developed separately and independently of the objects themselves. There is no need, and indeed no possible way for our ideas of and judgments about objects (states of affairs etc.) to mirror, in any way, the objects themselves. This seems to be what advocates of correspondence as correlation are supporting. Our concepts and subsequently our languages develop independently of the structure of the world we experience. (By independently I mean not contingent upon, I do not mean isolated from.). We create concepts that represent states of affairs without resembling them structurally in any way. Kantian Correspondence Correspondence for Analytic Judgments In previous chapters, I have argued that any account of transcendental idealism that does not recognize the unique nature of Kant's analytic/synthetic relationship is likely to incorrectly characterize his conception of truth. I have said that for Kant, any judgment is analytic so long as by some manner of conceptual analysis, the predicate can be
An Idealist Correspondence Theory
135
found in the subject. There are purely logical analytic truths (and falsehoods), and material analytic truths (and falsehoods). However, before the truth conditions and truth-makers for these two kinds of analytic judgments can be specified, more must be said about the nature and limits of the Kantian analytic/synthetic distinction. First, Kant's analytic/synthetic distinction is not a distinction between logical and non-logical judgments. Whether a judgment is analytic or synthetic is a function of the relationship between the judging subject, what is judged, and the object of the judgment. In analytic judgments, the relationship to the object is mediate. In synthetic judgments, the relationship to the object is immediate. (A8) If to know the truth or falsity of her judgment an agent need only come to fully understand the concept involved, then the judgment is analytic. In analytic judgments, the relationship between the judgment and its object is mediate. One only needs to know the concept that represents the nature of the object to know if the judgment is true. That is to say, one need only know the nature of the object in general. There is no need to go beyond the concept of the object to a specific instantiation to determine the truth of the judgment. In Kant's terms, the predicate is thought through identity with the subject concept. (A7/B11) As I have said, there are material and purely logical analytic judgments. An example of the former is the judgment "All bachelors in Cleveland are unmarried." The truth of this judgment can be deter-mined by a mediate relation to the object bachelor. In knowing the nature of bachelors, as represented in the concept 'bachelor', one can determine that this judgment is true. In purely logical judgments, an agent only needs to know that the judgment is of such a form that the concept ofthat form entails the truth of any instantiations of it. "All bachelors are bachelors" is an example of a purely logical truth. To know that this judgment is true, it is not even necessary to know the subject concept as it represents the nature of the object of the judgment. It is only necessary to know that the judgment has the form of a tautology, namely 'all A's are A Y , and that to be a tautology is identical with always being true. In other words, one need not know even the nature of a concept to know that it is identical with itself.
13 6 Kant 's Correspondence Theory of Truth However, since Kant maintains that all judgments involve the subsuming of a predicate under a concept, he cannot account for all logically true (or logically false) judgments. Consider the judgment "Either there are bachelors in Cleveland or there are not." As this judgment does not involve the attribution of a predicate to a concept, Kant's view cannot account for it. Nevertheless, if his logic were modified to include all logical truth and falsities, Kant would claim that they were true according to the same truth conditions. Namely, that the concept of an object in general, not even a particular object, involves certain internal relationships. A modified version of Moore's correspondence as logical relation view works best as a conception of truth for Kant's analytic judgments. The correspondence, which determines the truth or falsity of analytic judgments, is not a logical relation in the broadest sense, but rather in the narrow sense of an identity relationship. Note that we will not speak of facts, but of the content of judgments and the content of concepts as they represent the nature of things. Thus, the judgment "All bachelors in Cleveland are unmarried" is true in virtue of the fact that the relationship between the subject 'bachelors' and the predicate 'unmarried' is identical to a relationship between the nature of what it is to be a bachelor and the property of being unmarried. Similarly, the judgment "All bachelors are bachelors" is true in virtue of the fact that the relationship between subject and predicate in the judgment is identical to a relationship that holds within the nature of a given thing. Unlike empirical judgment, analytic judgments do not describe states of affairs per se. Analytic judgments depict a relation of ideas. In addition, both the relationship described in the judgment, and the nature that is represented in the concept are non-physical and non-empirical. Hence, there is not the same difficulty with there being an identity relationship between an analytic judgment and its truth-maker as for empirical judgments and their objects. As I have argued, Kant thinks that in an analytically true judgment, the predicate is thought through identity with the subject. The essence of the correspondence as logical relation account is this identity relationship. Therefore, the correspondence theory which best applies to analytic judgments is a modified version of Moore's correspondence as a logical
An Idealist Correspondence Theory
13 7
relation account. Moreover, given the intent and standard for confirmation of analytic judgments, an identity relationship makes more sense than a congruence or correlation relationship. That this type of correspondence is not drastically different from certain coherence accounts, I do not deny or reject. Since Kant does not intend analytic judgments to be substantive in the sense of extending our knowledge beyond conceptual analysis, the coherence/correspondence dichotomy is not important here. Although I think a correspondence account better captures the relationship Kant has in mind between analytic judgments and natures, I do not think that the difference between this notion of correspondence and coherence is significant. Correspondence for Empirical Judgments The touchstone whereby we decide whether our holding a thing to be true is conviction or mere persuasion is therefore external, namely, the possibility of communicating it and of finding it to be valid for all human reason. For there is then at least a presumption that the ground of the agreement of all judgments with each other, notwithstanding the differing characters of individuals, rests upon the common ground, namely, upon the object, and it is for this reason that they are all in agreement with the object—the truth of the judgment thereby being proved. (A 820-821/B848-849)
As I have argued that a substantive coherence theory will not work as a conception of truth for transcendental idealism, I will now focus on the nature of the substantive correspondence theory that I think is required by the text and the essential doctrines of a coherent interpretation of transcendental idealism. Truth for empirical judgments will be necessarily quite different from truth for analytic judgments. Kant asserts that an analytic relationship must be purely formal, i.e. it must extract all that is particular to a given concept. Our empirical judgments are never purely formal, they are always substantive. The only judgments with a purely formal criterion for truth are analytic a priori judgments in which the predicate is contained within the subject. (A 6-7/ B 10) Empirical judgments are neither a priori nor substantive. Thus, Kant must reject Moore's account of correspondence. He makes this point in the Logic section of the Critique of Pure Reason;
13 8
Kant 's Correspondence Theory of Truth If truth consists in the agreement of knowledge with its object, that object must thereby be distinguished from other objects; for knowledge is false, if it does not agree with the object to which it is related, even although it contains something which may be valid of other objects. (A58 B82-83)
Likewise, Kant would have to reject a correspondence as correlation account for empirical judgments. Empirical judgments entail a far stronger claim than correspondence as correlation implies. Kant intends that for each true empirical judgment, the structure of the relationship contained within it is structurally isomorphic with the corresponding state of affairs. We can infer from the direct one to one objective validity relationship suggested by Kant that he has this sort of isomorphism in mind for empirical judgments. The error, which quite obviously is the cause of this mistaken venture, and which indeed excuses though it does not justify it, lies in employing the understanding, contrary to its vocation, and in making object, that is, possible intuitions, conform to concepts, not concepts to possible intuitions, on which alone their objective validity rests. (A289/B345)
According to Kant, a particular judgment has validity only insofar as it is related to its object as a possible intuition. Possible intuitions are particulars, and thus, it is a particular set of relations (in the intuition) that determines the truth or falsity of the requisite judgment. For a judgment to be true, the relation between the subject and predicate terms in the judgment, and the relationship between the object of the judgment and its properties must be congruent. Since Kant's correspondence conception of truth for empirical judgments cannot be correspondence as logical relation or correspondence as correlation, it must be correspondence as congruence. An empirical judgment is true just in case the relationship it depicts is mirrored by the relationship between the subject and object terms as they obtain in the world of appearances. This applies to two kinds of empirical objects, objects of inner and outer sense, and objects of only inner sense. Both kinds of objects have sufficient objective reality (which they gain from determination of the undetermined manifold of empirical sensibility by transcendental judgments) to serve as truth-makers for empirical judgments.
An Idealist Correspondence Theory
139
The empirical truth of appearances in space and time is, however, sufficiently secured; it is adequately distinguished from dreams, if both dreams and genuine appearances cohere truly and completely in one experience, in accordance with empirical laws. (A492/B521)
That Kant only sees coherence as a negative, although important, condition of truth is demonstrated by the fact that he thinks the truth of empirical judgments can be secured even if both dreams and genuine appearances cohered with empirical laws. This is because Kant subscribes to the four-term relationship between subject and object depicted in Figure Three. Our judgments are only true if they correspond to appearances, but those appearances only have knowable properties in virtue of their being endowed with those properties by the transcendental employment of the understanding in the original unity of apperception. This holds for appearances of both inner and outer sense. For Kant, external empirical objects are constructed the finite rational mind, nevertheless, they are also outside of him. We know this because we perceive the world spatially, and according to Kant, space is our outer sense. (A 26/B42) Thus, you are justified in believing snow is white just in case the empirical object snow has the property of being white. If the judgment relationship and the phenomenal relationship are congruent, the judgment is true. Correspondence to a spatial object (an object that is part of the outer sense of the finite subject) is not the only way in which experience can confirm the truth of our empirical judgments. There are also objects of experience that are only temporal, that is, they are only determined by our inner sense. Take the empirical judgment "I feel happier now than I did yesterday" as an example of an empirical judgment that does not purport to correspond to a spatial object. It is a comparative report describing a change in my emotional state. Like all truth bearing empirical judgments, this judgment is true just in case it corresponds to what is real. If it is the case that my emotional state today is happier than my emotional state yesterday, then my judgment is true. To summarize, truth for empirical judgments is purely a matter of correspondence as congruence between those judgments and the appearances that they purport to describe. That the knowable qualities of these
140 Kant's Correspondence Theory of Truth appearances are dependent on the finite subject is not an issue in regard to empirical truth. This dependence is only revealed upon transcendental reflection, and as such, is properly an issue for the analysis of transcendental truth. Correspondence for Synthetic A Priori Judgments Establishing a theory of truth for synthetic a priori judgments is far more complicated than establishing one for empirical or analytic ones. Kant is quite clear about the criteria for the latter two, he is, however, less obvious in his treatment of the truth conditions for the former. Nevertheless, he does provide us with information from which to infer what the truth conditions must be. One of these bits of information concerns the role of the principle of contradiction. Now in our critical inquiry it is only with the synthetic portion of our knowledge that we are concerned; and in regard to the truth of this kind of knowledge, we can never look to the above principle [the principle of contradiction} for any positive information, though, of course, since it is inviolable, we must always be careful to conform to it. (A152/B191)
The principle of contradiction is a necessary, but not sufficient condition for all synthetic judgments. The next question to answer is, "What are the elements which, in conjunction with the principle of contradiction are all necessary and jointly sufficient for determining the truth of synthetic a priori judgments?" Constitutive and Regulative Synthetic A Priori Judgments. Before giving an account of the truth conditions and truth-makers for synthetic a priori judgments, it is important to distinguish between constitutive and regulative judgments. The former are truth-bearers and the latter are not. Therefore, the account of truth I will provide will only be intended to apply to constitutive transcendental judgments. Constitutive synthetic a priori judgments have truth-values. A constitutive judgment is one that employs a constitutive principle of the understanding. There is a further subdivision of constitutive synthetic a priori judgments into mathematical and dynamic. For the moment, I will focus on dynamic constitutive judgments.
An Idealist Correspondence Theory
141
The dynamic constitutive principles of the understanding, listed in the "System of all Principles of Understanding," are those that employ the categories. They are constitutive in that they are necessary conditions for the possibility of experience. It is only by having their form determined according to these principles that appearances may become possible objects for us. As conditions for the possibility of experience, these transcendental judgments are the source of the transcendental truth that makes empirical truth possible. (A146/B185) Moreover, dynamic constitutive judgments make knowledge of mathematics and physics possible. Mathematics (especially geometry) and physics are predicated on relations in space and time. The principles determine space and time and thereby what appear in these, and therefore are conditions for the possibility of any kind of knowledge predicated on space and time. I will specify the truth conditions and truth-maker for such judgments shortly. Regulative synthetic a priori judgments are non-cognitive, they are merely rules. They serve only to guide us in our thought. A regulative judgment employs a principle of the greatest possible continuation and extension of experience, allowing no empirical limit to hold as absolute. Thus it is a principle of reason that serves as a rule, postulating what we ought to do in the regress, but not anticipating what is present in the object as it is in itself, prior to all regress. (A509/B537)
Regulative judgments, then, are mere rules and as such are not truthbearers. Some examples of regulative judgments are "Man must be free," God must exist," and "The soul must be immortal." These judgments regulate the way we think. In these cases, they explain our moral responsibility and certain implications that must follow from it. (A337/B395) The fact that regulative judgments are non-cognitive does not imply that all non-cognitive judgments are regulative. Kant maintains that we can think whatever we please, so long as we do not contradict ourselves. (Bxxvii/) Only knowledge requires truth. It is certainly not the case that whatever does not provide us with knowledge regulates our behavior as do moral judgments. I can think, for instance, that there may be an inverted earth where everything is the opposite of what obtains in our
142 Kant 's Correspondence Theory of Truth world of experience. I am not contradicting myself, so this may be an object of thought for me. It is also a non-cognitive judgment because it does not have real possibility (real possibility being the possibility of obtaining in our world of experience). It is not, however, a regulative judgment. It neither guides nor purports to guide behavior or belief. Therefore, only a special sub-set of non-cognitive judgments are regulative. There are scholars who think that both constitutive and regulative synthetic a priori judgments are just rules, and as such, cannot be truthbearers. Jonathan Bennett subscribes to a version of this view. He considers ascending functions of the understanding and reason. An ascending function is one in which the activity is an activity of uniting and interlinking rather than inferring from pairs of premises. They are inferences which proceed upward from experience to its conditions. In other words, they are inferences from the laws that operate empirically to the laws that must operate for experience to be possible. Ascending functions establish rules of the basis of observation. The Ideal Gas Laws are examples of rules that are established by an ascending function of reason. While a truth claim can be made concerning whether such rules apply, the rules themselves are non-cognitive. According to Bennett, those judgments that Kant distinguishes as regulative and synthetic are both ascending in this sense. As ascending functions of reason, all synthetic a priori judgments are rules. (Bennett 1974, 261-262) What Kant calls constitutive judgments are actually just rules for constructing appearances. Thus, they cannot be truth-bearers. However, Bennett's suggestion can be rejected because the transcendental truth that is founded in transcendental judgments is the foundation of all truth for Kant. He explicitly claims that transcendental judgments are cognitive. Due to their role in constructing objects of possible experience, constitutive transcendental judgments have objective validity. Having objective validity is, for Kant, the necessary condition a judgment must fulfill to be a cognitive. All our knowledge falls within the bounds of possible experience, and just in this universal relation to possible experience consists that transcendental truth which precedes all empirical truth and makes it possible. (A146/B185)
An Idealist Correspondence Theory
143
Not only is the truth of transcendental judgments possible according to Kant, it is necessary. Truth for Transcendental Judgments. I think that the nature of transcendental truth for Kant is best understood by means of an analogy. Recall the preferred metaphysical reading of the phenomena/noumena distinction. According to that interpretation, we can say that things in themselves exist, but only analogously, because, strictly speaking, existence as actuality is a category and as such applies only to appearances. Thus, we can say that things in themselves exist in a way similar to the way appearances exist. However, this can only be a problematic judgment for us because we cannot have substantive knowledge of things as they are in themselves and hence cannot specify the similarity. An analogy can also be helpful in understanding the nature of transcendental truth. Let us begin with what the regulative principles of the understanding require us to think about the relationship between God and things as they are in themselves. God is an intuitus originarius, that is to say, his intellect creates as it represents. When he represents an object 'x' to Himself, that object comes into being as a thing in itself. (A584/B612) Although God need not, and does not, judge (i.e. subsume particulars under concepts), his intuitions are 'true' (meaning they specify what is actually the case) simply in virtue of the fact that the objects of His intuitions come into being because He intuits them. In other words, the form and content of the objects of His intuitions correspond to His intuition because His intuition creates their form and content. Finite subjects perform an analogous act of creation in the process of transcendental judgment. They do not create things in themselves by making judgments about them, but they do create the form of appearances by making transcendental judgments. This creation is only analogous to divine creation because finite subjects must start with an object and a given. The given is the thing in itself as transcendent object and the object is the transcendental object as the concept of an object in general. With these raw materials, the finite subject creates the form of appearances and synthesizes them into a unified world of experience. It is in this act of creation that transcendental truth is established. For each transcendental judgment that determines an appearance, that judg-
144 Kant's Correspondence Theory of Truth ment is true. It is true because it corresponds to the object the form of which it has created, and it is necessarily true because it creates its objects such that correspondence will hold. This correspondence relationship is a correspondence as correlation relationship because the set of transcendental judgments corresponds to the created synthetic unity of appearances. Transcendental judgments determine the synthetic unity of all appearance in the original unity of apperception. Moreover, all individual appearances are determined by each of the pure principles of the understanding. Thus, transcendental judgments not only determine the form of appearance in general, they determine the form of all individual appearances. Similar to the way in which God's intuitions are true of what he intuits because he creates as he intuits, our transcendental judgments are true of appearances in virtue of the fact that we create the form of appearances with these judgments. The original unity of apperception, which is the determination of the synthetic unity of appearances, takes the matter of appearances and determines them in such a way as to make them accessible to individuals with understandings like ours. The correspondence between transcendental judgments and experience is the essence of transcendental truth, and it is a necessary condition for all empirical truth. It is what gives empirical judgments objective reality. In addition, the correlation between our transcendental judgments and their individual objects is necessary, not as a matter of any convention, but as the correlation which makes experience possible. Therefore, transcendental judgments are true in virtue of their correspondence as correlation to the objects they determine as appearances. One might object that since I am arguing that the form of appearances must match the form of our transcendental judgments, that I am really arguing for coherence and not correspondence. This is not the case, however, because of the nature of a Kantian empirical object. Coherence must be coherence between beliefs, propositions, principles etc. It makes no sense to speak of a physical object or event cohering with a principle. Although the pure principles of the understanding impose a form on our appearances, the form of our appearances is not itself a principle, it is the application of a principle. Moreover, it is the application of a principle to something material. Since appearances are not principles in
An Idealist Correspondence Theory
145
the world, the truth of judgments about them cannot be truth in virtue of mere coherence. Coherence is an important, although secondary criterion of truth for Kant. It is true that the form of our judgments, considered in abstraction as the principles that are applied to them must cohere with the transcendental conditions for the possibility of experience. However, what makes a particular judgment true, be it transcendental or empirical, is its correspondence to a particular that is given in sensibility. As I have argued, it is the material component of sensibility that differentiates between objects. Thus, it is the material component that determines if a particular judgment is true or false of its object. Therefore, correspondence to formally determined givens is the essence of truth for Kant and coherence is important, but secondary. Note that this account is different from Putnam's constructivist account because it does not imply that we create all aspects of appearances including their sensual aspects. This account also does not imply that we are directly intuiting things as they are in themselves. Out intuition of the experience is made possible by transcendental judgment are the sensual properties of the transcendent object as they have been determined by the original unity of apperception. We intuit them as appearances, not as thing in themselves. Truth for Mathematical Judgments. The conception of truth I have attributed to transcendental judgments must be altered to hold for another type of synthetic a priori judgment, mathematical judgment. The difference is that for a priori mathematical judgments, there is no sensual given beyond the pure intuitions of space and time. Mathematical objects are constructed according to the formal conditions for the possibility of experience. Mathematical judgments are concerned with pure and empirical intuitions, whereas transcendental judgments are concerned with the existence of objects and the relationships between objects and the finite understanding. (Bl 10) Mathematical objects are constructed in the way that Putnam suggests appearances are constructed. Mathematical objects are constructions of magnitude and figure, which presuppose the pure intuition of space. (A717/B745) That purely mathematical objects are constructed is
146 Kant 's Correspondence Theory of Truth evidenced by the fact that it is possible to engage in pure mathematics. Thus, mathematical judgments are true in virtue of the fact that they correspond to the objects they have created. This correspondence, however, is closer to correspondence as logical relation because they need rely on nothing sensual that is given in the transcendent object, just the form of experience in general. As such, the correspondence between mathematical judgments and their objects, unlike the correspondence between transcendental judgments and their objects, is not substantively different from coherence. Summary I have argued that transcendental idealism implies a conception of truth for each of the three kinds of truth bearing judgments. These are analytic judgments, empirical judgments, and constitutive synthetic a priori judgments. The truth conditions for all of these kinds of judgments are that they, in some way, correspond to their objects. The correspondence between analytic judgments and their objects is best construed as correspondence as logical relation, which I have described as an account that is not drastically different from a coherence conception. Correspondence for empirical judgments is a matter of correspondence as congruence. Correspondence for transcendental judgments amounts to a special variety of correspondence as correlation. Finally, correspondence for a priori mathematical judgments is correspondence as logical relation, which again is not drastically different from coherence. Thus, with some modifications, Kant correctly ascribes a substantive correspondence conception of truth to transcendental idealism. What is not specified is that it must be a tiered conception because both the nature of the object and the nature of the correspondence relationship change from judgment type to judgment type. That this must be the case is dictated by the need for consistency with the essential tenets of an accurate, positive reading of transcendental idealism. My view is an improvement over those considered in previous chapters for a number of reasons. First, it assumes the most reasonable interpretation of the phenomena/noumena distinction. Specifically, it holds that the phenomena/noumena distinction is a distinction between the set of things as its member or members are in the absence of our de-
An Idealist Correspondence Theory
147
termination of them (things in themselves), and that same set of things as its members are according to the formal and material conditions of experience (appearances). This interpretation is preferable because it neither undermines the importance of transcendental judgment nor commits Kant to the existence of two ontologically separate, non-interacting worlds. My view is also preferable to those previously considered because it compromises neither Kant's transcendental idealism, nor his empirical realism. Neither phenomenalism nor transcendental realism is an implication of my interpretation. In addition, my interpretation does not imply that transcendental truth is merely a negative condition of empirical truth. Transcendental truth is substantive in its own right and indeed provides the synthetic unity of appearances, which will become both the matter and form of all true empirical judgments. For this reason, my view better fits with a reasonable reading of transcendental idealism than those previously considered. Empirical realism is preserved in my interpretation because the reality of empirical objects is not compromised. Their objectivity is guaranteed because the transcendent object is given to us as the material condition for all experience, and the formal conditions of the understanding take what is given to the sensibility and determine it in a way that it is knowable for us. The necessity of the matter of appearance for explaining the public objectivity of the world guarantees that there is something to the objectivity of our appearances beyond the coherence of our judgments. Moreover, since it is the matter of our judgments that differentiates them (in addition to their mere temporal and spatial designations); the correspondence aspect of truth is the essence of truth. Correspondence, not coherence, is the primary nature of truth for Kant. Transcendental idealism is preserved because, according to my view, all of the non-sensual knowable properties of empirical objects are dependent on finite subjects. There would be no knowable appearances if there were no finite subjects, thus, the empirical world, as we can know it, is transcendentally ideal. Finally, my interpretation is consistent with Kant's remarks in the Critique. I affirm his description of his view as a correspondence view
148 Kant's Correspondence Theory of Truth without undermining his rejection of empirical idealism or transcendental realism. Because my view is faithful to the requirements of transcendental idealism that are set out in the text and because it best supports an accurate and positive reading of transcendental idealism, it is the preferable interpretation of what Kant means by 'truth'.
SC BIBLIOGRAPHY Allison, Henry E., Kant's Transcendental Idealism: An Interpretation and Defense. Yale University Press, New Haven and London: 1983. , Idealism and Freedom: Essays on Kant's Theoretical and Practical Philosophy. Cambridge University Press, New York: 1996 , The Kant-Eberhard Controversy. Johns Hopkins University Press, Baltimore: 1973. Austin, J.L., "Truth" in Proceedings of the Aristotelian Society. Supplementary Issue, 1950. Beck, Lewis White, Kant Studies Today. Open Court Inc., LaSalle: 1969. , (editor) Proceedings of the Third International Kant Congress. Kluwer Academic Publishers, Dordrecht Holland: 1972. Bennett, Jonathan, Kant's Dialectic. Cambridge University Press, London: 1974. den Ouden, Bernard (editor), New Essays on Kant. Peter Lang, New York: 1987. Devitt, Michael, Realism and Truth. Basil Blackwell Inc., Cambridge and Oxford: 1984. Dummett, Michael, Truth and Other Enigmas. Harvard University Press, Cambridge: 1978. Ewing, A.C., A Short Commentary on Kant's Critique of Pure Reason. Guyer, Paul, Kant and the Claims of Knowledge. Cambridge University Press, Cambridge and New York: 1987. , (editor) The Cambridge Companion to Kant. Cambridge University Press, Cambridge and New York: 1992. Hahn, Lewis Edwin (editor) The Philosophy ofP.F. Strawson. Open Court Publishing, Chicago and LaSalle, 111.: 1998. Heyting, Arend, Intuitionism: An Introduction. Amsterdam-North Holland Publishing Company, Amsterdam: 1971.
150 Kant 's Correspondence Theory of Truth Hintikka, Jaakko, Knowledge and the Known: Historical Perspectives in Epistemology. Kluwer Academic Publishers, Dordrecht: 1974. Hofmeister, Heimo, "The Problem of Truth in the Critique of Pure Reason", Proceedings of the Third International Kant Congress, Kluwer Academic Publishers, Dordrecht Holland: 1972. , with Merrill Hintikka The Logic of Epistemology and the Epistemology of Logic: Selected Essays. Kluwer Academic Publishers, The Netherlands: 1989. , Logic, Language Games and Information: Kantian Themes in the Philosophy of Logic. Clarendon Press, Oxford: 1973. Kant, Immanuel, The Critique of Pure Reason. Translated by Norman Kemp Smith. Bedford/St. Martin's Press, New York: 1929. , Logic. Translated by Robert S. Hartman and Wolfgang Schwarz. Dover Publications Inc., London: 1988. , Prolegomena to Any Future Metaphysics. Translated by Paul Cams. Hackett Publishing Company, New York: 1977. More, G.E., Some Main Problems in Philosophy. Brill Academic Publishers, Boston: 1953. Parrini, Paolo (editor), Kant and Contemporary Epistemology. Kluwer Academic Publishers, Dordrecht, Boston, London: 1994. Paton, HJ., Kant's Metaphysics of Experience; volumes I and II. The Macmillan Company, New York: 1936. Posy, Carl J. (editor) Kant's Philosophy of Mathematics: Modern Essays. Kluwer Academic Publishers, Dordrecht, Boston, London: 1992. , Kant on Causality, Freedom, and Objectivity. Edited by William A. Harper and Ralf Meerbote. University of Minnesota Press, Minneapolis: 1984. Putnam, Hilary, The Many Faces of Realism. Open Court Publishing, LaSalle,IL: 1987. , Philosophical Papers, Volume 1: Mathematics, Matter and Method. Cambridge University Press. Cambridge: 1975. , Philosophical Papers, Volume 3: Realism and Reason. Cambridge University Press. Cambridge: 1983. , Reason, Truth, and History. Cambridge University Press. Cambridge: 1981.
Bibliography
151
Rescher, Nicholas, Kant's Theory of Knowledge and Reality: A Group of Essays. University Press of America Inc., Washington, D.C: 1983. Robinson, Hoke, "Two Perspectives on Kant's Appearances and Things in Themselves," in Journal of the History of Philosophy volume 32:3 John Hopkins University Press, Baltimore: July 1994. Russell, Bertrand, Human Knowledge: Its Scope and Limits. Routledge, New York: 1948. Schmitt, Frederick, Truth: A Primer. Perseus Books/Westview Press, Boulder, San Francisco, Oxford: 1995. Smith, Norman Kemp, A Commentary to Kant's Critique of Pure Reason. Brill Academic Publishers, Boston/Palgrave Macmillan New York: 1923. Soames, Scott, Understanding Truth. Oxford University Press, New York, Oxford: 1999. Strawson, P.F., The Bounds of Sense: An Essay on Kant's Critique of Pure Reason. Routledge, London and New York: 1966. , Entity and Identity and Other Essays. Clarenden Press, Oxford: 1997. Swing, Thomas K., Kant's Transcendental Logic. Yale University Press, New Haven and London: 1969. Walker, Ralph, The Coherence Theory of Truth: Realism, AntiRealism, Idealism. Routledge, London and New York: 1989. , Kant: The Arguments of the Philosophers. Routledge & Kegan Paul, London, Henley, and Boston: 1978. , "Synthesis and Transcendental Idealism," Kantstudien. 76 (1); 14-27. 1985.