Kant's Inferentialism: The Case Against Hume (Routledge Studies in Eighteenth-Century Philosophy) 9781138913080, 9781315691602, 1138913081

Citation preview

Kant’s Inferentialism

Kant’s Inferentialism draws on a wide range of sources to present a reading of Kant’s theory of mental representation as a direct response to the challenges issued by Hume in A Treatise of Human Nature. Kant rejects the conclusions that Hume draws on the grounds that these are predicated on Hume’s theory of mental representation, which Kant refutes by presenting objections to Hume’s treatment of representations of complex states of affairs and the nature of judgment. In its place, Kant combines an account of concepts as rules of inference with a detailed account of perception and of the self as the locus of conceptual norms to form a complete theory of human experience as an essentially rule-governed enterprise aimed at producing a representation of the world as a system of objects necessarily connected to one another via causal laws. This interpretation of the historical dialectic enriches our understanding of both Hume and Kant and brings to bear Kant’s insights into mental representation on contemporary debates in philosophy of mind. Kant’s version of inferentialism is both resistant to objections to contemporary accounts that cast these as forms of linguistic idealism and serves as a remedy to misplaced Humean scientism about representation. David Landy is Associate Professor of Philosophy at San Francisco State University, United States.

Routledge Studies in Eighteenth-Century Philosophy

1 Naturalization of the Soul Self and Personal Identity in the Eighteenth Century Raymond Martin and John Barresi

7 Aesthetics and Morals in the Philosophy of David Hume Timothy M. Costelloe

2 Hume’s Aesthetic Theory Taste and Sentiment Dabney Townsend

8 Hume’s Difficulty Time and Identity in the Treatise Donald L.M. Baxter

3 Thomas Reid and Scepticism His Reliabilist Response Philip de Bary

9 Kant and the Cultivation of Virtue Chris W. Surprenant

4 Hume’s Philosophy of the Self A E Pitson

10 The Post-Critical Kant Understanding the Critical Philosophy through the Opus postumum Bryan Wesley Hall

5 Hume, Reason and Morality A Legacy of Contradiction Sophie Botros 6 Kant’s Theory of the Self Arthur Melnick

11 Kant’s Inferentialism The Case Against Hume David Landy

Kant’s Inferentialism The Case Against Hume David Landy

First published 2015 by Routledge 711 Third Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2015 Taylor & Francis The right of David Landy to be identified as author of this work has been asserted by him in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Landy, David. Kant’s inferentialism : the case against Hume / David Landy. — 1 [edition]. pages cm. — (Routledge studies in eighteenth-century philosophy ; 11) Includes bibliographical references and index. 1. Kant, Immanuel, 1724–1804. 2. Hume, David, 1711–1776. I. Title. B2798.L293 2015 121'.3—dc23 2014048745 ISBN: 978-1-138-91308-0 (hbk) ISBN: 978-1-315-69160-2 (ebk) Typeset in Sabon by Apex CoVantage, LLC

This book is dedicated to those without whom it would not be possible, much less actual: Rick, Judy, Madeline, Violet, and, more than anyone else, Margo.

This page intentionally left blank

Contents

Acknowledgments Notes on the Texts

ix xi

Introduction

1

1

Hume’s Theory of Mental Representation

19

2

Two Objections to Hume’s Theory of Mental Representation

52

3

The A-Deduction and the Nature of Intuitions

107

4

The Object of Representation

173

5

Self and World in the Analogies of Experience

198

6

The Inferential Self

234

Postscript on Transcendental Idealism

276

Index

305

This page intentionally left blank

Acknowledgments

Portions of the following chapters have been previously printed as follows. My thanks to these publishers for allowing me to reprint this material here. PREFACE “The Premise That Even Hume Must Accept,” Self, Language, and World: Problems from Kant, Sellars, and Rosenberg. Eds. Jim O’Shea and Eric Rubenstein. Atascadero, CA: Ridgeview Publishing Co., 2010: 28–46. CHAPTER 1 “Hume’s Theory of Mental Representation,” Hume Studies, 38, 1 (April 2012): 23–54. CHAPTER 2 “Sellars on Hume and Kant on Representing Complexes,” European Journal of Philosophy, 17, 2 (August 2009): 224–246. “A Sellarsian Kantian Critique of Hume’s Theory of Concepts,” Pacific Philosophical Quarterly, 88, 4 (December 2007): 445–457. CHAPTER 3 “Inferentialism and the Transcendental Deduction,” Kantian Review, 14, 1 (March 2009) 1–30. CHAPTER 6 “A Rebuttal to a Classic Objection to Kant’s First Analogy,” History of Philosophy Quarterly, 31, 4 (October 2014): 331–345.

x

Acknowledgments

POSTSCRIPT ON TRANSCENDENTAL IDEALISM “What Incongruent Counterparts Show,” European Journal of Philosophy, 21, 4 (December 2013): 507–524. Portions of this manuscript were written with the generous support of the San Francisco State University Presidential Award and the National Endowment for the Humanities Summer Stipend. The book also benefitted greatly from the feedback I received on it from the students in my Fall 2013 graduate seminar on the first Critique, especially Tyler Olsson and Aaron Franklin. As always, Margo Landy provided an enormous amount and variety of support. Any faults that remain after the contributions of so many helpful readers are of course all my own.

Notes on the Texts

For citations from Kant’s works other than the Critique of Pure Reason I have employed the convention of listing the volume and page number of the Akademie edition followed by the title and page number of the translation used. For the first Critique, I follow the common convention of citing only the page numbers from the A and B German editions. For consistency’s sake, I have made use of the Cambridge Edition of the Works of Immanuel Kant for all of the English translations included. My thanks to Cambridge for giving me permission to use this material. I employ the following titular abbreviations. Correspondence. Trans. and Ed. Arnulf Zweig. Cambridge: Cambridge University Press, 1999. Theoretical Philosophy Theoretical Philosophy After 1781, Ed. Henry Allison and Peter Heath. Trans. Gary Hatfield, Michael Friedman, Henry Allison, and Peter Heath. Cambridge: Cambridge University Press, 2002. Notes and Fragments Notes and Fragments, Ed. Paul Guyer. Trans. Curtis Bowman Paul Guyer and Frederick Rauscher. Cambridge: Cambridge University Press, 2005. Logic Lectures on Logic, Trans. and Ed. J. Michael Young, Cambridge: Cambridge University Press, 1992. Metaphysics Lectures on Metaphysics, Trans. and Ed. Karl Ameriks and Steve Naragon. Cambridge: Cambridge University Press, 1997.

Correspondence

For citations from Hume’s A Treatise of Human Nature and Enquiry Concerning Human Understanding I employ the standard convention of citing the book, chapter, section, and paragraph number from the Clarendon edition, followed by the page number from the Selby-Bigge/Nidditch edition.

This page intentionally left blank

Introduction

THE PREMISE THAT CANNOT BE DENIED Legend has it that the way that Kant’s Transcendental Deduction operates is by beginning with a premise that even the most ardent skeptic—Hume— would accept, and then showing that accepting this premise already commits one to a whole host of other robust philosophical theses: the validity of the Categories, immediately, and that which comprises the rest of the first Critique, mediately. Just what this premise is, whether it must be accepted, and how it is that by accepting it one becomes committed to all of these other theses has long been the subject of much scholarly and philosophical debate. One aspect of this debate has been a concern over exactly how much is to be included in such a premise. This is because, prima facie, Kant seems to run into the following obvious structural difficulty. The more robust this premise is, the more plausible it is that much follows from it, but also the less likely it is that the skeptic will be forced to accept it. Conversely, the less robust the premise is, the more likely that the skeptic will agree to it, but the less plausible it is that anything very interesting follows from it. So, various interpreters over the years have tried to strike a balance between these two competing approaches to the Deduction: some making the Premise-ThatCannot-Be-Denied more robust and spending their time showing why skeptics are nonetheless committed to it, and others making it less robust and spending their time showing how accepting it still commits one to various of Kant’s other philosophical theses.1 One central thesis of this book is that all of this hand-wringing has been for naught. The legend surrounding the Transcendental Deduction is nothing but a myth, although like many such stories, it does have as its origin fact. While the argument of the Transcendental Deduction does begin with a premise that Kant holds that nobody can deny, and Kant also holds that the validity of the Categories and that which comprises the rest of the Critique do follow from this premise (with proper supplementation), the interpretive thesis that it is with the Premise-That-Cannot-Be-Denied that his argument

2

Introduction

begins is simply false. Kant is not entitled to, nor does he take himself to be entitled to, any such premise. The-Premise-That-Cannot-Be-Denied must be earned, and it is in the sections of the Critique that precede the Transcendental Deduction that Kant does this work. In particular, it is in the Transcendental Aesthetic and Metaphysical Deduction where Kant argues that the theory of mental representation that serves as Hume’s grounds for rejecting the Premise-That-Cannot-Be-Denied is inadequate to its task, and that it must be replaced by Kant’s own particular brand of what we would now call inferentialism: in Kant this takes the form of the thesis that concepts are rules for relating intuitions to one another by subsuming them under conditions that prepare them to be used in various syllogisms. This point brings out another central thesis of this book: that both Hume and Kant are first and foremost engaged in projects aimed at giving a philosophical account of the nature of mental representation. Reading Hume as engaging in a project of this sort fell out of favor in the second half of the twentieth century, although recently scholars have recognized the importance that a theory of mental representation must play for Hume and have begun investigating this aspect of Hume’s philosophy anew.2 The reason for the de-emphasis of this aspect of Hume’s philosophical system is likely a combination of a number of factors including but not limited to a reaction against the association between Hume and the Logical Positivists, a general resurgence in analytic metaphysics following the innovative work of Kripke and Lewis, and the work of interpreters such as Fogelin and Stroud on Hume’s skepticism. The twin reactions against this final sort of reading of Hume are also likely causes: one of these casts Hume as a naturalistic “scientist of man” and the other, the so-called New Hume approach, casts him as a robust metaphysical realist.3 Kant’s Critique likewise enjoys a long and rich history of being understood in a variety of ways (perhaps as many ways as there are trends in contemporary philosophical thinking). Since just the middle of the twentieth century it has prominently been read as a treatise in epistemology, metaphysics, cognitive science, and philosophy of science.4 Unlike Hume scholarship, though, the tradition of reading the Critique as including as an essential part a philosophical account of mental representation enjoys a steady and robust history, especially recently.5 The current study occupies a place in that tradition the explication of which must begin with the great twentieth-century philosopher Wilfrid Sellars. Sellars is, perhaps, more highly regarded for his contributions to contemporary fields such as the philosophy of mind and philosophy of language, but his influence on historical scholarship on Kant is no less important. Most prominently, Sellars instigates a debate about Kant that is very much at the forefront of Kant scholarship today: that over the nature and reach of conceptual representation in Kant’s theory of mental representation. Sellars himself takes Kant to be confused on the issue, although he presents what he takes to be the proper extrapolation of Kant’s most central commitments on the issue.6

Introduction

3

McDowell, engaging with Sellars, argues for a thoroughly conceptualist reading of Kant. He holds that for Kant all representation is conceptually structured. McDowell’s reading of Kant has become the target of a great deal of criticism from Kant scholars such as Hanna, Ginsborg, and Allais, all of whom argue in their own ways that we must find a place in Kant interpretation for non-conceptual representations as well.7 Meanwhile, there has all along been a quiet movement of scholars back toward understanding and defending some version of Sellars’s Kant.8 As I understand that interpretation, its most important thesis is that for Kant all representations of a complex state of affairs as complex must be conceptually structured. This, however, does not preclude the possibility of non-conceptual representation because some representations will not be called upon to represent anything as complex. I will argue that distinguishing between these two kinds of representations—cognitions on the one hand, and sensations on the other—one can chart a path that will satisfy the demands of both the conceptualist and non-conceptualist alike. So, it is a version of this return-to-Sellars line of interpretation that I will defend here, although I will eschew Sellars’s claim that Kant was confused about these matters. I hold instead that some of the consequences of these theses that Sellars thought Kant missed he did not, and that some of the places where Sellars parts from Kant are so much the worse for Sellars. Longuenesse’s Kant and the Capacity to Judge will also figure prominently in what is to follow. That is because Longuenesse’s focus there is squarely on a topic that is of primary importance here: the nature of concepts, intuitions, and judgments. While it will become clear that I agree with much of what Longuenesse says—most prominently that Kant holds that concepts are something very much like inferential rules—there are also several important points of disagreement between us. For example, as I will discuss in Chapter 2, Longuenesse holds that such rules represent immanent universals, but I argue that this cannot be how Kant understands them because it leaves him without an answer to the problem of the unity of the proposition, which serves as one of his most important arguments against Hume. Longuenesse and I also agree that intuitions must have a conceptual structure (contra the non-conceptualists above), but we disagree about how this structure is “imported” into intuitions, when its proper first application is in judgments (Chapter 3). Finally, it is my intention to give an account of the sorts of rules that concepts are—rules of inference that are valid but not in virtue of their logical form—that I believe fills an important lacuna in the literature on Kant, including Longuenesse’s own work. All of this, I argue, creates a picture of Kant’s theory of mental representation that is importantly different in key respects from any previous interpretation of Kant and is an incredibly powerful theory in its own right. Of course, understanding Hume and Kant as each presenting a philosophical account of mental representation has important consequences for understanding Kant’s arguments against Hume. On at least one reading

4

Introduction

of the so-called Premise-That-Cannot-Be-Denied, Hume’s most powerful objection to it is that it purports to make a claim about a single subject of experience persisting through time and that, according to the theory of mental representation that he defends in the Treatise, we can represent no such thing. To Hume, the-Premise-That-Cannot-Be-Denied cannot be denied not because it is so compelling but rather because it is literally nonsensical. There is nothing there to deny. It follows from Hume’s theory of mental representation that our idea of the self is an idea of a bundle of perceptions, and we cannot form an idea of a single subject of experience persisting through time. So, any premise that purports to have such an idea as part of its content is absurd. If this is Hume’s objection to Kant, it makes Kant’s project in the Critique much more difficult than has previously been supposed and that much more impressive if it succeeds. Before he can even so much as begin to attempt to prove the truth of the Premise-That-Cannot-Be-Denied, he must first demonstrate its representational possibility. That is, his first task must be to show that we can and how we can represent ourselves as single subjects of experience persisting through time. Only then can he proceed to argue for his claim that such a premise cannot be denied and that the robust philosophical theses that constitute the remainder of the Critique follow from it. What I hope to show here is that this is precisely Kant’s methodology. Among the other things that Kant does in the Transcendental Aesthetic and Metaphysical Deduction is to demonstrate that the theory of mental representation that Hume presents in the Treatise fails, and to put the lessons learned from the failures of Hume’s theory toward replacing that theory with one of his own. Thus, by the time the Transcendental Deduction starts, Kant has already undermined the grounds on which Hume would deny the Premise-That-Cannot-Be-Denied and has given some positive reason for thinking that not only was Hume’s particular theory wrong but also that its replacement will be one that can accommodate the kind of mental representations Kant’s work in the Critique will require. THE STRUCTURE OF THE CASE AGAINST HUME My argument for this reading of the dialectic between Hume and Kant will take place over the course of six chapters. In the first chapter, I will show that the most well-known of Hume’s arguments employ not only his famous Copy Principle but also what I will call the Representational Copy Principle. This latter principle states that a perception is of that of which it is a copy. On its own, this would be a woefully inadequate theory of mental representation because it would leave Hume unable to account for either of two important phenomena: misrepresentation and complex representation. Thus, I demonstrate how Hume amends this version of the Representational Copy Principle to account for our more sophisticated representational

Introduction

5

capacities. With this more robust theory in hand, Hume goes on to show that some of the ideas at the core of his predecessors’ philosophical system have a much more mundane content than those philosophers had supposed. In particular, he argues—beginning with his own theory of mental representation as a premise—that we can have no idea of necessary connection (or that our idea of necessary connection is just an idea of constant conjunction), that we can have no idea of the external world (or that our idea of the external world is just an idea of certain of our perceptions), and that we can have no idea of the self (or that our idea of the self is just an idea of a bundle of perceptions). These three theses together form the hard core of Hume’s representational idealism and are the targets at the forefront of Kant’s refutation of Hume. With Hume’s arguments before us, the challenge that Kant faces in the Critique is set, and so in the second chapter I present Kant’s arguments against Hume. In particular, I argue that Kant meets Hume’s challenge by demonstrating that the theory of mental representation that grounds Hume’s conclusions is untenable. Kant must demonstrate that Hume’s account of mental representation fails to adequately capture some feature of mental representation that Hume himself would agree that it ought to explain. Thus, Kant’s rejection of Hume cannot, as a superficial reading of Kant’s arguments would have it, begin with a rejection of Hume’s conclusions, although Kant does at times make it seem as though this is how he will proceed. To be successful, Kant cannot object to Hume on the grounds that we can represent a single subject of experience, or necessary connections, or the external world. Instead, Kant must meet Hume on his own terms and show that Hume’s theory fails to meet the goals that Hume himself sets for it. Thus, as I will show, Kant begins his refutation of Hume’s account in the Transcendental Aesthetic with a critique of Hume’s account of complex representation, of representations of complex states of affairs as complex. Once Hume amends the Representational Copy Principle his theory of complex representation is that a complex of representations represents the items represented by each of its component representations as being arranged in the way that those representations are arranged. So, for example, to represent a spatial complex, one forms a picture of it in which the elements of the picture are arranged as the items pictured are. Similarly, to represent a temporal complex, one forms a movie of it in which the elements of the movie are arranged in the same temporal order as the events represented are, etc. It is to this amended version of Hume’s theory that Kant objects. His argument, initiated in the Aesthetic, continues through a number of other parts of the Critique and is that the content of any representation structured in only this way will necessarily be indeterminate. Such representations cannot represent complex states of affairs as complex because they are incapable of specifying what complex state of affairs they represent. This charge is only the first part of Kant’s critique of Hume’s theory of mental representation. In the remainder of the second chapter, I outline

6

Introduction

what moves are available to Hume within the constraints of his theory to repel Kant’s attack, and I present other arguments in response to each of these moves. The most straightforward reply to be made on Hume’s behalf is to leverage his theory of general ideas as a way of disambiguating such representations. Not just any succession of representations will count as a representation of a succession, on this line, but only those that are also subsumed under the general idea ‘succession.’ Not just any picture in which one figure is taller than another will be a picture of a figure as being taller than another but only those that are also subsumed under the general idea ‘taller than.’ Kant, of course, has a radically different theory of “general ideas,” but again, in order to motivate that theory, he must first show where Hume’s theory fails. In this case, Kant argues that in order for Hume’s theory of general ideas to do the work of saving his theory of complex representation, Hume must be able to account for how a representation is “subsumed” under the relevant general idea. That is, Kant demands that Hume provide an account of predication. I present this demand in the context of the problem of the unity of the proposition, a problem that was clearly at the forefront of both Hume’s and Kant’s thinking. Here, the dialectic takes another twist because Hume explicitly rejects the claim that there is any such thing as predication at all. He holds that a judgment, typically conceived of as the joining of two representations, a subject and a predicate, is correctly construed as a single idea, properly enlivened. This does not release Hume from his obligation of explaining how his theory of general ideas interacts with his theory of representation more generally, and I spend the remainder of this chapter presenting a dialectic of moves suggested by Hume’s and Kant’s theories, ultimately concluding that Kant’s arguments against Hume are a success. This refutation of Hume’s theory of mental representation is the first step in Kant’s freeing himself of the constraints of Hume’s conclusions. Before he can proceed with the argument of the Deduction, though, which will begin with a premise that directly contradicts Hume’s conclusion about the representation of the self and will proceed to rejecting Hume’s other important conclusions about the representation of necessary connection and the external world, Kant must first tentatively outline what the theory of mental representation is that he would put in place of the one he has just rejected. Beginning in the Metaphysical Deduction, Kant presents just such a theory, and he presents it in a way that demonstrates precisely how his theory can solve the very problems that undid Hume’s: the problem of forming complex representations of complex states of affairs as complex and the problem of the unity of the proposition. Kant’s solution is not, as one might have expected, to reject Hume’s picture theory of representation, but it is rather to retain the thesis that our mental representations function by being pictures, while replacing both the elements and structure that Hume held compose such pictures. For Hume, our mental pictures are composed of bits of sensory data structured by the

Introduction

7

various kinds of associations that we form between them. Kant replaces the elements of Hume’s pictures (minima sensibilia) with intuitions and replaces Hume’s structures (matter-of-factual associative relations) with the normative structure of concepts-qua-inferential-rules. Kant’s own theory addresses the first of his objections to Hume’s account—that Hume’s pictures do not represent determinate complex states of affairs—by replacing the structures that Hume employs with a set of structures both rich and specific enough to guarantee a one-to-one correspondence between the relations that structure our picturing and the relations that are thereby pictured. Because normative inferential relations can be created and changed as needed, they can also be used to create pictures of whatever relations it is that are in need of picturing. The inferential licenses, forbearances, etc., that in part constitute the content of a concept such as ‘larger than’ can be made to match all and only those relations that constitute the relation of being larger than. Likewise, Kant’s inferentialist theory of mental representation also avoids the second objection that he raises to Hume’s account. The problem of the unity of the proposition is essentially the problem of how to combine independently meaningful expressions in a way such that the resulting combination is not merely an aggregate of expressions but is a judgment that makes a claim. In conceiving concepts as inferential rules, Kant abandons the thesis that concepts are independently meaningful expressions in need of being combined in judgments with intuitions (singular, determinate representations). Rather, Kant casts concepts as markers of the inferential relations in which intuitions stand to one another. That is, a judgment makes a claim about the world in virtue of the fact that it is the means by which intuitions are related to one another to form a picture of the world as a whole. Using this theory to solve these problems puts at least two items immediately on the agenda for Kant and for this study of Kant. Firstly, since intuitions are the elements in Kant’s picture theory, one needs an account of what an intuition is, how it represents, and what it represents. Secondly, since concepts-qua-inferential-rules are the structure in Kant’s picture theory, one likewise needs an account of what a concept is, how it represents, and what it represents. Each of these is explained over the course of the following two chapters (3 and 4), which engage the A-Deduction and B-Deduction respectively. In the third chapter, I begin with the contrast between Hume’s and Kant’s accounts of how we represent complex states of affairs as complex. I begin by showing Kant’s deep commitment to the thesis that all such representations are only possible via the deployment of concepts. This commitment has important consequences for how we understand the account of intuitions that Kant gives in the A-Deduction because he is there also explicit that intuitions are representations of complex objects as complex. Combining these two theses one arrives at the conclusion that intuitions must themselves be conceptually structured. Of course, this creates a bit of a new puzzle since according to the theory outlined above it is intuitions that are

8

Introduction

the elements of Kant’s pictures, and it is difficult to see how they themselves are also pictures in this sense. The solution is to see that the elements that compose intuitions are what Kant calls sensations and that these are not, strictly speaking, structured by concepts but do have a structure that models that of concepts. The role of sensations in intuitions provides the crucial link between Kant’s inferentialism about concepts, and thus his picture theory more generally, with perception. This in turn allows for Kant to give a robust sense to his claim that mental representations picture the world. Because our inferences are traceable back to “certain appearances that come before us,” the picture that we form using such appearances serves as an explanation of those very appearances, an explanation that, for Kant, has an ineliminable ontological component. That is the topic that of the next chapter, which continues the discussion of the nature of intuitions and adds crucial details concerning the nature of the concepts-qua-inferential-rules that link them. In particular, I there argue that the best way to understand the various claims that Kant makes about intuitions, objects, necessity, concepts, and the world as governed by causal laws is by taking him to hold that an intuition is a representation of an object as the necessary connection of its parts and that we represent such objects by employing rules of inference that are valid not just in virtue of their logical form but also in virtue of the content of the representations in the judgments that they involve. So, for example, in licensing the inference from 1. This elephant tail in front of me is grey, to 2. If I turn my head, there will be a grey elephant body in front of me, we thereby picture an elephant as the necessary connection of its tail and body. Of course, sadly, not all elephant tails are connected to elephant bodies, and so there is a sense in which this is not a valid inference at all. That is, the concept ‘elephant’ might license this inference, and since there are tail-less elephants, there is sense in which we ought to abandon the use of that concept. That is, we can discover that the pictures we use to represent the world are incorrect, or that they provide inadequate explanations of the order and coherence of experience. In such cases, we change the inferences we regard as valid and thereby change the systems of concepts that we use to picture the world. So, Kant’s account of mental representation has rich resources for explaining how scientific theories, explanations of the causal structure of the world, can change, a surprising result for a philosopher who is all too often criticized for giving the particular details of Newtonian science too privileged an epistemic place. The following chapter, 5, explores these resources in more detail by turning to the First and Second Analogies, which are not only where Kant is

Introduction

9

often portrayed as most directly responding to Hume but also where he delineates what he takes to be the regulative principles that govern the adequacy of our representations of the world. Kant’s goal in the Analogies is to present the specific form that these meta-inferential rules will take for creatures like us, creatures whose forms of intuition are space and time. In detailing how Kant’s arguments in the first two Analogies interact with his inferentialist theory of concepts, I present two novel defenses of his conclusions. First, I contend that Kant’s argument in the First Analogy does not consist in the simple quantifier-scope fallacy of which he has been accused and that this charge arises because of misinterpretation of his goal there. Kant intends to demonstrate the conditions not only for representing time as a unity of past, present, and future but also the unity appropriate to having a single standard of time’s passage. Next, I defend a reading of Kant’s argument in the Second Analogy as aimed at the conclusion that we can only represent the determinate order of temporal events by representing such events as being governed by specific causal laws (as opposed to representing them as merely governed by some law or other). This defense draws on the details of the theory of mental representation that is defended in earlier chapters and specifically on Kant’s answer to the problem he finds in Hume regarding determinate complex representation. What the arguments and theories culled from the examinations of the Transcendental Deduction and Analogies yield is a sophisticated inferentialist theory of concepts that has at its core the thesis that we represent the world by picturing the lawful relations of alterations of the states of a single sempiternal substance. We form a picture of that substance by placing representations of the parts of it, intuitions, into inferential relations with another via applying concepts to them in judgments, which concepts serve to locate those intuitions in a system of inferences that we refine as we learn more about the world that is so pictured. Thus, I take Kant’s inferential theory of concepts to yield a kind of scientific realism. What is pictured is substance, the real in appearance, and by replacing one picture of substance (one system of concepts-qua-inferential-rules) with another, we form increasingly accurate pictures of that which is real. At this point in the dialectic, Kant has earned back two of Hume’s three important conclusions. Because the rules of inference that structure our mental pictures, on Kant’s account, serve to represent objects in the world as necessarily connected, and because they thereby represent such objects as distinct from our perception of them and as continuing to exist when we are not perceiving them, Kant now has on offer a theory that demonstrates precisely how we can resist Hume’s conclusions regarding the ideas of necessary connection and the external world. That leaves just Hume’s thesis about the self: that we can have no idea of a single subject of experience persisting through time, and that is the subject of the final chapter. The sixth chapter addresses the question of how to use Kant’s theory of mental representation to account for the representation of the self. Focusing

10

Introduction

on what Kant says about the transcendental unity of apperception in the Deductions and on his arguments against the Rational Psychologist in the Paralogisms, I argue that, unlike representations of objects, the representation of the self, for Kant, is a representation not of any object but rather of just the representational properties of the ‘I think.’ Thus I side with those scholars who defend the claim that the ‘I think’ is a purely formal representation and against those who hold that what is represented by the ‘I think’ is essentially a substance (noumenal or phenomenal). While representations of objects bear an essential relation to “certain appearances that come before us,” i.e., to perception, the representation of the self does not. I give cash value to the formalist’s contention by explicating the notion of a formal representation in terms of the inferences that the representation ‘I think’ licenses, forbids, etc. In particular, in the Paralogisms, Kant presents three analytic propositions that together exhaust everything that can be said about the self, and each one of these turns out to concern only the inferential properties of the representation ‘I think.’ Namely, the ‘I think’ can never be predicated of any other representation, it is representationally simple, and it is univocal within a given subject’s thought. Any representation meeting these three criteria, for Kant, is a representation of a single subject of experience persisting through time, and any representation that fails to meet any one of them is not. Kant is a kind of functionalist, that is, about the representation of the self. Classical functionalism is the thesis that a mental state is the mental state that it is in virtue of its causal role. Kant’s thesis, by contrast, is that the self (not just a particular mental state) is a self in virtue of its inferential role. Thus, the unity of the self, the transcendental unity of apperception that Kant argues in the Deduction is a necessary precondition for all experience, is the unity of the subject of certain inferential norms, and to achieve this unity is just to be subject to certain meta-norms. Specifically, it is to accompany one’s representations with a further representation that has the inferential properties listed above. With this final piece of Kant’s theory articulated, the dialectic between Hume and Kant is completed. Hume begins with his theory of mental representation. Using that theory he concludes that we cannot so much as represent necessary connections, the external world, or the self. Kant begins his refutation of Hume by first arguing that Hume’s theory of mental representation fails on its own terms. He then replaces that theory with one of his own. Finally, he shows how his new theory can account for precisely the representations that Hume’s theory predicted we could not have. In tracing this dialectic, my hope is to recapture the genuine and philosophically robust arguments that one of the world’s greatest thinkers made against one of the world’s other greatest thinkers. I hope to show that not only does Kant address Hume head on in a way that has not been understood before but also that his arguments against Hume succeed and that what we are left with is a powerful theory of mental representation worthy of standing side by side with even the most sophisticated contemporary accounts.

Introduction

11

WAS HUME KANT’S TARGET IN THE CRITIQUE? While the arguments that I consider over the course of the book are themselves strong evidence of the historical accuracy of understanding Kant as responding to Hume, it will be worth pausing for a moment here to consider that question in a more focused way, especially as this has long been a matter of some controversy. I will begin by reviewing the state of the discipline regarding what Kant seems to have read of Hume and when, and what this implies about Kant’s ability and possible inclination to aim his objections at Hume’s arguments and conclusions. After that I will consider the arguments of two recent Kant scholars—Gary Hatfield9 and Eric Watkins10—for the conclusion that we ought not to read Kant as responding to Hume. Finally, I will briefly discuss two prominent recent studies that share the approach to Kant through Hume that I take here: Henry Allison’s11 and Paul Guyer’s.12 To begin, Hume’s Enquiry was first published in 1748 and was translated into German in 1755. Kant owned a copy of this translation at the time of his death, and so it seems reasonable to suppose that he owned and read it well before the completion of the Critique.13 Thus, Kant would have had sustained access to that version of Hume’s account of our idea of causation, which for Kant forms the centerpiece of Hume’s challenge (and which he thought Hume ought to have extrapolated to encompass at least the other Categories, and one presumes the concept of the self as well.)14 Kant also would have had access to section 1.4.5 of the Treatise, “Conclusion of this Book,” via Hamann’s translation of it, which was published in a Königsberg newspaper in 1771. Thus, Kant would have read at least these passages of Hume’s, which contain brief statements of his accounts of causation, the external world, and the self, as well as a suggestion that he considers each of these ideas not only illegitimate but also meaningless. Here is Hume on our concept of the self: Nay, even to these objects we cou’d never attribute any existence, but what was dependent on the senses; and must comprehend them entirely in that succession of perceptions, which constitutes our self or person. T 1.4.7.3; SBN 266, emphasis added And on the imagination’s role in the representation of cause and effect and the external world: ’Tis this principle, which makes us reason from causes and effects; and ’tis the same principle, which convinces us of the continu’d existence of external objects, when absent from the senses. T 1.4.7.4; SBN 266

12

Introduction

And again on causation, this time with an emphasis on the meaninglessness of ideas of necessary connection: And how must we be disappointed, when we learn, that this connexion, tie, or energy [between causes and their effects] lies merely in ourselves, and is nothing but that determination of the mind, which is acquir’d by custom, and causes us to make a transition from an object to its usual attendant, and from the impression of one to the lively idea of the other? Such a discovery not only cuts off all hope of ever attaining satisfaction, but even prevents our very wishes; since it appears, that when we say we desire to know the ultimate and operating principle, as something, which resides in the external object, we either contradict ourselves, or talk without a meaning. T 1.4.7.5; SBN 267–268, emphasis added In addition to the Enquiry and this section of the Treatise, Kant would also likely have had access to the extended engagement with and citations from the Treatise in the German translation of Beattie’s An Essay on the Nature and Immutability of Truth, in Opposition to Sophistry and Scepticism, which is known to have been available to him in the Königsberg university library, as well as those in various reviews of the Treatise in German periodicals.15 In Beattie, Kant would have read quotations from the Treatise such as the following, which concern Hume’s accounts of the self qua a bundle of perceptions, the incomprehensibility of the external world as anything but certain of our perceptions, and of our idea of cause qua the idea of a certain habit of mind, respectively: though that author [Hume], in his Treatise of Human Nature, has asserted, yea, and proved too, according to his notions of proof, that the human soul is perpetually changing; being nothing but “a bundle of perceptions, that succeed each other with inconceivable rapidity,” and are (as he chuses to express it) “in a perpetual flux.”16 “Philosophers,” says he, “have distinguished between objects, and perception, of the senses; but this distinction is not comprehended by the generality of mankind.”17 [. . .] when we think we perceive our mind acting on matter, or one piece of matter acting upon another, we do in fact perceive only two objects or events contiguous and successive, the second of which is always found in experience to follow the first but that we never perceive, either by external sense, or by consciousness, that power, energy, or efficacy, which connects the one event with the other. By observing that the two events do always accompany each other, the imagination acquires a habit of going readily from the first to the second, and from the second to the first and hence we are led to conceive a kind of necessary connexion between them. But in fact there is neither necessity nor power in the objects we consider, but only in the mind that

Introduction

13

considers them and even in the mind, this power of necessity is nothing but a determination of the fancy, acquired by habit, to pass from the idea of an object to that of its usual attendant.18 So, while it is certainly true that Kant did not have direct access to the detailed arguments of the entire Treatise, he would have had the opportunity for a robust examination of the abbreviated arguments from the Enquiry and many of the conclusions from the Treatise, as well as secondhand accounts of much of the important reasoning from the parts of the Treatise that he had not been able to read himself. All of this combined with Kant’s own explicit descriptions of his project as inspired by and directed at Hume19 would certainly seem to suffice for an at least prima facie case for reading the Critique as at least in part a response to Hume. If such an interpretation produces robust philosophical and historical outcomes, as I can only hope that the current study does, that will be more evidence still. That is a brief sketch of the positive historical grounding of the case for reading Kant as responding to Hume and a promissory note for a positive philosophical grounding. The next item on the agenda is to consider briefly some recent arguments against such an interpretation. One such argument comes from Gary Hatfield, who proposes that Kant’s main aim the Critique is a negative one: to limit the scope of knowledge claims such as to refute the ambitions of speculative metaphysicians. This is in contrast to the positive project that many commentators have seen as Kant’s primary aim: to justify the application of the pure a priori concepts of the understanding to objects of possible experience. Accordingly, Hatfield casts Kant as taking Hume as an ally against the speculative metaphysicians and precisely not as presenting a skeptical threat to commonsense and science, and thus as standing in need of refutation. Hatfield announces that part of his motivation for advancing this interpretive thesis is to save Kant from precisely the dilemma with which this introduction began. He reasons that if the Deduction is aimed at refuting the skeptic, then it must begin with a premise that the skeptic would accept. If anything interesting follows from this premise, then it will not be one that the skeptic accepts, and if it is such as to be acceptable to the skeptic, then nothing interesting can follow from it. Certainly, abandoning the attempt to understand the Critique as a whole as purporting to refute the skeptic is one way out of this dilemma. I have already proposed the outlines of my own way through it, which is not entirely dissimilar. The difference is that my proposal does not aim to take the skeptic out of the sights of the entire Critique but only out of the Deduction in particular. The Deduction is not aimed at Hume because by the time the Deduction begins, Hume has already been refuted. The premises that Hume has to accept on this scenario are only those regarding the criteria of success that he has adopted for his own theory of mental representation. Kant shows that Hume’s theory fails to meet these criteria and so is free to abandon the conclusions to which

14

Introduction

Hume’s theory leads him and start afresh in the Deduction. At that point, the goal is not to refute the skeptic anymore but rather to show that those concepts that Hume rejects as impossible to possess are not only possible but also that their application to objects of possible experience is entirely justified. Of course, this is a response to Hatfield only insofar as the project of this book is a success. If it is, then I will have addressed Hatfield’s motivation for seeking a way to read Kant and Hume as allies. That, however, does not address the evidence he presents for his interpretation. This evidence consists mostly in a close reading of parts of the A-edition of the Critique where Kant states that neither ordinary nor scientific knowledge stands in any need of defense by philosophy. As Guyer points out,20 this claim is compatible with Kant’s holding that such a justification is nonetheless possible, and that Hume’s skepticism leads philosophy to a position from which it would be impossible, and thus as seeing the need for a corrective. Furthermore, Hatfield concedes that in the Prolegomena and B-edition Kant is concerned with refuting the skeptic but argues that this is due more to pressure from Kant’s philosophical audience than to any real conviction of Kant’s that such a defense was necessary.21 That, however, requires both refusing to take Kant at his word when he testifies in those later pieces that this was part of his original intention and supposing that, again despite Kant’s own statements to the contrary, that the Prolegomena and B-edition were not mere reframings of the same material but were rather drastic reconceptualizations of the entire Critical project. That is certainly a live interpretive option, but if an interpretation can be found that takes Kant more at his word, this would seem to be a better option. Again, my hope is that this study can provide at least the outlines of one such interpretation. Finally, Eric Watkins has offered several reasons for thinking that Kant was not out to refute Hume, the most germane of which is that Kant does not offer an argument that proceeds from premises that Hume accepts to a conclusion that Hume rejects.22 Again, the dialectic that I have briefly described above is meant to address precisely this kind of concern about Kant’s arguments. If one goes hunting for the Premise-That-Even-HumeMust-Accept in the Deduction, or even farther along in the Analogies, as Watkins does, one is bound to fail in one’s search. This is because, by the time one reaches the Deduction, Hume has already been refuted. Given the nature of this refutation—Kant’s demonstrating that Hume’s theory of mental representation cannot account for certain phenomena that are clearly among its explananda—one can cast that argument in the form that Watkins demands, although in a rather roundabout way. One would begin by assuming that Hume’s theory of mental representation is adequate (for the purposes of reductio), take as another premise that certain phenomena must be accounted for by any adequate theory of mental representation, and proceed by showing that Hume’s theory is not adequate by this standard. So, again, the best argument that Kant really does take himself to be arguing

Introduction

15

against Hume is to show exactly how his arguments against Hume work. It is to that work that I will turn before one final introductory consideration. A final word is also in order about two prominent recent studies that do take Hume to be Kant’s target in the Critique: Allison, Custom and Reason and Guyer, Knowledge, Reason, and Taste. While each of these important books shares the approach to Kant through Hume of the current study, each also differs from what will follow in fundamental ways that take the three studies down widely divergent paths. In both cases, it is the author’s reading of Hume where this difference begins. Guyer reads Hume’s arguments as reaching conclusions about our knowledge of necessary connection, the external world, and the self. As I have suggested above, I understand Hume’s arguments as being much more radical: as reaching conclusions about our ability to represent such things. So, whereas Guyer interprets the dialectic between Hume and Kant as being concerned first and foremost with epistemology and its metaphysical underpinnings, I read it as one regarding theories of mental representation. Chapter 1 presents my interpretation of Hume along these lines and a defense of the particular theory of mental representation that I attribute to him. The rest of the book, of course, presents and defends my reading of Kant as responding to this Hume. Given this early and fundamental departure from Guyer, I will not have the opportunity to engage his text much during the rest of the book. Unlike Guyer, Allison does understand Hume as offering a theory of mental representation as a key component of his arguments concerning necessary connection, the external world, and the self. Unfortunately, he quickly abandons on Hume’s behalf the thesis at the center of this theory: that a representation of a complex state of affairs is just a complex of representations. He does so on the grounds that it is incompatible with Hume’s Copy Principle and that it does not allow Hume to account for representing a complex as complex. In the fourth section of Chapter 1, I demonstrate that with a small emendation of the letter of Hume’s theory, though not its spirit, Hume can use a version of the Copy Principle to give an account of the content of complex ideas. In Chapter 2, I present Kant’s critique of this theory, which does depend, in part, on an argument that Hume’s theory ultimately fails on this score. I do not, however, take that to be evidence that Hume did not hold such a theory, nor that his theory did not represent a worthwhile attempt to account complex representation using the austere resources that Hume grants himself. So, rather than abandon Hume’s claim, I argue that it is this claim, and the corresponding failure of Hume’s theory, that provides the impetus for Kant to provide the radical new account of mental representation that he does. I think we do Hume no favors by dismissing his theory as quickly as Allison does, and we do Kant no favors by neglecting his arguments against that theory and the structure that such arguments impose on the theory that he presents as a response to it. Thus, as with Guyer’s book, the approach that Allison’s book takes also diverges significantly early from

16

Introduction

the one taken here, and not much interaction between the two is possible from that point on. In comparing these three accounts, it will again be the fruitfulness of each interpretation as a whole that will, in the end, I think, decide which is best. So, it is to the work of presenting my interpretation that I now turn. NOTES 1. A clear and concise catalogue of many of the various possible positions here can be found in Van Cleve, Problems From Kant, 79–84. Notable attempts to navigate the Deduction with an eye toward this particular difficulty can be found in Wolff, Kant’s Theory of Mental Activity, 105–17; Strawson, Bounds of Sense, 85–117; Allison, Kant’s Transcendental Idealism, 137–40; and Engstrom, “Transcendental Deduction and Skepticism.” Castañeda, “Role of Apperception,” rejects entirely the need to make any concessions intended to persuade the Humean skeptic at all. And, of course, there are those who deny the veracity of the legend entirely and who argue that Kant and Hume are actually on the same side after all. See Wolff, “Hume’s Theory of Mental Activity,” and Kuehn, “Kant’s Transcendental Deduction.” Rosenberg, “Transcendental Arguments Revisited,” argues that the key to resolving the issues concerning the Premise-That-Cannot-Be-Denied lies in understanding the Transcendental Deduction as a practical argument that has, as its first premise, not a statement that cannot be denied but rather an intention that is constitutive of what it is to be the kind of creature that Kant takes us to be. I argue in Landy, “Premise That Hume Must Accept,” that while this suggestion helps Kant in several ways, it does not address the problem of the nature of the representation that is at the crux of the The-Premise-That-Cannot-Be-Denied. 2. Cf. Garrett, “Hume’s Naturalistic Theory of Representation,” and Schafer, “Hume’s Unified Account of Mental Representation.” 3. Cf. Winkler, “New Hume”; Stroud, Hume; Fogelin, Hume’s Skepticism; Strawson, Secret Connexion. 4. Cf. Allison, Kant’s Transcendental Idealism; Guyer, Kant and the Claims of Knowledge; Brook, Kant and the Mind; Kitcher, Kant’s Transcendental Psychology; Friedman, Kant and the Exact Sciences. 5. Cf. Strawson, Bounds of Sense; Bennett, Kant’s Analytic; Sellars, Science and Metaphysics; McDowell, Having the World in View; Longuenesse, Kant and the Capacity to Judge; Hanna, Kant and the Foundations of Analytic Philosophy. 6. Sellars, Science and Metaphysics. 7. Hanna, “Kant and Nonconceptual Content”; Ginsborg, “Was Kant a Nonconceputalist”; Allais, “Non-Conceptual Content.” 8. Rosenberg, Accessing Kant; O’Shea, Kant’s Critique. 9. Hatfield, “Prolegomena and Critiques”; Hatfield, “What Were Kant’s Aims in the Deduction.” 10. Watkins, Kant and the Metaphysics of Causality. 11. Allison, Custom and Reason. 12. Guyer, Knowledge, Reason, and Taste. 13. Guyer, Knowledge Reason, and Taste, 5; Warda, Immanuel Kants Bucher, 50; Hatfield, “Prolegomena and Critiques,” 186, n6. 14. Ak 4:260; Theoretical Philosophy, 57.

Introduction

17

15. Guyer, Knowledge, Reason, and Taste, 6, n5; Kuehn, Scottish Common Sense in Germany, 169. 16. Beattie, Essay, 53. 17. Beattie, Essay, 162. 18. Beattie, Essay, 195–96. 19. Ak 4:260; Theoretical Philosophy, 57. 20. Guyer, Knowledge, Reason, and Taste, 10. 21. Hatfield, “What Were Kant’s Aims in the Deduction.” 22. Watkins, Kant and the Metaphysics of Causality, 2005.

REFERENCES Allais, Lucy. “Kant, Non-Conceptual Content and the Representation of Space.” Journal of the History of Philosophy 47 (2009): 383–413. Allison, Henry. Kant’s Transcendental Idealism. New Haven: Yale University Press, 1983. Allison, Henry. Custom and Reason in Hume: A Kantian Reading of the First Book of the Treatise. Oxford: Clarendon, 2008. Beattie, James. An Essay on the Nature and Immutability of Truth, in Opposition to Sophistry and Scepticism. Edinburgh: Denham & Dick, 1805. Bennett, Jonathan. Kant’s Analytic. Cambridge: Cambridge University Press, 1966. Brook, Andrew. Kant and the Mind. Cambridge: Cambridge University Press, 1994. Castañeda, Hector. “The Role of Apperception in Kant’s Transcendental Deduction of the Categories.” Noûs 24 (1990): 147–57. Engstrom, Stephen. “The Transcendental Deduction and Scepticism.” Journal of the History of Philosophy 32 (1994): 359–80. Fogelin, Robert. Hume’s Skepticism in the Treatise of Human Nature. London: Routledge Kegan & Paul, 1985. Friedman, Michael. Kant and the Exact Sciences. Cambridge, MA: Harvard University Press, 1992. Garrett, Don. “Hume’s Naturalistic Theory of Representation.” Synthese 152 (2006): 301–19. Ginsborg, Hanna. “Was Kant a Nonconceptualist?” Philosophical Studies 137 (2008): 65–77. Guyer, Paul. Kant and the Claims of Knowledge. Cambridge: Cambridge University Press, 1987. Guyer, Paul. Knowledge, Reason, and Taste: Kant’s Response to Hume. Princeton: Princeton University Press, 2008. Hanna, Robert. Kant and the Foundations of Analytic Philosophy. Oxford: Oxford University Press, 2001. Hanna, Robert. “Kant and Nonconceptual Content.” European Journal of Philosophy 13 (2005): 247–90. Hatfield, Gary. “The Prolegomena and Critiques of Pure Reason.” In Kant Und Die Berliner Aufklarung: Akten Des Ix. Internationalen Kant-Kongresses, edited by Volker Gerhardt, Rolf-Peter Horstmann, and Ralph Schumacher, 185–208. Berlin: Walter de Gruyter, 2001. Hatfield, Gary. “What Were Kant’s Aims in the Deduction?” Philosophical Topics 31 (2003): 165–98. Hume, David. A Treatise of Human Nature. Edited by L. A. Selby-Bigge. New York: Oxford University Press, 1974.

18

Introduction

Hume, David. A Treatise of Human Nature. Edited by David Fate Norton and Mary J. Norton. New York: Oxford University Press, 2000. Kant, Immanuel. Critique of Pure Reason. Translated by Paul Guyer and Allen W. Wood. Cambridge: Cambridge University Press, 1998. Kant, Immanuel. Theoretical Philosophy After 1781. Edited by Henry Allison and Peter Heath. Translated by Gary Hatfield, Michael Friedman, Henry Allison, and Peter Heath. Cambridge: Cambridge University Press, 2002. Kitcher, Patricia. Kant’s Transcendental Psychology. New York: Oxford University Press, 1990. Kuehn, Manfred. “Kant’s Transcendental Deduction: A Limited Defense of Hume.” In New Essays on Kant, edited by Bernard den Ouden, 47–72. New York: Peter Lang, 1987. Kuehn, Manfred. Scottish Common Sense in Germany, 1760–1800: A Contribution to the History of Critical Philosophy. Kingston, ON: McGill-Queen’s University Press, 1987. Landy, David. “The Premise That Even Hume Must Accept.” In Self, Language, and World: Problems from Kant, Sellars, and Rosenberg, edited by Jim O’Shea and Eric Rubenstein, 28–46. Atascadero, CA: Ridgeview Publishing Co., 2010. Longuenesse, Beatrice. Kant and the Capacity to Judge. Translated by Charles T. Wolfe. Princeton: Princeton University Press, 1998. McDowell, John. Having the World in View. Cambridge, MA: Harvard University Press, 2009. O’Shea, James. Kant’s Critique of Pure Reason: An Introduction and Interpretation. London: Acumen Press, 2012. Rosenberg, Jay. “Transcendental Arguments Revisited.” Journal of Philosophy 72 (1975): 611–24. Rosenberg, Jay. Accessing Kant: A Relaxed Introduction to the Critique of Pure Reason. Oxford: Oxford University Press, 2005. Schafer, Karl. “Hume’s Unified Account of Mental Representation.” European Journal of Philosophy 21 (2013). doi: 10.1111/ejop.12022. Sellars, Wilfrid. Science and Metaphysics. Atascadero, CA: Ridgeview Publishing Company, 1967. Strawon, Galen. The Secret Connexion. Oxford: Oxford University Press, 1989. Strawson, Peter. The Bounds of Sense. London: Methuen Ltd., 1966. Stroud, Barry. Hume. London: Routledge & Kegan Paul, Ltd., 1977. Van Cleve, James. Problems From Kant. New York: Oxford University Press, 1999. Warda, Arthur. Immanuel Kants Bucher. Berlin: Breslauer, 1922. Watkins, Eric. Kant and the Metaphysics of Causality. Cambridge: Cambridge University Press, 2005. Winkler, Kenneth. “The New Hume.” The Philosophical Review 100 (1991): 541–79. Wolff, Robert Paul. “Hume’s Theory of Mental Activity.” The Philosophical Review 69 (1960): 289–310. Wolff, Robert Paul. Kant’s Theory of Mental Activity. Cambridge, MA: Harvard University Press, 1963.

1

Hume’s Theory of Mental Representation

The goal of the current chapter will be to demonstrate that some of Hume’s most important conclusions in the Treatise are first and foremost conclusions about what the human mind can and cannot represent. That is, they are conclusions about what our ideas can be ideas of. In particular, the three most important of Hume’s conclusions for current purposes can be expressed as follows. (NC) We have no idea that is an idea of a necessary connection. (EW) We have no idea that is an idea of the external world. (SSE) We have no idea that is an idea of a single subject of experience persisting through time. The first thing to notice about these conclusions is that there is an air of paradox about them. Each claims that we have no idea with a certain kind of content. Of course, it is natural to think that in order to understand each of these claims, one must understand the words that comprise them, and in order to understand those words, one must have ideas with the corresponding content. So, in order to understand the claim that we have no idea that is an idea of necessary connection, for example, one must have an idea of necessary connection. If not, there is no way to understand what is being claimed here. Thus the air of paradox. It is for this reason that Hume presents his arguments and conclusions in both of two ways. Hume’s goal in the Treatise is not just to articulate his own theory of human nature but also to show that the theories of his predecessors cannot possibly be right. In order to accomplish the latter task, he must engage his predecessors, and at times that engagement requires him to write in a way that belies his own theoretical commitments. So, for instance, he has to write as if the idea of necessary connection is perfectly intelligible, in order to demonstrate that we cannot possibly have such an idea. It is well known that Hume often puts aside his own theoretical commitments to “speak with the vulgar.” What is less noted is that he does the same in order to “speak with the philosophical.” Hume writes in at least three voices throughout the Treatise: in the voice of a scientist of man (which is what

20

Hume’s Theory of Mental Representation

Hume takes himself to be), in the voice of the lay person, and in the voice of his philosophical predecessors.1 The above conclusions are all expressed in the last of these voices, even though in the voice of the scientist of man, they are strictly speaking nonsensical. Of course, Hume’s conclusions can also be said in the scientist’s voice. When carefully articulating his own views, he writes more strictly and keeps to showing not that we do not have an idea of necessary connection, for example, but rather that our idea of causation is just an idea of constant conjunction, etc. He argues not that we have no idea of a single subject of experience persisting through time but rather that our idea of the self is just an idea of a bundle of perceptions. He argues not that we have no idea of the external world but rather that our idea of the external world is just an idea of certain of our perceptions. So, each of the conclusions above can be rendered more strictly as a positive thesis about what our ideas are ideas of, but in order to engage his predecessors, Hume is instead willing to use these somewhat paradoxical formulations. Here are a few examples of places at which Hume uses an idiom that makes it clear that these representational theses are exactly the kinds of conclusions that his arguments are meant to establish. During the course of his arguments concerning necessary connection he writes that, We never therefore have any idea of power. T 1.3.14.11; SBN 161, emphasis added About the external world he writes, To begin with the senses, ‘tis evident these faculties are incapable of giving rise to the notion of the continu’d existence of their objects, after they no longer appear to the senses. T 1.4.2.3; SBN 188, emphasis added Finally, about the self he writes, If any impression gives rise to the idea of the self, that impression must continue invariably the same, thro’ the whole course of our lives; since self is suppos’d to exist after that manner. Pain and pleasure, grief and joy, passions and sensations succeed each other, and never all exist at the same time. It cannot, therefore, be from any of these impressions, or from any other, that the idea of self is deriv’d; and consequently there is no such idea. T 1.4.6.2; SBN 251–2, emphasis added. We will have occasion to return to each of these passages farther on, but it is worth noting here at the outset that Hume’s idiom is an explicitly representational one. In each case his concern, as stated, is with the idea of

Hume’s Theory of Mental Representation

21

such-and-such, and his conclusions are all that we cannot have certain ideas with this controversial content. Supposing that Hume is concerned with determining what our ideas can and cannot represent brings with it certain advantages. To begin, consider the following puzzle concerning the precise status of Hume’s Copy Principle that has been the subject of a good deal of hand-wringing amongst Hume scholars in recent decades. Recall that the Copy Principle states that all simple ideas are copies of some simple impression and that it figures prominently in many of Hume’s most famous and powerful arguments. On the one hand, if the Copy Principle is a mere empirical generalization, it lacks the authority to be used to refute the claims of Hume’s predecessors that we have such controversial ideas as those of necessary connection, the external world, or the self. It would seem that each of these, rather than being undermined by the Copy Principle, would be counterexamples to it. On the other hand, the only alternative to the Copy Principle’s being an empirical generalization would seem to be that it is an a priori principle. This alternative is unattractive for at least two reasons. Firstly, accepting it severely undermines Hume’s commitment to pursuing a purely empirical science of man. Secondly, Hume explicitly denies that there can be any a priori principles regarding the causal connections between ideas, and the Copy Principle has an explicit causal component. So, the Copy Principle cannot be an a priori principle. Various attempts have been made by recent commentators to avoid spearing Hume on the horns of this dilemma.2 What I will argue here is that all of these efforts have been misplaced because there is an important sense in which it is not the Copy Principle that is meant to do this work in Hume’s arguments at all. The Copy Principle is a claim about the matter of factual relations of simple ideas to their corresponding simple impressions: the former are all copies of the latter. That is, simple ideas all exactly resemble and are caused by some simple impression. This kind of claim alone, however, cannot be all that is in play in Hume’s arguments. As I have suggested, Hume’s conclusions are all regarding what our ideas are ideas of. We do not have an idea of necessary connection. We do not have an idea of the external world. We do not have an idea of the self, Etc. What the Copy Principle earns Hume is that we do not have an idea that is a copy of, for instance, a necessary connection. That does not establish that we do not have an idea that is of a necessary connection, however, without the additional premise that our ideas are of that of which they are copies. This premise is what I will call the Representational Copy Principle and is, I will argue, the premise that does all the heavy lifting in Hume’s purported refutations of his predecessors.3 Thus, it is not the Copy Principle that stands in need of dialectical support—it might be a mere empirical generalization—but the Representational Copy Principle that does. Of course, not much would be gained if the Representational Copy Principle faces the same fatal dilemma that the Copy Principle does, and so I will further argue that it does not. In particular, I will

22

Hume’s Theory of Mental Representation

argue that the problems that face understanding the Copy Principle as an a priori principle do not apply to understanding the Representational Copy Principle as one. In particular, rather than violating Hume’s commitment to empiricism, the Representational Copy Principle qua a priori principle simply expresses this commitment. Furthermore, since the Representational Copy Principle is not itself a thesis that claims that any particular causal connections actually obtain, that it is an a priori principle does not violate Hume’s condemnation of purportedly a priori knowledge of such causal connections. My procedure here will be as follows. In the first section I will present and critique Don Garrett’s influential solution to the dilemma concerning the status of the Copy Principle. I will there draw what I take to be a crucial distinction between a perception’s pictorial content and its representational content, and argue that Garrett’s reading of the Copy Principle concerns only the pictorial content of perceptions, but that what is needed for Hume’s purposes is an argument concerning their representational content. In the next section, I will present three of Hume’s most important arguments from the Treatise, noting the crucial role that the Representational Copy Principle plays in each of these arguments. In the third section, I will demonstrate how the Representational Copy Principle does not fall prey to the same objections that the Copy Principle does. In the final section, I make an important modification to the Representational Copy Principle concerning complex ideas that will be of crucial importance to us going forward with Kant’s critique of Hume. GARRETT’S DEFENSE OF THE COPY PRINCIPLE Before we look at Garrett’s defense of Hume’s use of the Copy Principle, there are a few pieces of business that require our attention. First of all, there is the definition of the Copy Principle. The Copy Principle states that every simple idea is a copy of some simple impression. The key notion here is that of being a copy, and Hume is fairly clear about just what this entails. For x to be a copy of y requires that two conditions be met. These conditions are each necessary for x to be a copy of y, and together are jointly sufficient for that. The first condition is that x must be caused by y. Of course, ‘cause’ must be construed in the proper Humean way here, so that for x to be caused by y is for x and y to be constantly conjoined, and for y to always precede x. So, when Hume sets out to prove the Copy Principle in the opening pages of the Treatise, he observes that exactly these two parts of the causal condition are met. I first make myself certain, by a new review, of what I have already asserted, that every simple impression is attended with a correspondent idea, and every simple idea with a correspondent impression. From this

Hume’s Theory of Mental Representation

23

constant conjunction of resembling perceptions I immediately conclude, that there is a great connexion betwixt our correspondent impressions and ideas, and that the existence of one has a considerable influence upon that of the other. Such a constant conjunction, in such an infinite number of instances, can never arise from chance; but clearly proves a dependence of the impressions on the ideas, or of the ideas on the impressions. That I may know on which side this dependence lies, I consider the order of their first appearance; and find by constant experience, that the simple impressions always take the precedence of their correspondent ideas, but never appear in the contrary order. T 1.1.1.8; SB 4 Correspondent impressions and ideas are constantly conjoined, and the former always precede the latter. Thus, Hume can conclude that impressions are the cause of ideas (in the proper Humean sense of ‘cause’). The second condition that must be met for x to be a copy of y is that x must exactly resemble y. Again, here is Hume in the opening pages of the Treatise offering evidence that this condition is met in the case of ideas and impressions. The first circumstance, that strikes my eye, is the great resemblance betwixt our impressions and ideas in every other particular, except their degree of force and vivacity. T 1.1.1.3; SB 2 Of course, Hume goes on to limit his resemblance thesis to simple ideas and impressions only, and so correspondingly limits the Copy Principle to just these as well. It is important to note here another restriction in scope that Hume places on the resemblance thesis at the end of this quotation. An idea can exactly resemble an impression even if the degrees of force and vivacity of the two are different. This is because the exact resemblance thesis concerns specifically what Hume elsewhere calls the circumstances of these perceptions, and what I will call their pictorial content. The circumstance, or pictorial content, can best be explicated by way of an analogy. Consider the picture on the following page. The pictorial content of this picture consists of four black lines of equal length arranged at ninety-degree angles to one another against a white background. That is in what the picture consists. For another picture to exactly resemble this one, it would also have to consist in four lines of this length arranged at ninety-degree angles to one another against a white background. The pictorial content of an image, including impressions and ideas, is constituted entirely by intrinsic features of that image. This is a point that I have made elsewhere about Hume’s exact resemblance thesis in order to explicate the notions of force and vivacity, which are not part of the pictorial content of perceptions precisely because they

24

Hume’s Theory of Mental Representation

Figure 1.1

are not intrinsic features of a perception.4 What concerns us here, however, is a slightly different contrast. Notice that in describing the image above, we made no reference to what that image is an image of. That is, we described the intrinsic features of that picture but did not mention, for instance, that it is a picture of a building as seen from directly above, or a book seen from straight on, etc. We can call what a picture is a picture of the representational content of that image, and it will be important for what follows to notice that it is, pre-theoretically, at least possible for the pictorial content of an image and the intentional content of that same image to come apart. For instance, there are certain abstract paintings that certainly have pictorial content (a bunch of red, yellow, and blue paint splashes on a white canvas) but which do not have any representational content (these paintings are not paintings of anything). Conversely, in typical cases, words have no relevant pictorial content (they are not iconic representations; ‘dog’ does not look like a dog) but do have representational content (they are representations).5 Given the ambiguities surrounding the notion of representation (not to mention ‘idea,’ ‘perception,’ et al.) in Modern philosophy, it is worth spending some time to get as clear as possible on the distinction between these two kinds of content. Hume uses ‘circumstances’ and I will use ‘pictorial content’ to mean those intrinsic features of an image that make it the image that it is. So, the color and arrangement of the lines in the picture above are

Hume’s Theory of Mental Representation

25

part of the pictorial content of that image, but that the image is made of pixels (as it is as I type this) or of ink on paper (as it is in the hardcopy of this book) is not part of this pictorial content.6 A picture drawn on hemp paper can have the same pictorial content as a picture drawn on papyrus. Etc. The image consists of four black lines of such-and-such a length arranged in a square against a white background, and this is its pictorial content. Now contrast the pictorial content of an image with its representational content, which we will explicate further in a moment. This typographical mark ‘≈’ has as its pictorial content two zig-zagging lines each placed one on top of the other but does not thereby have any representational content. It does not mean anything in written English, it is not being used to picture anything, etc. Of course, it could be so used: we might introduce this mark as the new symbol for hydrogen; we might use it to picture waves approaching the shoreline, etc. Such uses would imbue that mark with representational content. Pictorial content is, at least prima facie, independent of representational content and refers only to the intrinsic features of an image, qua image, whether or not these features come to be used to form a representation or not. To turn now to representational content, or to what a representation is a representation of, we can begin by noting that since at least as early as Descartes—think “objective reality”—the notion of representational content has been a difficult one to explicate clearly. This is in part because in Descartes’s writing the notion of an idea—the primary vehicle of representational content—is an infelicitous conglomeration of phenomenological, epistemological, ontological, and representational theses. Many of these confusions remain in Hume’s work, and while Kant goes to some length to detangle some of them, he also has infelicities of his own. In what follows, I will do my best to extricate what is useful for current purposes and leave behind what is not. We can begin with an example. Suppose I form the following explicit judgment (say, by thinking it): ‘That rat is fat.’ Suppose further, though, that there is no rat in the immediate vicinity, and that it is my dimly lit cat that is the cause of this thought. Finally, suppose that my cat is not fat (he is quite svelte). One can, I think, discern in this example at least the following elements, each of which is related in its own way to the term ‘representation.’ (a) The judgment itself (be it a mental item, or act, etc.), as well as each of its elements, are representings.7 (b) The cat that I took to be a rat is what we might call the de facto referent of the phrase ‘that rat.’8 (c) What I thought I was seeing, a rat, is what I will say is represented by the phrase ‘that rat.’9 (d) Finally, something (the rat that is represented, the cat that is the de facto referent) is represented as being fat (even though nothing in the immediate vicinity is fat at all).

26

Hume’s Theory of Mental Representation

Both Hume and Kant note the intimate relation between (c) and (d), although each understands this relation differently. While the judgment here predicates ‘fat’ of the rat, the phrase that appears as the subject of that judgment, ‘that rat’ appears to have a predicative structure of its own. This can be brought out by noticing the close tie between that phrase and the judgment, ‘That is a rat.’ In fact, one of the foci of the next chapter will be Hume’s attempt to reduce representations with the form of (d)— judgments that attribute a property to a thing—to representations with the form of (c)—representations that represent a propertied-thing, as we might put it (a fat rat). Given the close relation between these two senses of representation, for the purposes of the current chapter, we can follow Hume in treating these as of a piece, although in later chapters we will have to follow Kant in distinguishing the two. So, I will use ‘representational content’ to mean what is represented by a representing—in contrast to its de facto referent—and what that which is represented is represented as.10 In the above example, the representational content of ‘that rat’ has a rat as its representational content, a rat is what that phrase is a representation of. The judgment ‘that rat is fat’ has as it representational content that that rat is fat. It is a judgment that represents the rat as being fat. Similarly, if I claim on Hume’s behalf that no idea has the representational content ‘necessary connection,’ what that claim will amount to is that no representing has as its represented a necessary connection, nor does it represent its represented as being a necessary connection (or as being necessarily connected to any other represented). Returning to Hume’s Copy Principle, it states that all ideas are copies of impressions, i.e., that all ideas are caused by, and have exactly the same pictorial content as, some corresponding impression. Important for us to note is that, as formulated here, the Copy Principle does not speak at all toward the representational content of impressions or ideas. It is merely a thesis concerning the causal relations between impressions and ideas and the relation of their pictorial content. It is the former relation that is of particular concern to those who have worried about the epistemological status of the Copy Principle in the Treatise, and we are now in a position to turn to Garrett’s defense of that status. Garrett’s defense is aimed first and foremost at critics of Hume such as Antony Flew, who claims that while the evidence that Hume offers in favor of the Copy Principle justifies its use as a defeasible empirical generalization, Hume’s actual use of it implies that it is a necessary truth. . . . such sentences as “all our ideas [. . .] are copies of our impressions [. . .]” [are] ambiguous: most of the time they are taken to express a contingent generalization; but at some moments of crisis [Hume] apparently construes them as embodying a necessary proposition. Such manoevres have the effect of making it look as if the immunity to falsification of

Hume’s Theory of Mental Representation

27

a necessary truth had been gloriously combined with the substantial assertiveness of a contingent generalization.11 The idea here is that if the Copy Principle is, as it seems to be in the opening pages of the Treatise, a mere empirical generalization, then Hume’s use of it to argue that we do not have ideas such as those of necessary connection, the external world, or the self cannot be justified. These should be seen as counterexamples to that principle rather than as violations of it. On the other hand, what would make the Copy Principle strong enough to play this role is if it were not an empirical generalization but a necessary proposition. It cannot, however, be necessary because the Copy Principle asserts that a certain causal relation holds between impressions and ideas, and all causal relations are contingent for Hume. Thus, Hume’s use of the Copy Principle in the Treatise is unwarranted. Garrett’s defense of Hume’s use of the Copy Principle centers on the claim that rather than mysteriously grant the Copy Principle the status of a necessary truth, and thus violate some of his deepest commitments to empiricism, Hume grants the Copy Principle the status of an empirical generalization with a good deal of evidence in its favor and uses it as one among many pieces of evidence weighing against the claim that we have certain controversial ideas: . . . there is no need to interpret Hume as maintaining that it is either a priori or necessary that every simple idea has a corresponding simple impression. He need only maintain that we have found this to be the case, thereby raising a reasonable expectation that the search for an original impression for a problematic idea will shed light (due to the greater clarity and vivacity of impressions) on whether the idea really exists and, if it does, on its nature.12 According to Garrett, Hume treats the Copy Principle as a well-grounded empirical generalization. It has a good deal of evidence in its favor, but it is neither necessary nor a priori. Impressions are more forceful and vivacious than ideas, and so if we have some evidence that every idea is a copy of some impression, it seems prudent to seek out the original impression for particularly obscure ideas in order to gain a better understanding of them. If we cannot find such an impression, given that the idea was questionable to begin with, we have some good evidence that we do not really have such an idea. Furthermore, Garrett points out that the Copy Principle is not the only piece of evidence in play in the debate over these controversial ideas. In each of these cases, admitting a counterexample to the Copy Principle would mean not merely violating the Resemblance Thesis but violating

28

Hume’s Theory of Mental Representation it in such a way as to allow nonimagistic ideas that could not, even in principle, resemble impressions. It would thus require the admission of an entirely distinct representational faculty, and hence a very serious modification in the cognitive psychology that Hume thinks he finds otherwise well supported by experience.13

The evidence that the Copy Principle provides against these controversial ideas is supplemented by the evidence that Hume has for his cognitive psychology as a whole. As Garrett mentions here, that cognitive psychology is one according to which all cognition is accounted for in terms of perceptions, which are themselves all images of various kinds (including sights, sounds, smells, tastes, etc.). Images, of course, all have pictorial content— this is what makes them images—and Garrett’s point here is that the ideas in question, were they to exist, would have to be such that they lacked pictorial content. The implausibility of such an idea, combined with the implausibility of a violation of the Copy Principle, and the inherently unclear quality of such ideas, amounts to enough evidence, for Hume, to reject the claim that such ideas exist. So Garrett’s position is as follows. The Copy Principle, which states that all simple ideas are caused by and have exactly the same pictorial content as their corresponding impression, is a well-supported empirical generalization. Hume never treats it as either necessary or a priori. The work the Copy Principle does in the Treatise is merely to support Hume’s claim that we can have no ideas of necessary connection, the external world, or the self. In further support of this claim is Hume’s similarly well-founded cognitive psychology, which purports to explain all human cognition in terms of images, mental items with pictorial content. I find this account of the use of the Copy Principle reasonable and compelling. I will now argue that it is still not enough to establish Hume’s conclusions. Here is the gist of my argument. What Garrett’s interpretation of Hume’s use of the Copy Principle earns for Hume are the theses that we have no ideas (a) with certain problematic pictorial content and (b) without any pictorial content at all. From these theses alone, however, it does not follow that we do not have ideas of necessary connection, the external world, or the self. What is needed to establish those claims is a thesis about the representational content of ideas. If we accept Garrett’s defense of the Copy Principle, what we earn for Hume is that we do not have ideas that are images consisting of necessary connections, the external world, or the self. This is because, as we have seen, the Copy Principle is a thesis about the causal connection between impressions and ideas and the relation of the pictorial content of these. It is not a principle about representational content.14 What Garrett gets from Hume is the claim that we do not have ideas with certain pictorial contents. That is an important step in Hume’s argument, but it cannot be the final step. What is still needed to make those arguments

Hume’s Theory of Mental Representation

29

good is a thesis that links the pictorial content of an idea to its representational content. It is via such a thesis that Hume earns for himself his conclusions that we cannot have ideas of necessary connection, the external world, the self, etc. The most plausible such thesis is what I have called the Representational Copy Principle, which states that the representational content of any perception is that of which it is a copy.15 To put it another way, the Representational Copy Principle states that a perception is of that of which it is a copy. It is this thesis that Hume uses to move from the claim that we do not have an idea with a certain pictorial content to the claim that we do not have an idea with certain, corresponding representational content. In particular, Hume uses it to move from, for instance, 1. ‘We have no idea that it is a copy of a necessary connection’ to 2. ‘We have no idea that is an idea of a necessary connection.’ (2) is what Hume is after in the Treatise. (1) is merely a necessary step along the way. (1) is what Garrett earns Hume. Hume earns (2) via the Representational Copy Principle. Now, to understand this point, we must understand what each of (1) and (2) are claiming, and this task seems particularly thorny when it comes to (1). What would it mean for an idea to be a copy of a necessary connection? At first blush, that might seem to be some sort of category mistake. We earlier cashed out ‘exact resemblance’ in terms of pictorial content—those intrinsic features of an image that make it the image that it is—but necessary connections (and the external world and the self), not themselves being images, do not themselves have pictorial content. They are not pictures, or images, after all. So, (1) seems trivially true. To see that it is not, we need to alter slightly our definition of exact resemblance. To do so, an example will help. Consider a standard office copier. What it produces are copies just in case these are caused by the original (they are not produced ex nihilo) and exactly resemble that original (they exactly replicate all the relevant intrinsic features of the original). Notice that even if the original is not something with pictorial content—we noted earlier that words, for instance, are not iconic representations and so do not have pictorial content—it can still be copied. This is achieved, roughly, just in case the copy reproduces certain relevant intrinsic features of the original. The copy has to replicate the shapes on the page, for instance, but can be made of a different kind of paper. So, our earlier definition of ‘exact resemblance’ was a simplified instance of this more general one. The precise details need not concern us here; a rough idea is all that Hume employs, and should suffice. So, (1) states that no idea is caused by and exactly resembles a necessary connection. We can grant that Hume presents enough evidence of various

30

Hume’s Theory of Mental Representation

kinds to establish that much. What (1) does not establish, however, is that we do not have an idea of a necessary connection. To earn that conclusion what Hume needs is a theory of what determines the representational content of ideas. This theory will have to be such that his claim that we have no idea that is a copy of a necessary connection is sufficient to establish his further claim that we have no idea of a necessary connection either. My suggestion, of course, is that the Representational Copy Principle is the most plausible candidate to play this role. To see what is at issue here, it might help to consider some alternative representational principles to the one that I have proposed Hume employs. So, consider instead a twentieth-century theory of representational content: Jerry Fodor’s. Being a twentieth-century philosopher rather than an eighteenth-century philosopher, Fodor abandons Hume’s commitment to imagism, the thesis that all mental items are images of some kind or other. Fodor thus constructs his theory of content as a theory of how certain items (e.g., brain-states) come to represent other items (e.g., cows). Cows cause “cow” tokens, and (let’s suppose) cats cause “cow” tokens. But “cow” means cow and not cat or cow or cat because there being cat-caused “cow” tokens depends on there being cow-caused “cow” tokens, but not the other way around. “Cow” means cow because [. . .] noncow-caused “cow” tokens are asymmetrically dependent upon cowcaused “cow” tokens.16 According to Fodor, it is the fact that any token of ‘cow’ that is caused by something other than a cow is asymmetrically dependent upon tokens of ‘cow’ that are caused by cows that ‘cow’ comes to represent cows. The particulars of this theory are not our concern here. The point here is that this is an example of a theory that, if Hume held it, would undermine the move from the application of the Copy Principle to the conclusion that we do not have the controversial ideas at issue in the Treatise. Let me explain. Suppose that we accept Garrett’s defense of Hume’s use of the Copy Principle as it stands and grant that Hume is justified in making something like the following argument. 1. We have no impression the pictorial content of which includes a single subject of experience persisting through time.17 2. All simple ideas are copies of some simple impression. (The Copy Principle) 3. So, we have no idea the pictorial content of which includes a single subject of experience persisting through time. (1, 2) If, as I claim he does, Hume holds the Representational Copy Principle, the argument would continue as follows.

Hume’s Theory of Mental Representation

31

4. So, we have no idea the representational content of which is a single subject of experience persisting through time. (3, the Representational Copy Principle)18 Now suppose that instead of the Representational Copy Principle, Hume held Fodor’s theory of representational content. In that case, 4 would not follow. Whether or not we have an idea with such-and-such representational content, on Fodor’s line, is a matter of what ideas asymmetrically depend on what worldly items. Pictorial content, and thus copying, is not a factor in determining representational content at all. So, from the fact that we do not have an idea that has the pictorial content “self,” it does not follow that we do not have an idea that has the representational content “self.” We would still have an idea of the self, on Fodor’s line, if that idea bore the appropriate asymmetrical dependency to a self, regardless of what the pictorial content of our ideas is. The point here is not that Hume could have held Fodor’s theory of representational content. Rather, it is that what theory of representational content Hume does hold matters. In fact, his conclusions that we do not have ideas of necessary connection, the external world, or the self depend on what theory of representational content Hume holds. The Copy Principle qua a thesis about the pictorial content of our ideas does not establish anything about the representational content of those ideas without supplementation by an appropriate theory of representational content. What I will argue is that just such a theory is readily available to, and employed by, Hume. It is the Representational Copy Principle that explicitly licenses exactly the move from conclusions about pictorial content established by the Copy Principle to conclusions about corresponding representational content, and there are numerous instances throughout the Treatise where we find Hume using the Representational Copy Principle in just this way. Before I do that, however, I want to make what is at issue here salient in one more way. So, consider again Garrett’s defense of Hume’s use of the Copy Principle. The Copy Principle states that every simple idea is a copy of some simple impression. Recently, we have been considering a worry that arises from Hume’s predecessors claiming that despite what the Copy Principle might purport to establish, they nonetheless find themselves with such controversial ideas. Armed with the Representational Copy Principle we can now see that this claim of theirs is ambiguous. On the one hand, they might be claiming to have an idea with a certain pictorial content, and on the other they might be claiming to have an idea with a certain representational content. The Copy Principle can most plausibly be used to refute only the former claim. In that case, the dialectic would unfold roughly as follows. Hume’s predecessors claim to be able to form mental images of things like necessary

32

Hume’s Theory of Mental Representation

connections, the external world, the self, etc. Hume leverages his wellsupported, but defeasible, Copy Principle to show that they can form no such mental pictures. While this dialectic does seem to make Hume’s argument strong enough, it also portrays him as arguing against a straw man. As portrayed here, Hume’s argument can only ever be successful against an opponent that already accepts that all thinking is imagistic. Certainly no rationalists would be willing to make this concession, and it is arguable that neither of Hume’s most important fellow empiricists, Locke and Berkeley, would either.19 All of these philosophers could agree with Hume that we can form no images of these controversial ideas but could consistently maintain that we could have mental representations with this controversial representational content nonetheless. If, on the other hand, we see Hume’s opponents as making a claim about their ability to form mental representations with certain controversial representational content, Hume’s use of the Representational Copy Principle engages his predecessors head on. Their claim then amounts to one that we can form mental representations of such things as necessary connections, the external world, the self, etc. Hume’s refutation of this thesis then has two parts: an analysis of what it is for an idea to have representational content and a well-supported empirical generalization to the effect that these conditions are not met in the controversial cases. On this reading, the status of the Copy Principle is much more comfortable. Even on Garrett’s reconstruction it is still awkward to use even a wellsupported empirical generalization to prove to people that they do not have ideas with a certain pictorial content that they claim to have. Here, the Copy Principle is relieved of that burden. Hume’s opponents are not disputing his claim that their ideas do not have a certain pictorial content. That can be a point of agreement, especially once Hume’s study establishing the Copy Principle is complete. What is at issue is something that we would expect to be much less transparent to an individual thinker: not the “phenomenology” of their ideas but their representational content. Once we separate these two notions of content out, it seems clear that what Hume is after is a conclusion about the latter kind, and that the Representational Copy Principle is exactly what he will use to reach it. Having now argued that Hume must employ a theory of representational content, the next item on my agenda will be to show Hume actually doing so. That is, in the next section, I will present selections from three of Hume’s most important arguments in the Treatise and show that each one of these more or less explicitly makes use of not just the Copy Principle but also the Representational Copy Principle. At that point, we will have established that Hume ought to employ the Representational Copy Principle and that he does employ it, but not how he can employ it. That is, we will not yet have addressed how his use of the Representational Copy Principle avoids falling prey to a straightforward iteration of the dilemma that the Copy Principle faced. Thus, in the third section, I will argue that

Hume’s Theory of Mental Representation

33

for Hume the Representational Copy Principle is a priori, necessary, and the result of an analysis of the idea ‘representational content,’ and that he can defend each of these positions. Before that, however, it is on to the arguments. HUME’S USE OF THE REPRESENTATIONAL COPY PRINCIPLE I will begin with a passage from Hume’s argument in 1.4.6 for the conclusion that we do not have the kind of idea of the self that his predecessors have supposed that we have. For from what impression can this idea be deriv’d? This question ‘tis impossible to answer without a manifest contradiction and absurdity; and yet ‘tis a question, which must necessarily be answer’d, if we would have the idea of the self pass for clear and intelligible. T 1.4.6.2; SBN 251 Notice that what Hume demands from such philosophers is that they produce the impression from which the idea is copied and that what is at stake in meeting this demand is the very intelligibility of that idea. That is, we cannot so much as make sense of an idea of the self, if there is no impression of which it is a copy. This is a much stronger claim than merely that we cannot have such an idea. The notion of an idea with that representational content is unintelligible unless we find the impression from which such an idea is copied. This stronger claim would be true only if there is something about the very notion of an idea’s representational content that implied that it is a copy of some impression. This is exactly what the Representational Copy Principle states. Hume’s argument continues, It must be some one impression, that gives rise to every real idea. But self or person is not any one impression, but that to which our several impressions and ideas are suppos’d to have a reference. If any impression gives rise to the idea of the self, that impression must continue invariably the same, thro’ the whole course of our lives; since self is suppos’d to exist after that manner. Pain and pleasure, grief and joy, passions and sensations succeed each other, and never all exist at the same time. It cannot, therefore, be from any of these impressions, or from any other, that the idea of self is deriv’d; and consequently there is no such idea. T 1.4.6.2; SBN 251–2, emphasis added Here we see the structure that we outlined earlier with both the Copy Principle and the Representational Copy Principle being employed.

34

Hume’s Theory of Mental Representation 1. Every idea is a copy of some impression. (the Copy Principle) 2. Since our impressions are varied, there is no enduring, simple impression that has an unvaried pictorial content. 3. Thus, there is no enduring, simple idea that has an unvaried pictorial content. (1, 2) 4. Thus, there is no idea of the self. (3, definition of ‘self’ given by predecessors, the Representational Copy Principle)

Hume’s argument is intended to show that we do not have an idea the representational content of which is a single subject of experience persisting through time. His methodology is to show, first, that we cannot have an impression with the appropriate pictorial content; next, that we cannot therefore have an idea with the appropriate pictorial content; and finally, that we cannot therefore have an idea with that representational content. We will see this same methodology employed in at least two other cases. Hume’s argument about the idea of the external world is less straightforward than his argument about the idea of the self. This argument proceeds via process of elimination. Hume first argues that this idea can only be a product of either the senses, reason, or the imagination. He then gives a two-part argument that this idea cannot originate with the senses, followed by an argument that it cannot originate with reason, and finally an explanation of how the relevant idea that we do have comes from imagination, and how it is an idea with a very different representational content from that which his predecessors proposed. What is of interest to us here is the first of these stages. Here is Hume’s argument for the conclusion that our idea of a being that continues to exist when it is no longer perceived cannot originate with the senses. To begin with the senses, ‘tis evident these faculties are incapable of giving rise to the notion of the continu’d existence of their objects, after they no longer appear to the senses. For that is a contradiction in terms, and supposes that the senses continue to operate, even after they have ceas’d all manner of operation. T1.4.2.3; SBN 188 Notice that once again Hume is explicitly concerned with the notion of continued existence, an idea with continued existence as its representational content. The key to interpreting this quick argument for that conclusion is to decipher what the contradiction to which Hume appeals here is. The obvious candidate is something like, “The senses sense what they do not sense.” This contradiction alone, however, is not enough to license the conclusion that the senses do not produce an idea of an object that continues to exist when it is not sensed. It is possible, that is, that although the senses do not sense what is unsensed, they still cause an idea to come into existence that itself is an idea of something unsensed. The conclusion that this is not

Hume’s Theory of Mental Representation

35

possible does follow, though, if Hume is employing the Representational Copy Principle. This is because if the senses do not sense what is not sensed, then whatever ideas are copied from the data of the senses cannot have as their representational content anything that is unsensed. The only ideas that can be copied from the data of the senses will necessarily be ideas whose content is sensed precisely because the representational content of an idea is that of which it is a copy. Thus, the senses cannot produce an idea whose content is an object that continues to exist when it is no longer perceived. The second part of Hume’s argument regarding the senses also relies on the Representational Copy Principle. That argument is for the conclusion that an idea of a being that exists distinctly from oneself also cannot originate with the senses. Here is that argument. Notice that it again explicitly begins with a concern with the representational content of perceptions, here with what our impressions are impressions of. That our senses offer not their impressions as the images of something distinct, or independent, and external, is evident; because they convey to us nothing but a single perception, and never give us the least intimation of any thing beyond. A single perception can never produce the idea of a double existence, but by some further inference either of the reason or imagination. T 1.4.2.4; SBN 189 The argument here is fairly straightforward. Our senses produce single simple impressions. Any ideas that trace their roots to the senses, therefore, have as their representational content only such simple impressions, not “any thing beyond.” Again, this conclusion only follows if we suppose that an idea is of that of which it is a copy. Otherwise, the fact that an idea is a copy of a single impression does not at all imply that it cannot be of anything other than this impression. Hume is employing the Representational Copy Principle here as well. Finally, there is Hume’s argument concerning our idea of necessary connection. Again, rather than go through the entire argument at length, it will suffice to focus on a key passage in which Hume is explicitly engaged in the negative portion of his argument, where he argues that we cannot have an idea of necessary connection. Here is Hume’s concise refutation of the suggestion that our idea of necessary connection is the idea of a power or efficacy. All ideas are, deriv’d from, and represent impressions. We never have any impression, that contains any power or efficacy. We never therefore have any idea of power. T 1.3.14.11; SBN 161 Here Hume is at his most explicit about the use of the Representational Copy Principle. Notice that in the first sentence here Hume distinguishes

36

Hume’s Theory of Mental Representation

two claims: that all idea are derived from impressions (the Copy Principle) and that all ideas represent impressions (the Representational Copy Principle). In the next sentence, he is careful to restrict his claim to the pictorial content of our impressions. Rather than assert that we never have any impression of a power or efficacy, he instead writes that we never have any impression that contains any power or efficacy. That is, the claim is that there is no impression that has as part of it a power or efficacy. Finally, Hume concludes from these three premises—the Copy Principle, the Representational Copy Principle, and an observation about the pictorial content of our impressions—that our ideas cannot have as their representational content ‘power.’ That is, he concludes that we never have any idea of power. Again, this argument proceeds using not only the Copy Principle but also the Representational Copy Principle. My argument up to this point has been as follows. The Copy Principle is a thesis that concerns the pictorial content of ideas and impressions but not their representational content. In the various arguments throughout the Treatise concerning certain controversial ideas such as those of necessary connection, the external world, and the self, Hume needs a premise that addresses not just the pictorial content of these ideas but also their representational content. We have seen that Hume needs some theory of representational content to make his arguments valid, and the Representational Copy Principle would validate exactly the move that we have seen that Hume needs: from a conclusion that we cannot have ideas with certain pictorial content to a conclusion that we cannot have ideas with a certain corresponding representational content. We have also now seen that Hume seems throughout these arguments to employ exactly this principle. That said, there is still an important item on our agenda. We began this chapter with the following dilemma. Either Hume treats the Copy Principle as an empirical generalization, in which case he cannot use it refute his predecessors’ claims that we have certain ideas, or he treats it as necessary and a priori, in which case he must hold that some causal connections are necessary and a priori (which he explicitly denies) and thus violates his commitment to empiricism. We have since seen that Garrett provides Hume with a plausible way through the horns of this dilemma but that his solution cannot be all that there is to story, given that we also need to account for Hume’s use of the Representational Copy Principle. That is, the Copy Principle alone does not suffice for rejecting these controversial ideas. We also need the Representational Copy Principle, and so Hume would seem to face another iteration of the same dilemma. Either the Representational Copy Principle is necessary and a priori or empirical, etc. Here Garrett’s solution will not work, both because Hume does not actually marshal evidence for the Representational Copy Principle as he does for the Copy Principle and because it is unclear what such evidence could be. What I will propose is that the Representational Copy Principle (if true) is, unlike the Copy Principle, necessary and a priori. I will also argue that,

Hume’s Theory of Mental Representation

37

unlike the Copy Principle, this is not bad. The objections to construing the Copy Principle as necessary and a priori were that doing so would require holding that some causal connections were necessary and a priori, and that that would violate Hume’s commitment to empiricism. I will show that holding that the Representational Copy Principle is necessary and a priori does neither of these. A HUMEAN ANALYSIS OF ‘REPRESENTATION’ Consider, then, the first possible objection: that Hume is committed to the claim that no causal connections are necessary or a priori. This poses a problem for holding that the Copy Principle is necessary and a priori because the Copy Principle states, in part, that all ideas are caused by some impression. If that were necessary and a priori, it would clearly violate Hume’s commitment. Notice, however, that the Representational Copy Principle allows for no such parallel objection. What the Representational Copy Principle states is that a perception is of that of which it is a copy. This does imply that if a perception is of something, then it was caused by that something, but it does not assert that any such causal connections actually obtain. Whether such causal connections obtain, and so whether any of our perceptions have representational content, is entirely contingent and a posteriori. What is necessary and a priori here is simply what determines the representational content of a perception. Hume has no in-principle grounds for rejecting that claim. In fact, I have argued, he accepts it. Next, consider the second possible objection here: that holding that the Representational Copy Principle is necessary and a priori violates Hume’s commitment to empiricism. Again, this poses a problem for the Copy Principle because the Copy Principle asserts that certain matter of factual relations hold between impressions and ideas: that the latter are caused by and exactly resemble the former. Clearly, a commitment to empiricism precludes accepting that either of these is necessary or a priori. Again, though, the Representational Copy Principle does not imply that either of these relations actually holds. What it states is that if and only if these relations do hold do our ideas have representational content. In fact, not only does holding the Representational Copy Principle not violate the spirit of empiricism, it is actually an expression of exactly that commitment. ‘Empiricism’ is a term that is applied broadly and as such can be construed in a number of different ways. One of those ways, though, is plausibly via a commitment to concept empiricism, to the thesis that the representational content of our perceptions is determined entirely by its relation to experience.20 The Representational Copy Principle is one very straightforward way of cashing out this commitment. The representational content of all of our perceptions is determined entirely by that perceptions being a copy (of some experience).

38

Hume’s Theory of Mental Representation

Of course, all of this leaves open the question of just how it is that the Representational Copy Principle can be necessary and a priori. To put it another way, Hume’s Fork leaves open two possibilities vis-à-vis any proposition: it is either a matter of fact or a relation of ideas. Since the Representational Copy Principle is clearly not a matter of fact, it must be a relation of ideas. It would seem to be, therefore, legitimate to ask, ‘What ideas, and what relation?’ Luckily, Hume’s own theory gives very specific instructions for how to answer such questions: via his account in 1.1.7 of general ideas. Hume’s account of general ideas is well-worn territory in Hume scholarship, so I do not want to go into much exegetical detail about it just now (although a more detailed investigation of it will be called for in the next chapter). Rather, I will simply present the theory in Hume’s own words and try to extract as noncontroversial an interpretation of the relevant passage as possible. Here is that passage. When we have found a resemblance among several objects, that often occur to us, we apply the same name to all of them, whatever differences we may observe in the degrees of their quantity and quality, and whatever other differences may appear among them. After we have acquir’d a custom of this kind, the hearing of that name revives the idea of one of these objects, and makes the imagination conceive it with all its particular circumstances and proportions. But as the same word is suppos’d to have been frequently apply’d to other individuals, that are different in many respects from that idea, which is immediately present to the mind; the word not being able to revive the idea of all these individuals, only touches the soul, if I may be allow’d so to speak, and revives that custom, which we have acquir’d by surveying them. They are not really and in fact present to the mind, but only in power, nor do we draw them all out distinctly in the imagination, but keep ourselves in a readiness to survey any of them, as we may be prompted by a present design or necessity. T 1.1.7.7; SB 20–1 When we hear a certain word repeatedly in the presence of various ideas that all resemble one another, we come to associate that word and those ideas. Subsequently, upon hearing that word, we call to mind some, but not all, of these resembling ideas, and stand ready, so to speak, to recall the rest. This is how the general term gains its meaning. Hume relies on this theory at numerous points throughout the Treatise to provide a methodology for giving the proper analysis of the controversial ideas that his predecessors have claimed to have. ‘Causation’ gets analyzed in terms of ideas constantly conjoined. ‘External world’ gets analyzed in terms of the constancy of resembling ideas. ‘Self’ gets analyzed in terms of a bundle of perceptions. Thus, our first questions regarding the Representational Copy Principle must be what term is to be analyzed here and what ideas will be the result

Hume’s Theory of Mental Representation

39

of this analysis. Of course, Hume does not explicitly address this question. My proposal is that he does so implicitly. So, our answers to these questions will not be, strictly speaking, Hume’s answers, but rather the answers that Hume could and would give, were he to take the question up. That said, since we have already demonstrated in the first section that Hume needs to have some theory of representational content in order to make his arguments good, it will suffice here to show that the Representational Copy Principle is a possible and plausible candidate. That is, what I want to show here is merely that a Humean analysis of representational content could consistently yield the Representational Copy Principle as its result. What we are looking for, then, is an idea, or set of ideas, that could plausibly be said to be that which we associate with the term ‘representational content’ (or its synonyms). My suggestion is that the set of ideas constituted by those ideas that are copies of something—that are caused by and exactly resemble some single thing—fits this mold perfectly for Hume. The first thing we should note is that this set can only be comprised by perceptions. That is, whatever the general idea of representational content will be, it can consist in only perceptions simply because, according to Hume’s theory of general ideas, all such ideas are collections of associated perceptions. The second thing we should note is that it is, at least on a first pass, plausible to suppose that this set will be comprised of pairs of perceptions: the perception that has some representational content and the perception that is the proper object of that content.21 With these two conditions in place, it is also worth noticing that if these pairs of perceptions are to come to play this role, they will need to be previously associated with one another. That is, only a pair of associated ideas could come to be linked together closely enough to form the further association between this pair and the term ‘representational content’ being analyzed. Finally, we should note that the ways that this pair of ideas can come to be associated are fairly limited. Hume holds that there are only three such kinds of association: resemblance, contiguity, and cause and effect.22 As we have already seen, Hume’s analysis of ‘copy’ relies on exactly two of these kinds of associations: x is a copy of y just in case x exactly resembles and is caused by y. So too, then, will our Humean analysis of ‘representational content’: x has y as its representational content just in case x exactly resembles and is caused by y. That is, according to this analysis, our general idea of representational content consists of the set of perceptions that are copies of one another. If all of this is right, then the Representational Copy Principle can be cast as a necessary a priori principle that is discovered via the analysis of our idea of representational content in accordance with Hume’s theory of general ideas, and the final hurdle to interpreting the Representational Copy Principle as a necessary a priori principle will thereby have been cleared. The idea here is this. While the Copy Principle may well be merely a wellsupported empirical generalization, the Representational Copy Principle

40

Hume’s Theory of Mental Representation

has a different status; it is not an empirical proposition at all. One does not determine that an idea represents that of which it is a copy simply by examining some collection of ideas. Nothing about these ideas, considered by themselves, would indicate that this is what determines their representational content. What is required here is not an observation of ideas but a theoretical account of them. That theory can be supported by observation, but it does not itself consist in mere observation. Nor does it consist of a generalization of such observations. Hume’s claim is neither that this or that idea has been observed to be of that of which it is a copy, nor that all ideas can be so observed. Rather it is a proposal for how to understand the very notion of representation. In fact, such analyses are intimately familiar to anyone who has studied the Treatise, although not much attention has been given to determining precisely what their status in that work is.23 Consider, for example, Hume’s presentation of his two definitions of ‘causation.’ ‘Tis now time to collect all the different parts of this reasoning, and by joining them together form an exact definition of the relation of cause and effect, which makes the subject of the present enquiry. This order wou’d not have been excusable, of first examining our inference from the relation before we had explain’d the relation itself, had it been possible to proceed in a different method. But as the nature of the relation depends so much on that of the inference, we have been oblig’d to advance in this seemingly preposterous manner, and make use of terms before we were able exactly to define them, or fix their meaning. We shall now correct this fault by giving a precise definition of cause and effect. T 1.3.14.30; SBN 169 What Hume is about to do after this paragraph is define ‘the relation of cause and effect.’ What he has done before it is spend a great deal of time detailing the inference, or association, that we tend to make between the idea of a cause and the idea of its effect. The reason for his doing the latter is that his “definitions” will make explicit reference to this inference. That is, Hume must present a good deal of the theoretical apparatus that he will employ in his definition of ‘causation’ before he can actually present that definition. The definition only makes sense and becomes at all plausible within the context of the theoretical account of human nature of which it is a part. Here, then, is (one of) Hume’s definitions of ‘cause and effect.’ A cause is an object precedent and contiguous to another, and so united with it, that the idea of the one determines the mind to form the idea of the other, and the impression of the one to form a more lively idea of the other. T 1.3.14.31; SBN 169

Hume’s Theory of Mental Representation

41

This “definition” is an odd one indeed. It is certainly not what one would find if one looked up ‘causation’ in the dictionary, nor does Hume think that it is. He clearly takes himself to be offering up a new account of ‘causation.’ In what sense, then, is it a definition of that term? Correspondingly, in what sense is the analysis above a definition of ‘representation’? The answer seems to be that these are both analyses of the ideas at hand from within the theoretical apparatus with which Hume is operating. Hume is speaking here in the voice of the scientist of man. Hume is speaking neither with the vulgar nor with his predecessors in such cases. Instead, he is demonstrating the ability of the theoretical apparatus that he has proposed—his account of human nature as consisting of impressions, ideas, association, etc.—properly to account for such ideas. This proper account will, of necessity, reconstrue some of these terms as having meanings different from the ones that either the vulgar or the philosopher takes them to have. Thus Hume’s “definitions” are definitions in the sense of being accounts of what the terms at hand really mean. Thus, if true, such definitions will be necessary and a priori in just the way that Hume’s theory allows: they will state the relations of ideas with which such terms are associated. ‘Causation’ is associated with ideas “precedent and contiguous to another, and so united with it, that the idea of the one determines the mind to form the idea of the other, and the impression of the one to form a more lively idea of the other.” ‘Representation’ is associated with ideas, one of which is a copy of another. These are analyses of these terms from within Hume’s theoretical apparatus, which if accurate, will turn out to be mere tautologies. If Hume’s theory is correct, then such analyses will turn out to be what we really meant by such terms all along, even though neither the common man nor the philosopher was in a position to recognize this (because neither yet had the correct account of human nature in hand). Thus, the dialectical position that Hume’s Representational Copy Principle stands in with respect to his predecessors is very different from that in which Garrett casts the Copy Principle as standing. According to Garrett, the Copy Principle is a well-confirmed empirical generalization. It stands fairly straightforwardly open to counterexample, so long as that counterexample is itself sufficiently well supported. The Representational Copy Principle, however, qua necessary and a priori principle, enjoys a more secure dialectical position. In order to refute it, it would not be sufficient to produce a counterexample taken from either the common understanding of ‘representation’ or from an alternative philosophical understanding of that term. Hume’s defense of the Representational Copy Principle purports to replace such understandings with a more comprehensively explanatory account of ‘representation.’ So, what undermining the Representational Copy Principle would require is either a counterexample that is articulable from within Hume’s own theoretical apparatus or a refutation of that theory itself. As we will see in the next chapter, Kant undertakes to present objections with

42

Hume’s Theory of Mental Representation

precisely these forms. Before we can get to that, however, we must make one final observation and consequent emendation of the Representational Copy Principle. COMPLEX IDEAS AND MISREPRESENTATION At this point, we must consider a final important emendation that must be made to the Representational Copy Principle before moving forward. As we noted earlier, Hume restricts the Copy Principle to simple impressions and ideas. The content of this claim from early on in the Treatise is only that every simple idea is copied from some simple impression. Correspondingly, then, if it is only simple ideas that are copied from simple impressions, the Representational Copy Principle must, it seems, similarly be applicable only to simples. That is, since the Representational Copy Principle states that a perception is of whatever it is a copy of, and only simple ideas are copies of anything, then it would seem to follow that only simple ideas are of anything. This, however, cannot be right. First of all, any plausible theory of mental representation must to be able to account for the representational content of thoughts such as that of a dog. Hume clearly does so by casting them as complex ideas, and so he must have some account of how such ideas represent what they do. He certainly writes as if he does. I can imagine myself such a city as the New Jerusalem, whose pavement is gold and walls are rubies, tho’ I never saw any such. I have seen Paris; but shall I affirm I can form any such an idea of that city, as will perfectly represent all its streets and houses in their real and just proportions? T 1.1.1.5; SBN 3 This is only one of a vast number of instances in which Hume indicates that he takes complex ideas to have representational content. What we need is an account of how he explains such cases that is consistent with his use of the Representational Copy Principle as that which determines such content. This quotation also brings out a second objection to the current reading of Hume. Any plausible account of mental representation must be able to accommodate the fact that we sometimes misrepresent. At first glance, the Representational Copy Principle makes this impossible. If a perception is of that of which it is a copy, then the object of any perception with representational content is always guaranteed to exist (since it had to be the cause of the idea of it) and likewise must be exactly as it is represented to be (since the idea of it must exactly resemble it). So, such representations can never go wrong. As we saw in the above passage, though, Hume is clearly aware

Hume’s Theory of Mental Representation

43

of the phenomenon of misrepresentation: we can represent New Jerusalem even though it has never existed, and our representation of Paris does not correspond exactly to the details of the actual city. What is also indicated by the above quotation is that Hume takes these two problems to be connected: it is our ability to form complex ideas with representational content that leads to our making such errors. Recall the context in which Hume writes the above. He has just proposed for the first time that all of his ideas exactly resemble some impression. He then notices these two kinds of examples and uses the distinction he has previously drawn between simple and complex ideas to restrict the scope of his claim to just simple ideas. The implication certainly seems to be that it is the fact that such ideas are complex that keeps them from exactly resembling some impression and that this is not the case with simple ideas. Simple ideas cannot misrepresent; only complex ideas can. Our focus, then, should clearly be on the nature of complex ideas, so here is Hume’s first enumeration of the distinction between simple and complex perceptions. Simple perceptions or impressions and ideas are such as admit no distinction or separation. The complex are contrary to these, and may be distinguished into parts. Tho’ a particular colour, taste, and smell are qualities all united together in this apple, ‘tis easy to perceive they are not the same, but are at least distinguishable from each other. T 1.1.1.2; SBN 2 Complex perceptions are those that have simple perceptions as their parts. They are aggregates of simple ideas.24 These are claims about the composition of complex ideas and can straightforwardly be extended to their pictorial content. That is, if complex ideas are literally composed of simple ideas, then the pictorial content of a complex idea—which we can remember is itself constituted by the intrinsic features of a perception—must also be, in some way, determined by the pictorial content of the simple ideas from which it is composed. Consider our earlier example. The pictorial content of this picture consists of four black lines of equal length arranged at ninety-degree angles to one another against a white background. The picture is composed of four black lines and a white background. It is a complex image with those as its parts. Of course, it is only the picture that it is because those four black lines are arranged a certain way against the white background. That is, complex pictorial content, while constituted by a straightforward aggregation of simple(r) parts, is also determined by the specific arrangement of those parts. This is a square-shaped figure, after all, not a trapezoidal picture, or a nonpolygonal figure, etc. So, complex ideas have simple ideas as their parts and have certain arrangements of these simple ideas as their pictorial content.

44

Hume’s Theory of Mental Representation

Since what we are looking for on Hume’s behalf is a way of moving from facts about an idea’s pictorial content to conclusions about its representational content, it is natural to assume here that just as the pictorial content of a complex idea is more than a mere sum of the pictorial content of its parts, so will be the representational content of such an idea. That is, complex ideas do not have as their representational content merely the simple impressions from which they are composed, but they also represent these impressions as being arranged a certain way. This is captured in the Humean slogan that a representation of a complex is just a complex of representations. We represent complexes of impressions by forming complexes of representations of these impressions. Consider, for instance, Hume’s account of the origin of our representations of spatial complexes. The table in front of me is alone sufficient by its view to give me the idea of extension. This idea, then, is borrow’d from, and represents some impression, which this moment appears to the senses. But my senses convey to me only the impressions of colour’d points, dispos’d in a certain manner. If the eye is sensible of any thing farther, I desire it may be pointed out to me. But if it be impossible to show any thing farther, we may conclude with certainty, that the idea of extension is nothing but a copy of these colour’d points, and of the manner of their appearance. T 1.2.3.4; SBN 34, emphasis added

Hume’s Theory of Mental Representation

45

Our complex idea of a spatial complex comes to have that representational content by being a collection of simple ideas of colored points arranged in a way that exactly resembles the arrangements of the spatial complex being represented. We represent the relation that some simple impressions stand in to one another by arranging simple representations of each of these impressions into the same relation. We represent a as being next to b by placing an idea of a next to an idea of b. The idea of a spatial complex is nothing more than a spatial complex of ideas.25 Hume is clear that our representation of temporal complexes works in exactly the same manner. The idea of time being deriv’d from the succession of our perceptions of every kind, ideas as well as impressions, and impressions of reflection as well as of sensation, will afford us an instance of an abstract idea, which comprehends a still greater variety than that of space, and yet is represented in the fancy by some particular individual idea of a determinate quantity and quality. T 1.2.3.6; SBN 34 Our idea of time is “deriv’d from the succession of our perceptions.” Hume’s thought is that we represent two items as being related, now temporally, by placing them in a temporal relation to one another. That is, for example, we represent one thing as happening before another by having a representation of the former followed by a representation of the latter. So, whereas we represent a spatially complex state of affairs by forming a kind of picture in our mind’s eye, we represent a temporally complex state of affairs by forming a kind of movie there. More generally, then, we can see that for Hume a complex representation represents a complex state of affairs by representing the elements of that state of affairs and arranging these representations in the same relations with one another as those in which the elements of the state of affairs represented stand in to each other. To represent two items as being next to one another, we place ideas of these items next to one another; to represent two items as succeeding one another in time, we have an idea of one of these items succeeding an idea of the other, etc. This is essentially the resemblance aspect of the Representational Copy Principle as it applies to complexes. It would seem that for a complex idea to represent, it must exactly resemble its object. We know this can’t be right, so let’s try beating around a neighboring bush for a while. Exact resemblance is a symmetric property. Not only do complex ideas exactly resemble the complex states of affairs that they represent, but those states of affairs also exactly resemble those ideas. The states of affairs, though, do not represent the ideas. A picture of a cat next to a picture of a dog might represent a cat as being next to a dog, for Hume, but an actual cat’s sitting next to an actual dog does not represent a picture of a cat as

46

Hume’s Theory of Mental Representation

being next to a picture of a dog. So, as in the case of simple ideas, there must be something more to representation than just exact resemblance. Once again, causation seems to fit this role perfectly. In a paradigm case of correct representation, for Hume, a picture is created of the object represented, and this picture is created as a result of an encounter with this object. ‘Tis evident, that the memory preserves the original form, in which its objects were presented, and that wherever we depart from it in recollecting any thing, it proceeds from some defect or imperfection in that faculty. (T 1.1.4.3; SBN 9) In the creation of a memory, a paradigm case of a representation, the original form is preserved. That is, the form of the memory is meant to correspond to the form of the object of that memory, not coincidentally, but as a result of the creation of that memory having gone correctly. Memories are meant to exactly resemble their objects as a result of being caused by those objects. It is only when this causal process has gone wrong that a memory fails to exactly resemble its object, and an idea that just so happened to resemble some object would not thereby be a representation of that object. There is a crucial causal process that links a memory to its object. Unfortunately, our recent investigations have now seemed to have led us back to a version of the Representational Copy Principle: complex representations represent that which they exactly resemble and which is their cause. Of course, as we have already seen, complex ideas are not always copies of that which they represent. Hume’s examples of Paris—caused by, but not exactly resembling Paris—and New Jerusalem—exactly resembling, but not caused by New Jerusalem—demonstrate precisely these points. So, neither exact resemblance nor causation can be necessary conditions for complex representation. Furthermore, neither alone can be a sufficient condition since, in addition to the case that we have already seen in which x exactly resembles y, but does x not represent y—a cat does not represent a picture of a cat—there are also obviously cases in which something does not represent its cause. (A window’s breaking does not represent the baseball that causes it.) What we must determine, therefore, is what the roles of exact resemblance and causation are here, since they both do seem to play some role in Hume’s account. To do this we can return to our paradigm case of correct representation. As we recently observed, when an accurate memory is formed, the representation both exactly resembles and is caused by its object. A correct representation is a copy of that which it represents.26 An incorrect representation, then, will be one that, in some sense, fails to be a copy. Of course, it cannot be just anything that happens not to be a copy, but rather it must be something that, again in some sense, ought to be a copy but is not. Exact resemblance and causation enter into Hume’s picture as the two ways that

Hume’s Theory of Mental Representation

47

such a failure might occur. A representation ought to exactly resemble and be caused by its object. A misrepresentation misrepresents in virtue of failing in one of these two ways. One can have an idea that is caused by Paris but that does not exactly resemble Paris. One can have an idea that exactly resembles New Jerusalem but that is not caused by New Jerusalem. So, if the representational content of a complex idea is that of which it ought to be a copy, what we need now is an account of how it is that something comes to have as its proper function to be a copy. What we need is an account not of how a complex idea comes to have the representational content that it has but rather of how it is that it has representational content at all. The key here is to notice again that complex ideas are composed of simple ideas and that simple ideas are straightforwardly copies of that which they represent. More specifically, as we have seen, complex ideas are arrangements of simple ideas. Complex ideas are arrangements of simple representations. That, I suggest Hume holds, is sufficient for making such a complex idea itself a representation. Again, we must take seriously the Humean slogan that a complex of representations is a representation of a complex. A spatial complex of representations is, for Hume, a representation of a spatial complex. A temporal complex of representation is a representation of a temporal complex. Etc. It is in virtue of being a complex of representations that a complex representation has representational content. A complex representation is formed, for Hume, precisely by arranging simple representations into a complex. To summarize, then, simple perceptions have representational content in virtue of their being copies. Their representational content is that of which they are copies. A complex perception has its representational content in virtue of its being an arrangement of simple perceptions. As such, it represents the simple objects that are the representational content of its simple parts as being arranged in the way that its own parts are arranged. Complex ideas purport to be copies of their corresponding complexes of impressions. They are accurate insofar as they are such copies but can misrepresent in either of the two ways that they can fail to be such copies: either by not exactly resembling their objects (Paris) or by not being properly causally connected to them (New Jerusalem). What this means for Hume’s arguments is that his tasks are slightly more difficult than I had earlier shown. To show that we do not have the controversial ideas at issue, it is not enough to show that we do not have any ideas that are not copies of such-and-such. That only suffices in the cases in which the idea is supposed to be simple. What Hume also needs to show, given our recent conclusion, is that we do not have any ideas that have as their proper function to be copies of such-and-such. That is, he needs to show that we do not have any ideas that could be copies of such-and-such, if only they hadn’t failed to exactly resemble or be caused by their representational content. In the case of necessary connection, for example, what Hume has to show is that we do not have any idea that could be a copy of a necessary

48

Hume’s Theory of Mental Representation

connection. He does so by pointing out that no matter how distinct ideas are arranged, they can always be separated and so could never be arranged in a way that exactly resembles a necessary connection. Rather than being of necessary connections, such ideas are of constant conjunctions. In the case of the external world, he would need to show that we do not have an idea that could be a copy of the external world. He does so by pointing out that since the simple components of complex ideas are all of impressions, they could not be copies of anything distinct from such impressions, or of anything that continues to exist independently of such impressions. Rather than being of an external world, such ideas are of arrangements of impressions. In the case of the self, Hume has to show that we do not have any idea that could be a copy of a single subject of experience persisting through time. He does so by pointing out that what would be required to form an idea of the self would be a perception with a pictorial content that cannot be had. Rather than being of the self, such ideas are of a bundle of perceptions. These, then, are the conclusions with which Hume leaves Kant. (NC) We have no idea that is an idea of a necessary connection. (EW) We have no idea that is an idea of the external world. (SSE) We have no idea that is an idea of a single subject of experience persisting through time. Hume establishes each thesis via an argument that crucially employs this more complex version of the Representational Copy Principle: complex ideas represent that which their simple parts represent as being arranged in the way that those simple parts are arranged.27 Each argument then proceeds to show that because of the nature of the kinds of pictorial content that our ideas can have, the controversial representational content in question could not possibly be represented in this way. Given the kinds of pictorial content that our ideas can have, there are only certain kinds of representational content that they can have. This is the kind of inference that the Representational Copy Principle licenses, and it is that principle that leads Hume to accept each of these radical conclusions. If Kant is going to reject these conclusions, then, his first order of business must be to refute Hume’s Representational Copy Principle. As we will see in the next chapter, that is exactly what we find Kant doing. NOTES 1. Cf. T 1.4.3.9; SBN 222–3. 2. Garrett, Cognition and Commitment argues that the Copy Principle is not a mere empirical generalization but a very well-supported such generalization. Landy, “Hume’s Impression-Idea Distinction” argues that the Copy Principle is not merely a well-supported empirical generalization but also plays a crucial explanatory role vis-à-vis the distinction between impressions and ideas that gives it not just empirical but also a more general scientific-theoretical support as well.

Hume’s Theory of Mental Representation

49

3. In Landy, “Hume’s Theory of Mental Representation,” I dubbed this principle the Semantic Copy Principle. The switch here to the idiom of ‘representation’ is meant to clarify that while it will be important to connect the truth conditions of a given representation to its representational content, it is certainly possible for this content to be determined (in part or entirely) in a way that is distinct from these conditions (as it is for both Hume and Kant). 4. Landy, “Hume’s Impression-Idea Distinction.” 5. Of course the written word ‘dog’ does have some pictorial content (there is something that it is to look like that written word), and so it follows that an idea of this word is a representation of the word. This must be carefully distinguished, however, from what the word represents qua a word. In that respect, too, its pictorial content comes into play. Hume’s theory of general representation makes it clear that the pictorial content of words (spoken, written, etc.) are essential to their meaning, although in a way that differs from the one we have been examining. It is the pictorial content of a word that allows different instances of it to be recognized as the same, and it is this perceived repetition that subsequently causes the associations to form that make possible the use of that term as a general representation. 6. It might be the case that in different contexts, the standards for an image’s having the same pictorial content might change. E.g., in class, I might draw a square on a whiteboard to make the point I’m making here and draw the same square on piece of paper later in my office, while if the course I was teaching was an art class, the difference in medium would be enough to require calling the two squares different images. 7. For Hume, mental representings are perceptions, mental images, and as I will argue farther along, they will, in fact, all be ideas. For Kant there are a variety of kinds of mental representings including judgments, sensations, intuitions, concepts, and ideas, all of which represent in distinct ways. 8. The notion of a ‘referent’ in this sense is a complicated and controversial one. It might be helpful here to think of Donnellan’s famous example of ‘the man in the corner drinking a martini.’ That phrase can be used to refer to a particular man in the corner, even if the liquid in his glass is, unbeknownst to the speaker, water. Donnellan, “Reference and Definite Descriptions.” 9. We can put aside, for the moment, the contribution that ‘that’ makes here, although this will be of some significance in later chapters. 10. It is worth noting here that for all that I have just written about representings and representeds none of this commits one to anything like an “intentional object” account of representational content. That is, one need not hold that a representing has representational content in virtue of standing in some particular relation to that which is represented by it. E.g., one need not think that a representing such as ‘that rat’ is only meaningful if it bears some relation to an actual, or non-actual, rat. (As we will see in later chapters, Kant’s theory of mental representation provides an example of how to account for representational content in a way that avoids precisely such a commitment.) 11. Flew, Hume’s Philosophy of Belief, 25–6. 12. Garrett, Cognition and Commitment, 49. 13. Garrett, Cognition and Commitment, 49–50. 14. It is worth mentioning here that Garrett does present his own account of Hume’s theory of mental representation in Garrett, “Hume’s Naturalistic Theory of Representation.” Since, however, Garret takes Hume to be more of a realist than I do, he is not concerned to show that arguments of this form are valid. In fact, his position is exactly the contrary: he predicates his account of Hume’s theory of intentional content on the claim that Hume does hold that we can represent such things. As I have already briefly pointed out, Hume’s explicit statements on these matters are more in line with my own

50

Hume’s Theory of Mental Representation

15.

16. 17. 18.

19. 20. 21.

22. 23.

24. 25.

reading. Farther along, I will also show that Hume can extricate himself from certain difficulties that Garrett raises in support for his view. This is a crude formulation of the Representational Copy Principle that will do for present purposes. A more careful rendering of it would have to allow for incorrect representation (which leveraging a strict application Hume’s account of copying clearly does not). The refinement that would be needed here would parallel Hume’s own refinement of the Copy Principle in light of the simple-complex distinction. Since most, if not all, complex ideas are not exact copies of anything, we would account for their intentional content as a function of the intentional content of their parts. Thus, a complex representation would represent the items represented by its simple components as being arranged in the way that these simple components are arranged in the complex idea. This would allow for Hume’s claim that falsehood results from the failure of ideas to “conform” to or “agree” with their objects. Each simple idea would conform to its object, but the complex idea that is formed by the association of these simples would represent the arrangement of the objects of those simples incorrectly. This addresses the argument Garrett makes against reading Hume as employing the Representational Copy Principle at Garrett, “Hume’s Naturalistic Theory of Representation,” 308. Fodor, Theory of Content, 91. “There is no impression constant and invariable” (T 1.4.6.2; SBN 251). In fact, Hume’s argument concerning the self is much more explicitly concerned with intentional content from its outset. Hume begins with an instantiation of the Representational Copy Principle: “If any impressions gives rise to the idea of the self, that impression must continue invariably the same, thro’ the whole course of our lives; since self is suppos’d to exist after that manner” (T 1.4.6.2; SBN 251). If there is to be an idea of the self, it must exactly resemble some impression. This is just the application of the Representational Copy Principle to the idea of the self. Of course, Hume famously goes on to deny that there is any such impression and so to conclude that we do not have any idea of such a self. Locke’s account of substance, for instance, is non-imagistic; Berkeley’s account of our “notion” of God is as well. This is one of five kinds of empiricism to which Garrett lists Hume as being committed. Garrett, Cognition and Commitment, 29–38. As we divvied the conceptual territory earlier during our explication of the notion of representational content, there would seem to be three, rather than two, perceptions in the immediate vicinity that ought to be considered here: the perception that is doing the representing, the perception that is represented by that representing, and the perception that is the de facto referent of that representing. For the moment, it will be easiest to limit our discussion to cases in which the second and third of these coincide: cases where what is represented is also the de facto referent, i.e., cases of correct representation. We will turn our attention to misrepresentation in the next section. T 1.1.4.1; SBN 11. I have elsewhere offered accounts of two such analyses: the impression-idea distinction and, with Alan Nelson, the simple-complex distinction. See Landy, “Hume’s Impression-Idea Distinction,” and Nelson and Landy, “Qualities and Simple Ideas.” In fact, they are aggregates of some finite number of smallest possible simple ideas. T 1.2.1.2–3; SBN 26–7. This instantiation of the principle will, of course, strike the reader’s ear as quite odd: we don’t normally think of our ideas as being spatially located at all (except maybe in the roundabout sense that we think of our ideas as

Hume’s Theory of Mental Representation

51

bearing some intimate relation to our brains, which are themselves spatially located). There are a few different ways that this can be handled on Hume’s behalf. One would emphasize the importance of keeping in mind that Hume is concerned first and foremost with our idea of space, which he analyses as being an idea of certain perceptions and a particular relation in which they stand in to one another. This line would represent a kind of deflationary interpretation of Hume’s claim. (Cf. Garrett, “Cognition and Commitment”; Falkenstein, “Hume on Manners of Disposition”; Coventry, “Hume’s System”.) Another would emphasize that Hume does think of ideas as images of various kinds and so in some cases as having genuinely spatial features (such as color, extension, motion, etc.) This line would require a more robust defense of Hume’s claim. 26. “Wherever ideas are adequate representations of objects, the relations, contradictions and agreements of the ideas are all applicable to the objects” (T 1.2.2.1; SBN 29). 27. This is itself a kind of Representational Copy Principle in two ways. Firstly, the intentional content of the simple representations that form such complex representations is itself determined by the Representational Copy Principle. Secondly, in cases of complex ideas that successfully represent, such ideas will be a copy of—will exactly resemble and be caused by—the arrangement of the objects that are the intentional content of the simple ideas that compose them.

REFERENCES Coventry, Angela. “Hume’s System of Space and Time.” Logical Analysis and History of Philosophy 13 (2010): 76–89. Donnellan, Keith. “Reference and Definite Descriptions.” The Philosophical Review 77 (1966): 281–304. Falkenstein, Lorne. “Hume on Manners of Disposition and the Ideas of Space and Time.” Archiv für Geschichte der Philosophie 79 (1997): 179–201. Flew, Antony. Hume’s Philosophy of Belief. London: Routledge and Kegan Paul, 1961. Fodor, Jerry. The Theory of Content and Other Essays. Cambridge, MA: MIT Press, 1990. Garrett, Don. Cognition and Commitment in Hume’s Philosophy. Oxford: Oxford University Press, 1997. Garrett, Don. “Hume’s Naturalistic Theory of Representation.” Synthese 152 (2006): 301–19. Hume, David. A Treatise of Human Nature. Edited by L. A. Selby-Bigge. New York: Oxford University Press, 1974. Hume, David. A Treatise of Human Nature. Edited by David Fate Norton and Mary J. Norton. New York: Oxford University Press, 2000. Landy, David. “Hume’s Impression-Idea Distinction.” Hume Studies 32 (2006): 119–39. Landy, David. “Hume’s Theory of Mental Representation.” Hume Studies 38 (2012): 23–54. Nelson, Alan and Landy, David. “Qualities and Simple Ideas: Hume and His Debt to Berkeley.” In Primary and Secondary Qualities: The Historical and Ongoing Debate, edited by Lawrence Nolan, 216–38. Oxford: Oxford University Press, 2011.

2

Two Objections to Hume’s Theory of Mental Representation

If the conclusions of our previous chapter are correct, then Hume’s three most important claims from the Treatise that (NC) We have no idea that is an idea of a necessary connection, (EW) We have no idea that is an idea of the external world, and (SSE) We have no idea that is an idea of a single subject of experience persisting through time, are all accepted on the grounds that an analysis of the idea of representational content shows that the controversial content of the ideas in question is impossible. According to Hume, we simply cannot form representations of any such things. We represent complex states of affairs by forming complex pictures, and we cannot form complex pictures of necessary connections, the external world, or the self. That Hume’s conclusions concern the nature and limits of our representational capacities is of crucial importance to anyone attempting to reconstruct Kant’s arguments against Hume. This is not least because on a fairly standard and widely accepted reconstruction of that dialectic, the bulk of Kant’s argument against Hume occurs during the course of the Transcendental Deduction of the Pure Concepts of the Understanding and has the following form. That argument begins with a premise that even the most ardent skeptic—Hume—would have to accept. This premise is often cast as one about the existence and nature of a single, simple subject of experience persisting through time.1 According to this reading, it then proceeds by showing that accepting this premise already commits one to a whole host of other robust philosophical theses including ones asserting the reality of the external world and of the existence of necessary connections. The problem with this reconstruction of Kant’s argument is that it casts Hume as a skeptic, and that, we have seen, is a misrepresentation. A skeptic is someone who holds that a certain proposition, while it might be true, cannot be known to be true. This is not Hume’s position regarding claims about necessary connections, the external world, or the self. He does not hold that such propositions cannot be known, although they might be true.

Two Objections to Hume’s Theory of Mental Representation

53

He holds that such propositions are “unintelligible,” “contradictory,” and “absurd.” Hume argues that we can have no ideas of any of these things, and so the propositions in which these ideas appear are, in a word, meaningless.2 Hume’s conclusions are not epistemic but representational. His claims are not that we cannot know such things to exist but rather that the very ideas of such things are impossible. So, if Kant is to engage Hume, he cannot begin with a premise that claims anything about a single, simple subject of experience persisting through time. Hume has already argued that we cannot have an idea with that representational content, and so no premise with such an idea as part of its content could ever be one that Hume must accept. He has already provided arguments for rejecting the very notion of such a premise. Thus, if we are to understand Kant’s arguments as aimed at and engaging Hume, those arguments must start well before such a premise ever surfaces. Furthermore, if Kant does leverage such a premise in his arguments, to do so legitimately, he must first refute Hume’s arguments that no such premise is possible. I will argue here that he does exactly this. In particular, I will argue that the sections prior to the Transcendental Deduction in the Critique, the Transcendental Aesthetic and the so-called Metaphysical Deduction, contain Kant’s most powerful objections to Hume’s theory of mental representation, which is the proper and necessary first target in Kant’s project. That is, before Kant can use a premise about the self to prove anything about the external world and necessary connections, he must show that (and how) we can legitimately employ such notions. He must show that Hume’s theory of mental representation is wrong, and he must replace that theory (at least schematically) with one of his own. These are exactly the two tasks that Kant undertakes in preparation for the Transcendental Deduction. The current chapter will be a discussion of Kant’s arguments against Hume’s theory of mental representation and a brief, big-picture look at Kant’s own theory. Subsequent chapters will be a more thorough and detailed examination of this theory. As we will see, in the sections we will be examining, Kant presents two powerful considerations that, while not explicitly billed as being aimed at Hume, are nonetheless raised at the exact points in the Critique at which Kant needs to give some argument against Hume and serve to undermine precisely the part of Hume’s arguments that serve as the most serious obstacle to Kant’s own project at these moments. Thus the interpretive argument of this chapter will be a bit unusual, and it is worth taking a moment now to plot its structure. It is uncontroversial, because Kant is explicit on the point, that it was the recollection (erinnerung) of Hume that awoke him from his dogmatic slumbers and that it was Hume’s conclusion concerning necessary connection that Kant took to be particularly insightful and in need of address.3 Furthermore, Kant took the argument leading to Hume’s conclusion about necessary connection to be only a single instance of a more general form of argument that could be used to call into question the legitimacy of a variety of similar representations.

54

Two Objections to Hume’s Theory of Mental Representation So I tried first whether Hume’s objection might not be presented in a general manner, and I soon found that the concept of the connection of cause and effect is far from being the only concept through which the understanding thinks connections of things a priori; rather metaphysics consists wholly of such concepts. Ak 4:260; Theoretical Philosophy, 57

In the previous chapter we were able to confirm Kant’s claim about Hume’s arguments. The argument for the conclusion that we can have no idea of necessary connection is an instance of a form of argument that can be used to call into question the legitimacy of a variety of similar ideas. The mark of that form of argument is Hume’s reliance on his theory of mental representation as its starting point. We further saw that not only did Hume use that form of argument in attempting to demonstrate that we have no idea of necessary connection, but he did, in fact, also apply it to the ideas of the external world and the self. Given these facts, with a liberal use of the principle of charity, we seem to be able to make the following argument. 1. Kant takes Hume to present a form of argument that can be used to call into question the legitimacy of certain representations. 2. Hume does present such a form of argument. 3. This form of argument is one that moves from Hume’s theory of mental representation to conclusions about these representations. 4. Kant sets out in the Critique to refute the form of argument that he finds in Hume. 5. Therefore, we should expect to find Kant, in the Critique, marshaling considerations against Hume’s theory of mental representation. Now, what we do not find in the Critique is a set of arguments explicitly aimed at refuting Hume’s theory of mental representation, and that is where the work of this chapter will take over. What I hope to show here is primarily that what we do find there is (a) Kant making arguments that can be used to refute Hume’s theory of mental representation and (b) Kant presenting his own theory of mental representation, which theory exactly meets the demands presented in these two objections. So, in addition to presenting the textual evidence for interpreting Kant as responding to Hume, I will also explore a line here that is more speculative: it will be an exploration of how to marshal the resources that Kant does explicitly present to make the arguments against Hume that Kant declares that he makes. I aim to show that these are arguments that are available to Kant, and that both the arguments that he constructs and the theory that he presents in response to these, are precisely what he would need were Hume’s theory of representation his target (as he declares that Hume’s arguments are).

Two Objections to Hume’s Theory of Mental Representation

55

To turn to the details of this chapter, then, I will present two objections on Kant’s behalf to Hume’s theory of mental representation. Each of these objections is meant to show that Hume’s theory of mental representation is inadequate to one of the tasks that Hume admits it must perform. The first is meant to show that Hume’s theory cannot adequately account for how we represent complex states of affairs as complex. The second is meant to show that Hume’s theory cannot account for the nature of judgments, representations that something is the case.4 In both arguments, Kant’s ultimate conclusion is that in order to form such representations, one must make use of certain conceptual resources for which Hume’s theory does not allow. The arguments we are about to examine, then, are each arguments against Hume’s theory by being arguments for the necessity of introducing conceptual structure (as Kant conceives it) into the kind of representations at hand. We will begin with Kant’s argument against Hume’s account of complex representation. REPRESENTATIONS OF COMPLEXES AS COMPLEX Recall that Hume’s theory of mental representation has at its core the Representational Copy Principle, which states that an idea is of that of which it is a copy. In the last chapter we delved into some of the details of how Hume works this theory out, but here we can employ this simplified version of the theory. A complex representation represents a complex state of affairs by being a copy of it. In particular, a complex representation is an arrangement of more simple representations, which represents the objects of those simple representations as being arranged in the way that the simple representations are arranged in the complex idea. A spatially complex state of affairs is represented by a spatially complex picture of it; a temporally complex state of affairs is represented by a temporally complex movie of it; in general for Hume, a representation of a complex is just a complex of representations. While complex representations of space and time are not the only representations that are structured this way for Hume, these two kinds of representations provide a helpful starting point for our study of Kant’s arguments against Hume because, of course, these are the subject of their own section of the Critique, the Transcendental Aesthetic, and this is where Kant makes his first stand against Hume. For those arguments to hit their mark, it must be the case that both Hume and Kant agree that representations of space and time (a) are complex representations and (b) represent a complex state of affairs as complex. We have already seen that these are both theses to which Hume is committed, so we can now turn to Kant. Here he is in the opening section of the Transcendental Aesthetic introducing its subject matter. The undetermined object of an empirical intuition is called appearance. I call that in the appearance which corresponds to sensation its matter,

56

Two Objections to Hume’s Theory of Mental Representation but that which allows the manifold of appearance to be ordered in certain relations I call the form of appearance. A20/B35

An appearance is that which is represented by an empirical intuition. Such appearances contain a manifold, or a complex of parts. So, what is represented by an empirical intuition is represented as complex. That which makes possible this representational content is the form of appearance, and its function is to represent the elements of the manifold contained in such a representation as related to one another in some way. Kant will go on to argue that our representations have two such forms: space and time. I.e., that which is represented in an empirical intuition is a complex of items related to one another spatially and temporally (in the case of outer appearance, only temporally in the case of inner experience). In fact, this is a thesis that Kant repeats in his third argument regarding space, which while it emphasizes the unity of space, does so by way of pointing out that this unity depends on the priority of the whole of space to “the manifold in it.” For, first, one can only represent a single space, and if one speaks of many spaces, one understands by that only parts of one and the same unique space. And these parts cannot as it were precede the single allencompassing space as its components (from which its composition would be possible), but rather are only thought in it. It is essentially single; the manifold in it, thus also the general concept of spaces in general, rests merely on limitations. A24/B39 While we do not represent space by adding together representations of individual spaces, building from the bottom up as it were, what is represented by a representation of space is nonetheless represented as a complex, as having “a manifold in it,” which are “parts of one and the same unique space.” The form of space is an a priori unity, but this form, as we saw above, just is “the manifold of appearance [. . .] ordered in certain relations.” The representation of space is a representation of a complex of spaces (even if their unity precedes their individual representation). Similarly, in the fourth argument concerning space, Kant contrasts the representation of space, which is an intuition, with what is represented by concepts. [N]o concept, as such, can be thought as if it contained an infinite set of representations within itself. Nevertheless space is so thought (for all the parts of space, even to infinity, are simultaneous). A25/B40

Two Objections to Hume’s Theory of Mental Representation

57

While concepts can be applied to a potentially infinite number of different representations, they do not “contain” this infinity within themselves. They are not representations of this infinity. Space, on the other hand, does contain such an infinity: “Space is represented as an infinite given magnitude” (A25/B40). The representation of space is a representation of something as (infinitely) complex. Space is “thought as if it contained an infinite set of representations within itself.” Kant clearly agrees with Hume that what is represented by spatial and temporal representations are complexes. Our next question, then, should be whether he also agrees that these representations themselves are complex. The answer to this question is, for the most part, the focus of Kant’s attention in parts of the Critique other than the Transcendental Aesthetic and is a question to which we will devote considerable attention in subsequent chapters. A relatively straightforward answer, however, can be culled from the Aesthetic as follows. Recall that sensations are that in the representation which corresponds to the complex matter in that which is represented. Here is a quick argument that Kant presents concerning sensation and the nature of the representations of space and time. Since that within which the sensations can alone be ordered and placed in a certain form cannot itself be in turn sensation, the matter of all appearance is only given to us a posteriori, but its form must all lie ready for it in the mind a priori, and can therefore be considered separately from all sensation. A20/B34 Kant’s argument here is for the a priority of space and time qua the form of that which is represented by an empirical intuition, but it is only the first premise of this argument that concerns us just now. That first premise is that representations of space and time themselves consist of a complex of sensations with a certain form (related in a certain way), and that this form cannot itself be just another sensation. So, Kant is clearly thinking of the representations of space and time as themselves being complex representations, of consisting in a complex of sensations related to one another in some as of yet unspecified way, which arrangement represents space and time themselves, as complex of relations of their parts (particular spaces and times). So Kant and Hume are in agreement on at least this much. At least spatial and temporal states of affairs are complex and are represented by representations that are themselves complex, that contain parts related to one another in a certain way. For Hume, these relations will themselves be spatial and temporal relations. Kant rejects exactly this aspect of Hume’s theory. In fact, the first argument concerning space in the Transcendental Aesthetic does just this.5 [I]n order for me to represent [items in space] as outside one another, thus not merely as different but as in different places, the representation

58

Two Objections to Hume’s Theory of Mental Representation of space must already be their ground. Thus the representation of space cannot be obtained from the relations of outer appearance through experience, but this outer experience is itself first possible only through this representation. A23/B38

Because Hume holds that we represent spatially complex states of affairs simply by copying these states of affairs in a complex of ideas, he is free to hold that we can derive our idea of space (which, for Hume, is just a set of associated ideas of spatial complexes) from “the relations of outer appearance.” That is, Hume can hold that the idea of space is nothing more than a collection of ideas of spatial complexes, which ideas are themselves mere spatial complexes of ideas. Kant, on the other hand, holds that it is only by having a distinct representation of space that we can have representations of spatial relations. So, for Kant, if we want to represent two items as being spatially related to one another, it is not enough that we place our representations of these items into spatial relations with one another. This would not yet involve any representation of space, and without such a representation, no two items can be represented as spatially related. For Kant, such a complex of representations would be merely that: a couple of ideas that happen to be sitting next to one another. In order to combine such representations into a single representation of the items represented as being related to one another, more is needed. We must add to these a representation of the space in which they are both situated. Of course, as far as Hume is concerned, this suggestion of Kant’s is literally absurd. This is because Hume holds that our idea of space is entirely relational. That is, our idea of space (or our idea of a particular space) is just a complex idea consisting of some number of ideas spatially arranged. We can leverage these particular ideas into an idea of space in general, but even that idea of space is nothing more than a collection of ideas of these particular spaces. Confronted with this passage from Kant, Hume would insist that the representation of space itself, apart from any objects “in” space, is entirely void of content. In fact, it is this very charge that Kant next confronts, albeit without much explicit argumentation. One can never represent that there is no space, although one can very well think that there are no objects to be encountered in it. It is therefore to be regarded as the condition of the possibility of appearances, not as a determination dependent on them, and is an a priori representation that necessarily grounds outer appearances. A24/B39 While these passages are typically read as being aimed at the Newtonian ontology of space and its accompanying Lockean epistemology, they can also clearly (and perhaps unsurprisingly) be read as aimed directly at

Two Objections to Hume’s Theory of Mental Representation

59

Hume’s theory of spatial representation. Kant’s premises here are that we cannot represent items as spatially related without a distinct representation of the space in which these items exist, and that it is entirely possible to represent this space independently of representing such items. Both are also claims that directly confront the theory of representation that we have seen Hume present. This immediately raises two questions: what are Kant’s arguments for these claims, and what is the nature of this representation of space itself that he claims, contra Hume, is not only possible but necessary? Neither of these questions receive their answers in the Transcendental Aesthetic. This is because each requires Kant to argue for and develop his claim that a manifold of representations can only be represented as a manifold using concepts. That is, they require Kant to show that complex representation is a conceptual affair. In the Aesthetic, however, Kant’s announced aim is to isolate sensibility by separating off everything that the understanding thinks through its concepts so that nothing but empirical intuition remains. A22/B36 Of course, if it is right that conceptual representation is essential to representing complex states of affairs as such, Kant cannot, strictly speaking, do what he says here. He cannot, that is, entirely account for our representations of space and time without, in some sense, bringing in the understanding. Kant himself admits this farther along when he is reconsidering the unity of the representations of space and time after having completed the work of the Transcendental Deduction: But this synthetic unity [“of the manifold, outside or within us”] can be none other than that of the combination of the manifold of a given intuition in general in an original consciousness, in agreement with the categories, only applied to our sensible intuition. Consequently all synthesis, through which even perception itself becomes possible, stands under the categories, and since experience is cognition through connected perceptions, the categories are conditions of the possibility of experience, and are thus also valid a priori of all objects of experience. B160–161 We will have reason to discuss this passage in greater detail in Chapter 3, but for now what is worth noting about it is that Kant points out that in order to represent space and time, one must form a single unified representation that represents the manifold of spaces as related to one another. Forming such a representation, he points out, requires not just the deliverances of sensibility, but also the action of the understanding. Kant is even more explicit and clear on this point elsewhere. For instance, in reporting on his

60

Two Objections to Hume’s Theory of Mental Representation

findings in the Critique in his prize essay, What Real Progress Has Metaphysics Made in Germany Since the Time of Leibniz and Wolff?, he writes: For we can represent a determinate space to ourselves no otherwise than by drawing it, i.e., by adding one space to the other, and so also with time. Now the representation of a composite, as such, is not a mere intuition, but requires the concept of a compounding, so far as it is applied to the intuition in space and time. So this concept (along with that of its opposite, the simple) is one that is not abstracted from intuitions, as a part-representation contained in them, but is a basic concept, and a priori at that—in the end the sole basic concept a priori, which is the original foundation in the understanding for all concepts of sensible objects. Ak 20:271; Theoretical Philosophy, 363 The representations of space and time are representations of complexes as complex. As such, they require “compounding,” which in turn requires the use of concepts. So, one cannot, at least with respect to the representations of space and time, “isolate sensibility by separating off everything that the understanding thinks through its concepts so that nothing but empirical intuition remains” because these representations are only possible through the use of concepts. It is also worth noting at this point that one of the most important conclusions that Kant reaches in the Aesthetic is that the representations of space and time are intuitions and therefore not concepts. To be clear, the thesis that I will advance on Kant’s behalf is not that the representations of space and time are concepts but rather that, while space and time are intuitions, intuitions themselves necessarily have a conceptual structure. The details of this particular thesis would take us too far afield of our current pursuits, but we will return to them in the next chapter. For now, it must suffice to say that intuitions are determinate singular representations of complex states of affairs as complex and as such require contributions from both sensibility and the understanding. So, again, as we have seen, Kant’s task in the Aesthetic is more nuanced than it first appears. Given that, what Kant can do is focus our attention on aspects of our representations of space and time that do not explicitly involve concept use. In doing so, however, Kant must necessarily take for granted certain conclusions about the nature of complex representation that he does not actually earn until farther along. Specifically, he must treat his crucial antiHumean conclusion here—that the representation of space is necessarily distinct from the representation of items in space—as a premise. To see the arguments that support these premises, we must turn our attention back to Hume. This is because the best way to bring out Kant’s answers to these Humean questions is to see how Kant’s theory is a response to certain, very specific failures inherent in Hume’s.

Two Objections to Hume’s Theory of Mental Representation

61

Once again, then, for Hume it is both necessary and sufficient for representing a complex state of affairs as complex that one places representations of the elements of these states of affairs in the same relations with one another as those in which these elements stand in to each other. So, for example, a representation of y as coming after x just is a representation of x followed by a representation of y. Hume’s theory of mental representation, then, has at least two parts: a claim that this kind of representation is necessary for representing complexes and a claim that this kind of representation is sufficient for representing complexes. Perhaps the best way to deal with the necessity claim is by presenting an alternative to it, which we will do when we turn to Kant’s view, so for now, we can focus our attention on the sufficiency of this relation of representations. The sufficiency claim is that all that one needs to do to represent some complex state of affairs is place representations of the elements of that state of affairs into the same relation that those elements stand in to each other. It is not difficult to find counterexamples to this claim, and in fact Hume himself inadvertently offers one such example. As he points out, The mind is a kind of theatre, where several perceptions successively make their appearance; pass, re-pass, glide away, and mingle in an infinite variety of postures and situations. T1.4.6.4; SBN 252 There is a countless succession of representations that pass through our minds. Clearly, though, not every such succession of representations is a representation of a succession. There is obviously a difference between, say, representing one’s appetizer as coming before one’s main course and lazily daydreaming about first a dog, then a cat, then a mat. The latter is not a representation of a succession at all. It is, however, a succession of representations. Therefore, a succession of representations cannot be sufficient for representing a succession. Of course, Kant famously employs a similar counterexample in the Second Analogy, which is standardly understood as the place in which Kant’s attention is focused most squarely on Hume. Thus, e.g., the apprehension of the manifold in the appearance of a house that stands before me is successive. Now the question is whether the manifold of this house itself is also successive, which certainly no one will concede. A191/B236 Suppose I find myself representing first the roof of a house, then its façade, then its foundation. That is a succession of representations. As Kant points out here, though, the question remains open whether this manifold is itself sufficient for representing these parts of the house as themselves occurring

62

Two Objections to Hume’s Theory of Mental Representation

successively, and that is a thesis that we should want to deny. In thinking of first the roof, then the façade, then the foundation, I do not thereby think of there existing first a roof, then a façade, then a foundation. Therefore, a succession of representations cannot be sufficient for representing a succession. So, Hume’s account fails as an account of all that is needed to represent a complex state of affairs as complex. It is important to see, though, precisely why it fails, and a further example will help bring this out more clearly. Consider the following complex state of affairs, in which the figure on the left is larger than the figure on the right.

Figure 2.1

On Hume’s account, the way to represent this state of affairs is by forming a representation of each of these figures and by having the representation of the figure on the left be larger than the representation of the figure on the right. So, one can represent the figure on the left’s being larger than the figure on the right by having a picture that exactly resembles the picture above appear before one’s mind’s eye. The real problem with this account is that according to it, if one has such a picture before one’s mind’s eye, one represents not just the figure on the left’s being larger than the figure on the right, but also the figure on the left’s being to the left of the figure on the right, and the two figures being congruent, etc. The fundamental problem with Hume’s account is that by having relations between items be represented by placing representations of these items into the same relations, it gives rise to crucial ambiguities. Any two representations necessarily stand in more than one relation to one another, and so any particular complex representation pictures the objects of its components as standing in all of those relations to one another. Thus, no complex representation is a representation of any particular and determinate complex state of affairs. Or, to put it another way, because every complex representation is a representation of so many complex states of affairs, none is a representation of any particular one of these.

Two Objections to Hume’s Theory of Mental Representation

63

This is what we saw a moment ago in the case of successive representations that did not represent a succession of items. Some such successions will, but others will not, count as representations of successions. The ideas of the dog, the cat, and the mat only just so happen to stand in a relation of succession to one another, but because all relations of ideas are also representations of relations, they end up being hijacked, so to speak, by Hume’s theory into also being a representation of a succession. Hume’s theory is too inclusive. In the case of ‘larger than’ above, a complex of ideas exactly resembling the picture above again represents far too much to be a representation of that picture as being any particular way. The problem is that because Hume holds that representing items must be placed in exactly the same relations as the items represented, every complex representation becomes an instantiation of far too many such relations. Every temporal sequence is a representation of temporal sequence. Every spatially structured representation becomes a representation of a spatial complex. Every pair of resembling representations becomes a representation of resembling objects. The structures that Hume employs for uniting complex representations into representations of complexes are simply not adequate for disambiguating what is being represented. The question now is what can do this work. The proper first place to look for an answer to this question is within Hume’s own theoretical system, and one place within that framework that naturally suggests itself is Hume’s account of abstract ideas. The thought here would be that one way of disambiguating complex representations would be to classify them using determinate general terms. So, the suggestion would go, in the case above, the representation of these two figures would be a representation of the one’s being larger than the other, not just in virtue of copying that relation but also because that representation is itself an instance of the kind ‘larger than.’ More specifically, using Hume’s theory, this representation would be disambiguated by being associated with other pictures that are themselves properly associated with the term ‘larger than.’ The very same picture, then, could also function as a representation of two figures being next to one another by being associated with other pictures that are themselves properly associated with the terms ‘next to,’ ‘congruent to,’ etc. As I said, this strikes me as the proper first place to look for a solution to the problem raised above. For the moment, though, since Hume’s theory of abstract ideas will be the focus of the next section, I would like to turn briefly to a different solution: Kant’s. This is because, while investigating Hume’s resources for dealing with these problems is one natural way to start seeking a solution, another is by addressing these problems more directly. Specifically, since it is the structures of complex representations that are at the root of the troubles with Hume’s theory, one way to address those troubles would be to replace these structures with ones that are more adequate to the task at hand, and this is exactly what Kant does.

64

Two Objections to Hume’s Theory of Mental Representation

CONCEPTS-QUA-INFERENTIAL-RULES With this, we have finally arrived at Kant’s thesis that all representations of complex states of affairs as complex are conceptually structured. That is, on Kant’s account, instead of placing representations of the elements of complex states of affairs into the same relations as the relations to be represented, we place them in what Wilfrid Sellars later called counterpart relations and, more particularly, counterpart inferential relations.6 Here is the basic idea. For Hume, to represent x as standing in relation R to y, we place a representation of x in relation R to a representation of y. The general schema looks like this. ‘x’R‘y’ represents xRy. The idea ‘x’ represents x; the idea ‘y’ represents y; and the fact that the representations ‘x’ and ‘y’ stand in relation R to one another represents the fact that x and y stand in that same relation to each other. Kant’s proposal is to make the simple, but radical move of forming such pictures not by placing ‘x’ and ‘y’ into the same relation as that represented but in a different one that will act as a counterpart, and representation of, this relation. Kant’s general schema, then, looks like this. ‘x’R*‘y’ represents xRy. To represent x and y as next to one another, we do not place ‘x’ next to ‘y,’ but rather we relate ‘x’ and ‘y’ to one another in a way that thereby represents x and y as being next to each other. Schematically, we place ‘x’ and ‘y’ next-to* one another. That Kant has such counterpart relations in mind for his own theory of mental representation is not a thesis that Kant explicitly states in the Critique (although we will see that there is evidence that the theory he does present there has this form), but he does address precisely the kind of example that we have been discussing elsewhere, for example, in his notes to the body of Meier’s Auszug aus der Vernunftlehre. Meier, like Hume has a theory of mental representation that requires a complex representation to have the same structure as that which it represents. In his notes on Meier, Kant presents an argument against this kind of theory and goes on to articulate his own alternative, which is one that employs relations that are the counterparts of the relations represented, rather than the very same ones, which is just what we have presented on Kant’s behalf as an alternative to Hume. The author purports that the representation of a thing that is to be found in the soul has the same sort of similarity with the represented thing as a painting has with the depicted object. But I assert that this is false, and prove it thus. When I see a house, then according to this

Two Objections to Hume’s Theory of Mental Representation

65

opinion there is a depiction of the house in my soul which is similar to the represented house. Now since similar things differ only with regard to their magnitude, a tiny house is depicted in my soul which, however small it is, must still occupy some space—which is impossible. [. . .] What is it then in the representation that is in agreement with the represented things? Since the representation borrows its ground from the represented thing, it agrees with the latter in that it is composed out of its partial concepts in the same way that the whole represented thing is composed out of its parts. E.g., one can say that the notes of a musical piece are a representation of the harmonic combination of the tones— not as if a note were similar to a tone, but because the notes have a combination among themselves like that of the tones themselves. Ak 16:76; Notes and Fragments, 34 A representation “agrees with” what it represents “in that it is composed out of its partial concepts in the same way that the whole represented thing is composed out of its parts.” Given the argument that Kant presents against Meier, we cannot understand this as saying that a representation and that which it represents share a structure in virtue of having their parts standing in the same relation to one another. Rather, what the example of the way in which notes represent tones shows is that it is in virtue of standing in a counterpart relation to one another that complex representations represent complex states of affairs. We place one note higher on a musical stave than another to indicate that the tone represented by the first is of a higher pitch than that represented by the second. That the notes (the representations of the tones) are placed in a spatial relation represents that the tones (represented by the notes) stand in a certain musical relation. Kant, unlike Hume and Meier, clearly thinks of representation as functioning via counterpart relations.7 In fact, Kant also makes it clear here that the counterpart relations that he think plays this role are specifically conceptual ones: “[a representation] agrees with the latter [its object] in that it is composed out of its partial concepts in the same way that the whole represented thing is composed out of its parts.” It is in virtue of being conceptually structured that a complex of representations represents the parts of its object as being determinately related to one another. That is Kant’s picture-theory of conceptual representation at the most general level in a nutshell. Of course, in the example above, it seems that we have merely swapped one relation (higher pitched) for another (above). If this suggestion is to succeed as an improvement on Hume’s theory, the relations into which we put such representations must be sufficient for disambiguating the complex representations thereby formed. It would not do, for instance, simply to switch out spatial relations for temporal ones, or bigger than relations for congruent ones. What is needed is a set of representing relations that is both rich enough for there to be one such relation for every relation represented

66

Two Objections to Hume’s Theory of Mental Representation

and yet specific enough for each representing relation to stand for only one relation represented. Again, Kant’s theory is tailor-made to provide just this. What I will argue is that Kant’s theory is one in which inferential relations between judgments are the primary structure of complex representations. The basic outline is this. For Hume, simple ideas are the fundamental unit of representation in part because they are the most straightforwardly determinate representations: a simple idea represents that of which it is a copy. Complex representations of complex states of affairs are formed by arranging these simple ideas, and this arrangement represents the objects of these simple ideas as arranged in the same way that the ideas are arranged in the complex. Oversimplifying for the moment, what plays the role played here by simple ideas, for Kant, is intuitions. Intuitions are the most straightforwardly determinate representations in Kant’s system: an intuition represents an object.8 Complex states of affairs are represented as complex by placing intuitions into inferential relations with one another. This is accomplished via judgment, wherein a concept is applied to an intuition. The role of the concept in a judgment is to relate one intuition to others. It is to license, forbid, and require certain other judgments to be made. It is to relate intuitions to one another inferentially to create a picture of the complex state of affairs that is the world of experience. We can find evidence that this is Kant’s view both in the Critique and in his other writings. We can begin by bringing together some of these other disparate texts to form a clear picture of Kant’s theory of concepts-qua-inferentialrules. Considering these latter texts first will allow for a more in-depth look at Kant’s theory than can be garnered by looking at the Critique alone, where Kant has many more items on his agenda. This, in turn, will make discerning these details in what Kant does say in the Critique that much easier. What these texts will show is that Kant (a) takes concepts to be rules, or to have serving as a rule as their essential function; (b) understands a rule as “asserting under a condition”; (c) understands “asserting under a condition” as playing the role of a major premise in a syllogism; and therefore (d) understands concepts as being identical to, or essentially constituted by, the role they play in inference. To begin, consider some of the many places at which Kant equates concepts with rules. [S]ensibility gives the mere material for thought, but the understanding rules over this material and brings it under rules or concepts. Ak 9:37; Logic, 547 Considered from the viewpoints indicated, a cognition will thus be logically perfect as to quantity if it has objective universality (universality of the concept or of the rule). Ak 9:38; Logic, 548

Two Objections to Hume’s Theory of Mental Representation

67

A rule is a concept under which much, a manifold of representations, is contained. Ak 24: 693; Logic, 431 That concept of reason which combines the greatest particular unity with this universal agrees with possible experience and is to that extent a correct rule. Ak 18:221, 5553; Notes and Fragments, 239 As indicated, these are only a small sample of the many places where Kant makes very similar remarks about the close tie between concepts and rules.9 Our next question, then, must be regarding the nature of a rule. Here is Kant in the Logic. A rule is an assertion under a universal condition. Ak 9:121; Logic, 615 Every rule has its condition, under which it can be affirmed or denied. [. . .] If something is subsumed under a condition, then this is as much as to say that the predicate of the subsumption applies; to subsume means to cognize that something is contained under the condition of the rule. Ak 9:93; Logic, 39010 So, since a concept serves as a rule, a concept serves as an assertion under a universal condition. That is, a concept serves as the condition by means of which one understands that to which the concept is applied. So, finally, we must ask after how it does this, what it is to identify something as standing under a universal condition, and what this “condition” is a condition of. Here is Kant at the start of the Transcendental Dialectic and again in the Logic. [R]eason in its logical use seeks the universal condition of its judgment (its conclusion), and the syllogism is nothing but a judgment mediated by the subsumption of its condition under a universal rule (the major premise). A307/B364 The combination of that which is subsumed under the condition with the assertion of the rule is the inference. Ak 9:121; Logic, 615

68

Two Objections to Hume’s Theory of Mental Representation

A syllogism is the subsumption under a universal condition. That, however, is exactly the role that we just saw assigned to a concept by way of its serving as a rule. A concept serves to locate that to which it is applied in a system of syllogisms. It gives the condition under which the intuition to which it is applied is to be understood, and we can now see that this condition is the condition of drawing an inference. It is the condition under which one can move from one intuition standing under a concept to another one standing under that same, or a related concept. The appearance of conjugation is a subsumption of a given representation under the general capacity for arranging sensations. The function of this capacity is the concept of the understanding, and its conditions make the rules for the transition from one representation to another. Ak 18:18, 4882; Notes and Fragments, 197 A concept makes the rules for the transition from one representation to another: it serves as a rule for connecting distinct representations to one another via inferences. For example, to represent something using the concept ‘body,’ one provides the condition under which one can draw an inference to judgments employing related concepts, such as ‘divisible’ or ‘shape.’ Insofar as something falls under the concept ‘body,’ it will also fall under the concept ‘divisible.’ If I say: a body is divisible, this means the same as: Something x, which I cognize under the predicates that together comprise the concept of a body, I also think through the predicate of divisibility. Ak 17:616, 4634; Notes and Fragments, 149 Notice that what ‘a body is divisible’ means is that if one thinks of something using the concept ‘body,’ one also thereby thinks of it using the concept ‘divisibility.’ Part of what it is to be the concept ‘body’ is to play a role in inference whereby one can draw the inference from ‘this is a body’ to ‘this is divisible.’ The role of a concept is essentially tied to linking intuitions to one another via inference. It provides the condition for drawing an inference from one judgment—the application of a concept to an intuition—to another. (And recall that “the understanding can make no other use of these concepts than that of judging by means of them” [A68/B93]. That is, concepts just are rules of inference from one judgment to another, or ways of relating intuitions to one another inferentially.) Note further that this form is not limited to analytic judgments but extends to synthetic ones as well. An example of a synthetic proposition is, To everything x, to which the concept of body (a+b) belongs, belongs also attraction (c). Ak 9:111; Logic, 607

Two Objections to Hume’s Theory of Mental Representation

69

Thus in the appearance x, in which a is a concept, there must be, in addition to what is thought through a, conditions of its specification which make necessary a rule whose function is determined through b. Ak 17:665, 4680; Notes and Fragments, 172 In the first quotation, the concept ‘body’ serves as the major premise in an inference, not to a judgment that applies one of the concepts that are themselves part of the concept ‘body’ to that something but to a judgment that applies the concept ‘attraction’ to them. Thus, applying the concept ‘body’ here represents the object to which that concept is applied as standing under a condition, which condition commits one to an inference to the conclusion that anything so represented also exhibits an attractive force.11 More significant, however, is the second quotation in which Kant puts his claim even more strongly: to use one concept, a, to represent an object there must be a rule that connects that object to a distinct concept b as well. Consider again the example of notes on a musical stave. Placing a single note on a stave does not represent the corresponding tone as being of any determinate pitch without at least potentially relating that tone to other tones. Placing notes on a stave represents the relation of tones to one another. Thus, a single note, a single mark on a page, does not serve its representative purpose unless it occupies a place in a structure that relates its object to other, at least potential, objects of representation. A similar line of reasoning underlies Kant’s claim in the second quotation above. In representing any object using one concept, a, one thereby also relates that object to, again at least a potential, other object that is represented using a different concept, b. E.g., in representing something as a body, I thereby relate it to some other, at least potential, object represented as exerting an attractive force. Thus, can we make sense of the following passage from another part of Kant’s notes. A concept, by means of its universal validity, has the function of a judgment. It is related to other concepts potentialiter. The actual relation of one concept to others as a means for their cognition is the judgment. [. . .] A judgment is the unity of a concept out of the relation (connection) of different concepts. Ak 16:630, 3045; Notes and Fragments, 59 Concepts are “related to other concepts potentialiter” in the sense that a concept serves its function in a judgment by relating the intuition to which it is applied in that judgment to other intuitions in other judgments via potential syllogisms involving other related concepts. For example, applying the concept ‘body’ to an intuition relates it potentialiter to the concept attraction (and thereby to intuitions themselves subsumed under that concept). Thus also is a judgment “the unity of a concept out of the relation (connection) of different concepts.” A judgment applies a concept to an intuition by relating

70

Two Objections to Hume’s Theory of Mental Representation

that intuition, via the essential connections of concepts to one another, to other intuitions falling under these other concepts. Consider also Kant’s repeated claim that concepts are “predicates for possible judgments.” (For example, Kant says this twice at A69/B94.) In light of this most recent passage from his notes, the sense of ‘possible’ here becomes twofold: a concept is both essentially tied to its role in judgments, possible and actual, but it also, even in a single judgment, serves as a predicate in some other possible judgment, i.e., serves to subsume the intuition to which it is applied under a universal condition by means of which it can be connected to other intuitions via some possible syllogism. Finally, consider one last passage from the Logic, wherein Kant writes about concepts in an explicitly inferential idiom. As one says of a ground in general that it contains the consequence under itself, so can one also say of the concept that as ground of cognition it contains all those things under itself from which it has been abstracted. Ak 9: 96; Logic, 594 A ground in general contains the consequence under itself. E.g., if one asserts that all As are Bs, this serves as the major premise in an inference that will have as its minor premise the subsumption of something as being an A and as its conclusion that that thing is also a B. The ground contains this consequence under itself insofar as it is part of the content of that judgment that anything that meets its condition (being an A) will also be subsumed under its consequent (will be a B). What Kant is pointing out here is that the relation of a concept to an intuition is an instance of this same relation. Concepts are the grounds of cognition. A cognition, as we have been understanding it, is the representation of a complex as complex. So, a concept serves as the inferential rule that structures such representations. In the most straightforward case, as we have just seen, a concept serves in this role by connecting intuitions to one another via the potential judgments in which it is a predicate. Thus does a concept contain “all those things under itself from which it has been abstracted.” A concept subsumes intuitions under itself (both those from which it is abstracted and those to which it can otherwise be applied) as ground and consequent. It serves as a rule connecting intuitions to one another via syllogisms. So, what the texts I have presented show is that (a) Kant holds that we represent complex states of affairs as complex by placing representations of the components of these states of affairs into counterpart relations with one another and (b) that these counterpart relations are the inferential relations that partly constitute the content of any concept. The latter thesis was established by showing that Kant holds that concepts are rules, rules are conditions, and such conditions are conditions for constructing syllogisms, for inference.

Two Objections to Hume’s Theory of Mental Representation

71

Of course, while Kant does appear to hold all three of these theses, he also discusses concepts in more traditional terms, e.g., as proceeding via marks: From the side of the understanding, human cognition is discursive, i.e., it takes place through representations which take as the ground of cognition that which is common to many things, hence through marks as such Ak 9:58; Logic, 564 and as having their ontogenesis in “Logical Actus of comparison, reflection, and abstraction” (Ak 9:94; Logic, 592). It is these latter texts that have led some scholars to believe that Kant holds a much more traditional, e.g., Lockean, theory of concepts as abstract, nonpictorial, general representations. For example, Walker suggests that Kant “is unable to cut himself altogether free from the Lockean view that empirical concepts are somehow given to us from outside and therefore do not need the category-governed activity of the mind.”12 As he points out, though, “This is quite incompatible with his recognition that at every level of classification is something we ourselves do.”13 On the Lockean view, while a concept may serve as a rule of inference, it is not itself such a rule but rather can serve in this capacity because it is an abstract, nonpictorial, general representation. So, for example, the concept ‘body’ represents either a universal such as body-hood or bodyness, or some set of particular bodies, and it is in virtue of the nature of this universal or these particulars that the inference from, say, ‘This metal is a body’ to ‘This metal is divisible’ is valid. Bodies, conceived in this way, would bear the marks the of body-hood, and it would be through a process of comparison, reflection, and abstraction that we discover these marks and use them to form such abstract, nonpictorial general representations. On this line, we would take away everything from the particular representation of this or that body and be left with a representation that consists only of the marks common to all bodies. The first thing to note here is that Kant would have been well aware of Berkeley’s criticism of such a Lockean position (which Hume, of course, also presents in the Treatise). We will have occasion to revisit that criticism in the next section during the course of our investigation of Hume’s theory of general representation, but the gist of it is that since everything that exists is fully determinate, mental representations are also determinate. If, however, mental representations are fully determinate, then they cannot also be abstract. So, the objection goes, whatever it is that is represented by concepts, e.g., that which is common to several things, cannot be represented as such by a mental representation that is itself abstract. Hume, of course, goes on to cast general representation in terms of the human disposition to associate certain resembling ideas with certain words. Kant recognizes the inadequacy of such a view and, to oversimplify for the moment, replaces Hume’s associations with normative-inferential relations.

72

Two Objections to Hume’s Theory of Mental Representation

Recognizing that Kant was well aware of this line of argument, Longuenesse argues that while Kant does not hold that concepts are abstract, nonpictorial representations—she agrees that concept are rules—he nonetheless retains the Lockean view that the process by which the concepts come to have the representational content that they do is perceptual. According to Longuenesse, Kant holds that we form concepts by coming to perceive the immanent universals that are contained within the objects of experience and that it is these universals that such concepts represent. Again, I must postpone a fuller consideration of this view until the next section, where the apparatus of the problem of the unity of the propositions will prove useful in seeing where this Lockean approach to Kant goes wrong. My diagnosis there will be that Longuenesse makes a mistake in thinking that concepts, for Kant, have any object of their own, immanent universals, rather than serving to represent the relations between objects of intuition. I will also have occasion there to more thoroughly draw out the inconsistencies (to which Walker alludes above) between such a view and several of Kant’s other central commitments regarding the nature of concepts. Still, even if the case against a Lockean reading of Kant’s theory of concepts can be secured, these Lockean-sounding texts remain. Perhaps the most sustained attempt to make these texts consistent with the thesis that concepts are rules is Pippin, Kant’s Theory of Form. The gist of his interpretive line is that while one function of the understanding is to unite manifolds of intuition according to rules, another of its functions is to itself generate these rules, and it does this precisely by seeking marks in intuitions that might signal a basis for conceiving them under the same condition. That is, the understanding uses the acts of comparison, reflection, and abstraction as the basis for the rules that it prescribes to itself for how such manifolds ought to be united. Thus, while that which is being represented provides the ultimate ground of how a representation comes to be structured, the structure itself is nonetheless a product of spontaneity: it is, in fact, the inferential structure that we will detail over the next several chapters. On this line, concepts are rules for uniting manifolds of intuitions—and so not Lockean abstract ideas—and represent the objects of such intuitions as related to one another in determinate ways—and so have no distinct objects of their own. Of course, the philosophical viability of this thesis will depend on being able to make sense of the claim that representations are grounded in that which they represent in the face of the fact that the acts that are to provide this grounding are themselves subject to the rules of the understanding. Pippin ultimately deems that project as doomed to failure, and we will have reason to revisit that pessimistic conclusion in the next chapter. To summarize, then, there are passages in Kant’s writing, especially in the Logic, that seem to imply that he holds something like a Lockean theory of concepts as both abstract, nonpictorial representations, and as being passively derived from the objects of experience. He cannot hold the first of these theses because he is aware of the criticisms of such a view leveled by Berkeley

Two Objections to Hume’s Theory of Mental Representation

73

and Hume, and he repeatedly stresses that concepts are rules. He cannot hold the second of these theses because this would leave him no answer to the problem of the unity of the proposition—of which he was also well aware— and he repeatedly stresses that concepts do not represent any distinct kind of objects of their own. I will discuss each of these points in more detail in the following section and will engage Pippin on the issue of the relation of marks and comparison, reflection, and abstraction in the following chapter. In the meantime, though, notice lastly that the account that I have been sketching gives an explicit answer to a central question that Kant must address: what is the role of concepts in representing complex states of affairs? That answer is that concepts-qua-inferential-rules relate intuitions to one another to form a picture of such states of affairs. If one takes concepts to be abstract, nonpictorial, general representations, this answer is no longer available. While such representations might “serve” as rules of inference, their primary function will be similar to that of intuitions. A concept does not represent a single object on this line, but it does represent either a universal or a collection of objects. The natural understanding of such a theory would be that the elements pictured are both objects and either universals or collections of objects, and this has the immediate undesirable consequence of forcing concepts to play an uncomfortable dual role, as representing both an element of experience—a certain universal or collection of particulars—and the structure of experience—the relations that these elements stand in to that which is represented by an intuition. Understanding concepts as inferential rules and as representing objects of intuition as related in particular ways (to form a picture of the world) straightforwardly avoids this arguably untenable situation. CONCEPTS-QUA-INFERENTIAL-RULES IN THE CRITIQUE Having sketched Kant’s theory of representation via counterpart relations and concepts-qua-inferential-rules as it runs through some of his texts other than the Critique, we can now turn to Kant’s specific employment of it in the Critique itself as a remedy to the problem that he discovers in Hume’s (and Meier’s) theory of mental representation. We will find Kant’s first description of his theory in the Critique in the Metaphysical Deduction, but before we turn to that, a word is in order of the place of that section in the broader argumentative context of the Critique. The purpose of the Metaphysical Deduction is, as Kant indicates, to provide a clue to the discovery of the pure a priori concepts of the understanding. This clue comes from the nature of judgment, which, as Kant understands it—an understanding that we will see Hume reject in the next section—is the application of a concept to an intuition. Since no representation pertains to the object immediately except intuition alone, a concept is thus never immediately related to an object, but

74

Two Objections to Hume’s Theory of Mental Representation is always related to some other representation of it (whether that be an intuition or itself already a concept). Judgment is therefore the mediate cognition of an object, hence the representation of a representation of it. A68/B93

The language of a judgment as a mediate representation is similar to that in the following note in which Kant outlines the notion that such judgments will have different forms. A judgment is the mediate cognition of one representation through other representations. The relation of mediate representation to the immediate one is (the relation in the judgment or) the form; the subject is the immediate representation, the predicate the mediate one. Ak 16:631, 3047; Notes and Fragments, 59 Intuitions are immediate representations of objects; concepts mediate representations of them. The way that a concept relates to an intuition in a judgment is the form that the judgment takes. For example, the heading ‘Quantity’ in the table of logical forms of judgment represents the different ways that the concept ‘S’ relates to an intuition in judgments with the forms: a) All xs that are S are P, b) Some xs that are S are P, and c) This x that is S is P. This provides a clue to the discovery of the pure a priori concepts of the understanding because the Categories are simply the concepts of these forms.14 So, for example, the concepts under the corresponding heading of ‘Quantity’ in the table of categories—unity, plurality, and totality—are just the concepts of intuitions that take the forms a’) All-such, b’) Some-such, or c’) This-such. Finally, it is the Categories, derived in this way from the logical forms of judgment, that Kant will, in the Transcendental Deduction, go on to argue are applicable to all and only objects of possible experience. (More on this in the next chapter). This, then, is why we find Kant, in the Metaphysical Deduction, presenting (very quickly) his account of judgment—the relation of a concept to an intuition—and thereby his account of concepts themselves. It is from this account of judgments, concepts, and intuitions that Kant will derive the Categories. Having quickly run through the place of the Metaphysical Deduction in the Critique, we can now turn to the business of moving through it more

Two Objections to Hume’s Theory of Mental Representation

75

carefully to determine whether the theory of concepts-qua-inferential-rules that we have lately been attributing to Kant can be found there as well. To do that, we can return to the passage just quoted. Here it is again. Since no representation pertains to the object immediately except intuition alone, a concept is thus never immediately related to an object, but is always related to some other representation of it (whether that be an intuition or itself already a concept). Judgment is therefore the mediate cognition of an object, hence the representation of a representation of it. A68/B93 There is room for debate here regarding precisely what Kant means by the claim that intuitions are the only representations that pertain to the object directly. That is a debate to which we will return in the next chapter.15 For the time being, what is noteworthy about this passage is the conclusion that Kant draws from this thesis: while intuitions represent objects, concepts by contrast do not represent any objects but instead relate to intuitions (or other concepts). A concept is a mediate cognition of objects. By this, Kant does not mean to imply that concepts do not represent objects simpliciter, but rather that (a) they do not represent any distinct object (such as a universal) and (b) what a concept represents is not an object but the relations of the objects represented by intuitions. Contrast this with the Lockean account of concepts that we were just considering. According to that account, just as ‘Socrates’ is a name for Socrates, so ‘wise’ is a name for the universal wisdom. Kant’s point here, as we will see in greater detail in the next section, is that it is a mistake to construe concepts as performing the same representative function as intuitions do, in this way. Intuitions represent single, determinate objects. Concepts represent these objects as related to one another but do not thereby represent any further object. E.g., one need not suppose that there is an object, wisdom, that ‘wise’ names in order to make sense of the judgment ‘Socrates is wise’ (and, in fact, one would be better off supposing that there is not). So Kant continues with an example, So in the judgment, e.g., “All bodies are divisible,” the concept of the divisible is related to various other concepts; among these, however, it is here particularly related to the concept of body, and this in turn is related to certain appearances that come before us. These objects are therefore mediately represented by the concept of divisibility. A68/B93 The concept of divisibility does not represent any object of its own but rather represents the objects represented by the intuitions to which it is applied as all being divisible. This explains the sense in which a concept is a metarepresentation: concepts represent the objects represented by intuitions as being related in various ways by placing representations of those objects

76

Two Objections to Hume’s Theory of Mental Representation

(intuitions) into counterpart relations with each other, by locating them in a system of inferences. Notice in the above quotation that Kant makes the point that in the judgment, “All bodies are divisible,” “the concept of the divisible is related to various other concepts,” in addition to being specifically related to the concept ‘body.’ As we have already seen, what Kant has in mind here are the inferential connections between the concept ‘divisible’ and other concepts, such as ‘shape’ and ‘penetrability.’ The concept ‘divisibility’ is a rule, a condition of inference, a potential major premise in syllogisms that allows one to move from ‘body’ through ‘divisibility’ to one of these other concepts. To apply the concept ‘divisibility’ to an intuition is to locate that intuition in a network of such syllogisms, which network represents those intuitions as standing in certain relations to one another, which all together forms a picture of the world. All judgments are accordingly functions of unity among our representations, since instead of an immediate representation a higher one, which comprehends this and other representations under itself, is used for the cognition of the object, and many possible cognitions are thereby drawn together into one. A68/B93, emphasis added ‘Unity’ is one of Kant’s terms for signaling a representation of a complex as complex. Thus, his description of judgments here is exactly meant to be a description of such a representation. What we should notice about this description is that despite the fact that in its most basic form a judgment is the application of a concept to a single intuition, Kant here claims that the essential role of a judgment is to unite “many possible cognitions.” These many possible cognitions are the many intuitions that are united via the inferential rules linking one judgment to another. (Recall that concepts are all “related to other concepts potentialiter”.) Thus, Kant tells us that a concept just is “the predicate for a possible judgment” (A69/B94). A concept, by its very nature, is an interjudgmental function of unity of intuitions. Concepts are the inferential links that Kant uses to replace Hume’s structure to form representations of complex states of affairs as complex. In a slightly different context, Kant makes it more explicit that this is the move he is making. Thus we think of a triangle as an object by being conscious of the composition of three straight lines in accordance with a rule according to which such an intuition can always be exhibited. A105 A triangle is a complex object with three intersecting line segments as its parts. Our thinking of a triangle as such, then, is exactly a representation of a complex as complex. Kant is here detailing how we form such a

Two Objections to Hume’s Theory of Mental Representation

77

representation according to his new theory. Putting aside for the moment that Kant is writing here about the complexity internal to an intuition, we can see that the same kind of unity is effected here as it is in the case of representing intuitions themselves as related to one another.16 One represents a triangle as consisting of three straight lines forming a closed figure by uniting representations of each of those lines according to a rule. As Kant tells us a few lines later, a concept is something “that serves as a rule,” and as we saw earlier, it serves as a rule by being a predicate for possible judgments, which is to set the condition under which an inference can be drawn relating one representation to another. That is, the representation of this triangle as a determinate complex of line segments is made possible via the employment of inferential rules. In particular, the representations of each line segment are united by applying the concept ‘triangle’ to them. This unites these representations into a complex representation of a complex state of affairs (in this case a complex object) not by placing them in the same relation as that which is represented (by placing the representations end to end) but rather by relating them inferentially. For instance, the rule here might be something like, if one reaches the end of the first line, one is licensed to infer that another line forms an angle with the first at just this point. So, what we can see in the Metaphysical Deduction is the theory that we have extracted from Kant’s other texts—of inferential rules as counterpart relations—being put to use by Kant at exactly the moment that we would expect: just after having presented reasons in the Aesthetic for thinking that Hume’s theory of mental representation is wrong, and just before the Transcendental Deduction, where he, Kant, will use the theory that he means to replace Hume’s to demonstrate the validity of the concepts that Hume had rejected. According to that theory, as it appears across the Notes and Fragments, the Logic, the Prolegomena, and the Critique, is that the most fundamental way of filling in the scheme ‘x’R*‘y’ represents xRy. is as follows. {intuition-of-x} relates inferentially via concept C to {intuition-of-y} represents xRy. What is important here is not so much the introduction of intuitions as the new relata in this scheme, although we will spend a good deal of time investigating that change in the following chapters, but rather the introduction of counterpart inferential relations. This change places two items immediately onto the top of our agenda. The first is to see, at least in brief for now, how such relations would function in this role, and the second is to see, again in brief for now, why we might expect such relations to be an improvement.

78

Two Objections to Hume’s Theory of Mental Representation

As for this former task, then, consider the following passage from the A-Deduction, which mirrors the ones that we examined earlier from the Logic and Notes and Fragments: the concept of body serves as a rule for our cognition of outer appearances by means of the unity of the manifold that is thought through it. However, it can be a rule of intuitions only if it represents the necessary reproduction of the manifold of given intuitions, hence the synthetic unity in the consciousness of them. Thus in the case of the perception of something outside of us the concept of body makes necessary the representation of extension, and with it that of impenetrability, of shape, etc. A106, emphasis added Kant here tells us that “the concept of body serves as a rule” and that it “makes necessary the representation of extension,” etc. We saw earlier that Kant also extends this portion of his account to include not only analytic judgments but also synthetic ones: e.g., a judgment in which the concept ‘body’ is related to the concept ‘attraction.’ Kant, of course, does not mean that whenever we employ ‘body’ in a judgment, we must also employ ‘impenetrability,’ ‘shape,’ ‘attraction,’ etc. in further judgments. He is not speculating about the associative tendencies of the human mind. As we have seen, by way of Kant’s notion of “asserting under a condition,” what he means is that the concept of a body serves as a rule of inference according to which if one is committed to a thing’s being a body, then one is thereby also committed to its being extended, being impenetrable, having a shape, etc. It sets the “condition” for drawing inferences from a judgment employing ‘body’ to further judgments employing these other concepts. It is the concept ‘body’ itself that licenses these inferences because, as Kant sees it, that concept just is (or is at least) a certain set of rules for how to place intuitions into relations with one another, which relations represent the objects of these intuitions as standing in certain relation to each other. To take a slightly different, and more relevant, example, consider a representation of a spatially complex state of affairs: a tree standing next to a house. How, on this account, do we represent this state of affairs? Well, on Hume’s account we would do so by placing a representation of the tree next to a representation of the house. We have already seen why such an account fails. On the account that Kant adopts, we represent the tree as being next to the house by licensing and forbidding certain judgments, or inferences. For instance, we license the inference from ‘this house is directly in front of me’ to ‘if I turn slightly, that tree will be directly in front of me.’ We forbid the inference from ‘this house is nearby’ to ‘there is no tree nearby.’ Etc. It is commonly accepted that one of the important changes that Kant institutes with his new theory of mental representation is introducing a normative element that was not present in his predecessors. We are now in a position to see exactly why he does this. Kant attempts to correct what

Two Objections to Hume’s Theory of Mental Representation

79

he sees as the failures of Hume’s theory by replacing the relations that structure Hume’s representations of complexes with a set of relations that is both rich and specific enough to allow for the determinate representation of any such complex state of affairs. Any nonnormative, or matter of factual, set of relations would necessarily fail to meet these requirements exactly because all such relations are simply too robust. Every spatial relation is also a temporal relation; every larger than relation is also a similarity relation; etc.17 Inferential relations, on the other hand, because they are not merely matter of factual but also normative, can be tailored to meet whatever representational demands we have of them. This is because a new inferential license, forbearance, etc., can always be put in place to represent the specific complex state of affairs at hand. The next to* relation will license, forebear, etc., certain inferences concerning what one ought to judge of two items if they are next to one another. The larger than* relation will involve very different inferences concerning what one ought to judge if one item is larger than another. While Humean nonnormative relations are fixed in what they can and do represent, inferential relations can be made to be unique by putting in place whatever allowances, disallowances, etc., that are needed to capture the unique features of the complex state of affairs at hand.18 There is a sense, then, in which Kant has not abandoned Hume’s account at all. Kant agrees with Hume that the way to represent complex states of affairs as complex is to picture them. Where he differs from Hume is in what structures he thinks are adequate for constructing these pictures. He replaces Hume’s spatio-temporal structures because he recognizes that spatio-temporal structures are not adequate to the task at hand. He chooses inferential structures for very good reason. If what is needed is a set of structures that can unambiguously picture one relation rather than another, if what is needed is a one-to-one correspondence between relations represented and relations doing the representing, inferential relations are tailor made for just this. Inferential relations are not fixed in the way that nonnormative relations are. We can always license a new inference upon discovering a new relation among objects. We can always forbid a certain inference upon discovering that some relation does not hold between objects. There can be as many relations between judgments as are necessary for representing relations among objects, so inferential structure is infinitely richer that the nonnormative one Hume employs.19 So, for Kant, representing a complex state of affairs as complex requires placing representations of the elements of these states of affairs into inferential relations with one another. It requires this because only a normativeinferential structure is rich enough to capture the many ways that the objects of representation can relate to one another. Since concepts serve as the inferential rules for so arranging such representations, it follows that, for Kant, any representation of a complex state of affairs as complex must be conceptually structured.

80

Two Objections to Hume’s Theory of Mental Representation

This, then, is Kant’s suggestion about how to solve the problem of the lack of determinacy in Hume’s account of complex representation. As we mentioned earlier, however, it is not yet clear that Hume’s theory, considered as a whole, is in need of such fixing. This is because we have not yet shown that Hume’s theory itself does not contain the resources for addressing this difficulty. The most promising avenue of exploration in this regard is the part of Hume’s theoretical apparatus meant to account for general ideas. We have more reason to pursue this strategy now than we did when it was first suggested. Then our reason was the hope that the account of general ideas could be used to individuate exactly which feature of a complex state of affairs is being represented by any complex idea by relating that idea to other similar ideas. Now we have the additional impetus of knowing that Kant’s solution to this problem is to marshal the resources of conceptual representation, and Hume’s theory of general ideas can be plausibly described as his own account of concepts. So, if Kant’s suggestion is that concepts are needed for complex representation, then it is even more important than ever to examine Hume’s own theory of concepts before leaving his account of mental representation behind. This, in turn, will also bring us to the second of Kant’s objections to Hume’s theory of mental representation: that it cannot account for the nature of judgment. THE PROBLEM OF THE UNITY OF THE PROPOSITION For Kant, a theory of concepts necessarily goes hand in hand with a theory of judgments. That is because, as we have seen, for Kant, a concept just is “a predicate for possible judgments.” The two are essentially interconnected. For Hume, this connection is much less strong in no small part because Hume does not really have a theory of judgment at all. What he has instead is the suggestion that the then-received view of judgment is woefully misguided and an account of how one can get by without such a theory. As we will see, however, the connection between this account and his account of general ideas is one that we will do well to forge on Hume’s behalf. We can begin with Hume’s account of general ideas and the way that it might at first seem to be able to play the role, described in the previous section, of disambiguating representations of complex states of affairs. This will lead us naturally to the issue of judgment as the most plausible vehicle by which one might come to affect that disambiguation. That in turn will lead us to Kant’s critique of Hume’s account of judgment (or what is supposed to take judgment’s place), and this will finally lead us once again back to Kant’s own inferentialist theory. We begin, then, with Hume’s theory of general ideas. Hume begins this section of the Treatise arguing with Berkeley against Locke that there is no such thing as an idea that is itself abstract. The thought here is that if all mental representation is, as we have seen, achieved

Two Objections to Hume’s Theory of Mental Representation

81

via copying, then for there to be an idea of something abstract—not this or that person, but of persons in general—the most straightforward way for such an entity to be represented would be for there to be an idea that is itself a copy of this abstract entity: an abstract idea. Hume rejects this suggestion on the grounds that the pictorial content that such an idea would have to have is impossible. It would either have to have the qualities of all persons, or the qualities of no persons, but neither of these are realizable in a single picture.20 (We can have a picture neither of a person that is both six feet tall and not six feet tall, nor of a person that is neither six feet tall nor not six feet tall.) Thus, Locke was wrong that general representations are abstract ideas. Of course, Hume would not deny that there are general representations, so to make his conclusion tenable, he needs to give an account of how these are possible. At first blush, it might seem that, given the failure of the Lockean account of general ideas, whatever account Hume will offer must be one that abandons copying as its fundamental representational principle. If there are no abstract ideas, the reasoning would go, there cannot be anything that is a representation of anything abstract (including anything general). In fact, this is not the case at all. As we will see in a moment, Hume’s account of general ideas is yet another instantiation of the Representational Copy Principle as it applies to complex ideas. That is, this part of Hume’s system also fits into the scheme that we have lately been using of representing xRy via a representation of the form ‘x’R‘y’. To see this, we need first to have Hume’s account in front of us, and it is brief enough to quote in full. When we have found a resemblance among several objects, that often occur to us, we apply the same name to all of them, whatever differences we may observe in the degrees of their quantity and quality, and whatever other differences may appear among them. After we have acquir’d a custom of this kind, the hearing of that name revives the idea of one of these objects, and makes the imagination conceive it with all its particular circumstances and proportions. But as the same word is suppos’d to have been frequently apply’d to other individuals, that are different in many respects from that idea, which is immediately present to the mind; the word not being able to revive the idea of all these individuals, only touches the soul, if I may be allow’d so to speak, and revives that custom, which we have acquir’d by surveying them. They are not really and in fact present to the mind, but only in power, nor do we draw them all out distinctly in the imagination, but keep ourselves in a readiness to survey any of them, as we may be prompted by a present design or necessity. T 1.1.7.7; SBN 20–1 When we encounter items that resemble one another in certain ways, we call them all by the same name. Because this name is frequently used in the

82

Two Objections to Hume’s Theory of Mental Representation

presence of these items, our minds form an association between the two such that whenever that name is used we call to mind some one, or a few, ideas of these items. Furthermore, we stand disposed to call to mind more such ideas (ideas of the items between which this resemblance was found) upon further prompting. So, we represent persons in general by forming certain associations between our ideas of particular persons. Namely, those ideas all resemble one another and are called to mind in the appropriate situations (primarily when the word ‘person’ is used, etc.) because of this resemblance.21 So far, it does not appear that copying plays any explicit role in this account at all. Given, however, how seriously we have seen Hume take the Representational Copy Principle in other areas, it is worth exploring whether this principle is implicitly at work here nonetheless, and I will suggest that it is. One way to see this is to ask what is represented by such general ideas. The obvious answer is that, for instance, the general idea ‘personhood’ represents personhood. That answer, however, simply pushes the question back a step because we now need to know what kind of thing personhood is. The obvious first candidate here is persons. That is, perhaps Hume’s theory is that the general idea ‘personhood’ is just a way of representing all particular persons. Thus, the mechanisms that Hume details here are merely that: mechanisms for calling to mind just the ideas needed to represent all and only persons. On this reading of his theory, that we rely on the resemblance of each of these ideas to one another is an incidental feature of the account. If it just so happened that persons didn’t resemble one another at all but were encountered all standing next to one another, we would form the general idea personhood via associations of contiguous items rather than resembling ones. It is this last point that is implausible. It does not seem to be a mere accident that it is resemblance that is at work in general ideas. In fact, a more plausible suggestion of what is represented by a general idea is exactly persons qua resembling. That is, according to Hume’s account, what prompts us to form general ideas in the first place is that “we have found a resemblance among several objects, that often occur to us.” Thus, another natural suggestion for what is represented by an idea prompted by such encounters is that it is these objects as resembling one another. So, it is not just that our general idea ‘personhood’ represents persons, but more specifically, it represents persons qua members of a set of objects among which we have found a resemblance. That is, it represents persons as persons. This suggestion also just so happens to fit nicely with the line of inquiry that we set out to pursue in this section. Recall that this was to see if general ideas could be used to disambiguate representations of complex states of affairs, if they could be used to make a representation more specific. If general ideas really do represent not just persons but persons as persons, and do so successfully, then they would be up to just this task. That is, they could be used to represent items as, for instance, larger than qua larger

Two Objections to Hume’s Theory of Mental Representation

83

than (as opposed to also representing them as congruous, next one another, simultaneous, etc.) Furthermore, it seems that Hume has just this kind of work in mind for general ideas. Here he is later in that section of the Treatise writing about distinctions of reason. Thus when a globe of white marble is presented, we receive only the impression of a white colour dispos’d in a certain form, nor are we able to separate and distinguish the colour from the form. But observing afterwards a globe of black marble and a cube of white, and comparing them with our former object, we find two separate resemblances, in what formerly seem’d, and really is, perfectly inseparable. After a little more practice of this kind, we begin to distinguish the figure from the colour by a distinction of reason; that is, we consider the figure and the colour together, since they are in effect the same and undistinguishable; but still view them in different aspects, according to the resemblances, of which they are susceptible. T 1.1.7.9; SBN 21–2, emphasis added Here Hume outlines how it is that we represent an object as being white as opposed to also being a globe. What is represented is disambiguated by being represented as resembling some other object represented. This, however is explicitly billed by Hume as being an instance of the use of general ideas. Thus, what is represented by a general idea is not just a group of particular objects but rather a group of particular objects as resembling one another (and, thereby, as having a specific relation to one another). Again, though, we have not yet seen copying play an explicit role here. It is, however, playing a role nonetheless. What we have tentatively established is that a general idea represents the items it represents as resembling one another. It is clear from Hume’s account of general ideas that these function via the association of ideas of these items. That association is the one formed in virtue of these ideas resembling one another. So, what we have is that certain objects are represented as resembling one another in virtue of the fact that representations of these items are related to one another via their own relation of resembling each another. That is, ‘x’ resembling ‘y’ represents x resembling y. General ideas are an instance of the Representational Copy Principle as it applies to complex ideas. Objects represented are represented as resembling one another by placing representations of these objects into the same resemblance relations with each other. This is a surprising but very tidy result.22 Hume’s theory of how general ideas represent turns out to be an instantiation of his theory of how complex ideas represent. Here it is not spatial relations or temporal relations that are structuring the representations, but

84

Two Objections to Hume’s Theory of Mental Representation

that ought to be expected since it is not spatial or temporal relations that are being represented. Rather, since it is resemblance relations that are being represented (persons are represented as resembling each other), it is resemblance relations that provide the structure. This makes general ideas copies of that which they represent in the same sense that complex ideas are copies of that which they represent. General ideas represent objects as resembling one another by placing representations of those objects into the same relation as the objects are represented as being in: resemblance relations. With that said, we can now return to the line of inquiry we set out to pursue in this section: can Hume’s theory of general ideas be used to disambiguate what is represented by a complex idea?23 As we have now seen, Hume seems to think that it can do exactly this work. He holds that the way to represent a white globe as white is exactly by employing the general idea ‘white,’ which represents that globe as resembling other white items. Likewise for representing the white globe as a globe, etc. To represent an object as of a certain kind, one employs a general idea of that kind. Notice that in order to articulate this point, I have had to leave this formulation vague. What Hume outlines above is a procedure whereby we begin to distinguish the figure from the colour by a distinction of reason; that is, we consider the figure and the colour together, since they are in effect the same and undistinguishable; but still view them in different aspects, according to the resemblances, of which they are susceptible. T 1.1.7.9; SBN 21–2 It is clear that to “view them in different aspects” we use the general ideas ‘white’ and ‘globe’ here, but it is not at all clear exactly how we use them. Pre-theoretically, one might assume that these two general ideas are predicated of the same thing in different judgments, but as we are about to see, that cannot be Hume’s answer at all, or if it is, it can only be so by seriously reconceiving the normal sense of ‘predicate’ (which would not be the kind of move with which Hume is unfamiliar or uncomfortable). This is because Hume argues that the standard model of judgment according to which a judgment is the application of a predicate to a subject is grossly mistaken. According to Hume, this portrays judgment as a conjoining of ideas, when a judgment is actually just a single idea with an appropriately high degree of force and vivacity. This error consists in the vulgar division of the acts of the understanding into conception, judgment and reasoning, and in the definitions we give of them. Conception is defin’d to be the simple survey of one or more ideas: Judgment to be the separating or uniting of different ideas: Reasoning to be the separating or uniting of different ideas by the interposition of others, which show the relation they bear to each other. [. . .]

Two Objections to Hume’s Theory of Mental Representation

85

What we may in general affirm concerning these three acts of the understanding is, that taking them in a proper light, they all resolve themselves into the first, and are nothing by particular ways of conceiving our objects. T 1.3.7.5n; SBN 97 Judgment is not the “uniting of different ideas,” for instance, the application of a predicate to a subject, but is actually just a “particular way of conceiving objects.” Hume is concerned in this section to account for the difference between merely entertaining a certain idea and believing it. Since he takes his predecessors to equate this distinction with one between conception and judgment, he is eager to show that judgment and belief are really nothing more than a way of conceiving. In particular, he holds that a belief may be most accurately defin’d, a lively idea related to or associated with a present impression. T 1.3.7.5; SBN 97 A belief is a single idea with a particularly high degree of force and vivacity, which is usually produced by the presence of a certain enlivening impression. To believe something is to have such an idea. To merely entertain that same thought is to have that idea but with a much lower degree of force and vivacity. So, belief, or judgment, does not consist in the conjoining of ideas, it does not consist of applying a predicate to a subject, but it is rather the having of a certain idea with an appropriate degree of liveliness. On its face, this would seem to leave us in a bit of a quandary. We were wondering how it is that a general idea makes particular complex ideas more determinate. The proposal we were entertaining was that the general ideas would be predicated of the particular complex idea in a judgment. We can now see that, by Hume’s lights, this is an unacceptable account. Judgments do not consist of predicates being applied to subjects. Judgments, or beliefs, are single ideas properly enlivened. So, what we need is a single idea that can do the work that we were doing by predicating general ideas of particular complex ones. Luckily, our recent investigation into the nature of general ideas presents us with just such an idea. Since general ideas are themselves mere instances of complex ideas, and since the particular complex ideas that we were seeking to disambiguate are literal parts of this more complex general idea, the obvious way to proceed is on the assumption that the single idea that we need is just the general idea itself. The thought here is this. Consider again one of our working examples: a complex idea in which one figure is larger than another. It might look like the figure on the next page. The problem that we encountered with such an idea is that while, according to Hume’s theory, it represents the figure on the left as larger than the figure one the right, it also represents that figure being to the right of the

86

Two Objections to Hume’s Theory of Mental Representation

other figure, and the two figures as being congruent, etc. We set out, therefore, in search of a way to disambiguate this representation and found that Hume’s way of doing so involves in some way the use of general ideas. One natural suggestion would be that the general idea is predicated of this particular idea, but we have seen that for Hume this is a mistake. The suggestion now is that the way this picture comes to represent just the one figure’s being larger than the other is simply by being a literal part of the general idea ‘larger than.’ It is simply by being part of this general idea that the particular idea comes to be a representation of one figure’s being specifically larger than the other. Put this way, this suggestion is open to a simple objection: this particular idea is not just part of the general idea ‘larger than.’ It is also, for instance, part of the general ideas ‘next to,’ ‘congruent,’ etc. So, it does not unambiguously represent the figure on the left being larger than the figure on the right because it also represents the figures as being next to one another, being congruent, etc. That is, just as this idea can be the trigger for our recalling other ideas that it resembles in virtue of all of these ideas containing items that are larger than one another, it can equally well be the trigger for recalling different groups of ideas that resemble one another in different ways (by all being pictures with congruent figures, or figures next to one another, etc.). So, merely being part of a certain general idea does not make this idea an unambiguous representation of the figure on the left as being larger than the figure on the right. Since it is also a part of other general ideas, it is equally a representation of those relations as well. Here Hume would surely point out that while such an idea is certainly a part of various general ideas in the way we have described, we have left out a crucial aspect of his account of how general ideas function. Namely, upon hearing a certain term, say ‘larger than,’ we are disposed to call to mind

Two Objections to Hume’s Theory of Mental Representation

87

only the ideas that are associated with that word. So, while the idea at hand might be part of a wide variety of general ideas, in a given context, only one of these general ideas will actually be thought. Upon hearing ‘larger than,’ one has this idea and is disposed to think of the other ideas that together form the general idea ‘larger than.’ One does not stand disposed, in this context, to think of ‘congruent’ ideas or ‘next to’ ideas. Thus, it is the dispositions one has in a certain context that do this disambiguating work. While the idea that we have been considering is a part of lots of general ideas, and so can potentially serve as a picture of lots of different relations, it is disambiguated when one of these general ideas is actually thought. Unfortunately, even this full-dress version of Hume’s account faces a serious problem. Consider the situation we have just described. A person hears the word ‘larger.’ They have an idea like the one we are considering and stand disposed to recall other ‘larger than’ ideas. I.e., they have the general idea ‘larger than.’ The problem here is that, as we have seen, in having the general idea ‘larger than,’ what is represented is not this figure’s being larger than this other figure but rather the resemblance of all of the objects of the ideas that make up this general idea. That is, according to the scheme for representing complexes with which we are working, a general idea represents the objects of its component ideas as being related in the way that those ideas are related in the complex (as resembling). So, in having the general idea ‘larger than’ one no longer represents the figures in the single idea as being related, but rather all ‘larger than’ ideas as representing resembling objects. This is all to say that since we have already determined that what the general idea ‘larger than’ represents is that certain objects resemble one another, simply having this idea cannot also represent the relations internal to these objects. General ideas represent the relations between the objects of their component ideas, not the relations that might so happen to inhere amongst the parts of the objects of those ideas. For instance, what the general idea ‘larger than’ represents is the fact that the objects of pictures like these

Figure 2.2

88

Two Objections to Hume’s Theory of Mental Representation

resemble one another. Representing that resemblance, however, is not to represent the objects of these pictures as bearing any particular relation to one another. What gets represented in the general idea is a resemblance relation between objects, not the relation of one of the parts of these objects (or states of affairs) as being larger than another of its parts. It merely represents the fact of this resemblance, not the way in which they resemble one another, nor the fact that each of these is a representation of one thing’s being larger than another. General ideas are, so to speak, too busy representing objects as resembling to also represent them as being larger than one another, or being congruent, etc. To see this more clearly, consider a general idea the components of which are simple ideas: the general idea of red dots. Each idea of a red dot is an idea of something simple. Furthermore, at least prima facie, the qualities redness and being a dot are, presumably, simple qualities. They are not relational in the sense that for something to be red is not, like something’s being larger than something else, a matter of standing in any relation to any other thing. Notice, though, that the general idea ‘red dots’ is itself an idea of something relational. It is the idea of red dots all standing in a resemblance relation to each other. If, however, this general idea represents something relational, then it cannot thereby represent something nonrelational. That is, what the general idea represents is not the red dot as red, or as a dot, etc. What it represents is that the red dot resembles other red dots, and whatever that earns Hume, it is not the disambiguation for which he had hoped. So, we are still left with the puzzle of how it is that a complex idea can come unambiguously to represent a determinate complex state of affairs. It cannot do this on its own. It cannot do so by merely being a part of a general idea. Of course, this last conclusion is one that we drew after exploring the relation between particular ideas and general ideas within the bounds determined by Hume’s account of judgment, or belief. It is now important to notice the role that those constraints played in drawing that conclusion. That merely having a general idea is not enough to disambiguate the content of a certain particular idea should not be surprising. The general idea has its content; the particular idea has its own. The only reason we found ourselves exploring that possibility, however, is because Hume’s account of judgment demanded that we find a single idea that could represent our two figures as related in particular way. Our first inclination in forming that representation was rather to represent those figures as related by predicating a general idea of the particular idea. This was ruled out only because of the idiosyncrasies of Hume’s theoretical apparatus. Now that we have seen that apparatus fail, however, it is worth revisiting this previous suggestion to see what, if anything, can be made of it on an otherwise Humean line. That is, what we will do next is explore how far we can get toward disambiguating Hume’s complex representations using his theory of general ideas, but also allowing ourselves to employ a theory of judgment that

Two Objections to Hume’s Theory of Mental Representation

89

involves the conjoining of ideas. Part of the motivation for this is to see what the best we can do on Hume’s behalf is, and part of it is to finally bring us back to Kant’s critique of Hume on judgment. The conclusion that Kant will reach via that critique is that in order to give an adequate account of the unity of judgments, one must conceive of them inferentially. As we have already seen, Kant holds that judgments are not merely single ideas, and that they are not mere conjunctions of multiple ideas. Rather they are metarepresentations of intuitions as being related to one another via conceptualinferential rules. Examining this extrapolated version of Hume will allow us to bring out just what Kant’s argument for this conclusion is. Consider then the judgment ‘x is larger than y.’ The first thing to notice about this judgment is that we already have an account from Hume of how many of its components come to represent what they do. ‘x’ will be a picture of x. ‘y’ will likewise be a picture of y. ‘Larger than’ is a general idea that represents certain objects as all resembling one another. What we do not have an account of is ‘is,’ but as Kant points out, there is something strange about taking the copula to represent anything at all. [A] judgment is nothing other than the way to bring given cognitions to the objective unity of apperception. That is the aim of the copula is in them: to distinguish the objective unity of given representations from the subjective. B142 What the copula does in a judgment is not represent any distinct object, but rather it marks the judgment as making a claim about what is the case. Judgments make claims about the way the world is, e.g., that one figure is larger than the other. So, we can transpose the question of what ‘is’ might represent into the question of what brings ‘x,’ ‘larger than,’ and ‘y,’ together into a judgment about what is the case. This question is what is known as the problem of the unity of the proposition, and it was a central philosophical problem of Hume’s day.24 In fact, Hume’s own theory of belief can straightforwardly be read as an attempt to address that problem: if judgments are reducible to single ideas, then there is no problem of the unity of the proposition. The Kantian critique that we are about to encounter centers on and emphasizes the difficulty of solving this problem. As we will see at the close of this section, again, Kant’s inferentialist theory of mental representation is tailor made to do just this. To make the problem of the unity of the proposition salient, we can turn our attention briefly to a twentieth-century analog of this dialectic, after which we will return to Hume and Kant. In its contemporary incarnation, this problem takes the form of the linguistic version of Bradley’s Regress that Wilfrid Sellars presents as an objection to the use of a Platonic metaphysics of abstract entities to account for the meaning of predicate

90

Two Objections to Hume’s Theory of Mental Representation

terms.25 The objection, as Sellars formulates it, is to the claim that predicative expressions come by the meaning they have by referring to abstract entities (or, worse still, that all expressions come by their meaning in this way). For current purposes, it will be terminologically helpful to call this kind of referring expression a name—even at the risk of blurring whatever fine-grained distinctions there may be between these two terms in ordinary use. The first step in the objection to this account of meaning is to notice that whatever a judgment is, it is at least more than a mere list of names. (1) Joan, Judy, Jeffrey, Jessica is not a judgment. Analogously, then, neither is (2) ‘x,’ ‘larger than,’ ‘y.’ To make even clearer that (2) is not a judgment, but a mere list, we can change the order in which the items on the list are placed. (3) ‘x,’ ‘y,’ ‘larger than’ now bears almost no resemblance to anything that we might mistake for a judgment. This mere list of names is quite obviously not a judgment, and simply reordering such lists so that they superficially resemble recognizable judgments cannot make them into such. The point here is that the Platonist’s attempt to account for the meaning of predicate terms as names of abstract entities leaves us in need of an additional account of how such names come together with names of objects to form judgments (rather than mere lists of names). As we have seen, Hume is explicit in eschewing the Platonist’s gambit of casting predicates as names of abstract entities. Instead, he construes general ideas as representations of resemblance relations. What all of this is meant to bring out, however, is that Hume falls prey to the problem of the unity of the proposition nonetheless. This is because while Hume does not cast general ideas as names of abstract entities, what he does cast them as is just as problematic. If ‘larger than’ is a representation of the relations of resemblance between various complexes of items, one of which is larger than the other, then ‘larger than’ functions enough like a name to generate just this problem. It names this complex of objects. The problem of the unity of the proposition, as Sellars’s presentation of it brings out, arises from construing all of the parts of a judgment as functioning as names, and Hume does this just as much as the Platonist does. This is clear in the case of the particular ideas ‘x’ and ‘y’ but is equally true of the general idea ‘larger than.’ ‘Larger than’ represents the resemblance relation amongst pairs of objects, one of which is larger than the other. It has that representational content independently of any of the other representations

Two Objections to Hume’s Theory of Mental Representation

91

in the judgments in which it figures, and so it functions enough like a name to raise this problem. Ultimately, it is also enough to make the problem intractable. As we saw, Hume cannot solve this problem by reducing judgments to single ideas. He also cannot solve it by construing the subject of a judgment as part of the predicate. What we can see now is that he really does not have many resources to address the problem at all. So long as he construes each of the constituents of a judgment as having the content that it does independently of all of the other constituents, he will be faced with the difficulty of uniting these pieces into a whole. Part of the problem here is that Hume places severe restrictions on the resources that he will allow himself to bring to bear on this problem.26 The only connections between ideas that he allows are relations of association: resemblance, contiguity, and a cause and effect. The only modifications of ideas that he allows (while keeping the content of these ideas the same) are in the degree of force and vivacity of such ideas. Unfortunately, neither of these resources provide much in the way of help toward solving the problem of the unity of the proposition. To see this, we can briefly examine each in turn. In a sense we have already seen that the associations of ideas will not be enough to unite ideas into a judgment. These are exactly the relations that Hume posits as joining the components of complex ideas, and so it is no surprise that his theory of judgment is one according to which a judgment just is a single idea (properly enlivened). Given that the only associations between ideas are those that make simple ideas into complex ones, any unity that is effected in a judgment can only be the unity of a complex idea. So, the unity of the proposition, for Hume, is not a unity of multiple ideas but rather the unity internal to a single idea (insofar as there is any such unity). However, as is now clear, complex ideas are not judgments. For Hume, a belief is always a belief that the object of the idea believed exists. This being the case, Hume’s theory of belief fails to account for exactly what we needed it to account for at the start of this section: the classification of complex states of affairs via general terms. A judgment, on this theory, can assert that the object of some general idea (a complex of objects that all resemble one another) exists, but it cannot represent that any particular idea is of such-and-such a kind, or that any particular idea represents such-and-such a complex. Notice that once Hume has limited himself to associations of ideas as providing the only possible structure of a judgment, and is thus compelled to cast judgments as single, complex ideas, it is exactly because the only resource left for him to draw on is degree of force and vivacity that he is then forced to say that all belief is belief in the existence of the object represented. There is nothing much more that varying the degree of force and vivacity of such an idea could do. It is wildly implausible, for instance, to think that such a variation could produce the different forms of judgment (a sample of which we briefly encountered in the previous section). Rather, since Hume’s primary concern is in the distinction between contemplating and believing, supposing that this difference is facilitated by the difference between less and more forceful and

92

Two Objections to Hume’s Theory of Mental Representation

vivacious ideas is a natural choice. Unfortunately, more is needed to capture other important features of judgments, but most of these features are ruled out as explananda by the reduction of judgments to single ideas. Without a solution to the problem of the unity of the proposition (and with it the notion of the various forms of judgment), Hume’s theory is left with no way to remedy two fatal failures. We had hoped that an appeal to general ideas might be able to address the problem of how to form determinate representations of complex states of affairs as complex. Without a coherent theory of judgment, however, there does not appear to be a way to use general ideas for this purpose. We simply cannot get general ideas hooked up with particular ideas in a way that we would need to in order to garner this relief. Furthermore, this failure to be able to produce a coherent account of judgment is a serious flaw in itself. Hume’s theory of mental representation includes his theory of belief, and as we have seen, that theory of belief is incapable of accounting for certain kinds of mental representation that we clearly possess. KANT AND THE PROBLEM OF THE UNITY OF THE PROPOSITION The question now is how to solve these problems. We saw earlier that the theory of concepts as inferential rules that Kant presents purports to be a solution to the first problem: how to make representations of complex states of affairs determinate. Interestingly, this same theory is also a solution to the second problem: the problem of the unity of the proposition. Remember that one way to frame that problem is as how to distinguish a mere list of objects from a judgment about those objects. We saw earlier that this problem arises when we understand each part of a judgment as representing its own kind of object. Thus, each representation seems to function as a kind of name, and a judgment ends up looking a lot like a list of names. Notice, though, that the theory we earlier presented on Kant’s behalf does not follow that formula. For Kant, a concept is not itself a “direct” representation of any object but rather functions within any particular judgment as a rule of inference connecting the intuition that appears as the subject of that judgment to other intuitions that appear in other possible judgments. In our current idiom, an intuition would function something like a name: it represents its object “directly.” A concept, however, does not. It does not represent objects directly but instead relates to them only mediately through intuitions. A concept is “the predicate for a possible judgment” (A69/B94). It licenses, forbids, etc., further judgments “concerning certain appearances that come before us” (A69/B93), intuitions. So, while for Hume, the problem of the unity of the proposition arises with the need to unite, for instance, the representations, ‘x,’ ‘y,’ and ‘is larger than,’

Two Objections to Hume’s Theory of Mental Representation

93

for Kant, there is one too many items on this list. ‘Is larger than’ is not a representation that represents any object. Rather, it is the connection between ‘x’ and ‘y’ that itself makes ‘x is larger than y’ a representation of x as larger than y. It is the structure that itself binds ‘x’ and ‘y’ together into a representation of the complex state of affairs x’s being larger than y. So for Kant, there is a sense in which there is no problem of the unity of the proposition. Judgments are not unities of independently representing items. The application of a concept to an intuition is not the joining of two representations but rather the placement of a single representation (the intuition) into a normatively-inferentially structured system of representations that together represent the complex state of affairs that is the world. (“Judgment is therefore the mediate cognition of an object, hence the representation of a representation of it.”) Thus, the unity of an intuition and a concept in a judgment is not the unity of two independent, object-level representations, but instead it is the unity of, as Kant describes it, a function, and its object. A concept takes an intuition as its input and outputs various other, inferentially related possible judgments. Whatever it takes to do this work, and we will investigate the details of this in the coming chapters, it is clearly not the same as what would be required to unite independently contentful representations, which is the hard core of the problem of the unity of the proposition. As was the case with our previous objection, we do not find Kant raising this objection explicitly as one that should trouble Hume. (Kant rarely mentions his opponents by name in the Critique.) Instead what we find is Kant concerning himself with exactly the problem that Hume confronts and offering his theory as a solution to that problem. As it turns out, we find him doing this in the same bit of text that we examined in the previous section. What I want to suggest is that this is not an accident. That Kant’s theory just so happens to provide solutions to these two problems with Hume’s theory, the theory that he announces is one of the primary target of the Critique, is evidence that he creates this theory, at least in part, with these two objections in mind and so presents it as a simultaneous solution to both. Recall that we saw Kant’s proposal to understand concepts inferentially: a concept is a rule for connecting intuitions appearing in various judgments. In the following quotation, Kant explicitly links that proposal with the problem of the unity of the proposition. It [a concept] is therefore the predicate for a possible judgment, e.g., “Every metal is a body.” The functions of the understanding can therefore all be found together if one can exhaustively exhibit the functions of the unity in judgments. A69/B94 The second sentence here is a dense one, but it will repay our efforts to unpack it. The first thing to notice is that Kant takes this sentence to follow from the preceding one: thus his use of ‘therefore’ (also). So, that the functions of the understanding can all be found together if one can exhaustively

94

Two Objections to Hume’s Theory of Mental Representation

exhibit the functions of the understanding in some way follows from the fact that concepts are predicates for possible judgments. As it stands, the structure of this argument is far from obvious, and this is in no small part because the conclusion itself is rather opaque. The ‘functions of the unity in judgments’ certainly concerns the unity of the proposition, and more specifically indicates, as we have seen, that Kant holds that there is more than one such unity. That is, Kant holds that propositions can be united in various ways, each of which corresponds to a different logical form that a judgment can take. So, the conclusion here is that if one can delineate the functions that serve to determine these various forms, then one can thereby discover the “functions of the understanding.” This last phrase is an important clue because earlier in Metaphysical Deduction, in introducing his discussion of concepts, Kant writes, [T]he cognition of every, at least human, understanding is a cognition through concepts, not intuitive but discursive. All intuitions, as sensible, rest on affections, concepts therefore on functions. By a function, however, I understand the unity of the action of ordering different representations under a common one. Concepts are therefore grounded on the spontaneity of thinking, as sensible intuitions are grounded on the receptivity of impressions. A68/B93 Concepts rest on functions, on the unity of the action of ordering different representations under a common one. We have recently seen how it is that Kant thinks that we order different representations under a common one: we use concepts to inferentially connect intuitions to one another. We represent ‘x’ as larger than ‘y’ by licensing, forbidding, etc., certain inferences from the judgment ‘x is larger than y,’ and thus these instances of intuitions of x and y to others employing further intuitions of them. ‘x’ and ‘y’ thus become ordered under the common representation ‘larger than.’ Concepts rest on inferential connections between judgments. This is the kind of function of unity on which concepts rest. Of course, these functions are, as Kant emphasizes above, functions of the understanding. That is, the understanding just is our faculty for employing such concepts, or for connecting intuitions via judgments in just this way. So, a “function of the understanding” is just a concept. Thus, Kant’s conclusion from the previous text is that if one can exhaustively exhibit the functions of the unity in judgments, then certain concepts can all be found together. Obviously, Kant does not think that we can find all concepts, empirical and a priori alike, simply by looking at the various forms of judgments. Rather, his project here is to abstract from all content of a judgment in general, and attend only to the mere form of the understanding in it. A70/B95

Two Objections to Hume’s Theory of Mental Representation

95

The idea here is that because there are only so many forms that judgments can take, there are also only so many forms that concepts can take. These forms of concepts are the pure a priori concepts of the understanding (these are meta-concepts). They are all the kinds of concepts that can play the role that concepts are meant to play in judgments. The last remaining question about this argument, then, has to be why Kant thinks that there is this tight connection between the forms that a judgment can take and the forms that concepts can take. The answer to that question, however, is given by the first premise of the argument! It is because concepts just are predicates for possible judgments that it follows that the various forms that a judgment can take will determine the various kinds of concepts that there are. Since “the understanding can make no other use of these concepts than that of judging by means of them” (A68/B93), the kinds of concepts that there are will be entirely determined by the kinds of judgments that there are. The kinds of judgments that there are, in turn, is itself entirely determined by the various functions of unity in judgments, i.e., ways that the elements of a judgment can be united. Consider Kant’s example: Every metal is a body. The problem of the unity of the proposition is the question of what joins the representation ‘every metal’ with the representation ‘is a body.’ As we have seen, Kant’s answer to this question is that ‘is a body’ does not represent in the same way that ‘every metal’ does. ‘Every metal’ is an intuition of all metals. ‘Is a body’ serves as an inferential license connecting that intuition and other intuitions, such as ‘this metal’ or ‘some bodies.’ The unity of this judgment is the unity not of two representations that each have their content independently of one another, but rather it is the unity brought about through the action of licensing, forbidding, etc., certain inferences involving a single representation (the intuition ‘every metal’). Kant’s short argument, then, is that since concepts can play this inferential role in various ways within a judgment—the concept ‘metal’ plays an inferential role in his example just as much as does the concept ‘body’; more on this later—judgments can likewise said to be united in all of these various ways. The pure a priori concepts of the understanding just are these various ways that concepts can function. The problem of the unity of the proposition is thus necessarily at the center of Kant’s discovery of the Categories. That discovery, in turn, is the centerpiece of Kant’s refutation of the three conclusions of Hume’s that we have been investigating. It is by discovering these concepts, and the role that they play in our cognitive lives, that Kant is able to systematically replace Hume’s theory of mental representation with his own. It is no accident that that project begins with Kant’s solution to the problem of the unity of the proposition. It is by addressing that problem, along with the problem of complex representation, that Kant is able both to demonstrate the failures of Hume’s theory and to replace that theory with one that is tailor made to rectify them.

96

Two Objections to Hume’s Theory of Mental Representation

So, again, we do not find Kant explicitly aiming such an argument at Hume. He does not proclaim that Hume produced a theory that would have trouble with the problem of the unity of the proposition or that his theory can remedy the failures of Hume’s. Rather, Kant produces a theory that does solve that problem, as well as the problem of complex representation, and then points out that it is that very theory that leads immediately to the possibility of refuting Hume’s three theses. That is, Kant raises a problem that is a problem for Hume’s theory of mental representation, solves that problem by proposing an alternative to that theory, and then uses that new theory to show that the conclusions to which Hume’s theory lead him are false. That form of presentation certainly makes a decent case that Kant can and does think of the problem of the unity of the proposition as an important objection to Hume’s theory of mental representation. If all of this is correct, then we should understand the problem of the unity of the proposition as a significant component of Kant’s critique of Hume’s theory of mental representation and as providing an important criterion of success for his own replacement theory. This, in turn, provides further evidence for understanding Kant as holding the kind of inferential theory of concepts that we outlined earlier because as we have just seen, this inferential theory of concepts is particularly well suited to address precisely that problem. By way of demonstration of this latter point, we can contrast that reading of Kant’s theory of concepts with that which Longuenesse presents in her excellent book, Kant and the Capacity to Judge. Longuenesse also attributes to Kant what she takes to be a radically new theory of concepts, which is similar in many ways to the one that I have proposed: most importantly, we both take concepts to have as their essential function serving as rules. Longuenesse’s account is, however, different in one crucial respect. I hold that the problem of the unity of the proposition motivates Kant to abandon the thesis that concepts represent their own special kind of object in favor of the view that concepts represent intuitions as related to one another, thereby forming a picture of the world in which the elements of the pictures are intuitions, and its structure is the inferential structure provided by concepts. On this line, concepts do not represent any objects of their own but represent the objects represented by intuitions as being related to one another. On Longuenesse’s line, by contrast, concepts do have their own particular object. Longuenesse proposes that concepts represent what she calls immanent universals. It will be worth following along with her presentation of this issue in order to bring out the significance of this difference between our two views. Here is the beginning of Longuenesse’s discussion of what is represented by the concepts deployed in experience.27 The “rule of apprehension” of which the concept is the universal representation is both immanent to the sensible, singular representation, and generated by the act of comparison. [. . .] The determination of the concept will result from the act of comparison, but the concept must already

Two Objections to Hume’s Theory of Mental Representation

97

be present in an “undetermined” state, that is, in an intuitive state, or more precisely, as a still unreflected, “obscure” rule for the synthesis of intuition. This is why apprehension is said to be “universal in itself”: it is universal insofar as the rule is already present there, in an unreflected state. It is universal in itself insofar as universality is immanent to it.28 Longuenesse and I agree that concepts are, or have as their essential cognitive function serving as, rules (of apprehension). Longuenesse, though, additionally holds that concepts can only be such rules if the understanding discovers these rules in objects themselves. That is, Longuenesse holds that concepts-qua-rules are only possible insofar as they are representations of immanent universals that exist unreflected and potentially in the objects of representation. This comes out in her discussion of Kant’s example of the so-called savage who encounters a house for the first time, where Longuenesse claims that his coming to have the concepts that he does is a “gradual dawning”: he slowly and by piecemeal discovers the universal (house-hood) that is immanent in the house all along. It is significant that Longuenesse claims that “the concept must already be present in an ‘undetermined’ state, that is, in an intuitive state.” While I will agree with Longuenesse that intuitions are conceptually structured, Kant is repeatedly clear that he assigns entirely distinct representative functions to intuitions and concepts. All intuitions, as sensible, rest on affections, concepts therefore on functions. By a function, however, I understand the unity of the action of ordering different representations under a common one. Concepts are therefore grounded on the spontaneity of thinking, as sensible intuitions are grounded on the receptivity of impressions. A68/B93 As I have cast this distinction, it is that intuitions represent objects directly, while concepts represent these objects as bearing certain relations to one another by relating representations of these objects to each other. What Longuenesse is claiming here is that concepts, in fact, function much more like intuitions than one would have expected. They, too, represent a kind of object, and apparently one that is itself the object of a kind of perception. For Longuenesse, what a concept represents is the universal immanent to the object of representation, and it does so by discovering this universal from an intuitive (perceptual) state that contains it. (Longuenesse does not address texts such as B134—“Combination does not lie in the objects, however, and cannot as it were be borrowed from them through perception”— where Kant appears to explicitly deny just this view.) Longuenesse continues by pointing out how this interpretation of Kant makes him radically different from his empiricist predecessors. As we have, she begins by comparing the thesis that she attributes to Kant to Hume’s theory of general representation, and then to Locke’s theory of abstract ideas.

98

Two Objections to Hume’s Theory of Mental Representation On the contrary, if for Kant too the universality of the concept lies in its use, that is, in its “application in comparison,” universality is nevertheless its proper form, and this form is adequate to what it represents— not a singular image, but a rule of apprehension, the generality of which results from the act of comparison.29

Here we see the same schema that we have recently been critiquing: while intuitions (or names, or ideas, etc.) represent “singular images,” concepts play a similar role but with respect to a different, but still singular, object of representation, here a “rule of apprehension.” Intuitions represent objects; concepts represent the universals that are immanent in them. This is precisely the position in which I have suggested Kant recognizes the fault and subsequently abandons. Just as Plato takes ‘Socrates’ and ‘wise’ both to be names—of a person and of the Form of wisdom—and just as Hume takes an idea of Socrates to represent Socrates and an idea of wisdom to represent the resemblance of all things called ‘wise,’ so Longuenesse’s proposal is that Kant takes both intuitions and concepts to represent their own kind of objects: objects proper and immanent universals.30 That this is Longuenesse’s approach is made even clearer in her comparison of Kant and Locke. But actually Kant goes well beyond reaffirming a Lockean position on universals. Locke denied any reality to universals: what is general or universal is not in things, but only in how we think them. For Kant, on the contrary, to say that concepts are strictly discursive (that they are never given as to their form, but can only result from acts of the understanding) does not mean that they do not represent anything real in things. [. . .] For Kant on the contrary, universals do indeed belong to the existence of things (they represent resemblances lending themselves to “rules of apprehension”), but are revealed in things only by the acts of “comparison, reflection, and abstraction” of the understanding.31 Longuenesse is certainly right to point out that Kant’s thesis that concepts must be constructed by the understanding does not imply that concepts do not reflect anything real about the empirical world. The question is how we understand what is so reflected by concepts.32 What is significant here, however, is Longuenesse’s commitment to the reality of these universals themselves. They “represent [something] real in things,” “belong to the existence of things,” and “represent resemblances.” On the interpretation that Longuenesse advances, Kant’s rejection of Locke’s nominalism about universals consists precisely in his acceptance of the thesis that universals are a kind of metaphysical entity that have a real existence in objects and that are the direct object of representation of concepts.33 Here is a very quick argument that Kant can hold no such position. Recall that we have already seen Kant tell us that the matter of appearance is that which corresponds to sensation.

Two Objections to Hume’s Theory of Mental Representation

99

I call that in the appearance which corresponds to sensation its matter, but that which allows the manifold of appearance to be ordered in certain relations I call the form of appearance. A20/B35 We do not represent Longuenesse’s immanent universals through sensation. So it follows that they cannot be the matter of experience. That fits with what Longuenesse points out in her discussion of Hume: universality is the form of a concept. The problem is that Longuenesse also claims that this form is “adequate to what it represents,” and she takes it to represent an immanent universal. So, a concept does also represent a certain matter. It discovers in an “intuitive state” the immanent universals that “belong to the existence of things.” Universality is the form of a concept, and for Longuenesse, it is in virtue of having this form that a concept represents its own special kind of object, an immanent universal, so this immanent universal is also the matter of the concept: it is that which the concept represents. In the very passage that Longuenesse is discussing from the Logic, though, Kant is explicit that what concepts represent is not any distinct kind of entity of their own but only the form that the objects of experience take. In every cognition we must distinguish matter, i.e., the object, and form, i.e., the way in which we cognize the object. If a savage sees a house from a distance, for example, with whose use he is not acquainted, he admittedly has before him in his representation the very same object as someone else who is acquainted with it determinately as a dwelling established for men. But as to form, this cognition of one and the same object is different in the two. With the one it is mere intuition, with the other it is intuition and concept at the same time. Ak 9: 33; Logic, 544–545 This passage is one of the infamous places in which how ones parses the ambiguity of ‘intuition’ makes a significant difference to how one understands Kant’s theory of mental representation, but putting those issues aside, what is clear here is that a concept does not represent any object of its own. Intuitions represent objects, whereas concepts represent the form that those objects take. It is not that Kant and the savage each represent there being two things in their immediate vicinity: an object and the universal house-ness in one case, and an object and, e.g., the universal structure-hood in the other. Both represent there being one object, even though they represent this object as having different forms in each case. It is a radical mistake, however, to take these forms to be their own kind of metaphysical entity, immanent or not. Here is yet another argument for that thesis. Recall that we earlier saw that Kant understands concepts as functioning as a kind of meta-representation. Since no representation pertains to the object immediately except intuition alone, a concept is thus never immediately related to an object, but

100

Two Objections to Hume’s Theory of Mental Representation is always related to some other representation of it (whether that be an intuition or itself already a concept). Judgment is therefore the mediate cognition of an object, hence the representation of a representation of it. [. . .] All judgments are accordingly functions of unity among our representations, since instead of an immediate representation a higher one, which comprehends this and other representations under itself, is used for the cognition of the object. A68–69/B93–94

We also saw that Kant takes this role that concepts play in judgments to exhaust their representative function. Concepts are predicates for possible judgments, and the understanding can make no use of a concept other than judging by means of it. The trouble for Longuenesse is in accounting for both of these facts about concepts. For Longuenesse, while a concept does provide a rule of apprehension, that is merely its form. The representational content of a concept, according to her understanding of Kant’s theory, is an immanent universal. Representing an immanent universal, however, does not make a concept a meta-representation. A concept is not a “representation of a representation,” but it is rather a representation of an imminent universal. As such, while Longuenesse’s casting of a concept as a rule of apprehension does prevent the understanding making any use of it other than judging by means of it, if what a concept represents is an entity that it discovers in an intuitive state, there is no reason why this something must be represented by a concept at all. To return to Longuenesse’s discussion of Kant’s empiricist predecessors, if Kant rejects Locke’s nominalism about universals in the manner that Longuenesse suggests, there is nothing preventing him from adopting the position that it is the universals immanent to mental representations that represent the universals immanent to their objects. Kant has no reason to bother with concepts as rules of apprehension at all when there is a much more direct way to make the form of a concept “adequate to what it represents.” Thus I believe that we must reject Longuenesse’s interpretation of Kant’s theory of concepts for several reasons. Concepts do not represent any matter of their own but only the form of appearance. If concepts did represent entities of their own, Kant would be left with no solution to the problem of the unity of the proposition. Concepts are meta-representations that relate to objects only indirectly by representing intuitions. The alternative that I have outlined can meet all of these conditions. Concepts do not represent any matter of their own because they instead represent the objects of the intuitions to which they are applied as standing in certain relations to one another. As such, concepts are meta-representations: they relate intuitions, which are “direct” representations of objects, to one another to thereby form a picture of the world. Finally, this theory of concepts-qua-inferential-rules avoids the problem of the unity of the proposition. Because the representational function of a concept is of an entirely different kind than that of an intuition—it is not “namelike”;

Two Objections to Hume’s Theory of Mental Representation

101

it does not represent its own object—the question of what makes a judgment different than a list of names does not arise for Kant’s theory. A judgment is an intuition linked to another intuition through the concept that serves as an inferential rule relating the two. (A concept is a rule; a rule is the assertion under a condition; the condition is a condition for constructing a syllogism.) CONCLUSIONS Here, then, is where we stand dialectically. Hume uses his own theory of mental representation—an extrapolation of the Representational Copy Principle—to establish the three most important conclusions of the Treatise. (NC) We have no idea that is an idea of a necessary connection, (EW) We have no idea that is an idea of the external world, and (SSE) We have no idea that is an idea of a single subject of experience persisting through time. I have suggested that Kant’s first order of business in rejecting these conclusions must therefore be to offer grounds for rejecting this key premise of Hume’s: his theory of mental representation. We have now seen two arguments meant to do just this, as well as a sketch of how Kant’s own theory of mental representation is specifically tailored to remedy these deficiencies in Hume. The first of Kant’s objections is that Hume’s theory is inadequate to the task of determinately representing complex states of affairs as complex. The second objection is that Hume’s theory of belief is unable to account for, or explain away, the unity of the proposition. The sketch that we have seen of Kant’s normative-inferentialist theory seems to hold great promise in addressing exactly these issues. Thus, the way is open for Kant to replace Hume’s theory with his own and to show that by doing so he can earn back the representations that Hume was forced to reject (as articulated in the three theses above). Of course, what we have seen of Kant’s theory has only been a very rough sketch, or a promissory note. Since our ultimate goal is to determine whether and how Kant’s theory really does adequately account for these representations, our first order of business must be to delve into the details of that theory to see exactly how it is supposed to work. What we know is that Kant will explain both the unity of the proposition and the nature of determinate complex representation by appealing to the nature of judgment and that he thinks of judgments as the vehicles by which intuitions come to be normatively-inferentially connected with one another. It makes sense, then, to begin our investigation of Kant’s theory with what at least seems to be its most fundamental part: the intuition.34 It is the nature of intuitions that occupies much of Kant’s attention in the A-version of the Transcendental Deduction, and so that will be the focus of the next chapter.

102

Two Objections to Hume’s Theory of Mental Representation

NOTES 1. Of course, there is a substantial debate in the secondary literature over this point with some interpreters taking the premise to be one about the self, others taking it to be about experience, and others still taking it to be about worldly objects. See Guyer, Kant and the Claims of Knowledge, 77 for a catalog of then recent such approaches. Guyer there also excoriates Kant for using as premises in the Deduction assertions of the necessity of certain synthetic a priori truths, which he notes makes the form of Kant’s argument particularly unconvincing. I must postpone a full discussion of this issue to the next chapter where we will see that Kant’s premises are meant to be analytic not synthetic, but it must suffice for now to note again that either way, since the Deduction occurs after much of the argumentative work against Hume has taken place, it is not surprising to find Kant there beginning with premises that Hume would not accept. 2. Again, we have to be careful here. This is the negative side of the Humean coin. The complementary positive side is that these ideas and propositions, rather than having the controversial content that his predecessors had supposed them to have, actually have the benign content that he delineates in the constructive parts of the Treatise. E.g., our idea of causation is an idea of constant conjunction; our idea of the self is an idea of a bundle of perceptions. 3. Ak 4:260; Theoretical Philosophy, 57. 4. A treatment of each of these issues—the nature of representing a complex as complex and the problem of the unity of the proposition—is likewise at the center of Dickerson, Kant on Representation and Objectivity, in which Dickerson offers a detailed reconstruction of Kant’s argument in the B-Deduction. In fact, as in the current study, Dickerson argues that these two problems are the primary issues that Kant wishes to address in presenting his theory of mental representation. Dickerson takes these two problems to be different versions of the same problem. In summarizing his own account, Dickerson writes, at the core of the B-Deduction is a problem—the problem of making intelligible the unity of complex representations—that is the representationalist parallel of the semantic problem of the unity of the proposition. (Dickerson, Kant on Representation and Objectivity, 1) Putting aside for the moment Dickerson’s account of Kant as a representationalist, what Dickerson is claiming more generally here is that these two problems differ only in that the question of complex representation is a question about mental representation, whereas the problem of the unity of the proposition is a question about linguistic representation (Dickerson, Kant on Representation and Objectivity, 2). While both problems do concern kinds of unity, and specifically kinds of unity of complex representations, Dickerson’s running together of the two problems in this way obscures differences between them and, importantly, the fact that each problem has both a mental and linguistic version. That is, as we are about to see, Kant is concerned with both the question of how human beings form complex mental representations of complex objects as complex and how they form judgments: roughly the mental parallel of at least a certain kind of proposition. That is, Kant articulates and addresses both of these problems as they pertain specifically to mental representation, and he offers importantly different, but related, answers to each of them. (The difference is the difference between Kant’s account of the structure of intuitions, on the one hand, and judgments, on the other. The intimate relation between these two

Two Objections to Hume’s Theory of Mental Representation

5.

6.

7.

8.

9.

103

kinds of mental representation and the differences between them will be the subject of Chapters 3 and 4.) So, Dickerson’s claim that these are two versions of the same problem, one linguistic, the other mental, cannot be correct. Allison, Custom and Reason also notices that Hume is just as much a target of Kant’s arguments in the Aesthetic as Leibniz, Newton, and Locke, who are typically cast as Kant’s main antagonists there. He also notes the role that Hume’s struggle with complex representation plays in those arguments, although his focus is more on the unity of such representations than their determinacy. That unity will be the focus of the following two chapters here. Kant’s version of inferentialism has received a good deal of attention in recent years, beginning with Sellars, Science and Metaphysics, and then prominently again with Brandom, Tales of the Mighty Dead and McDowell, Having the World in View. It is worth noting that this example could also be used as counterexample to the necessity half of Hume’s theory. I.e., since a musical stave is structured by spatial relations, but what it represents is musical relations, then it is not necessary that a representation be structured in the same way as that which it represents. Of course, Kantian intuitions represent objects as complex, so Kant has to give an account of how that works too. Suffice it to say that the story of how an intuition represents its complex object will be parasitic on the story of how multiple intuitions are combined to represent complex states of affairs as complex. That is, in the next chapter, we will see that intuitions themselves are conceptually structured. One of the more important disanalogies between how intuitions relate to one another and how intuitions are internally structured is that, while intuitions are themselves representations of determinate objects, the representations that are the constituents of intuitions— sensations—are not. Also compare the following three passages, from the Critique, the Logic, and the Notes and Fragments respectively, each of which makes what appears to be the same claim but which switch between the idiom of concepts and that of rules. All cognition requires a concept, however imperfect or obscure it may be; but as far as its form is concerned the latter is always something general, and something that serves as a rule. (A106) All cognition, and a whole of cognition, must be in conformity with a rule. (Ak 9:134; Logic, 628) The principle: Everything that is thought stands under a rule, for only through the rule is it an object of thinking. (Ak 17:660, 4678; Notes and Fragments, 168)

In the first passage Kant’s claim is that all cognition requires a concept that serves a rule. In the second and third, this claim becomes simplified to all cognition must be in conformity with a rule and that all thought stands under a rule. Kant does not hesitate to interchange ‘concept’ for ‘rule’ in passages that are otherwise strikingly similar because he takes concepts to be rules, or at least to function by “serving as a rule.” Many more such passages could be provided. 10. See also “If the understanding in general is explained as the faculty of rules, then the power of judgment is the faculty of subsuming under rules, i.e., of determining whether something stands under a given rule (casus datae legis) or not.” (A132/B171) 11. As we will see in subsequent chapters, it is not a mere accident that Kant here chooses a concept that is related to the concept ‘body’ by way of natural law.

104

12. 13. 14.

15. 16.

17.

18.

19.

20. 21.

22.

Two Objections to Hume’s Theory of Mental Representation As it will turn out, the specific function of concepts-qua-inferential-rules will be to relate intuitions to one another in a way that forms a picture of the necessary connections between worldly objects. Walker, Kant, 79. Walker, Kant, 79–80. Walker goes on to suggest that Kant only fully worked out the consequences of the latter view later in his career and outlines a threestage history of the development of Kant’s thought on this topic. Cf. “The given intuition must be subsumed under a concept that determines the form of judging in general with respect to the intuition, connects the empirical consciousness of the latter in a consciousness in general, and thereby furnishes empirical judgments with universal validity; a concept of this kind is a pure a priori concept of the understanding, which does nothing but simply determine for an intuition the mode in general in which it can serve for judging.” (Ak 4: 300; Theoretical Philosophy, 94) It must suffice for now to say that ‘directly’ here is not equivalent to ‘without the use of concepts’ but rather signals only that no cognition, which is a very particular kind of representation, stands between an intuition and its object. In fact, Kant is explicit about just this point. “The same function that gives unity to the different representations in a judgment also gives unity to the mere synthesis of different representations in an intuition, which expressed generally, is called the pure concept of understanding.” (A79/B104) It is interesting to note here that this is precisely the problem that moves Plato to move from the sensible world to the world of the Forms: nothing in the sensible world can perfectly represent, for instance, the large because all such things will also represent the small. It is worth noting here that a Lockean account of concepts according to which a concept is an abstract general representation can also claim this advantage over Hume. That will be because for the Lockean, each complex state of affairs will be represented by its own abstract general idea. The next section, on the unity of the proposition, will constitute, in part, an argument against such a position. A fuller discussion of the specific function of such inferential relations must be postponed to Chapters 4 and 5 where the object of our focus will be precisely what is represented by the pictures formed by the relating of intuitions to one another inferentially, and how different pictures so formed can relate to one another. In brief, Kant’s fullest picture of mental representation is that such pictures represent the world as subject to causal laws and that the imperative to change the inferences that structure these pictures is part of the content of the Categories of ‘Relation.’ T 1.1.7.4–6; SBN 19–20. There is a very serious concern here, much discussed in the recent literature on Hume, about what it means to say that we find such items to resemble one another. On the one hand, resemblance might be a real relation between these objects that we discover by encountering them. That goes against Hume’s explicit and strong commitment to the thesis that there are no real relations among objects. On the other hand, then, it might be that ‘resemblance’ merely names whatever associations we form amongst ideas of such objects. ‘Contiguity’ would name a different such association, and ‘cause and effect’ a third. Both of these positions have their attendant problems, but there is no need to settle the issue here. Baxter, “Abstraction, Inseparability, and Identity” presents the worry that stems from understanding resemblance in the first way. Garrett, Cognition and Commitment provides a means for reading Hume the second way. Schafer, “Hume’s Unified Account of Mental Representation” presents a different way of unifying the seemingly different ways that complex ideas and general ideas represent their objects.

Two Objections to Hume’s Theory of Mental Representation

105

23. Here one might be inclined to object that the answer is obviously ‘no’ on the grounds that since a general idea just is a complex idea, whatever defects there are in the latter must therefore also be present in the former. While I suspect that something like this is true, it will be worth our while to put aside this worry for the moment in order to explore the (very popular) proposal at hand in spite of it. 24. See Gibson, From Naming to Saying. 25. Sellars, “Naming and Saying.” 26. Attempts have been made in recent philosophical discussions of this problem to rely on syntax as that which joins otherwise independent representations in a judgment. See Gaskin, Unity of the Proposition. 27. Longuenesse is here writing about the conceptual structure internal to an intuition, but another thesis that she and I share is that this structure is identical to the structure that relates intuitions to one another via judgments. This will be the central topic of the next chapter, where I will again have occasion to contrast the details of my working out of this thesis with Longuenesse’s. 28. Longuenesse, Kant and the Capacity to Judge, 118. 29. Longuenesse, Kant and the Capacity to Judge, 119. 30. One might be tempted here by the following line. Every predicate of the form ‘is F’ can be nominalized into a subject term of the form ‘F-ness.’ Immanent universals can be cast as the semantic value of such nominalized predicate terms. Thus, Longuenesse could earn, in a deflated sort of way, both immanent universals and a reference-like function for concepts without succumbing to the objections that I have and will level against her view. Such a line is all well and good, but the important insight that Kant has about such machinations is precisely that the cash value of such a combination of such nominalized predicate terms and their corresponding immanent universals will always be parasitic on the role that the predicate plays in its original, ‘is F’ form, i.e., as a rule of inference. Terms such as ‘property’ and ‘universal’ are best understood not as naming a special kind of metaphysical entity but rather as meta-conceptual sortal shorthands for ‘is a predicate term.’ E.g., ‘F-ness is a property’ is given cash value by “ ‘is F’ is a predicate”. As we will see in the following chapters, Kant takes a parallel line on ‘object,’ which he thinks of as first and foremost in terms of the meta-conceptual sortal ‘object-concept,’ the criteria for which are given by the Categories. 31. Longuenesse, Kant and the Capacity to Judge, 120. 32. I will argue in Chapter 4 that it is by relating intuitions to one another via inferential rules that intuitions so structured picture objects in the world as being governed by causal laws. The crucial difference between my claim and Longuenesse’s is that I do not hold that concepts represent laws but rather that it is by relating intuitions to one another via concepts that we represent the objects represented by those intuitions as lawfully related. I.e., only intuitions function as elements in our picturing the world, and so only they have objects. 33. Another point worth noting here is that while Longuenesse presents how Kant differs radically from his empiricist predecessors, she does not mention any other figures from the history of philosophy, despite the fact that the subject of immanent universals dates back to at least Aristotle. (It is no coincidence that the version of the problem of the unity of the proposition that I presented earlier was aimed at a twentieth-century Platonist after all.) 34. As we will see in the following chapters, things are not nearly as simple as this. While there is a sense in which the intuition is fundamental for Kant, there is also a sense in which it is sensations, the matter out of which intuitions are built, that are fundamental, and yet another sense in which it is judgments themselves that are. Part of the work of the next two chapters will be to sort out these different senses.

106

Two Objections to Hume’s Theory of Mental Representation

REFERENCES Allison, Henry. Custom and Reason in Hume: A Kantian Reading of the First Book of the Treatise. Oxford: Clarendon, 2008. Baxter, Donald. “Abstraction, Inseparability, and Identity.” Philosophy and Phenomenological Research 57 (1997): 307–30. Brandom, Robert. Tales of the Mighty Dead. Cambridge, MA: Harvard University Press, 2002. Dickerson, A. B. Kant on Representation and Objectivity. Cambridge: Cambridge University Press, 2004. Garrett, Don. Cognition and Commitment in Hume’s Philosophy. Oxford: Oxford University Press, 1997. Gaskin, Richard. The Unity of the Proposition. Oxford: Oxford University Press, 2009. Gibson, Martha. From Naming to Saying. Oxford: Blackwell Publishing, 2004. Guyer, Paul. Kant and the Claims of Knowledge. Cambridge: Cambridge University Press, 1987. Hume, David. A Treatise of Human Nature. Edited by L. A. Selby-Bigge. New York: Oxford University Press, 1974. Hume, David. A Treatise of Human Nature. Edited by David Fate Norton and Mary J. Norton. New York: Oxford University Press, 2000. Kant, Immanuel. Lectures on Logic. Translated and edited by J. Michael Young. Cambridge: Cambridge University Press, 1992. Kant, Immanuel. Critique of Pure Reason. Translated by Paul Guyer and Allen W. Wood. Cambridge: Cambridge University Press, 1998. Kant, Immanuel. Theoretical Philosophy After 1781. Edited by Henry Allison and Peter Heath. Translated by Gary Hatfield, Michael Friedman, Henry Allison, and Peter Heath. Cambridge: Cambridge University Press, 2002. Kant, Immanuel. Notes and Fragments. Edited by Paul Guyer. Translated by Curtis Bowman, Paul Guyer, and Frederick Rauscher. Cambridge: Cambridge University Press, 2005. Longuenesse, Beatrice. Kant and the Capacity to Judge. Translated by Charles T. Wolfe. Princeton: Princeton University Press, 1998. McDowell, John. Having the World in View. Cambridge, MA: Harvard University Press, 2009. Pippin, Robert. Kant’s Theory of Form. New Haven: Yale University Press, 1982. Schafer, Karl. “Hume’s Unified Account of Mental Representation.” European Journal of Philosophy 21 (2013). doi: 10.1111/ejop.12022. Sellars, Wilfrid. “Naming and Saying.” Philosophy of Science 29 (1962): 7–26. Sellars, Wilfrid. Science and Metaphysics. Atascadero, CA: Ridgeview Publishing Company, 1967. Walker, Ralph. Kant. London: Routledge & Kegan Paul, 1978.

3

The A-Deduction and the Nature of Intuitions

In the previous chapter we uncovered one of the theses at the center of Kant’s arguments against Hume—that all representation of a complex state of affairs is conceptual—and I proposed that we understand this claim by combining it with what I take to be Kant’s theory of concepts-quainferential-rules. Over the course of the next four chapters I will strengthen the case for this understanding of Kant’s theory of concepts as well as show how it can be used to solve certain longstanding puzzles in Kant interpretation and stand on its own as a powerful account of the nature of human thought. I will begin here with the thorny issues surrounding the nature of the representations that Kant calls ‘intuitions’ and ‘sensations’ and their relation to one another. This is because what we have just seen is that two of Kant’s most important arguments against Hume are the immediate dialectical preface to Kant’s introducing his radically new theory of concepts-quainferential-rules. This theory is meant to provide the means to accounting for both the nature of representations of complex states of affairs and for the nature of judgments. To this point, however, we have only had a brief sketch of what that theory is, and the time has now come to start delving into its details. As we noted in the previous chapter, there is a sense in which Kant’s theory is of a piece with Hume’s: they are both theories according to which complex states of affairs are represented as such by forming pictures of such states of affairs. Hume’s pictures were composed of simple ideas related to one another by the very same kinds of relations that were represented by such pictures. In the very brief sketch that we gave in the previous chapter, Kant’s pictures had as their elements intuitions, and as their structure normative-inferential counterpart relations. The purpose of the current chapter will be to begin to elaborate on and clarify that brief sketch. In particular, it will be to detail the nature of these intuitions, which it will turn out, are not the most elementary form of representation for Kant, and which in fact have a conceptual structure of their own. In fact, this last claim—that intuitions are conceptually structured—is a matter of great controversy in the contemporary literature on Kant, especially among those who interpret Kant’s project as first and foremost one

108

The A-Deduction and the Nature of Intuitions

concerning the nature of representation. This debate proceeds simultaneously along two main lines, both of which can largely trace their origins to the twentieth-century philosopher Wilfrid Sellars. The first line is one that focuses its attention mainly on philosophical questions concerning the nature of perception and the role that concepts play in experience. This line intersects with Kant’s texts insofar as Kant engages with those very same philosophical issues and so is cast as an early proponent of one view or another, cited as providing evidence for these views, or looked to as providing some important insight into the issue. The second line is one that focuses its attention mainly on exegetical questions that emphasize understanding what Kant’s view on these questions is over looking to Kant for help in settling the contemporary debates. In many ways, it was Sellars who first introduced both the philosophical questions as we understand them today and the possibility of seeking answers, or at least guidance, to these questions in Kant, and so it is with a “return to Sellars” that I will approach the issue here.1 Sellars’s proposal is that intuitions, which are the representational end product of the perceptual process, are best understood as having the form ‘this-such.’ The ‘this’ indicates the role that our forms of intuition play in perception: they allow us to locate demonstratively the objects of perception from our perspective on them in space and time.2 The ‘such’ indicates that intuitions represent their objects as falling under some concept, that they are essentially conceptually structured. This is where the controversy begins along both of the lines mentioned above. On the one hand, philosophers of mind have objected to the philosophical implication of this view— that perception is essentially conceptual—on the grounds that it detaches perception, and thereby thought in general, from any worldly constraint.3 On the other hand, Kant scholars have objected to the exegetical claim— that Kant held that intuitions are essentially conceptual—on the grounds that Kant’s texts seem to draw a firmer distinction between the contributions of sensibility and understanding than this account of “hybrid” intuitions allows.4 Building on the work of the previous two chapters, what I will do here is develop Sellars’s original proposal in a way that can avoid both of these kinds of objections to it. To do this, I will turn my attention to the A-Deduction, where Kant is most explicit and detailed about his theory of perception. Before doing that, though, a word is order about the place of the Deduction in the Critique, its broad argumentative structure, and the role of Kant’s inferential theory of concepts in it. I will also briefly touch on what I take to be the relation of the A-Deduction, which is the main focus of this chapter, and the B-Deduction, which will be the main focus of the next. With that outline in place, the following section will be another short prelude, this time presenting textual evidence for attributing to Kant the thesis that any representation of a complex state of affairs as complex, including representations of a complex objects, must be conceptually structured. This

The A-Deduction and the Nature of Intuitions

109

thesis is essential to everything that follows, and so it will be important to have Kant’s commitment to it firmly established before proceeding. After that, we will turn our attention to the details of the A-Deduction, and using the complex representation thesis, we will extract from it an account of the nature of intuition and the role played in intuition by what Kant calls sensations. These two sections will constitute, at least in part, an adjudication of the exegetical debate about Kant on perception. It is at this point that we will turn our attention to a specific and important problem that arises for any approach to the relation of sensations to intuitions like the one I will detail here. Roughly, this is the problem of how non-conceptual representations can be united according to concepts-qua-inferential-rules, when they are not apperceived and they do not appear in judgments. Robert Pippin raises this issue against Sellars’s approach in his excellent book, Kant’s Theory of Form, and we will follow his presentation of it there. I will then present my solution to this problem and contrast it with the solution of Beatrice Longuenesse, whose approach to perceptual synthesis is otherwise very much like the one I will present here. With this problem addressed, the final section will then show that it is what Kant calls sensations that are the non-conceptual representations that are structured by the conceptualinferential activities of the understanding in the formation of an intuition, and that recognizing this provides the needed resources for at least beginning to adjudicate the philosophical debate between conceptualists and nonconceptualists outlined above. THE ARGUMENT OF THE TRANSCENDENTAL DEDUCTION As is well known, Kant’s Transcendental Deduction is his attempt to answer a certain quid juris.5 As Kant tells us, Jurists, when they speak of entitlements and claims, distinguish in a legal matter between questions about what is lawful (quid juris) and that which concerns the fact (quid facti), and since they demand proof of both, they call the first, that which is to establish the entitlement of the legal claim, the deduction. A84/B117 Kant’s Deduction concerns a kind of entitlement: the entitlement that creatures like us have to the use of a particular kind of concept: pure a priori concepts. These concepts can be helpfully contrasted with empirical a posteriori ones, which are concepts the content of which is derived from experience— in a sense with which we will not concern ourselves here, but which essentially involves connecting to the sensations that we will have reason to discuss farther along in this chapter—and the justification of the use of which is conducted via an appeal to this pedigree.

110

The A-Deduction and the Nature of Intuitions We make use of a multitude of empirical concepts without objection from anyone, and take ourselves to be justified in granting them a sense and a supposed signification even without any deduction, because we always have experience ready at hand to prove their objective validity. A85/B117

Empirical concepts are, more or less, the concepts with which Hume is most comfortable. They are concepts whose origins can be traced up to experience (although perhaps not in the straightforward way that Hume thinks they can). Pure a priori concepts, on the other hand, are those concepts that are not derived from experience and the justification of the use of which cannot, therefore, be a posteriori. It is Kant’s goal in the Transcendental Deduction to provide an a priori justification of such concepts. Among the many concepts, however, that constitute the very mixed fabric of human cognition, there are some that are also destined for pure use a priori (completely independently of all experience), and these always require a deduction of their entitlement, since proofs from experience are not sufficient for the lawfulness of such a use, and yet one must know how these concepts can be related to objects that they do not derive from any experience. A85/B117 Pure a priori concepts are the kinds of mental representations that we have seen Hume argue that we cannot possibly have. Kant, on the other hand, believes that we do make use of such representations, and the goal of the Transcendental Deduction is to justify this use. Since the use to which such concepts—and all concepts—are put is to be applied to objects, what must be shown in this deduction is that applying pure a priori concepts to objects is justified. The way that Kant sets out to demonstrate this justification is by showing this activity is an essential part of another activity, which is itself justified.6 Suppose, for instance, that I have been given permission to play baseball. Now suppose that the question arises whether I have permission to take an at bat. By showing that taking an at bat is an essential part of playing baseball, I thereby show that I have permission to take an at bat. That is, one could not have permission to play baseball without also having permission to take an at bat.7 Similarly, Kant’s plan in the Deduction is to show that the use of pure a priori concepts is an essential part of another activity that is itself justified and thereby to secure justification for the former activity. The activity of which Kant takes the employment of pure a priori concepts to be an essential part is the activity of conceiving of some representations as belonging to oneself. That is, Kant argues that employing pure

The A-Deduction and the Nature of Intuitions

111

a priori concepts is necessary for conceiving of one’s thoughts as one’s own. Here is Kant, early on in the B-Deduction, explicitly stating this as the condition to which he will appeal during the course of the Deduction. For the manifold representations that are given in a certain intuition would not all together be my representations if they did not all together belong to a self-consciousness; i.e., as my representations (even if I am not conscious of them as such) they must yet necessarily be in accord with the condition under which alone they can stand together in a universal self-consciousness, because otherwise they would not throughout belong to me. From this original combination much may be inferred. B131 The mental representations that are given in an intuition must all be someone’s mental representations. Kant is concerned with what conditions must be in place for the possibility of a person’s being justified in claiming his representations as his own. As the cryptic remark at the end of this quotation implies, it is from the conditions of this activity—of being able to claim one’s representations as one’s own—that Kant hopes to justify the activity of employing pure a priori concepts. Of course, claiming these representations as one’s own implies that we can represent the “one” whose representations these are, i.e., that we have an idea of a single subject of experience persisting through time.8 As we have seen, that idea is one that Hume argues that we cannot have. So, Kant’s starting with the premise that the activity of conceiving of ourselves as such a subject is hardly going to make this argument compelling to Hume. As we have also seen, however, Hume concludes that we do not have this idea because he holds the Representational Copy Principle. We have already encountered, in Chapter 2, Kant’s arguments from the Transcendental Aesthetic and Metaphysical Deduction against the Representational Copy Principle. So, if we take that theory of mental representation to have been refuted by the time we reach the Transcendental Deduction, Kant is by that time free to reject Hume’s conclusion that we can have no such idea. Of course, he will need to give his own account of the representation of the self, and he does precisely this, beginning in the Transcendental Deduction, and then more fully in the Paralogisms. For current purposes, it is enough to take note of the dialectical position here and postpone further discussion of this representation to Chapter 6. For now, we can make do with this quick and dirty version of Kant’s reasons for thinking that the activity of conceiving oneself as a single subject of experience persisting through time is legitimate: it is necessary; one must attribute one’s mental representations to oneself. Part of Kant’s insight here is to see that being able to claim one’s representations as one’s own is not as straightforward a process as some of

112

The A-Deduction and the Nature of Intuitions

his Modern predecessors thought it to be. Descartes, for instance, seems to think that the inference from a premise of the form (D1) [I think x] and [I think y] and [I think z] leads validly to a conclusion of the form (D2) [The I that thinks x] = [the I that thinks y] = [the I that thinks z]. That is, Descartes takes the fact that he can introspectively observe that he thinks x, and that he can introspectively observe that he thinks y, and that he can introspectively observe that he thinks z, to imply that it is one and the same thing, he, that has all of these thoughts.9 Descartes famously writes of his arrival at this conclusion that This is a considerable list, if everything on it belongs to me. But does it? Is it not one and the same ‘I’ who is now doubting almost everything, who nonetheless understands some things, denies everything else, desires to know more, is unwilling to be deceived, imagines many things even involuntarily, and is aware of thinks just as true as the fact that I exist, even if I am asleep all the time, and even if he who created me is doing all he can to deceive me? [. . .] That fact that it is I who am doubting and understanding and willing is so evident that I see no way of making it any clearer.10 Of course, those of us who have read our Hume find the matter to be significantly less clear. We know that this inference—from the existence of certain experiences to the identity of the subjects of these experiences—is fallacious. Putting the matter first-personally, as Descartes does, Hume writes, For my part, when I enter most intimately into what I call myself, I always stumble on some particular perception or other, of heat or cold, light or shade, love or hatred, pain or pleasure. I never catch myself at any time without a perception, and never can observe any thing but the perception. [. . .] The mind is a kind of theatre, where several perceptions successively make their appearance; pass, re-pass, glide away, and mingle in an infinite variety of postures and situations. There is properly no simplicity in it at one time, nor identity in different; whatever natural propension we may have to imagine that simplicity and identity. T1.4.6.3–4; SBN 252 What Hume points out here is that, when we introspect, we find exactly the matter that Descartes does—this or that perception—but that this is not sufficient to yield an experience of the self—something that endures through time and is the subject of these perceptions. As Kant describes Hume’s vain

The A-Deduction and the Nature of Intuitions

113

search for an experience of the self, “[the] identity of the subject, of which I can be conscious in all my representations, does not concern any intuition of the subject, whereby it is given as object” (B408). If, however, we can have the experiences that would justify our endorsing Descartes’s premise, but still lack the resources for supporting his conclusion, then his inference is fallacious. Because Hume thinks that an experience of the self is the only ground that could warrant the further premise needed to make the argument valid, when he fails to find such an experience, he famously rejects Descartes’s conclusion. Kant does not.11 What Kant sees is that, although Descartes’s inference is invalid, his conclusion is one that each of us is nonetheless entirely justified (and, in fact, which it is necessary for each of us) to accept. That is, we are each justified, according to Kant, in conceiving ourselves as single, unified subjects of experience persisting through time. This is just what it is to be able to claim various temporally dispersed representations as our own. It is to identify those representations as belonging to a single, unified self persisting through time. Kant takes the claim expressing this identification—that it must be possible for me to think all my representations collectively as mine, the principle of the analytic unity of apperception—to be analytic, and so takes our justification for holding it to be straightforward.12 His question is not whether we are justified in so thinking of ourselves but rather how we come to be so justified. The lesson that Kant learns from Hume, contra Descartes, then is not that we ought not to (or cannot) conceive of ourselves as single, unified subjects persisting through time but rather that our doing so cannot consist in an experience either of this persisting self or of the manifold of experiences that this subject has. It cannot consist in the former because we have no such experience. It cannot consist in the latter because such a manifold is not sufficient for constituting the idea of a single self that is the subject of the entirety of such a manifold. Here is Kant expressing both the problem that he finds in Descartes, along with the general outlines for his solution to it. For the empirical consciousness that accompanies different representations is by itself dispersed and without relation to the identity of the subject. The latter relation therefore does not yet come about by my accompanying each representation with consciousness, but rather by my adding one representation to the other and being conscious of their synthesis. Therefore it is only because I can combine a manifold of given representations in one consciousness that it is possible for me to represent the identity of the consciousness in these representations itself. B134 Kant’s explanation, then, of how it is possible to conceive of oneself as a single, unified subject of experience persisting through time is that this is possible only if one can ‘combine a manifold of given representations in

114

The A-Deduction and the Nature of Intuitions

one consciousness.’ What Kant sees is that while (D2) does not follow from (D1), it does follow from (K) I think [x + y + z].13 That is, while introspectively observing a manifold of various representations is not sufficient for conceiving of oneself as a single, unified subject persisting through time, forming a single representation, the content of which includes a manifold of representations, is sufficient.14 Otherwise put, he sees that we would be justified in inferring that one and the same thing experiences all of x, y, and z if we were justified in thinking that one and the same thing thinks something else whose elements included x, y, and z. Again, if x, y, and z were three parts of a single cognition had by a single individual, then it would follow trivially that the I that thinks x is the same as the I that thinks y and the same as the I that thinks z. (D2) is what Kant calls the analytic unity of apperception, (K) is what he calls the synthetic or transcendental unity of apperception, and as he famously puts his discovery, “the analytical unity of apperception is only possible under the presupposition of some synthetic one” (B134). It is worth noting here that Kant’s claim is not merely that the synthetic unity of apperception (K) can provide a means for conceiving of oneself as the single subject of experience persisting through time: it is that this is the only means for doing so. The analytic unity of apperception is only possible under the presupposition of some synthetic one. Whether Kant is entitled to this much stronger claim has been the subject of debate among Kant scholars, but it will suffice here to note what Kant himself appears to take as his grounds, namely, the failure of the two approaches that we have recently been considering: Descartes’s and Hume’s. Namely, this thoroughgoing identity of the apperception of a manifold given in intuition contains a synthesis of the representations, and is possible only through the consciousness of this synthesis. For the empirical consciousness that accompanies different representations is by itself dispersed and without relation to the identity of the subject. The latter relation therefore does not yet come about by my accompanying each representation with consciousness, but rather by my adding one representation to the other and being conscious of their synthesis. Therefore it is only because I can combine a manifold of given representations in one consciousness that it is possible for me to represent me identity of the consciousness in these representations itself B133 It is because “the empirical consciousness that accompanies different representations is by itself dispersed” that the “thoroughgoing identity of the apperception” is only possible “by my adding one representation to the

The A-Deduction and the Nature of Intuitions

115

other and being conscious of their synthesis.” That is, Kant concludes that the synthetic unity of apperception is necessary for securing the analytic unity of apperception via an argument from elimination. Kant follows Hume in noticing the fallacy in Descartes’s thinking and thus eliminates the approach of what Kant will later call the Rational Psychologist.15 He likewise follows Hume in noticing the inadequacy of introspection alone (empirical consciousness) in grounding the identity of the experiencing subject. It is from the failure of these two approaches, which Kant takes to be representative of all the extant options, that Kant concludes that his own solution is the only possible one. Since neither the rationalist nor the empiricist can account for the unity of experiencing subject, as far as Kant is concerned, his transcendental approach is the only remaining possibility. Therefore it is only because I can combine a manifold of given representations in one consciousness that it is possible for me to represent the identity of the consciousness in these representations itself, i.e., the analytical unity of apperception is only possible under the presupposition of some synthetic one. B133–134 It is only because one can form a single cognition the contents of which contain a manifold of representations that one can represent the identity of the subject of such a manifold. As we should now expect, the cognition that Kant thinks plays this role in our thought is exactly the kind of cognition that necessarily employs pure a priori concepts. Thus, Kant’s strategy can now be fleshed out a bit more. Kant sets out to justify our use of pure a priori concepts. His plan is to show that our use of pure a priori concepts is an essential part of our engaging in another practice that is itself justified. This practice is that of conceiving of ourselves as single, unified subjects of experience persisting through time. Following Hume, he argues that doing this is not possible via an experience of such a self because we have no such experience. He further follows Hume in thinking that being introspectively aware of each member of a manifold of experiences is also not sufficient for these purposes. Most recently we have seen Kant notice that it would suffice for so conceiving ourselves to have a single cognition, the contents of which are the set of experiences in need of uniting. If it is true that we can only have such a cognition by employing the pure a priori concepts, and the rest of Kant’s arguments here are sound, then he will have found the justificatory argument for which he is searching. So, the last piece of strategy that needs outlining here is the role of the pure a priori concepts of the understanding in such a representation. We can begin, as Kant does, with the clue to the discovery of these concepts: the logical function of the understanding in judgments. As Kant understands a judgment, it is essentially the application of a concept to an intuition (sometimes directly, sometimes less directly).

116

The A-Deduction and the Nature of Intuitions In every judgment, there is a concept that holds of many, and that among this many also comprehends a given representation, which is then related immediately to the object. A68/B93

The Table of the Logical Forms of Judgment is meant to present a catalog of all of the forms that this application of a concept to an intuition can take, i.e., all of the different ways that a concept be related to an intuition in a judgment. For example, the forms listed under Quantity delineate the different ways that a concept ‘S’ can be applied to an object ‘x’ in judgments of the forms: a) All xs that are S are P, b) Some xs that are S are P, and c) This x that is S is P. The forms listed under Quality delineate the different ways that a concept ‘P’ can relate to an object ‘x’ in judgments of the forms: d) This x that is S is P, e) This x that is S is not P, and f) This x that is S is not-P. The move to the Table of Categories is precipitated by noticing that if concepts are to be applied to intuitions in these ways, intuitions will themselves have to have a certain logical form or structure (trivially, that form that allows concepts to relate to them in the way that they do). The pure a priori concepts of the understanding are just the concepts of what forms intuitions must have if they are to play the role in judgments that they do. For example, since a judgment can be universal, particular, or singular as in (a), (b), and (c) above, then the intuitions that represent the object ‘x’ must have the forms of unity, plurality, and totality, i.e., intuitions must have the forms: a’) All-such, b’) Some-such, or c’) This-such. Correspondingly, if judgments can be affirmative, negative, or infinite as in (d), (e), and (f) above, then the intuitions that represent the object ‘x’ must have the forms of realities, negations, or limitations, i.e., intuitions must have the forms: d’) Being P e’) Not being P, or f’) Being not-P.

The A-Deduction and the Nature of Intuitions

117

Kant sets out in the Deduction to show that representing oneself as a single subject of experience requires having a single complex cognition that represents a complex state of affairs as complex. Such a representation is an intuition, and he precedes the Deduction by demonstrating that, given the role that intuitions play in judgments, they can only take the forms listed in the table of categories. Thus, if Kant can indeed show that representing oneself as the single subject of experience requires having a single complex cognition that represents a complex state of affairs as complex (an intuition), he will thereby demonstrate the validity of the Categories because these are just the concepts of the forms that such intuitions must take. As I understand them, these goals and methods are roughly the same in both the A-Deduction and B-Deduction, although Kant’s approach in each differs. In the A-Deduction, Kant’s focus is on the role that the understanding plays in such representations. His focus is thus on the representative structure internal to an intuition. (The remainder of this chapter will be dedicated to detailing the relation of this internal structure to concepts-qua-inferential-rules.) In the B-Deduction the detailed results of the investigation from the A-Deduction are condensed into a brief mention of the figurative synthesis (synthesis speciosa) in §24, and Kant’s focus is instead trained not on the nature of the representing that is an intuition but instead on what is represented by it—an object—and the reasons why it is only by representing objects as he understands them that the synthetic unity of apperception is possible. (Chapter 4 will be an explication of this aspect of the Deduction and its relation to Kant’s theory of concepts-quainferential-rules.) Thus I differ once again with Longuenesse, this time with regard to the relation of the two Deductions. On Longuenesse’s view, the A-Deduction lacks the argumentative force that is needed to complete Kant’s proof of the validity of the Categories, which force is added by the supplements to his argument that Kant makes in the B-Deduction. Specifically, she argues that the argument that shows that one can move from the table of judgments to the table of categories via the concept of an object (that which is represented by an intuition) is omitted in the A-Deduction and only included in the B-Deduction.16 As I have indicated above, Kant already offers such an argument in the Metaphysical Deduction and so does not need to repeat it in the Transcendental Deduction. That Kant does repeat those conclusions in the B-edition version does not imply, as Longuenesse herself notes in passing, that those conclusions are not already at work implicitly in the A-Deduction. As Kant himself testifies, his rewrites for the B-edition do not change the contents of the A-edition but only the form of exposition. Concerning this second edition, [. . .] I have found nothing to alter either in the propositions themselves or in their grounds of proof, or in the form and completeness of the book’s plan. Bxxxvii

118

The A-Deduction and the Nature of Intuitions

Furthermore, Kant specifically addresses the rewriting of the Deduction and reports only having improved “the obscurity in the Deduction of the Concepts of the Understanding” (Bxxxviii). Finally, Kant himself describes the Deduction has having two sides, an objective and subjective one, the latter of which he describes as important to the Deduction’s main argument but not essential to it. This inquiry, which goes rather deep, has two sides. One side refers to the objects of the pure understanding, and is supposed to demonstrate and make comprehensible the objective validity of its concepts a priori; thus it belongs essentially to my ends. The other side deals with the pure understanding itself, concerning its possibility and the powers of cognition on which it itself rests; thus it considers it in a subjective relation, and although this exposition is of great importance in respect of my chief end, it does not belong essentially to it. [. . .] In view of this I must remind the reader in advance that even in case my subjective deduction does not produce the complete conviction that I expect, the objective deduction that is my primary concern would come into its full strength. Axvi In keeping with his described aim of making the Deduction less obscure, the most striking difference between the A-version and the B-version is precisely that the subjective side of the Deduction, that which “deals with the pure understanding itself, concerning its possibility and the powers of cognition on which it itself rests,” i.e., the threefold synthesis, is excised. So, given Kant’s own testimony as to what he changed in rewriting the Deduction, and the fact that his actual procedure in modifying it seems to fit with his original conception of the Deduction, we should expect the same broad argumentative structure to appear in each version of the Deduction, with changes only to the level of obscurity of the presentation. Perhaps the best way to show that this is the case is to outline how that broad argumentative structure is instantiated in each Deduction. Therefore, below, I present an outline of the argument of the Deduction in my own words with the corresponding passages from the A-edition17 and B-edition18 version below each step. 1. All representations of a complex state of affairs as complex are conceptually structured. (This thesis establishes that neither the transcendental unity of apperception nor the representation of an object is the product of pure sensibility, and thus both need to be constructed by the understanding—in order to make the analytic unity of apperception possible, as it turns out.) “All cognition requires a concept.” A106

The A-Deduction and the Nature of Intuitions

119

“Yet the combination (conjunctio) of a manifold in general can never come to us through the senses [. . .] for it is an act of the spontaneity of the power of representation.” B129 2. One must represent oneself as the single subject of a manifold of representations. “That which should necessarily be represented as numerically identical cannot be thought of as such through empirical data. There must be a condition that precedes all experience and makes the latter itself possible, which should make such a transcendental presupposition valid.” A107 “[The] principle of the necessary unity of apperception is, to be sure, itself identical, thus an analytical proposition.” B135 3. Representing complex states of affairs as complex is necessary for conceiving oneself as the single subject of a manifold of experiences. “For this unity of consciousness would be impossible if in the cognition of the manifold the mind could not become conscious of the identity of the function by means of which this manifold is synthetically combined into one cognition.” A108 “Synthetic unity of the manifold of intuitions, as given a priori, is thus the ground of the identity of apperception itself.” B134 4. In particular, in order to conceive of oneself as the single subject of a manifold of experience, one must represent an object (which is a complex of its parts). “Thus the original and necessary consciousness of the identity of oneself is at the same time a consciousness of an equally necessary unity of the synthesis of all appearances in accordance with concepts. [. . .] Further, we are now able to determine our concepts of an object in general more correctly.” A108–9 “An object, however, is that in the concept of which the manifold of a given intuition is united.” B137

120

The A-Deduction and the Nature of Intuitions

5. Therefore, it is necessary that we represent objects. 6. Since a judgment is the application of a concept to an intuition—the means by which we unite manifolds of representation—it is the vehicle through which we form such representations. “Only in this way does there arise from this relation a judgment, i.e., a relation that is objectively valid. [. . .] “It, the body, is heavy” which would be to say that these two representations are combined in the object.” B142 7. Therefore, it necessary that we make judgments. 8. Since the Categories are the logical functions for judgment, they are employed in every judgment. “Now I assert that the categories [. . .] are therefore also fundamental concepts for thinking objects in general.” A111 “But now the categories are nothing other than these very functions for judging, insofar as the manifold of a given intuition is determined with regard to them.” B143 9. Therefore, it is necessary that we employ the Categories.19 “and they [the Categories] have a priori objective validity” (A111) “Thus the manifold in a given intuition also necessarily stands under the categories” (B143) There are obviously some subtleties omitted here, and several of these steps stand in need of their own defense, but what we find in the two Deductions is a different instantiation of precisely the same broad argumentative structure. In order to conceive of oneself as the single subject of a manifold of representations, one must form a single representation, an intuition that has as its elements this manifold of representations. Forming such a representation requires deployment of the Categories. Therefore, since representing oneself as the single subject of a manifold of representations is necessary (and therefore legitimate), so is deploying the Categories. What is importantly different about the version of this argument that appears in the B-Deduction is the shift in focus from the logical analysis of the intuition that represents an object and its relation to representation of

The A-Deduction and the Nature of Intuitions

121

the identity of the self—comprised mostly by the presentation of the threefold synthesis—to the role of the object so represented. This fits with Kant’s portrayal of the Deduction as having two sides, one of which is important but inessential and which concerns the operation of the understanding, the other of which comprises the core of the argument of the Deduction and which concerns the objects of representation. In rewriting the Deduction for the B-edition, Kant excises the inessential matter concerning the operation of the understanding and thereby shifts the emphasis to the object of representation. With the broad argumentative structure of the Deduction now in place, there is only one more piece of business in need of address before we will be ready to turn to its details. As I have already suggested, the thesis that drives much of the argument of the Deduction is that any complex representation of a complex state of affairs (or complex object) must be conceptually structured. The next section will be a presentation of some of the textual evidence that Kant took himself to be committed to this thesis. It will also provide an occasion for presenting textual evidence for a related thesis: that the representations that are the conceptually structured elements that compose an intuition are what Kant calls sensations. After we have these two pieces of exegesis established, we will be well positioned to turn to the A-Deduction itself. KANT ON REPRESENTING COMPLEXES AS COMPLEX The goal of this section will be to briefly marshal evidence for two claims that will serve as premises in the arguments of the following two sections. The first is that Kant holds that all representations of complex states of affairs as complex consist of representations of the elements of the states of affairs related to one another via concepts. That is, the first thesis for which I will present evidence is that all representations of complexes as complex are conceptually structured. The second thesis is that insofar as intuitions are representations of complex states of affairs as complex—a claim that itself will be defended in the next section—the representations that are the conceptually structured elements of intuitions are what Kant calls ‘sensations.’ Both are purely exegetical claims, and my intention in this section is to present only textual evidence in their defense. A more robust defense of each along philosophical interpretive lines will come in the following sections and chapters as the pieces of the interpretation of the Critique and its relation to Hume’s Treatise fall into place. We can begin, then, with the first thesis: that all representations of complexes as complex are conceptually structured. Here is a passage from the Metaphysical Deduction in which Kant contrasts what is given by sensibility, a manifold of representations, with what is produced by the understanding, a representation of this manifold as a manifold.

122

The A-Deduction and the Nature of Intuitions Now space and time contain a manifold of pure a priori intuition, but belong nevertheless among the conditions of the receptivity of our mind, under which alone it can receive representations of objects, and thus they must always also affect the concept of these objects. Only the spontaneity of our thought requires that this manifold first be gone through, taken up, and combined in a certain way in order for a cognition to be made out of it. I call this action synthesis. By synthesis in the most general sense, however, I understand the action of putting different representations together with each other and comprehending their manifoldness in one cognition. A77/B102–3

Receptivity—our capacity for being affected in such a way that we find ourselves with a manifold of representations—produces only such a manifold. In itself, such a manifold is not yet a representation of a complex state of affairs as complex. While receptivity merely contains a manifold of representations, spontaneity (or the understanding, or our faculty for using concepts) goes through and combines these representations to form a cognition. The process wherein this combination is carried out is called ‘synthesis,’ and synthesis, finally, is the action of putting different representations together and comprehending their manifoldness. Synthesis, that is, is the formation of a representation of a complex as complex. This is what Kant means by “comprehending their manifoldness.” A cognition is a representation of a manifold (complex) as a manifold (complex). The synthesis that produces these cognitions is distinctly a function of the spontaneous understanding and therefore a product of conceptual activity. Thus, we can conclude that for Kant any representation of a complex state of affairs as complex will be a conceptual representation. Now compare the last sentence of the above quotation with another from the Metaphysical Deduction, describing the nature of concepts. All intuitions, as sensible rest on affections, concepts therefore on functions. By a function, however, I understand the unity of the action of ordering different representations under a common one. A67/B93 Synthesis is the “action of putting different representations together with each other and comprehending their manifoldness in one cognition.” A concept rests on “the unity of the action of ordering different representations under a common one.” The two processes are essentially the same: a manifold of representations is, through some action of the subject, combined and ordered to form a single cognition that represents that manifold as a manifold via the concept common to all of its elements. So Kant’s thesis in the first passage is that all representations of a manifold as a manifold (a complex as a complex) are produced by the imposition of a conceptual structure on that manifold.

The A-Deduction and the Nature of Intuitions

123

Scholars who deny that all representations of a complex as complex are conceptually structured have resisted the claim that synthesis is an essentially conceptual affair. They argue on various grounds that there must be a kind of non-conceptual synthesis that is prior to conceptual synthesis. For example, Allais and Hanna both argue that the representation of spatio-temporal complexes, because they belong to sensibility, must be non-conceptual. In the previous chapter we saw that while Kant does attempt in the Transcendental Aesthetic to isolate the contributions of sensibility to the representations of space and time, and so does there adopt an idiom that suggests that such representations are non-conceptual, he later revisits these texts and clarifies his views in a way that explicitly denies that even the representations of pure intuitions of Space and Time are, in fact, conceptually structured. Here is one of these passages. But space and time are represented a priori not merely as forms of sensible intuition, but also as intuitions themselves (which contain a manifold), and thus with the determination of the unity of this manifold in them (see the Transcendental Aesthetic). Thus even unity of the synthesis of the manifold, outside or within us, hence also a combination with which everything that is to be represented as determined in space or time must agree, is already given a priori, along with (not in) these intuitions, as condition of the synthesis of all apprehension. B160 The parenthetical remark in the final sentence of this passage makes an important, albeit by now familiar, point. Intuitions (determinate singular representations of a complex as complex) of space and time are only possible if there is a synthesis of the manifold presented through sensibility. A manifold of representations is not sufficient for representing a manifold. The passage continues: But this synthetic unity can be none other than that of the combination of the manifold of a given intuition in general in an original consciousness, in agreement with the categories, only applied to our sensible intuition. Consequently all synthesis, through which even perception itself becomes possible, stands under the categories, and since experience is cognition through connected perceptions, the categories are conditions of the possibility of experience, and are thus also valid a priori of all objects of experience. B160–161 All synthesis, including the synthesis that is required for representing space and time as complex, stands under the categories. Since the categories are just the concepts of the relations of concepts to intuitions in judgments, it follows that all synthesis is conceptual synthesis. What it means for a

124

The A-Deduction and the Nature of Intuitions

representation to fall under the categories is precisely that it has the conceptual form that would allow it to appear in a judgment. While this passage seems to be a univocal endorsement of the complex representation thesis, it is also accompanied by an infamous footnote, which seems to support with equal univocality the thesis that Kant endorses a kind of synthesis that is pre-conceptual. Space, represented as object (as is really required in geometry), contains more than the mere form of intuition, namely the comprehension (Zusammenfassung) of the manifold given in accordance with the form of sensibility in an intuitive representation, so that the form of intuition merely gives the manifold, but the formal intuition gives unity of the representation. B160–161n Thus far, there is nothing objectionable or unfamiliar: to represent space or time as the object of an intuition, or as a complex, a synthesis of the manifold presented by sensibility is required. Kant’s continuation, however, takes a confusing turn: In the Aesthetic I ascribed this unity merely to sensibility, only in order to note that it precedes all concepts, though to be sure it presupposes a synthesis, which does not belong to the senses but through which all concepts of space and time first become possible. B160–161n Here it certainly sounds as if Kant is proposing that while the unity of a representation of space or time is not made possible by sensibility alone, it is nonetheless the result of a non-conceptual synthesis. The key claim here is that this synthesis is what makes possible the concepts of space and time. This would seem to imply that this synthesis must precede these concepts and so cannot depend on them. So, the tension that this footnote creates is between the following two claims. (a) Representations of space and time are the result of a synthesis (which Kant has just proclaimed in the main text is subject to the categories and therefore conceptual). (b) This synthesis is nonetheless what makes concepts of space and time possible. Keeping these claims in mind, we can proceed to the final sentence of the footnote, which purports to give an explanation of the compatibility of these claims. For since through it (as the understanding determines the sensibility) space or time are first given as intuitions, the unity of this a priori

The A-Deduction and the Nature of Intuitions

125

intuition belongs to space and time, and not to the concept of the understanding (§ 24). B160–161n. The explanation begins by noting that the synthesis that he mentions in the previous sentence is a way that “the understanding determines the sensibility” and is how space and time are “first given as intuitions.” That would make sense on the reading of this synthesis as conceptual: intuitions are determinate representations of complex objects because they consist of the matter provided by sensibility structured by the concepts of the understanding. Kant goes on, however, to conclude from this that “the unity of this a priori intuition belongs to space and time, and not to the concept of the understanding.” Several things are puzzling about this claim. Firstly, the beginning of this sentence attributes this synthesis to the combination of the sensibility and the understanding, but Kant concludes from this that it belongs to space and time alone and not to the concept of the understanding. Secondly, it is unclear precisely what is at issue in claiming that the synthesis “belongs to” (angehört) space and time alone. That could mean that the understanding is not involved in the synthesis at all; it could mean that while the synthesis is the result of a combination of sensibility and understanding, sensibility plays a greater or more important role; etc. Finally, it is unclear how what Kant says here is meant to explain the claim made in the previous sentence. How does the attribution of this synthesis to space and time, and not to the understanding, explain the claim that representations of space and time are the result of a synthesis that is not merely sensible, but that also makes the concepts of space and time possible? What I want to suggest is that all of these puzzles are solvable on the simple supposition that Kant is not here introducing an entirely new representative faculty distinct from both the sensibility and the understanding, but instead, having just emphasized both in the main text and the beginning of the footnote the role of the understanding in representing space and time, he is now allowing the pendulum of emphasis to swing back the other way again. That is, in the Aesthetic, Kant attempts to isolate the contributions of sensibility to the representations of space and time. Here, in the Analytic, he emphasizes the role of the understanding, and concepts, in those representations. Lest his reader forget, however, that the understanding must operate on the matter given by the sensibility, he re-emphasizes the role of the sensibility. This synthesis “belongs to” space and time alone in the exact same sense in which, in the Aesthetic, Kant argues that the representations of space and time are not concepts but intuitions. We have seen that this does not imply that such representations are entirely non-conceptual, but rather it means only that they include an ineliminable contribution from sensibility. Thus, while it is a synthesis of the deliverances of sensibility that make the concepts of space and time possible, it is equally true that these concepts make that synthesis possible. One cannot use the concepts of space and time unless one has spatial and temporal forms of intuition; but one

126

The A-Deduction and the Nature of Intuitions

cannot have such forms of intuition, or formal intuitions of space and time, without employing the concepts of space and time. The advantages of understanding this footnote in this way are several. First of all, we can take Kant at his word that all representations of a complex state of affairs are conceptual. (We will return to that thesis in a moment.) Second, we can avoid having to posit a third, mysteriously unmentioned, faculty that is required for forming cognitions in addition to the sensibility and the understanding. Third, as none of the proponents of understanding this synthesis as being non-conceptual has been able to articulate in what such a synthesis would consist, other than by via negativa, understanding this footnote as being compatible with the complex representation thesis makes available the explanation that we have been giving of all synthesis as proceeding via concepts-qua-inferential-rules. Finally, all of this combines to allow us straightforwardly understand Kant’s famous claim: The same function that gives unity to the different representations in a judgment also gives unity to the mere synthesis of different representations in an intuition, which expressed generally, is called the pure concept of understanding. A79/B104 Insofar as the previous footnote appears to hedge Kant’s claim that all representations of complexes as complex are conceptual, this one appears to double down on that claim with at least as much force in the opposite direction and give additional succor to the claims that (a) intuitions themselves are such representations and are conceptually structured and (b) that the representations of space and time (which are, of course, pure intuitions) are products not only of sensibility but also of the understanding. Recall also the passage from Kant’s prize essay that we had occasion to examine in the previous chapter that makes exactly this point again. For we can represent a determinate space to ourselves no otherwise than by drawing it, i.e., by adding one space to the other, and so also with time. Now the representation of a composite, as such, is not a mere intuition, but requires the concept of a compounding, so far as it is applied to the intuition in space and time. So this concept (along with that of its opposite, the simple) is one that is not abstracted from intuitions, as a part-representation contained in them, but is a basic concept, and a priori at that—in the end the sole basic concept a priori, which is the original foundation in the understanding for all concepts of sensible objects. Ak 20:271; Theoretical Philosophy, 363 Kant is explicit here that to represent any determinate space at all, one must draw it, and that drawing a space requires conceptual structuring. (It is no

The A-Deduction and the Nature of Intuitions

127

accident that, as we will see in the next section, the example of drawing plays a central role in the section of the A-Deduction on the synthesis of recognition in the concept.)20 Here is yet another place where Kant states the thesis that what sensibility provides is merely the matter of cognition, but that it takes structuring by concepts to represent a complex state of affairs as complex, this time from his unpublished notes written during the time when he was composing the Critique. We know any object only through predicates that we can say or think of it. Prior to that, whatever representations are found in us are to be counted only as materials for cognition but not as cognition. Hence an object is only a something in general that we think through certain predicates that constitute its concept. In every judgment, accordingly, there are two predicates that we compare with one another, of which one, which comprises the given cognition of the object, is the logical subject, and the other, which is to be compared with the first, is called the logical predicate. Ak 17:616, 4634; Notes and Fragments, 149 Notice that here Kant explicitly applies the complex representation thesis to intuitions themselves, not only the intuitions of space and time but now also the intuitions of objects. Concepts operate in judgments in at least two ways: they serve as the logical predicate of the judgment, connecting intuitions to one another via inferential rules (as we saw in the previous chapter), and they also serve to structure these intuitions themselves.21 An intuition is a “logical subject” insofar as the representation of its object requires the deployment of concepts. Again, though, the key thesis here is that the sensibility provides only manifolds, whereas it is the role of the understanding, of concepts, to use such manifolds to represent complex states of affairs as complex We have seen Kant say as much in the Metaphysical Deduction, in the prize essay, and in his notes. Here is yet another passage, this time from the B-Deduction, that presents almost the exact same dialectic. The manifold of representations can be given in an intuition that is merely sensible, i.e., nothing but receptivity, and the form of this intuition can lie a priori in our faculty of representation without being anything other than the way in which the subject is affected. Yet the combination (conjunctio) of the manifold in general can never come to us through the senses, and therefore cannot already be contained in the pure form of sensible intuition; for it is an act of the spontaneity of the power of representation, and, since one must call the latter understanding, in distinction of sensibility, all combination [. . .] is an action of the understanding, which we would designate with the general title

128

The A-Deduction and the Nature of Intuitions synthesis in order at the same time to draw attention to the fact that we can represent nothing as combined in the object without having previously combined it ourselves, and that among all representations combination is the only one that is not given through objects but can be executed only by the subject itself, since it is an act of its self-activity. B13022

All combination is an action of the understanding, which is a capacity for using concepts, and the activity of which is called synthesis. Receptivity, or sensibility, produces only a manifold of representations: it is the way in which the subject is affected. To represent anything as combined, this representation must be produced by an activity of the subject itself, an activity of understanding, a conceptual activity. In the idiom that we have been using, all bringing together a manifold of representations into a single representation of a complex state of affairs as complex is conceptual. We can represent nothing as combined in the object without this act of the subject itself. No complex object can be represented as complex without the conceptual activity of the representing subject. These are expressions of exactly the thesis that we have been attributing to Kant, and which we are about to see is the key to interpreting the A-Deduction and to thereby understanding the nature of intuitions qua conceptually structured representations of complex objects as complex. These passages are merely a few of those in which Kant declares his commitment to this thesis explicitly and emphatically.23 Combined with the philosophical motivations of the previous chapter, the textual evidence strongly indicates that that Kant placed this thesis at the very center of his entire theory of mental representation. Before moving on to the second of the exegetical theses with which we will be concerned here—that it is what Kant calls ‘sensations’ that are the components of conceptually structured intuitions—it will be instructive to pause for a moment to consider at least one more passage that commentators often cite as evidence that Kant holds that we can represent complex states of affairs as complex without the deployment of concepts. The following passage is cited by both Hanna and Allais to that end, although both Hanna and Allais excise most of the explanatory material that appears between the first and last sentence here.24 The categories of the understanding, on the contrary, do not represent to us the conditions under which objects are given in intuition at all, hence objects can indeed appear to us without necessarily having to be related to functions of the understanding, and therefore without the understanding containing their a priori conditions. [. . .] For appearances could after all be so constituted that the understanding would not find them in accord with the conditions of its unity, and everything would then lie in such confusion that, e.g., in the succession of appearances nothing would offer itself that would furnish a rule of synthesis

The A-Deduction and the Nature of Intuitions

129

and thus correspond to the concept of cause and effect, so that this concept would therefore be entirely empty, nugatory, and without significance. Appearances would nonetheless offer objects to our intuition, for intuition by no means requires the functions of thinking. A89/B122–A90/B123 What Hanna and Allais focus on there are the first and last sentences, which in isolation can seem to indicate that Kant holds that we can represent at least some complex states of affairs—Hanna calls these ‘rogue objects,’ Allais, ‘particulars’—as complex without the deployment of concepts. The categories of the understanding, on the contrary, do not represent to us the conditions under which objects are given in intuition at all, hence objects can indeed appear to us without necessarily having to be related to functions of the understanding, and therefore without the understanding containing their a priori conditions. [. . .] Appearances would nonetheless offer objects to our intuition, for intuition by no means requires the functions of thinking. The key to making sense of this passage is in understanding what Kant means when he says that in the circumstances that he describes in the middle portion of the passage “objects are given in intuition,” “objects can appear to us” without concepts, and “appearances would nonetheless offer objects to our intuition.” Given the evidence that I have recently presented that Kant holds that all complex representation of complex states of affair (including the representation of complex objects) are conceptually structured, what he cannot mean by these phrases is that such objects and appearances are represented as complex. In fact, he says as much in the portion of the passage that both Hanna and Allais omit: For appearances could after all be so constituted that the understanding would not find them in accord with the conditions of its unity. As we have seen, the “unity” of the understanding is the unity of a single representation of a complex state of affairs as complex. Without this unity, the elements of a representation remain “different perceptions by themselves [. . .] encountered dispersed and separate in the mind.” As we will see farther along, Kant does consider such perceptions representations, but in a very different sense than he does cognitions. Specifically, such representations do not represent complex states of affairs as complex. This latter kind of representation does require unity (contra Hume) and therefore does require the deployment of concepts. What is made clear by examining the above passage in its proper context is that all that Kant is doing there is pointing out the possibility, to which both the conceptualist and the non-conceptualist alike can agree, of an object’s producing states in the

130

The A-Deduction and the Nature of Intuitions

experiencing subject, sensations, which are not themselves subject to the Categories. That is, the world could be such that it would not meet the conditions necessary for representing it as complex, and if it were such, it would not thereby cease to produce sensations in us, although these would in that case not be capable of being synthesized into representations of complex objects as complex (intuitions). This most recent discussion has brought us to the point at which we must turn to the second exegetical thesis that we are to consider in this section: the thesis that the representations that constitute the manifold that is united by concepts in an intuition are what Kant calls sensations. To that end, we can return to a passage that we examined a moment ago. Consider again the first sentence of the passage from B130, with the interjectory clause removed. The manifold of representations can be given in an intuition that is merely sensible [. . .] without being anything other than the way in which the subject is affected. Yet the combination (conjunctio) of the manifold in general can never come to us through the senses, and therefore cannot already be contained in the pure form of sensible intuition. B130 This is of interest because Kant here uses a phrase very similar to one that he uses in the beginning of the Transcendental Dialectic, in the so-called step ladder (Stufenleiter) of representations, where he is presenting more carefully and explicitly than he does anywhere else the differentia for all the various kinds of mental representations that he takes human beings to be capable of forming. The genus is representation in general (repraesentatio). Under it stands the representation with consciousness (perceptio). A perception that refers to the subject as a modification of its state is a sensation (sensatio); an objective perception is a cognition (cognitio). The latter is either an intuition or a concept (intuitus vel conceptus). The former is immediately related to the object and is singular; the latter is mediate, by means of a mark, which can be common to several things. A320/B377, emphasis added The complete picture of Kant’s taxonomy, then, is that intuitions and concepts are both cognitions. From the first passage we considered, we know that, “[a] manifold [must] first be gone through, taken up, and combined in a certain way in order for a cognition to be made out of it.” That is, cognitions, and therefore both concepts and intuitions, are representations of a manifold as a manifold. Given the complex representation thesis that Kant advocates, this implies that both of these representations are conceptually structured. In order for that to be true, in order for cognitions to have

The A-Deduction and the Nature of Intuitions

131

conceptual structure, they must have elements that form this structure. In the case of concepts, if our reading of Kant up to this point is correct, these elements are intuitions. Since, however, intuitions are likewise cognitions, and therefore likewise structured, they too must have parts. By putting the passage from the B-Deduction together with this differentia passage, we can now see just what these parts are: they are a manifold of representations that are “the way the subject is affected.”25 They are perceptions that “refer to the subject as a modification of its state.” That is, they are what Kant calls sensations. Here is a passage from Kant’s notes that outlines precisely this role for sensations. The primary elements of our cognitions are sensations. This is what one calls those representations in which the mind is regarded as merely passive, acted upon by the presence of an object. They comprise the matter, as it were, of all our cognition. For the form is given subsequently by the soul’s own activity. Ak 15:268, 619; Notes and Fragments, 483 Sensations are representations in which the mind is regarded as merely passive: they are mere modifications of the state of the subject. They are also the elements of our cognitions (which, of course, can be either intuitions or concepts), which are combined by “the soul’s own activity,” the spontaneous understanding, the faculty for using concepts to form cognitions.26 This reading of the relation of sensations to intuitions makes clear the presentation of these terms that Kant offers at the start of the Transcendental Aesthetic, which we had occasion to consider briefly in the previous chapter. The effect of an object on the capacity for representation, insofar as we are affected by it, is sensation. [. . .] I call that in the appearance which corresponds to sensation its matter, but that which allows the manifold of appearance to be intuited as ordered in certain relations I call the form of appearance. A20/B35 Here Kant’s focus is on the object represented by an intuition, an appearance, rather than the representation itself, but a few points relevant to our current concerns can still be extracted from this passage. First, sensations are here equated with the effect of an object on the capacity for representation. This description matches well what we saw Kant earlier attribute to sensibility: a manifold of representations (not represented as such) that is the result of an object’s affecting the subject, which manifold is what is combined by the understanding to form a cognition. Second, the manifold of appearance described here is “intuited as ordered in certain relations,” i.e., an intuition represents a complex state of affairs as complex. Finally, the

132

The A-Deduction and the Nature of Intuitions

object represented here is represented as having its matter ordered in certain relations, and this matter is that in the object which corresponds to sensation. That is, the object represented is represented as a complex of parts via a representation that is itself a structured collection of sensations, which are themselves representations of these parts. All of this is to say that intuitions represent their objects as complex by being a structured arrangement of representations of the parts of such objects. That is precisely the reading of Kant’s theory of mental representation that we have been describing. While the passages that we have seen are not the only places where Kant articulates this theory of complex representation, they do give a fairly clear picture of how it is that Kant’s theory is supposed to work. Of course, all we have done thus far is notice that the elements that are put into a conceptual structure in order to form an intuition are sensations. We have not yet said anything about what these sensations are. In particular, since what we have been emphasizing on Kant’s behalf is that it is only conceptually structured representations that can represent complex states of affairs as complex, it would seem to follow, on pain of infinite regress, that whatever sensations are, they cannot be representations of complex states of affairs as complex. Yet it is clear from these passages that Kant does take sensations to be representations of some kind, and if our most recent suggestion is correct, that he takes them to be representations of parts of complex objects. Just how all of this gets worked out, however, is a topic that I will postpone until after our examination of the A-Deduction. That is because the A-Deduction is where Kant carefully delineates the nature of intuitions, and the most salient features of sensations are precisely those that Kant attributes to them as is necessary for completing this account of intuitions. What I hope to have done to this point is merely to present enough textual evidence that Kant holds the two theses announced at the outset—that representations of complex states of affairs are conceptually structured and that, insofar as intuitions are such representations, they have sensations as their components—to have made it plausible to attribute these theses to him going forward, keeping in mind that I expect these exegetical claims to be further supported by the interpretation of Kant’s theory as a whole that I will develop throughout the remainder of this study. Presuming that I have accomplished this much, I will now proceed to the A-Deduction. PERCEPTUAL SYNTHESIS The question we posed at the outset of this chapter was whether, and in what sense, for Kant, perception itself requires the deployment of concepts. Continuing the work from the previous chapter, what we have just tentatively established is that, for Kant, all representations of complex states of affairs as complex require concepts. We have also seen hints that Kant takes intuitions to be this kind of representation and that he takes sensations to

The A-Deduction and the Nature of Intuitions

133

be the elements that, when conceptually structured, make up an intuition. The purpose of this section will be, with our tentative conclusion in hand, to see the details of how these latter theses come into play in Kant’s theory of perception. These details are presented in the A-Deduction, which—like everything in Kant—has been understood in very different ways by different commentators. Strawson famously understands the A-Deduction as presenting a crude and inadequate empirical account of a certain kind of faculty psychology, and Kant’s idiom there certainly lends itself to this kind of interpretation. More recently, Kitcher has argued that Kant’s account is not a merely empirical psychology but a more genuinely transcendental theory, although still concerned with a certain kind of faculty psychology. I will here offer a different kind of account of the theory of perception that Kant offers in the A-Deduction. I take it to be neither an empirical account of how perceptions are formed nor a transcendental account of the nature of cognitive psychology. Rather, as I understand it, what Kant is after is an analysis of the very notion of a representation of an object (as we have seen Hume was before him), of the logic internal to such a representation. Of course, as we noted in Chapter 1, ‘representation’ is an ambiguous term, and it is important to be clear on which of its diverse meanings we here intend. The reader will recall that in Chapter 1 we found the following possible disambiguations of ‘representation’ as applied to the judgment ‘That rat is fat’ when what was actually being perceived was a cat, not a rat: (a) A representing: that which represents (‘that rat’ or ‘that rat is fat’). (b) A de facto referent: that object, if one exists, of which a judgment is either true or false (a cat). (c) A represented: that which a representing represents (that rat). (d) Finally, that which the subject of the judgment is represented as (a fat cat). So, the claim that Kant is concerned with an analysis of the notion ‘representation of an object’ could mean that he is analyzing the notion of a representing, of a de facto referent, of a represented, or representing as. Clearly these notions are closely related, and Kant will have something to say about all of them by the time both Deductions are complete. As I mentioned earlier, though, I take the primary difference between the two Deductions to be one of emphasis and angle of approach. The focus of the A-Deduction is on the nature of our representings. Kant there aims to explicate the role that the understanding plays in our representing objects. In the B-Deduction, his focus is on the nature of what is thereby represented. (More on that in the next chapter.) Of course, empirical and cognitive psychology also attempt to explicate the notion of a representing, but not in the way that Kant does in the A-Deduction. His approach there is more properly describable as a kind of logical analysis: Kant wants to determine what our representings

134

The A-Deduction and the Nature of Intuitions

themselves must be like in order to represent the kinds of things that they must represent (if we are to conceive of ourselves as single subjects of experience, etc.). He is not concerned with the empirical genesis of such representings, nor the cognitive faculties that are presupposed by our ability to represent in this way. Rather, Kant is interested in the representative structure that such representings must have, if they are to accomplish the work that they do, namely, representing complex objects as complex. His question is, given the kinds of creatures that we are, what structural features a representation of ours must have, if it is to be a representation of an object. The thesis that I will defend here is that Kant’s answer is that such representations must be conceptually structured complexes of sensations, united according to concepts-qua-inferential-rules. My strategy here will be to allow myself use of the claim that all representations of complex states of affairs as complex require conceptual structure and to investigate what parts of this process of synthesis require such representations. That will allow us to see exactly what the role of concepts are in Kant’s theory of perception, and thereby the precise nature of the product of perception: intuitions. My procedure for this section will be as follows. I will read along with Kant as he proceeds from one phase of perceptual synthesis to the next, noting what transpires at each stage, and paying particular attention to whether each stage requires or does not require representing a complex state of affairs as complex. If it does, we will pause to consider the particular role that conceptual representation plays at that stage. I begin, then, with the first phase of perceptual synthesis: the synthesis of apprehension in the intuition. This first phase requires forming a representation of the temporally successive elements that combine to form an intuition as temporally complex. That is, Kant’s theory of perception begins with the fact that we represent the world as being temporally structured, and in order to do so, we must represent our manifold of representations of that world as itself occurring in time. That is, Kant’s target in the A-Deduction is an account of that which is represented by an intuition, and implicit in such a representation is the fact that we are both sensorily passive and temporally discursive: that what is represented by an intuition is something experienced via sensing it over time. In order to represent the objects experienced this way, we must first parse this temporal stream of representings of them. For example, to represent a (temporally persistent) elephant, the experience of which consists of representings of first a tail, then a body, then an ear, then a trunk, we must first represent these representings as themselves being successive. Here is a key passage from the A-Deduction. Every intuition contains a manifold in itself, which however would not be represented as such if the mind did not distinguish the time in the succession of impressions on one another; for as contained in one moment no representation can ever be anything other than absolute unity. Now in order for unity of intuition to come from this manifold (as, say, in

The A-Deduction and the Nature of Intuitions

135

the representation of space), it is necessary first to run through and then to take together this manifoldness, which action I call the synthesis of apprehension, since it is aimed directly at the intuition, which to be sure provides a manifold but can never effect this as such, and indeed as contained in one representation, without the occurrence of such a synthesis. A99 Kant is fairly explicit here on a number of points that concern us. Firstly, notice that he begins this passage by announcing that he will be concerned with the representation of a manifold as such, i.e., with a representation of a complex state of affairs as complex, and that he takes this to be what is represented by an intuition. As should be clear, given what we have been arguing up to this point, we can already draw the conclusion that intuitions—the product of perceptual synthesis—are conceptually structured through and through, and we will get to the details of this position in a moment. It will be worth our while, though, to take things slowly here. To continue with the passage, notice secondly that Kant next claims that in order for the manifold to be represented as such, one must distinguish the elements of this manifold as ordered in time. The reason for this is fairly straightforward when we notice that one of the things with which Kant is particularly concerned in the Deduction is the diachronic identity of both subjects and objects of representation. That is, Kant is particularly concerned here with the way that we perceive objects as persisting through time. His point is that the first order of business in perceiving an object this way is to give order to our representations of its parts as themselves occurring in a temporal sequence. In order to understand our representations of a temporally persistent object as representations of such an object, we must conceive of them as being representations of first this part of the object, then this other part, then another. So, we must represent our representations as being temporally complex. Finally, notice again Kant’s explicit claim that an intuition—now of something as being not only temporally but also spatially complex—is only possible if the manifold that is contained in that intuition is somehow pulled together (combined) through an act of synthesis. He is setting out, that is, to determine exactly what is needed to have a single, complex representation (in this case an intuition) that represents a manifold as a manifold. So, Kant clearly claims here that the first phase of perceptual synthesis requires representing a complex state of affairs as complex. That is what the first phase of synthesis is. It is forming a representation of the various temporally successive elements that combine to form an intuition as temporally complex. As we have already established Kant’s thesis that representing a complex as complex requires that this representation be conceptually structured, it follows straightforwardly that intuitions must be conceptually structured. More specifically, it follows—since we have likewise already

136

The A-Deduction and the Nature of Intuitions

demonstrated that Kant is a kind of inferentialist about concepts—that the representations that are united in an intuition must be related to one another inferentially. Of course, what the synthesis of apprehension describes is merely the fact that in representing an object, we represent our perceptual experiences of that object as each occurring at different times, and we have already had a brief glimpse of how Kant thinks that this kind of representation is formed. In order to represent a series of things as existing at various times, for Kant, one needs to add to these a representation of time. That is, one must represent, in addition to each of the objects, a representation of the time at which they occurred. One must place these objects into the temporal sequence. As we have seen, one does this by licensing and forbidding certain temporal inferences, such as, the move from ‘This representation appeared at t1’ to ‘It was followed by this other representation that appeared at t2.’27 Still, the representation of these representations as being related in time is guided in the way that we have outlined by a concept, here the concept of time. It is the inferences that (at least in part) constitute the concept ‘time’ that guide the construction of the intuition that represents these representations as related to one another in time. We are now in a position to draw our first conclusion about whether the products of perceptual synthesis are conceptual. Synthesis of apprehension in the intuition, requires that a complex state of affairs—the temporal relations between the representations that compose an intuition—be represented as (temporally) complex. This is achieved by uniting these representations according to the concept-qua-inferential-rule ‘time.’ So much for the first phase of perceptual synthesis. Now onto the second: the synthesis of reproduction in the imagination. As we saw in our delineation of the previous phase, the representations that are combined to form an intuition typically occur across time. In this phase, those elements that have occurred prior to the final act of synthesis are reproduced so as to be available for combination. Here is the key passage for understanding this phase. Now it is obvious that if I draw a line in thought, or think of the time from one noon to the next, or even want to represent a certain number to myself, I must necessarily first grasp one of these manifold representations after another in my thoughts. But if I were always to lose the preceding representations (the first parts of the line, the preceding parts of time, or the successively represented units) from my thoughts and not reproduce them when I proceed to the following ones, then no whole representation and none of the previously mentioned thoughts, not even the purest and most fundamental representations of space and time, could ever arise. A102

The A-Deduction and the Nature of Intuitions

137

What Kant is pointing out here is that any synthesis, any combination of elements of a manifold into a single representation of that manifold as complex, requires that we have representations of the elements of that manifold available to us for combining. Consider again the example of the experience of an elephant. Suppose the manifold of sense with which one found oneself consisted of a representation of a short gray tail, a big gray body, a flat gray ear, and a long gray trunk. Kant’s point here is that to combine the representations of the various parts of the elephant into a single representation of an elephant, one must have all of these representations available for this combination. If, however, each of these representations is produced at a different time—say as one moves one’s eyes along the length of the elephant—there must be some way of reproducing the earlier representations once the later ones are available. Kant calls this the “transcendental faculty of the imagination.” He later tells us, in the B-Deduction, more about exactly what this faculty is. Imagination is the faculty for representing an object even without its presence in intuition. Now since all of our intuition is sensible, the imagination, on account of the subjective condition under which alone it can give a corresponding intuition to the concepts of understanding, belongs to sensibility; but insofar as its synthesis is still an exercise of spontaneity, which is determining and not, like sense, determinable, and can thus determine the form of sense a priori in accordance with the unity of apperception, the imagination is to this extent a faculty for determining the sensibility a priori, and its synthesis of intuitions, in accordance with the categories, must be the transcendental synthesis of the imagination, which is an effect of the understanding on sensibility and its first application (and at the same time the ground of all others) to objects of the intuition that is possible for us. B152 The transcendental imagination is bound by our faculty for receptive sensibility insofar as its products (representations) are all derived from this faculty. The transcendental imagination can only produce representations of the same kind as those received by sensibility. So, in our example of the elephant, the transcendental imagination produces, or in this case reproduces, the representations of a short gray tail, big gray body, and a flat gray ear that the sensibility has already received. When the transcendental imagination is working with pure, non-empirical representations, it produces these as Kant here points out, in accordance with the unity of apperception and the categories. In the case of the production of an empirical intuition, such as that of an elephant, it also does so in accordance with the concept ‘elephant.’ That is, the transcendental imagination reproduces specifically those representations because the intuition of which this reproduction is in the service is that of an elephant.

138

The A-Deduction and the Nature of Intuitions Since, however, if representations reproduced one another without distinction, just as they fell together, there would in turn be no determinate connection but merely unruly heaps of them, and no cognition at all would arise, their reproduction must thus have a rule in accordance with which a representation enters into combination in the imagination with one representation rather than with any others. A121

The concept ‘elephant’ is the rule that guides the transcendental imagination throughout its process. This is why Kant calls it an exercise of spontaneity and an effect of the understanding. The activities of the transcendental imagination proceed according to concepts. What we must do now is give cash value to this claim in terms of inference, and doing so should not be difficult. Again, if we are correct, the role of the concept ‘elephant’ is to serve as a rule for constructing representations of elephants. In the synthesis of apprehension in an intuition, certain representations are represented as following one another in time. We have already seen how this representation is formed under the guidance of the concept-qua-inferential-rule ‘time.’ Here in the synthesis of reproduction in the imagination certain of these representations are reproduced so as to be available for combination with others. Of course, not all of the representations that one has during any particular stretch of time will need to be reproduced for this purpose. Only those representations that are necessary for forming an intuition of, in our example, an elephant. So, for instance, the representation of the tree that was behind the elephant’s tail, or of the bird sitting on the elephant’s ear, etc., won’t need to be reproduced. Part of the job of the transcendental imagination will be to reproduce those, and only those, representations that are relevant to forming a complex representation of, in this case, an elephant.28 So, just as the first phase of synthesis is guided by the concept ‘time,’ so this second phase of synthesis will be guided by whatever concept will eventually structure the intuition being formed. In particular, the concept ‘elephant’ licenses certain inferences about what representations count as representations of elephants, and of how these representations must be related to one another (one can expect the elephant’s tail to be encountered next to its body, rather than its ear, for instance). What we can see now is that the transcendental imagination will use these inferential rules as a guide for determining which representations need to be reproduced in order to form an intuition of an elephant. Here, then, at this phase of synthesis, there is no representation of a complex state of affairs as complex. There is no single representation that represents a manifold as a manifold. The only representations being produced here are representations that themselves will be taken up by perceptual synthesis in its other phases. So, we have no reason for thinking that these representations are conceptually structured. Yet, concepts-qua-inferential-rules are still playing an important role, and

The A-Deduction and the Nature of Intuitions

139

a role that we can already see will structure the final product of perceptual synthesis of which the results of this phase of synthesis will be a part. These representations are produced by the transcendental imagination in accordance with the concepts-qua-inferential-rules that will eventually structure the product of their being synthesized, an intuition. It is the final stage of this structuring—the synthesis of recognition in the concept—to which we will now turn. For this stage of synthesis, the representations that we have thus far seen represented as occurring in time and reproduced by the imagination, we will now finally see be combined into an intuition. There are three key passages to examine. We can take each of these in turn, noticing how they fit into the picture that we have been sketching up to now. Here is the first. Without consciousness that that which we think is the very same as what we thought a moment before, all reproduction in the series of representations would be in vain. For it would be a new representation in our current state, which would not belong at all to the act through which it had been gradually generated, and its manifold would never constitute a whole, since it would lack the unity that only consciousness can obtain for it. If, in counting, I forget that the units that now hover before my senses were successively added to each other by me, then I would not cognize the generation of the multitude through this successive addition of one to the other, and consequently I would not cognize the number; for this concept consists solely in the consciousness of this unity of the synthesis. A103 What the previous two phases of perceptual synthesis have achieved is, first, representing a series of representations as following one another in time, and secondly, making these representations available for combination in an intuition. Thus far, these representations have not yet been combined into a single representation of a complex object. In the first phase of synthesis, these representations were combined to form a single representation of a temporally complex manifold as temporally complex. In the second phase, certain of these representations are reproduced in accordance with the concept that will eventually structure the intuition into which they will be combined. It is the last phase of synthesis that finally affects that combination. It takes the representations of the previous two phases and represents them as representations of the parts of a single object. It does this, as Kant points out, by forming a single complex representation of these representations as related to one another according to a rule (above, the rule for counting). Thus we think of a triangle as an object by being conscious of the composition of three straight lines in accordance with a rule according to

140

The A-Deduction and the Nature of Intuitions which such an intuition can always be exhibited. Now this unity of a rule determines every manifold and limits it to conditions that make the unity of apperception possible, and the concept of this unity is the representation of the object = X, which I think through those predicates of a triangle. A105

The way that we pull the representations of the parts of a triangle (here its three sides) together into a complex representation of a triangle as a triangle is by relating these representations to one another via a rule for constructing triangles. This rule “determines every manifold and limits it to conditions that make the unity of apperception possible.” That is, the rule determines how such representations must appear if they are to be representations of a triangle. A concept-qua-inferential-rule does exactly this. It determines what relations must hold between representations by licensing and forbidding certain inferences between these representations (namely, that one should find oneself with representations of such-and-such a kind in such-and-such circumstances). In the case of a triangle, for instance, the concept ‘triangle’ would license the inference from ‘This line segment is part of a triangle’ to ‘There are line segments that intersect each of the ends of this line segment that also intersect each other.’ It is such rules, Kant is saying, that necessarily guide the formation of a complex representation of a complex object as complex. In the final important passage below, Kant makes it clear that he is thinking along just these lines. All cognition requires a concept, however imperfect or obscure it may be; but as far as its form is concerned the latter is always something general, and something that serves as a rule. Thus the concept of body serves as the rule for our cognition of outer appearances by means of the unity of the manifold that is thought through it. However, it can be a rule of intuitions only if it represents the necessary reproduction of the manifold of given intuitions, hence the synthetic unity in the consciousness of them. Thus in the case of the perception of something outside of us the concept body makes necessary the representation of extension, and with it that of impenetrability, of shape, etc. A106 As Kant says here, the concept ‘body’ serves as a rule insofar as a manifold of intuitions is united according to it. This concept makes possible this combination by “making necessary” certain further representations, in particular, the representation of extension, impenetrability, shape, etc. As we noted in the previous chapter, Kant is not making an empirical claim here about the association of our representations ‘body,’ ‘extension,’ etc. Rather, the point he is making is that one represents a manifold of representations as

The A-Deduction and the Nature of Intuitions

141

being representations of a body—one forms a complex representation out of these representations which itself is a representation of a body as having parts, which parts are represented by these constituent representations— by placing these representations into inferential relations with one another. That is, one uses the concept ‘body’ to unite these representations by thinking of these representations as representations of the parts of a body. One does this by placing representations of these representations into certain inferential relations with one another. For instance, one licenses the inference from ‘This representation is of part of a body’ to ‘This representation is of part of something impenetrable.’29 The picture at which we arrive at the end of this final phase of synthesis is this. Intuitions are representations of objects. As such, they are representations of a complex state of affairs (the composition of the object) as complex.30 As such, they are formed by placing constituent representations into relations with one another that are analogous to, and governed by, the inferential relations that (at least in part) constitute whatever concept structures that intuition. So, in the case of an intuition of an elephant, the representations from which the intuition is composed are related to one another according to the concept ‘elephant’; in the case of an intuition of a body, they are related according to the concept ‘body’; etc. The concepts are themselves rules for making certain inferences, crucially including inferences about what representations one will find oneself with upon encountering this kind of object, and how these representations will be related to one another. Before we can proceed, with this picture in hand, to the issue of the nature of the representations that are united in this way in an intuition, there is one pressing worry that needs to be met. This concern is that what we have just outlined is a theory according to which these component representations are united in an intuition according to inferential rules, but inferential rules are rules for judging: they are licenses, forbearances, etc., that determine what judgments can be made given other judgments, etc. The problem is that the component representations with which we have been concerned never appear in judgments.31 They occur, and are united, at a pre-perceptual, prejudgmental level, so to speak. That being the case, then, it is unclear how the structure that unites these representations in an intuition could possibly be an inferential structure. Such representations are simply not of the right kind to be so united. Addressing this worry will be the subject of the next section, and we will return to the nature of the representations so united in the final section. PIPPIN’S PROBLEM As I mentioned at the outset of this chapter, the approach that I have been taking here to the nature of intuitions is a broadly Sellarsian one.

142

The A-Deduction and the Nature of Intuitions

In his excellent 1982 book, Kant’s Theory of Form, Robert Pippin raises an important objection to any such approach—his is aimed at Sellars’s own—that parallels the one with which we ended the previous section. Here is Pippin’s articulation of the challenge that any such approach must meet. Since conceptualizing is an indispensable element in all apperception, we must be able to describe “broad classes of states of consciousness none of the members of which are apperceived.” This requirement means that we must find a special way to explain the guidedness of empirical knowledge, dependent as it now is on nonapperceived (wholly nonconceptual) states of consciousness which are in no senses, however minimally, conceptually (but only physically) describable. Such a way of describing their role in explaining “Why does the perceiver conceptually represent a red (blue, etc.), rectangular (circular, etc.) object in the presence of an object having these qualities?” is to claim they only guide “from without,” strongly distinguished from any conceptual discrimination. (“From without” here means that while this manifold is a state of consciousness, it is not an object of consciousness.)32 According to the interpretive line that we have been pursuing, sensations are the non-conceptual components of conceptually structured intuitions. Since any representation that is a representation of a complex as complex must be conceptually structured, sensations are not representations of a complex as complex. As we will see in more detail in the next section, Kant has available to him a transcendental argument that such sensations exist—in Sellars and Pippin this takes the form of an explanatory hypothesis aimed at explaining why it is that we respond conceptually (correctly and incorrectly) to worldly objects as we do—but these sensations are never per se themselves the object of any representation. Still, they are meant to play, as Pippin puts it, a guiding role in the formation of fully fledged conceptually structured intuitions. They somehow influence the productive imagination in its construction of intuitions but are neither merely causes of this activity nor represented by the productive imagination (as, say, a map is used by a lost camper in guiding him to the correct path home). It is this final point that Pippin will leverage in his objection. Pippin’s worry about this approach is a worry about the relation of these non-conceptual representations (sensations) to the conceptual representations that they are meant to guide (intuitions). For one thing, we would have to ask about the “transition” from unapperceived sensory states to apperceived ones. That is, if such sense impressions guide “from without,” it is not yet clear how they guide at all (as opposed to merely determining our response).33

The A-Deduction and the Nature of Intuitions

143

Pippin’s point is this. On Sellars’s account, non-conceptual representations “evoke” conceptually structured intuitions in the subject. This makes it sound as though intuitions, like sensations, are the products just of receptivity. If having certain sensations leads straightforwardly to having certain intuitions, then it looks like there is no room for spontaneity to get its foot in the door. The relation between the two is a straightforward causal (or evoking) one. Pippin’s suggestion for responding to this objection is that we acknowledge that intuitions are not merely evoked by sensations but that they are also the product of the understanding’s uniting sensations according to rules. He then worries, however, that the pendulum has swung too far the other way. And this means that the whole issue of “guidedness” in Kant should be taken in his own special transcendental sense. We are so guided by our sensory experience only to the extent that we follow our own rules for what counts as guidedness. [. . .] [I]n a strict transcendental sense, such a restraining function played by sensation is still itself a result of our conceptual activity, and the “immediate” characteristics of sensibility “restrain” our conceiving only to the extent that we follow our own “rules” for what to regard as constraint.34 Pippin’s worry now is that if perceptual synthesis is guided by rules set forth by the understanding for how to unite sensations into intuitions, these sensations only provide a constraint on cognition (on the formation of intuitions) if these rules provide for a finite set possible outcomes. That is, if the understanding can unite sensations into intuitions any old way it pleases, then there is no operative constraint on perceptual synthesis at all. If, however, the understanding is limited to a finite set of ways that it can unite sensations into intuitions, then we are thrust back onto the other horn of this dilemma: the understanding is simply determined by the sensibility, and there is no sense to be made of the idea of rules, or spontaneity at all. Pippin’s worry is essentially the twentieth-century concern about rule following imported back into Kant.35 Pippin’s dilemma is clearly closely related to the issue that we have already encountered between the conceptualist and the non-conceptualist. There the non-conceptualist charges the conceptualist with reversing the order of explanation that is supposed to run from our experience of the world to our conceptual representations of it. The conceptualist, in turn, charges the nonconceptualist with drawing too close a connection between perception and conceptual representation, thereby leaving too little room for the spontaneity of conceptual activity on which Kant insists. The dilemma is also connected, though, to the problem that we have recently encountered of how an intuition can be conceptually structured when nothing in the intuition seems to be of the form that makes conceptual structuring possible. Pippin’s worry is that sensations are either the mere cause of intuitions, in which case they

144

The A-Deduction and the Nature of Intuitions

are not also the components of intuitions, or that they are the components of intuitions, but there is no viable story of how the understanding can move from sensations to intuitions via rules on pain of infinite regress. There is nothing about the sensations themselves, being non-conceptual, that constrains the application of rules to them. What I will argue now is that Pippin’s dilemma is a false one, predicated on the erroneous supposition that the only way for the imagination to unite sensations according to rules is by representing those rules and representing sensations as falling under them. It is this supposition that makes the second horn of Pippin’s dilemma seem to be the only alternative to casting sensations as mere causes. If they are mere causes, the thought goes, then there is no sense in which rules are applied to them, and so they cannot be the elements of a conceptually structured representation; but if they are more than mere causes, then they are subject to rules, represented as such, and therefore themselves already conceptual (and so cannot provide the non-conceptual constraint on representations that is supposed to be their primary role). Kant, I believe, saw a middle ground here, a way for intuitions to have a conceptual structure, with sensations as their non-conceptual components, without concepts-qua-inferential-rules being directly applied to them. It is worth noting at the outset that what will follow will be relatively speculative, although it takes its lead from an important passage in the Critique, and I will attempt to tie it to other of Kant’s text where possible. The argument in favor of the view I will present consists mostly of the facts that (a) it fits what texts there are of Kant’s on this subject, (b) it is a philosophically tenable solution to the puzzles at hand, and (c) it is better than nothing, which seems to be what is currently on offer in the secondary literature. That is, as we will see at the end of this section, among those who agree that intuitions are conceptually structured, there is no successful extant account of how they can be such,36 and among those who disagree that intuitions are conceptual, what is put in place of that conceptual structure is a non-conceptual one—the proponents of this line of interpretation appeal to Kant’s remarks about the synthesis speciosa, or figurative synthesis—but this ends up being nothing more than a via negativa, a kind of synthesis that is only ever described as not involving concepts. What I hope to do in this section, therefore, is to fill this lacuna in the literature by providing a genuinely informative account of what an intuition’s being conceptually structured means, which will both give a much-needed proof of this possibility and thereby shift the burden to the proponent of non-conceptual intuitions to offer a similarly informative account. The clue that Kant gives for solving this puzzle is in the following passage, which immediately follows the one we examined earlier from the Metaphysical Deduction on the nature of combination and synthesis. Synthesis in general is, as we shall subsequently see, the mere effect of the imagination, of a blind though indispensable function of the soul,

The A-Deduction and the Nature of Intuitions

145

without which we would have no cognition at all, but of which we are seldom even conscious. A78/B104 On its face, this is a very strange thing for Kant to write. As we have seen, much of the argument of the Transcendental Deduction turns on Kant’s claim that the unity affected by the application of concepts to a manifold of representations is necessarily capable of being accompanied by the ‘I think.’ Thus, it is very odd to find Kant saying here that synthesis, which we have seen just is the uniting of a manifold of representations via concepts, is a process of which we are seldom conscious. Of course, these claims are not incompatible: one can certainly be capable of becoming conscious of something even if one seldom does so. Still, the description of synthesis as a blind function would seem to indicate a stronger claim: that synthesis does not just typically occur without consciousness, but that its functioning is entirely independent of consciousness, even if we can sometimes become aware of this functioning. (Later we will see that sensations themselves have a similar feature: they are not apperceived but posited, and insofar as we do become conscious of them, they are no longer playing the role of the nonconceptual components of conceptually structured intuitings but are instead that which is represented by inner sense.) We can begin to understand this strange claim, and in turn the sense in which intuitions are united by concepts, if we recall that we have recently seen at least one example of the kind of work that the imagination does. Remember that the second phase of synthesis is the synthesis of reproduction in the imagination. There the imagination’s role was to reproduce specific representations so that these could be combined in the third phase of synthesis. The idea is that while the imagination does not apply a concept to these representations (that happens in the next phase), it still “looks to” this concept in order to determine what representations would need to be reproduced. What we need to recognize now is that “looking to” is exactly the wrong phrase to use to describe this process. While the imagination must reproduce representations in accordance with the relevant concept, in so doing it does not represent this concept to itself. It does not “look to” or consult the concept at all. Rather, the imagination is subject to a rule but is not the subject of the rule. Its behavior is rule governed without being rule following. This is the sense in which the imagination functions blindly.37 Next notice that in the passage above, Kant’s claim is not just that the synthesis of reproduction in the imagination is blind, but that synthesis in general is. Synthesis, the process of forming complex representations of complex states of affair, is one that is carried out blindly. So, it is not just that the component representations that will be included in (and excluded from) an intuition will be produced “merely” in accordance with the relevant rule, but that the application of the rule itself to these components must also be executed without the imagination’s consulting that rule.

146

The A-Deduction and the Nature of Intuitions

Now, the issue of the nature of rule following received a great deal of attention in the previous century, and while there is certainly little to report by way of consensus, since we are already taking a Sellarsian approach to Kant’s texts in other respects, it will be worth our while to consult his contribution that discussion as well. Compare, then, Sellars’s discussion of a dilemma similar to the one that we have extracted from Pippin. For it becomes clear that we have tacitly accepted a dichotomy between (a) merely conforming to rules: doing A in C, A’ in C,’ etc., where these doings “just happen” to contribute to the realization of a complex pattern. (b) obeying rules: doing A in C, A’ in C,’ etc., with the intention of fulfilling the demands of an envisaged system of rules. But surely this is a false dichotomy! For it required us to suppose that the only way in which a complex system of activity can be involved in the explanation of the occurrence of a particular act, is by the agent envisaging the system and intending its realization. This is as much as to say that unless the agent conceives of the system, the conformity of his behavior to the system must be “accidental.” Of course, in one sense of the term it would be accidental, for on one usage, ‘accidental’ means unintended. But in another sense, ‘accidental’ is the opposite of ‘necessary,’ and there can surely be an unintended relation of an act to a system of acts, which is nevertheless a necessary relation—a relation of such a kind that it is appropriate to say that the act occurred because of the place of that kind of act in the system.38 Sellars’s idea here is that there is a middle ground between merely conforming to a rule and obeying a rule, where the latter requires representing the rule to oneself and intending that one’s behavior conform to it. This middle ground is occupied by behavior of which it is appropriate to say that it conforms to a rule because of the rule, but not as a result of the fact that the person exhibiting the behavior consulted the rule, or otherwise deliberately shaped their behavior to be in accordance with it. So consider the behavior of a child first learning a language. Before such a child can ever be in a position to formulate the rules of that language to herself, she can be made to conform to those rules (through the attentive and judicious behavior of their parents, for example).39 The rules of the language that the child is learning play an explanatory role in the child’s behavior, and so it is not appropriate to say that the child merely conforms to those rules. There is a sense in which the child says ‘red’ because it is appropriate to say red in the presence of red things. The child is being trained to do just this. On the other hand, it is important to note that the child is not obeying the rules of the language in the sense that she consults those rules and uses them as a guide to shaping her behavior because ex hypothesi she is not capable of doing so. So, we

The A-Deduction and the Nature of Intuitions

147

might say that the child instantiates or models the rules (which is more than merely conforming to them) without obeying them. Here ‘instantiating’ and ‘modeling’ are appropriate because we can think of the rules of the language as providing a kind of picture of what ought to be the case (‘red’ ought to be said in the presence of red things and not otherwise), and the child’s behavior will be an instantiation or model of this picture. The normative relations among items represented in the meta-language of rules, if the child is well trained, will correspond to the relations among the child’s linguistic behavior and items in her immediate environment. To put these distinctions to use in understanding Kant’s text, what I want to suggest is that Kant intends by his use of the idiom of the imagination as blind (blinden) is that it instantiates the inferential rules that relate judgments to one another without either merely conforming to these rules or obeying them. Concepts-qua-inferential-rules relate intuitions to one another in judgments, thereby creating a picture of the world. Using the concept ‘elephant’ to license an inference from the judgment ‘This elephant tail in front of me is short and stubby’ to the judgment ‘If I turn my head, I should see a large gray elephant body’ creates a picture of the world as containing elephant tails and elephant bodies in it related to one another in various ways (at the very least as being adjacent to one another). Perceptual synthesis cannot be inferential because its elements are essentially sub-judgmental. The representations that make up an intuition will not be representations of elephant tails and long gray trunks as such and will never appear in such judgments. Again, this is why the syntheses that form this intuition are blind. The representations with which such processes operate are not representations of anything as anything, and the process by which these are united cannot therefore be rule-following processes. But the structure of inferential relations can nonetheless be instantiated or modeled by such syntheses, and in such a way that its doing so is neither merely accidental nor the result of the imagination’s obeying any inferential rules. Just as it is because the child is being trained to use the word ‘red’ that she says ‘red’ in the presence of red things and thus her utterances are instantiations of the rules governing ‘red,’ it is because what is represented by an intuition is, say, an elephant, that the syntheses enacted by the imagination are instantiations of the rules governing the concept ‘elephant.’ Those syntheses are neither mere conformities with those rules nor carried out in obeisance of them. Nonetheless, it is because what is being represented is an elephant that, e.g., a representing of a tail is retained, while a representing of, e.g., a tree is not. Of course there are important differences between the child learning the language and the imagination’s syntheses, and so we must be careful in interpreting this analogy. Most importantly, it is crucial to recall that Kant’s threefold syntheses are not descriptions belonging to faculty psychology but rather are accounts of the logic internal to the representations with which we find ourselves. They describe the logic of our most fundamental

148

The A-Deduction and the Nature of Intuitions

representations: intuitions. Thus, there is no training of the imagination to act in any way in forming intuitions. Rather, what needs to be the case in order for the inferential structure of judgments to be instantiated or modeled by the logical structure internal to an intuition is that the latter must be appropriately responsive to the former. E.g., if one changes the concepts that one employs at the level of judgments, this change must be mirrored by a corresponding change in the structure of the intuitions with which one finds oneself. It will be instructive here to consider an example.40 Imagine, then, a situation in which someone with little knowledge of particle physics enters a physics laboratory and notices some of the experimental equipment around the lab. This layman might find himself making the spontaneous judgment, ‘That box full of gas has a streak running through it,’ which would entail his having an intuition of the form ‘that-box-full-of-gas.’ That intuition, in turn, would have the internal logic appropriate to uniting a manifold of sensations according to the concept-qua-inferential-rule ‘box-full-of-gas,’ which, of course, has a structure that would be very different from an intuition structured by, say, the concept ‘cloud chamber.’ (E.g., only very specific gases can be used in cloud chambers, such chambers are used for very particular experimental purposes, etc.) Now consider, by contrast, the physicist whose lab this is. Having a great deal of experience with cloud chambers, one can imagine that such a physicist no longer finds herself making the judgment, ‘That box full of gas has a streak in it,’ but rather she might well find herself entering her lab and forming the spontaneous judgment, ‘Some sub-atomic particle has recently passed through that cloud chamber,’ or even ‘That electron’s path will need to be recorded.’ Correspondingly, then, the scientist will find herself with the intuition, ‘some-sub-atomic-particle,’ or ‘that electron.’ There are a number of important points to notice about these cases. The first is that it is the concepts with which the layman and the scientist operate at the level of judgment that determine what concepts operate at the level of intuition for each. It is because the scientist has come to master and employ the inferential commitments that constitute the concepts, ‘sub-atomic particle,’ ‘electron,’ and ‘cloud chamber,’ that the intuitions with which she finds herself are those of sub-atomic particles, electrons, and cloud chambers. It is because the scientist’s judgments are governed by the norms that constitute the use of concepts such as ‘cloud chamber’ that she can find herself with intuitions of the form ‘that cloud chamber.’ The inferential structure of the former is imported into the representational structure of the latter, and it is because this is so that her intuition has the proper conceptual form to be the subject of such judgments. The two structures are jointly necessary for conceptual representation and mutually reinforcing. The second thing to notice about these cases is that it is plausible to think that at some point the conceptual repertoire of the scientist was similar to that of the layman. That is, at some point before she had mastered the

The A-Deduction and the Nature of Intuitions

149

concepts ‘cloud chamber,’ ‘electron,’ etc., the scientist would have reacted to the items in her laboratory in much the way that the layman does, with spontaneous judgments such as ‘That box full of gas has a streak running through it.’ Of course, what changes is that she comes to master the norms governing the use of the more advanced concepts, and this change in the norms that she takes to govern her judgments affects a corresponding change in the conceptual form that her intuitions take. At some point during the course of her education, the scientist is taught to infer from ‘That box full of gas has a streak running through it’ that ‘Some sub-atomic particle has recently passed through that cloud chamber.’ At some point later in her education, she no longer needs to draw this inference because she simply finds herself with the latter judgment spontaneously occurring to her. Again, it is the scientist’s mastery of the inferential norms governing her judgments that precipitates the change in the logical structure of her intuitions. Finally, it is worth noting here that the difference between the layman and the scientist is not first and foremost a difference in their psychological makeup. Rather it is a difference in the conceptual norms that govern each of their representations. As we will see in greater detail in the next section, there must be a sense in which the layman and the scientist share a certain kind of psychological state: they each have a manifold of sensations similar to the other; they each are subject to a similar modification of their state. This is the psychological effect that the “box-full-of-gas” or the “cloudchamber” has on them. What makes the difference in the content of the representation (intuition) that they each form from this manifold is that it is governed by different norms in each case. The layman unites this manifold using the concept-qua-inferential-rule ‘box-full-of-gas,’ while the scientist does so using ‘cloud-chamber.’ What such examples show is the sense in which the conceptual structure internal to an intuition is responsive to, and thereby models or instantiates, the structure of concepts-qua-inferential-rules. Just as in the case of the language learner, it is because one’s judgments are governed by the use of the concept ‘box-full-of-gas’ that one is capable of finding oneself with the intuition ‘this-box-full-of-gas.’ When the norms governing one’s judgments change, say when one masters the concept ‘cloud-chamber,’ one’s intuitive capacities likewise change: one can now find oneself with the intuition, ‘this-cloud-chamber.’ The imagination does not need to consult the rules for applying the concepts ‘box-full-of-gas’ or ‘cloud-chamber’ in order to affect this change, but it is not thereby merely conforming to those rules in producing intuitions with such structures. Rather, what I want to suggest is that Kant’s blind imagination walks the middle ground outlined by Sellars. It neither merely conforms to a rule nor obeys a rule. Rather, the blind imagination models or instantiates concepts-qua-inferential-rules by producing intuitions that have the structure that they do because of the rules that govern judgment: it produces intuitions of the structure of which is responsive to changes in the structure of the rules governing judgment.

150

The A-Deduction and the Nature of Intuitions

The claim that the imagination is blind is one of two places in the Critique that Kant uses that term, the other being his famous claim that intuitions without concepts are blind (A51/B75). While the two uses are not precisely the same, they are clearly related. In the earlier and more wellknown instance, Kant’s point is that without conceptual structure, an intuition cannot represent anything as anything: sensations are merely a manifold of representations but without concepts do not constitute a representation of a manifold as a manifold. Similarly, on the account of the functioning of the imagination that we have been examining, the imagination functions blindly precisely because it does its work without representing anything as anything. It does not represent sensations as such, and it does not represent the rule according to which it combines these. This is why it is a rule-instantiating faculty and not a rule-obeying one. What ‘blind’ indicates in both passages is that without concepts, nothing can be represented as, although in the case of the blind imagination, as we have seen, its functioning blindly does not prevent it from producing a representation which is itself conceptually structured. This is precisely because while the imagination functions blindly, it is nonetheless functioning according to the rules of the understanding (concepts). The unity of apperception in relation to the synthesis of the imagination is the understanding, and this very same unity, in relation to the transcendental synthesis of the imagination, is the pure understanding. A119 The transcendental unity of apperception that is made possible by the representation of an object, an intuition, is both a product of the synthesis performed by the imagination and a product of the understanding, our faculty for using concepts. What I have suggested is that the way to make this identification of the synthesis of the imagination and the role of the understanding compatible with the claim that the imagination functions blindly and serves to unite sensations in an intuition is via the distinction between rule obeying and rule instantiating or rule modeling. The imagination operates blindly but according to rules nonetheless. It will be instructive at this point to compare the interpretation of the role of the blind imagination that I have been outlining with that which Longuenesse presents in Kant and the Capacity to Judge. This is because while much else in our respective interpretations of the A-Deduction is similar, this is a crucial point of difference, which will serve both to clarify the current interpretation and to highlight what I take to be missing from Longuenesse’s account. As I hope to show, Longuenesse also attempts to formulate a solution to the problem of the conceptual structure of intuitions, but the answer that she gives—that the imagination takes producing such a structure as its goal—misses Kant’s point in describing that faculty as blind. That is, Longuenesse attributes what amounts to full-fledged rule-following agency

The A-Deduction and the Nature of Intuitions

151

to the imagination, when the puzzle at hand is precisely that of accounting for this structure when such agency is impossible. (Again, because that agency would require sensations themselves to have the conceptual structure to appear in judgments, judgments such as ‘to achieve such-and-such goal, I must unite these sensations in such-and-such a way.’) Longuenesse and I are in agreement that intuitions come to be representations of objects as complex (manifold) only by themselves having a conceptual structure, and while Longuenesse does not attribute to Kant the specifically inferential theory of concepts that I have here, she does recognize a difficulty parallel to the one that we have been discussing for her own understanding of ‘intuition.’ For her, that difficulty takes the form of answering the question: “how can a discursive act produce a sensible, intuitive synthesis?” Here is her answer: The act of thinking whose result is judgment, because its goal is judgment, affects receptivity and thereby combines the sensible given with a view to judgments. [. . .] This is not to say that the logical forms themselves affect sensibility and produce forms of sensible combination. Rather, sensibility is affected, the sensible given is combined, by an act whose goal, and thus whose proper effect, is to reflect representations under concepts combined according to logical forms of judgment. What this act is, we don’t know says Kant. All we can say is that the activity of combination is oriented toward the forms of discursive combination in judgment. This provides at least a partial answer to the difficulty raised earlier: how can a discursive act produce a sensible, intuitive synthesis? The answer, I think, is that in fact it is not judgment, in its discursive forms, that affects sensibility. But the act of spontaneity which affects sensibility has judgment for its goal—that is, discursive combination of concepts according to logical forms.41 There are several important points to note about this solution as Longuenesse presents it. The first is that Longuenesse and I agree that the solution to this problem has to come from somehow importing the structure of judgments into intuitions. The second is that we are also in agreement that this cannot be done by making the representations internal to an intuition themselves the subjects of any judgment, but rather it can only be accomplished by subjecting these representations to a rule, the end product of which is a conceptually structured intuition that is itself suitable to appear as such a subject. So much for agreement. The next point to take away from this passage is that Longuenesse is clear, and emphasizes, that judgmental structure is imported into an intuition by an “act of spontaneity” that “orients itself towards” and acts with “a view to judgment,” and takes imposing this structure as its “goal.” The entirety of Longuenesse’s chosen idiom indicates that she envisions the role of the imagination here as being analogous to what we have been calling

152

The A-Deduction and the Nature of Intuitions

a rule-following agent (rather than a rule-conforming or rule-instantiating non-agent). The imagination is oriented toward judgment; its acts (rather than, say, behaviors) impose structure on the non-conceptual representations that compose an intuition; this act is an act of spontaneity, i.e., the understanding itself, rather than some proxy for it; it takes this imposition as its goal and acts, presumably by pursuing the means to this goal, to accomplish it. The question that we have been recently confronting, and one that Longuenesse takes herself to be answering, however, is precisely the question of how this is possible. How can the imagination import the conceptual/rule-governed structure of judgment into an intuition if the items so structured are not themselves represented by the imagination as subject to any such rules (which they cannot be on pain of infinite regress)? Longuenesse’s proposed solution seems to require the imagination to have in view both the judgments in which intuitions will eventually appear and the sensations that it must unite to form such intuitions. It seems to require, that is, that both judgments and sensations themselves be the subject of some further judgment, a judgment such as ‘To form an intuition of such-and-such kind, I must unite such-and-such sensations in such-and-such a way.’ The problem here, however, is precisely that because they are ex hypothesi prejudgmental, sensations cannot be subject to rules for judging. Longuenesse is certainly correct that it cannot be that judgment itself somehow imposes this structure on intuitions, but she does not seem to have much more of an answer of how it gets done.42 Notice that the idiom of agency is not merely a quirk of how Longuenesse presents her solution here but a theme that runs through her discussions of this issue throughout her study. Below is an excerpt from her discussion of Kant’s example from the Logic of the savage who sees a house but does not have the concept ‘house.’ In his apprehension there is no rule guiding him to privilege certain marks and leave aside others, so that a concept of a house might apply. [. . .] Only the “application in a comparison,” that is, the gradually dawning consciousness of a “rule of apprehension” common to the representation of various objects serving the same purpose, would pick out analogous mark and bring forth the concept of a house. [. . .] But this universality is brought about only by the act of comparison accompanied by acts of reflection and abstraction.43 Here again Longuenesse describes the importation of conceptual structure into an intuition as a “consciousness of a rule” and as a conglomeration of “acts” of the understanding. Adding that this act of consciousness is “gradually dawning” simply obscures the real issue here. Either the imagination imposes this structure by representing the relevant rule to itself and somehow applying that rule to sensations (which are neither apperceived nor represented as having the marks that would make the application of

The A-Deduction and the Nature of Intuitions

153

such a rule to them appropriate) or it imposes this structure, using this rule “blindly.” ‘Gradual dawning’ implies that there is some middle ground here in which the imagination is first blind, but then comes to see, but Longuenesse offers no explanation for what such a middle ground would be, nor how it is so much as possible, given the considerations with which we have recently been concerned. Moving from merely conforming to a rule to genuinely following a rule does not seem to be a matter of degree at all. Insofar as there is a middle ground here—such as the one presented above, rule instantiating—it is does not fall on a spectrum somewhere between rule conforming and rule following but is rather a distinct conceptual space all of its own. So, Longuenesse and I are in agreement that intuitions have a conceptual structure and that this conceptual structure is that described by the three syntheses in the A-Deduction. Where we differ is in what it means for intuitions to have this structure. Longuenesse takes this to mean that the imagination consults the rules that concepts impose on judgments and imposes these rules, or some analog thereof, onto the sensations that make up an intuition. I follow Pippin in thinking that this picture cannot be correct because it requires that sensations be apperceived and represented as being subject to such rules. If, as Kant tells us, this is only possible if they have the forms described by the Categories, and that they can have these forms only if they, in turn, are conceptually structured, then on pain of infinite regress, sensations cannot be the subject of any judgment, including judgments about how they ought to be united. Instead, I propose that we understand Kant’s remark that the imagination is a blind but indispensable function as gesturing at what I have called rule instantiating or rule modeling. Concepts-qua-inferential-rules create a certain structure among the judgments that they relate; the imagination imposes this same structure on sensations to form intuitions. It does not do so by forming judgments about sensations but rather by forming fullfledged conceptual intuitions, the structure of which is modeled on that of concepts-qua-inferential-rules. It neither merely conforms to these rules nor obeys them in its activity. Rather, it instantiates these rules: unites sensations in a way that models these rules, and does so because of those of those rules, but without representing either the rules or the sensations to itself as such. Finally, it is helpful to remember that just as the idiom that Kant employs in the A-Deduction rings of faculty psychology but is actually intended to convey a transcendental-logical analysis of the representational content of intuitions, so my talk here of the imagination “producing” intuitions with conceptual structure is likewise intended only to explicate that same logical structure. The blind imagination does not literally “construct” intuitions from sensations. Rather, we find ourselves with intuitions, representations of complex objects as complex, and the sense in which these are “constructed” by the blind imagination is only that a transcendental-logical

154

The A-Deduction and the Nature of Intuitions

analysis of the representational content of such intuitions reveals that their internal structure must model or instantiate the structure of concepts-quainferential-rules. The same function that gives unity to the different representations in a judgment also gives unity to the mere synthesis of different representations in an intuition, which expressed generally, is called the pure concept of understanding. A79/B104–5 As Sellars notices, the intuition ‘this-box-full-of-gas,’ is, after all, very closely related to the judgment ‘This is a box full of gas.’ We are now in a position to say in what that similarity consists: it is a similarity of structure. In the latter case, this structure is that imposed by concepts-qua-inferential-rules; in the former it is a structure modeled on those same rules. Having such a structure is what it is to be such an intuition. Up to this point, our focus has been on the form of intuitions: their structure. We have not had much to say about the matter of intuitions other than that this is the manifold of sensation, which itself is a manifold of representations of the parts of objects. As we noted early on, however, only conceptually structured representations can be representations of complex states of affairs as complex, and so there is some work that remains to be done to explain in what sense sensations are representations of the parts of complex objects. It is to that work that we will now turn in the next section. NON-CONCEPTUAL REPRESENTATION At the outset of this chapter we outlined two philosophical positions on perception, and correspondingly two exegetical positions on Kant’s account of perception. Conceptualists argue that all perceptual representation is conceptual representation, that to perceive worldly objects at all requires that we already have concepts of these objects, without which concepts we could not perceive these objects. The corresponding exegetical claim is that this is the view that Kant held. Non-conceptualists argue that conceptualism reverses the order of explanation: that what explains why we have the concepts that we do is that we have the perceptions that we do, and to make room for a constraint on cognition by the world requires that some of our perceptual representations be non-conceptual, that there be a kind of perceptual representation that is prior to, and independent of, concept use.44 The corresponding exegetical claim is that this is the view that Kant held. On this way of dividing the terrain, then, the theory we have now presented is thoroughly conceptualist. Intuitions, the products of the perceptual process, are through-and-through conceptually structured.

The A-Deduction and the Nature of Intuitions

155

Dialectically, then, the theory we have presented would seem to be subject to the objections of the non-conceptualist. As the reader might have already discerned from the fact that the current section concerns the nature of sensations, it is these sensations that will provide the means for meeting these objections. My procedure here will be as follows. I will begin with a discussion of the nature of these component representations and their place in Kant’s account of perception. Once I have this outline complete, I will use it to make my way through the objections of the non-conceptualists, noting where I can accommodate their concerns within Kant’s theory. The hard core of these accommodations will be that these component representations, which are preperceptual—prior to perceiving anything as anything—are non-conceptual. Thus, while we will agree with the conceptualist that all perceptions are conceptually structured, by the end of this section, we will also have presented an account that makes a place for the non-conceptualist’s contention that there must be a non-conceptual constraint on perception, and even that there must be non-conceptual representations. To begin our examination of sensations, we can consider the following question: if sensations do not represent complex states of affairs as complex, in what sense are sensations representations at all? A helpful way to answer this question is to depart, once again, from Kant’s text, and look briefly at one of Sellars’s. Here is a passage from Sellars explaining how something very much like Kant’s sensations fit into his own account of perception. Thus, the sense impression inference is an attempt to account for the fact that normal perceivers have conceptual representations of a red and rectangular object both (a) when they are being affected in normal circumstances by a red and rectangular object; and (b) when they are being affected in abnormal circumstances by objects which have other, but systematically related characteristics.45 For Sellars, sense impressions are an explanatory posit. In particular, what is posited is a stratum of mental states that have certain qualities and are related to one another in certain ways such that they cause us to respond conceptually to worldly objects in these predictable ways. As he puts it, [T]hese nonconceptual states must have characteristics which, without being colours, are sufficiently analogous to colour to enable these states to play this guiding role.46 We posit sense impressions, not as themselves being colored, but as having a property that is analogous to color, which property explains why we

156

The A-Deduction and the Nature of Intuitions

respond conceptually to that property in ways that are systematically connected not just to, say, red objects, but also to white objects in red light. That is, the sense impression inference is meant to account for what it is that is similar in experiences of red objects and white objects in red light without having to suppose that in all such cases there are mental entities that are literally red (as, e.g., Hume seems to have). It does so by supposing that there are certain mental states that stand as intermediaries between worldly objects and our conceptual responses to these. These mental states, sense impressions, are posited as having a structure that is both affected in predictable ways by worldly objects and in turn affects our conceptual faculties in similarly predictable ways. The first important point to take away from Sellars and back with us to Kant here is that these sense impressions, the mental states that Sellars posits, are the mere cause of our conceptual responses. These sense impressions do not themselves appear in judgments, warrant any—even defeasible— conclusions about the way the world is, or even appear to consciousness. Again, they are posited as mere proximate causal antecedents of conceptual responses to worldly objects. (Recall that this was the basis of one of the horns of the dilemma that we saw in the previous section Pippin pose to Sellars.) The second important point to take way is that even though they are mere causal antecedents of conceptual responses to worldly objects, there is a sense in which sense impressions are representations nonetheless. This sense can be brought out by noticing the contrast between the claims that sense impressions stand for worldly objects (which they do not) and that sense impressions stand in for worldly objects (which they do). That is, sense impressions, qua the proximate causal antecedents of worldly objects, are mental proxies for these objects. They represent the worldly objects that they do, not in the sense that a picture represents a scene, but rather in the sense that a lawyer represents his client. We respond conceptually—sometimes correctly, sometimes incorrectly—to worldly stimuli. What explains both of these kinds of response is that there is something in us that represents, stands in for, the worldly objects to which we are responding, and accounts for the ways in which we respond to them. Sense impressions, and importantly their qualities and relations, are posited as playing exactly this representative role. What I want to suggest is that sensations play a similar, although not identical, role in Kant’s theory of perception to the one played by sense impressions in Sellars’s account. Sensations, remember, are representations that refer to the subject as a modification of its state. The genus is representation in general (repraesentatio). [. . .] A perception that refers to the subject as a modification of its state is a sensation (sensatio). A320/B377

The A-Deduction and the Nature of Intuitions

157

Sensations are, in other words, mental states. Furthermore, as we have seen, Kant is clear that no non-conceptual mental state can represent a complex state of affairs as complex, and yet, as we have also seen, he is equally clear that he takes sensations to be a kind of representation. Below are passages from both his famous letter to his student Marcus Herz and an excerpt from his notes in which Kant both casts sensations as causal intermediaries between our minds and the world and makes explicit the sense in which he takes these to be representations. If a representation comprises only the manner in which the subject is affected by the object, then it is easy to see how it is in conformity with this object, namely, as an effect accords with its cause, and it is easy to see how this modification of our mind can represent something, that is, have an object. Ak 10:130; Correspondence, 133 Sensation represents individual objects insofar as they stimulate the senses. Ak 17:366, 3958; Notes and Fragments, 105 Sensations are merely “the manner in which the subject is affected by the object,” but Kant nonetheless casts them as a kind of representation in both passages. Furthermore, in both passages, he casts sensations as representing that which is their cause, although in neither does he make mention of sensations representing their causes as anything. The best way to understand these claims is exactly to recognize the distinction between representations that stand for their objects and representations that stand in for their objects. To do so, we would need to cast sensations as causal intermediaries between worldly objects and the conceptual representations of these objects as complex. Fortunately, I think we can do just this. Let’s return to our intuition of an elephant again. Earlier, we described the process of synthesis that produced this intuition as beginning with a manifold of sensations: short gray tail, big gray body, flat gray ear, long gray trunk. We noted at the time that while none of these sensations represented any of these parts of the elephant as parts (or as anything else, for that matter). What we can see now is the sense in which they are representations nonetheless. We encounter an elephant. This encounter produces certain modifications in our bodily state: we have a manifold of sensations. This manifold of sensations, in turn, causes us to respond to the elephant by producing a conceptually structured intuition, ‘this elephant,’ the elements of which are the very sensations (or new ones produced by the imagination) that triggered the synthesizing process in the first place. In this intuition these sensations do come to represent the parts of the elephants as parts. Pre-perceptually each sensation merely corresponds to some part of the elephant: each is a

158

The A-Deduction and the Nature of Intuitions

modification of the state of the subject produced by encountering some part of an elephant. As such, these sensations are the bodily representative, so to speak, of the parts of the elephant. For our intuition of the elephant to be the result of an encounter with an elephant, there must be something about the complex mental state that is our intuition of it that corresponds to such a piecemeal encounter. Kant posits sensations as precisely this something: The effect of an object on the capacity for representation, insofar as we are affected by it, is sensation. [. . .] I call that in the appearance which corresponds to sensation its matter, A20/B35 Sensations are the immediate result of encountering an object; they are “the effect of an object on the capacity for representation.” They also, “correspond to” the matter of such objects. That is, when one encounters an object, a manifold of sensations is produced in the experiencing subject, which sensations correspond to the parts of the object encountered. Recall the following passage that we encountered in the previous chapter.47 They serve as the mental representative of the parts of the object the function of which is essential to the process of synthesis that eventually produces an intuition of the object. We cannot synthesize elephant parts. We can synthesize sensations of elephant parts. Thus, sensations of elephant parts stand in for, represent, elephant parts in our cognitive lives. To think of oneself as encountering an object requires thinking of one’s intuition of that object as consisting of a temporally extended manifold of sensations produced in a systematic way that corresponds to one’s temporally extended encounter with the parts of this object.48 Compare the example of the elephant that we have been considering with the example of an “indistinct” representation of a house from the Jäsche Logic. We glimpse a country house in the distance. If we are conscious that the intuited object is a house, then we must necessarily have a representation of the various parts of this house, the windows, doors, etc. For if we did not see the parts, we would not see the house itself either. But we are not conscious of this representation of the manifold of its parts, and our representation of the object indicated is thus itself an indistinct representation. Ak 9:34; Logic, 545 To represent a house is to combine representations of the parts of the house. One cannot represent a house as a house without representing its various parts as being related in particular ways. It is, however, possible to represent the house without being conscious that one is representing these parts. This is precisely because the representations of the parts of the house are

The A-Deduction and the Nature of Intuitions

159

sensations—modifications of the state of the experiencing subject—which, while they are essential to accounting for the constitution of the intuition of the house, are not themselves directly apperceived. Sensations represent the parts of the house in the sense that they serve as the mental representatives of those parts, and as such are the necessary elements of the complex representation of the house as complex (even if we do not perceive them as such). In fact, this is precisely wherein Kant’s picture differs from the one presented by Sellars: Kant takes sensations to be not just the causal antecedents of intuitions but also their constituents. This, in turn, is where Kant enjoys an advantage over Sellars vis-à-vis responding to the non-conceptualist. For Sellars, sensations “guide from without”: sensations causally mediate between worldly objects and conceptual responses to these but do not provide any constraint on the content of the intuitions that they cause. For Kant, since intuitions are literally composed of sensations, and since the concepts-qua-inferential-rules that provide the structure of such intuitions are rules for uniting manifolds of sense,49 there is a clear way in which the content of intuitions is determined by sensations, and in turn by the world. An intuition of an elephant is complex representation consisting of sensations united in accordance with the concept-qua-inferential-rule for synthesizing precisely these kinds of representations, ‘elephant.’ The state of the subject that is the manifold of sensations produced by an encounter with an elephant corresponds to the matter of the elephant. This manifold of sensations represents, stands in for, the manifold of elephant parts encountered and is itself taken up and conceptually/inferentially structured to form the intuition, the content of which, therefore, consists of a representation of these elephant parts as standing in a certain relationship to one another.50 That is, it is because sensations correspond to the matter of the elephant that the intuition formed from such sensations represents an elephant. Returning to Sellars’s picture for a moment, we can say a little bit more about how such a process results sometimes in correct conceptual representations of the world and sometimes incorrect ones. The rules that the understanding uses to combine sensations are rules that govern the combination of specific kinds of sensations. So, if one kind of sensation is produced, say a sensation that is normally the sensation produced by a red object, the understanding will apply its rules to produce a representation of a red object. If, however, this sensation is produced as a result of the subject being affected by a white object in red light, the understanding will also apply its rules to produce a representation of a red object. So, for Kant, like for Sellars, sensations will play a role in explaining both our correct and incorrect conceptual responses to worldly objects. The difference between Sellars and Kant is that Sellars is hesitant, at least in Science and Metaphysics, to say anything more about the relation of sensations to intuitions than that the former is the cause of the latter. In fact, he argues that understanding

160

The A-Deduction and the Nature of Intuitions

sensations as being the literal constituents of intuitions is one version of what he calls the Myth of the Given: the idea that there could be nonconceptual representations that nonetheless represent something as being someway. [I]t is only if the manifold is mistakenly construed as belonging to the conceptual order that it makes sense to suppose that it, so to speak, bodily or literally becomes a part of the resulting intuitive representation. If it is, as I take it to be, non-conceptual, it can only guide ‘from without’ the unique conceptual activity which is representing of thissuches as subjects of perceptual judgment.51 Sellars’s thought here is that in order for the understanding to be able to unite a manifold of sensations into a representation of a complex state of affairs as complex, it must do so according to conceptual-inferential rules, and conceptual-inferential rules only apply to items already in the conceptual order. For example, intuitions are subject to rules of inference because they have the form this-such, the ‘such’ being a conceptual component that itself already locates intuitions within the inferential network formed by these rules. Since sensations do not have this conceptual component, they are the non-conceptual causal antecedents of intuitions, they cannot be part of an inferential structure, and thus they cannot be literal parts of conceptually structured intuitions. So, Sellars concludes that sensations merely “guide from without.” This would seem to leave Sellars with the question of what the parts of an intuition are, but he would reject that way of putting this question because it presupposes that representations of complex states of affairs as complex must have parts. As Sellars sees it, at least early on in his career, intuitions are, in a sense, simple representations. They have no parts. We simply find ourselves with intuitions—it turns out that we later discover that they are caused by sensations—and begin our cognitive lives from there. Such intuitions are not the representations that they are in virtue of having any internal structure but only in virtue of the place they hold in the network of inferential rules that tie these simple intuitions together with one another. Our picturing the world, for Sellars, begins with intuitions as the simple elements, not with sensations. Of course, we have seen that the case is different with Kant. The threefold synthesis is precisely the story of how intuitions come to have the internal structure that Sellars denies that they have. If this is to be a tenable position, Kant must have some response to Sellars’s argument that this is only possible by mistakenly taking sensations to belong to the conceptual order. Fortunately, we have already seen what this response of Kant’s will be. As it operates to unite sensations into an intuition, the imagination is a blind faculty. It does not consult the concept-qua-inferential-rule, compare sensations to that rule, and then unite them accordingly. Rather, it

The A-Deduction and the Nature of Intuitions

161

unites sensations in such a way that they have a structure that models or instantiates that provided by these rules, but it does so without having the represent either that rule to itself, nor those sensations as being subject to that rule. In retrospect, so to speak, we can explain these activities in terms of the relevant rule, and so these activities are, in that sense, rule following, but there is no need for the understanding itself, in its role of uniting sensations into intuitions, to be able to represent itself this way. Sensations can be united to form complex representations of complex states of affairs as complex (intuitions) without ever having themselves to be parts of the conceptual order.52 Of course, casting sensations as the elements of conceptually structured intuitions might seem to be in conflict with passages such as the following, where Kant appears to indicate that no representation stands between an intuition and its object. Since no representation pertains to the object immediately except intuition alone, a concept is thus never immediately related to an object, but is always related to some other representation of it. A68/B93 Kant certainly seems to be saying here that no representation stands in between an object and the intuition of that object, and that would certainly seem to imply that sensations cannot be causal intermediaries between objects and our conceptual representations of them (intuitions). On the other hand, we have already seen that Kant does take sensations to be the matter from which intuitions are formed, and that he also takes sensations to be a kind of representation. So, it is clear that there are representations that stand in between intuitions and their objects. The key to resolving this tension is to notice that Kant has previously restricted his discussion in this passage to cognitions. Concepts represent their object by relating intuitions to each other. This is the sense in which they are always related to some other representation (cognition) of the object. In that sense, they represent their objects only indirectly. Intuitions, by contrast, represent their objects “immediately.” That is, they represent their objects not by relating any other cognitions, representations of complexes as complex, to one another. An intuition is the most basic representation of a complex as complex, and therefore represents complex objects as complex “directly.” Intuitions are first-order representations, while concepts are second-order ones (representations of representations, as Kant puts it). Intuitions are composed of sensations, which are also representations, but only in the extended sense that we have just described: intuitions are composed of mental states that are the causal intermediaries between objects and our conceptual representations of them. Thus, none of what Kant is addressing in this passage concerns sensations, which as we have seen are not cognitions. Sensations are simply not under consideration in the context

162

The A-Deduction and the Nature of Intuitions

of the Metaphysical Deduction, which concerns the elements of judgment, and therefore not sensation. So, in fact, what passages like these show is that while intuitions and sensations are both representations for Kant, there is something fundamentally different about the sense in which this is true of each. Intuitions, like concepts, are cognitions: they represent their objects as complex. Sensations are not cognitions: they do not represent their objects as complex. Yet sensations represent nonetheless. Understanding sensations as mental proxies, as standing in for their objects rather than standing for them, allows us to make sense of this claim of Kant’s. As mental proxies, sensations will also be both the proximate causal antecedents of intuitions as well as their components. Worldly objects affect the subject; they produce a modification of its state. This modification does not represent anything as being anyway. To do this requires the deployment of concepts, and when sensations first make their appearance in the mind, concepts have not yet been deployed. Sensations yield no cognition at all; thus something must be added to them a priori if experience is to be possible. Beyond the a posteriori representation only the a priori representation from concepts can be added, and this can only be the connection (synthesis) insofar as it is determined a priori (for the mere comparison of sensations gives nothing except sensation, and no object). Ak 18: 393; Notes and Fragments, 310 Sensations are merely the way the subject is affected by the worldly objects that it encounters. Such modifications of the state of the subject are not, by themselves, cognitions: they have not yet been synthesized, combined, to form a representation of an object as a complex of its parts. Such encounters, however, do subsequently prompt the understanding to perform its perceptual synthesis, the end product of which is an intuition: determinate singular complex representations of objects as the complex of their parts. Another way to put this is that what our conceptually structured intuitions are, “ontologically” speaking, are unities of sensations.53 An intuition is, after all, a modification of the state of the experiencing subject. The key difference between an intuition and a sensation in this regard is that an intuition is, as we have seen, subject to conceptual norms. In fact, it is a unity of sensations that have become subject to such norms. Just as a pawn is “ontologically” just a piece of wood of a certain shape but is a pawn in virtue of being subject to the rules of chess, so an intuition is “ontologically” speaking composed of sensations but is a intuition by being subject to the norms of the concept that structure these sensations. Intuitions represent objects by being unities of manifold of sensations—these are the modifications of the state of the subject out of which such intuitions

The A-Deduction and the Nature of Intuitions

163

are composed—but that this is the case is something that we know via a transcendental investigation, or as I put it earlier, a logical analysis of the notion of representation. We can now see roughly how the compromise between conceptualists and non-conceptualists can be brokered. The conceptualist is right that all perceptions—products of perceptual synthesis—are conceptually structured. The non-conceptualist is right that there must be non-conceptual representations that provide for a worldly constraint on perception. The compromise is won by understanding that the non-conceptual representations that provide the worldly constraint on perception are not themselves perceptions. Rather, they are sensations, the mental proxies of worldly objects that guide the formation of intuitions, but which themselves do not represent anything as complex (which the products of perceptual synthesis all do). What concepts we form to represent the world will be constrained by what sensations we have insofar as we cannot conceive of those sensations as being united according to rules that do not apply to them. On the other hand, to represent any object as an object requires having a concept in place to unite representations (stand ins) of the parts of that object into a representation of that object as a complex of those parts. Even with this, though, we have not yet completed our investigation into the nature of intuitions. This is because while we have now established that and how intuitions represent complex states of affairs as complex, and the sense in which perception is conceptual while still accommodating input from the world, we have still not nearly said enough about the complex states of affairs that are thereby represented. In particular, what intuitions represent are complex objects, and we have not yet said anything about what it is to represent an object as opposed to a state of affairs. For Kant, this essentially involves the denial of all three of Hume’s conclusions. It requires representing the parts of the complex as necessarily connected; it requires representing the object as existing independently of our representation of it (as part of the external world); and it involves representing oneself as the single subject of such a representation, persisting through time. It is to these details that we will turn in the next chapter.

NOTES 1. Sellars’s work on Kant has largely been presented to recent audiences through McDowell’s engagement with it. As that engagement is in the context of McDowell’s own reading of Kant’s texts, this is not an ideal way for such audiences to try to understand Sellars, and the evidence seems to be mounting that many have been led to misunderstand Sellars’s work as a consequence of this form of presentation. (This is, of course, no fault of McDowell’s.) Other recent attempts to understand Kant in a broadly Sellarsian vein include Rosenberg, Accessing Kant and O’Shea, Kant’s Critique. 2. See Landy, “Incongruent Counterparts.”

164

The A-Deduction and the Nature of Intuitions

3. Among philosophers of mind who are also interested in Kant’s approach to these questions, the following are prominent reactions to this form of objection. McDowell, Having the World in View proposes that since there is no sense to be made of the world other than through the use of concepts, when we think of the world, we think of it as itself conceptually structured. Thus, a conceptually structured intuition is precisely what is needed to provide for a worldly constraint on thought. Ginsborg, “Was Kant a Nonconceptualist?” also presents a form of this objection. Her worry is that it seems to reverse the most plausible order of explanation for the origin of our empirical concepts: that we acquire such concepts from experience of objects in the world, not that we need such concepts in order to experience such objects. Her response is to propose that while Kant is right that there is a normative element in perception, this is not provided by the activation of any particular norm but rather by the requirement that perceptual representations be produced merely according to some rule or other and must be accompanied by a feeling of being constrained by a norm, although not necessarily any specific one. Hanna, “Kant and Nonceptual Content” argues, as we will here, that Kant’s use of the term ‘intuition’ is ambiguous between full-fledged conceptual representations of this kind and entirely non-conceptual representations. The plan for the current chapter is to present a theory that meets this form of objection and also demonstrates why each of these views would be inadequate in Kant’s view. 4. Each of the philosophers in the previous note could also be listed here as also advancing exegetical evidence from Kant’s texts as well. Additionally, though, Allais, “Non-Conceptual Content” proposes that Kant’s actual view is one according to which while concepts are necessary for representing objects, objects are not the only things that we perceive, or of which we form intuitions. We also perceive what she calls particulars, which can be represented without the use of concepts. Again, I hope to show here that this view is both inadequate to the philosophical work that Kant sees his theory as performing and a less than ideal exegetical match for Kant’s texts. 5. It is thanks to Henrich, “Kant’s Notion of a Deduction” that we know the full extent to which Kant modeled his deduction on the juridical practices of eighteenth-century German courts. 6. For a similar approach to a slightly different justificatory problem cf. Sellars, “Induction and Vindication” and Sellars, “First Principles.” For a similar approach to the Transcendental Deduction see Rosenberg, Accessing Kant. For a discussion of Rosenberg’s approach see Landy, “Premise That Even Hume Must Accept.” 7. Designated hitters, bench warmers, and rain-outs aside. 8. Kitcher, Kant’s Thinker argues that what she calls the togetherness of our mental representations (conceiving oneself as a single subject of experience persisting through time) and the mineness of our mental representations (attributing a manifold of mental representations to oneself) are different projects and that we must understand Kant as first and foremost engaged with only the former. I will argue in Chapter 6 that attributing a manifold of representations to oneself is precisely how one conceives of oneself as a single subject of experience persisting through time for Kant, and so these two projects cannot possibly come apart for Kant. It is because Kitcher denies that the ‘I think’ is a purely formal representation in this way, despite Kant’s repeated assertions of that thesis, that she is able pry these two projects apart. I will address Kitcher’s treatment of this issue, as well as others’, in Chapter 6 as well. 9. Of course, it is controversial whether Descartes actually makes this inference. It will suffice for present purposes to see that it is an inference with which Kant’s Modern predecessors are concerned, and to which Kant pays a good

The A-Deduction and the Nature of Intuitions

10. 11.

12.

13. 14.

165

deal of attention. For a more charitable interpretation of how Descartes moves from something like D1 to something like D2 via what he calls analysis, see Nelson, “Descartes’ Ontology of Thought.” Descartes, Philosophical Writings, 19. Of course, Hume does not really reject Descartes’s conclusion either. Famously, in the Appendix to the Treatise he writes, “But upon a more strict review of the section concerning personal identity, I find myself involv’d in such a labyrinth, that, I must confess, I neither know how to correct my former opinions, nor how to render them consistent.” (T App. 10; SBN 633) The principle for uniting the self that Hume sought, Kant thinks, is simply ‘These thoughts are mine.’ There is no reductive principle according to which we can identify all and only my thoughts. We can only so identify them by presupposing the self that has such thoughts. The thoughts that are mine are simply the ones that I have. This is the thesis that the representation ‘I think’ is representationally simple, which is one of three analytic propositions that we will see in Chapter 6 constitute the content of that representation. The nature of the ‘+’s here will be our topic later. For now these just signal some sort of combination of a manifold of intuitions occurring as part of a single cognition. Later we will see that this combination is inferential. Cf. Brook’s notion of global representation and single global object: Global representation = df: a representation that has a number of particular representations and/or their objects or contents as its single global object. Single global object = df: an intentional object that represents a number of intentional objects and/or the representations that represent them, such that to be aware of any of these objects and/or their representations is also to be aware of other objects and/or representations that make it up and of the collection of them as a single group. (Brook, Kant and the Mind, 33)

Brook also notes the special role that the Categories of relation play in such representations, i.e., that we form such global representations by representing the parts of the object represented as standing in necessary causal relations to one another, which is a thesis that will be front and center in Chapter 4. Furthermore, Brook follows Kitcher in noting the inadequacy of an associationist approach (e.g., Hume’s) to such representations on the grounds that “associations can be too promiscuous” (122). As will become apparent in a moment, where Brook and I part company is that Brook, like Kitcher, understands Kant as pursuing a project in cognitive psychology, and understands the A-Deduction as providing an account of our faculty psychology. I, on the other hand, understand the A-Deduction as providing a kind of logical analysis of the conditions of the representational content of a “global representation” and take Kant’s account of such representations to be essentially normative and inferential. 15. More on this in Chapter 6. 16. Longunesse, Kant and the Capacity to Judge, 56–58. 17. The A-edition version begins in the section, “On the synthesis of recognition of the concept” and carries over into the following section, “Provisional explanation of the possibility of the categories as a priori cognition.” As Longuenesse notes, in the A-Deduction version, Kant does not explicitly draw the connection between representing an object via applying concepts to intuitions in judgments and the deployment of the Categories, so there are no A-edition passages listed for steps 6 and 7 below.

166

The A-Deduction and the Nature of Intuitions

18. As it turns out, in addition to improving the clarity of the main argument of the Deduction, Kant also sets out to clarify some the consequences that he takes the various parts of that argument to have, and those are not of immediate concern just now. So, the argument with which we are concerned occurs in sections §15 through §20. 19. This form of argument is what Guyer, Kant and the Claims of Knowledge, 76 cites as “what is currently in danger of becoming the received interpretation” and attributes to Kitcher, Kant’s Transcendental Psychology, Allison, Kant’s Transcendental Idealism, Becker, Selbstbewusstein und Erfahrung, and McCann, “Skepticism and Kant’s B Deduction” (Guyer, Kant and the Claims of Knowledge, 432, n4). While Guyer’s fear of this becoming the received view may not have been realized—there appears to be no danger of anything concerning Kant garnering enough consensus to become “received”—he is certainly right that many prominent scholars find this form of argument in the Deduction, and I am happy to count myself in agreement with them. 20. Thus, as I have argued elsewhere on independent grounds—Landy, “Incongruent Counterparts”—the suggestion in Hanna, “Kantian Nonconceptualism” that Kant uses the example of incongruent counterparts to demonstrate that spatial representations are entirely non-conceptual cannot be right. 21. See also “A cognition serves an internal use insofar as it helps us to see the manifold in the object. It serves an external use when one wants to distinguish the object from others, or to see its agreement with others.” (Ak 24:836; Logic, 291) 22. Guyer, “Transcendental Deduction” argues that this passage is incompatible with much of what else Kant claims in the Transcendental Deduction and so should be excised from our understanding of the Critique. The reading presented here enjoys the benefit of making the passages in which such claims appear consistent with the rest of the Deduction, the relevant passages concerning synthesis, and the Critique more generally. So, I do not take this purported inconsistency as evidence of the unreliability of such passages, but rather I take Kant at his word that this is a thesis that he holds, and set out to show (here as well as elsewhere) that this is consistent with the rest of his views. 23. Here is a brief selection of further samples. The senses alone still do not make experience; rather, experience is the judgment of the understanding concerning the combination of our sensible representations. Sensible certainty, accordingly, is nothing but the certainty of the cognized connection of the senses. The senses and understanding constitute all our cognitions. The senses give appearance. The understanding connects it, and this makes experience and experience, then, is the cognized connection or unity of appearances. (Ak 24: 856, Logic, 308) All combinations are made through the mind. (Ak 17:668 4681; Notes and Fragments, 174) If therefore I ascribe a synopsis to sense, because it contains a manifold in its intuition, a synthesis must always correspond to this, and receptivity can make cognitions possible only if combined with spontaneity. (A97) But since every appearance contains a manifold, thus different perceptions by themselves are encountered dispersed and separate in the mind, a combination of them, which they cannot have in sense itself, is therefore necessary. There is thus an active faculty of the synthesis of this manifold in us, which we call imagination. (A120) (The imagination, as we will see in

The A-Deduction and the Nature of Intuitions

167

the next section, is just the understanding, the faculty for using concepts applied to the deliverances of sensibility.) Combination does not lie in the objects, however, and cannot as it were be borrowed from them through perception and by that means first taken up into the understanding, but is rather only an operation of the understanding. (B134–135) They are only rules for an understanding whose entire capacity consists in thinking, i.e., in the action of bringing the synthesis of the manifold that is given to it in intuition from elsewhere to the unity of apperception, which therefore cognizes nothing at all by itself but only combines and orders the material for cognition, the intuition, which must be given to it through the object. (B143) It is one and the same spontaneity that, there under the name of imagination and here under the name of understanding, brings combination into the manifold of intuition. (B162n) Now the representation of a composite, as such, is not a mere intuition, but requires the concept of a compounding, so far as it is applied to the intuition in space and time. So this concept (along with that of its opposite, the simple) is one that is not abstracted from intuitions, as a part-representation contained in them, but is a basic concept, and a priori at that—in the end the sole basic concept a priori, which is the original foundation in the understanding for all concepts of sensible objects. (Ak 20: 271, Theoretical Philosophy, 363) In my judgment everything depends on this: since, in the empirical concept of something composite (des Zusammengesetzten) the composition (Zusammensetzung) itself cannot be given or represented by means of mere intuition and its apprehension, but can only be represented by means of the self-active connection of the manifold given in intuition—that is, it can be represented only in a consciousness in general (which again is not empirical)—it follows that this connection and its functioning under a priori rules, rules that constitute the pure thought of an object in general (the pure concept of the understanding) must be in the mind. The apprehension of the manifold must be subject to this pure concept of the understanding insofar as it constitutes one intuition and insofar as it [the pure concept] constitutes the condition of all possible experiential knowledge of what is composite (or of what belongs to what is composite, i.e., something that requires a synthesis), experiential knowledge that is expressed by means of these principles. It is commonly supposed that the representation of a composite as such is given included with the representation of the apprehended manifold, and that thus it does not entirely belong, as however it really must, to spontaneity, etc. (Ak 11:376; Correspondence, 434–435) 24. Hanna, “Kant and Nonconceptual Content,” 259; Allais, “Non-Conceptual Content,” 387. 25. See also “sensation[,] how I am affected by the presence of an object” (Ak 24:44; Logic, 30); “Here, likewise, the laws of the understanding and of reason are, as it were, absent and not at home[;] there is simply no reflection, and instead certain sensations simply occur” (Ak 24:167; Logic, 132); “Sensation is the representation of our present condition insofar as it originates from the presence of a certain object” (Ak 24:235; Logic, 187); “A perception that is

168

26.

27.

28.

29.

The A-Deduction and the Nature of Intuitions merely related to the subject as its state is called sensation” (Ak 16:548, 2836; Notes and Fragments, 54). See also “Sensations, e.g., stimulation, excitement, are the matter of sensibility, intuition is its form” (Ak 24:807; Logic, 267) and “Experience is without doubt the first product our understanding brings forth as it works on the raw material of sensible sensations” (A1). Also, Ak 24:132; Logic, 103, Ak 24:752; Logic, 485, Ak 24:252; Logic, 202, Ak 24:752; Logic, 486. Obviously, the inferences that actually make up the representation of time will be far more complicated than this toy example indicates. In particular, they will be subject to the constraints that Kant outlines in the Analogies and Refutation of Idealism. These will be the subject of Chapter 5. Here we can see why the proposal in Ginsborg, “Was Kant a Nonceptualist?”—that the normative element in intuition formation is merely that the subject recognize that they are being guided by some norm or other but not by any specific norm—will not do. This second phase of synthesis requires not just some concept or other to guide it but rather a concept specific enough to guide the reproductive imagination in “selecting” which representations to reproduce. Recall that in the previous chapter we considered examples of Kant discussing such rules with respect to not only analytic judgments but synthetic ones as well. E.g., An example of a synthetic proposition is, To everything x, to which the concept of body (a + b) belongs, belongs also attraction (c). (Ak 9:111; Logic, 607) Thus in the appearance x, in which a is a concept, there must be, in addition to what is thought through a, conditions of its specification which make necessary a rule whose function is determined through b. (Ak 17:665, 4680; Notes and Fragments, 172)

30.

31. 32. 33. 34. 35. 36.

The example above, of the construction of a representation of a triangle, is another such instance, as Kant repeatedly stresses that the rules for such constructions, because they involve a priori intuitions of space, are necessarily synthetic. “But since every appearance [object] contains a manifold, thus different perceptions by themselves are encountered dispersed and separate in the mind, a combination of them, which they cannot have in sense itself, is therefore necessary” (A120). Objects are complexes of their parts. So, to represent objects, we must combine representations of these parts into a single representation of these complex objects as complex. Kant himself is clear on this point: “For we cannot judge by means of sensation, but we can by means of intuition.” (Ak 24:44; Logic, 31) Pippin, Kant’s Theory of Form, 48. Pippin, Kant’s Theory of Form, 50. Pippin, Kant’s Theory of Form, 51. Ginsborg, “Kant and the Problem of Experience” takes a similar approach to these issues. Ginsborg, “Was Kant a Nonconceptualist?” is an attempt at such an account but is inadequate to the task at hand. Roughly, Ginsborg’s solution is to understand the unity of conceptually structured intuitions as consisting of the subject’s associating certain sensations with one another and accompanying this association with a feeling of being governed by a norm (not a specific norm but merely some norm or other). In addition to the mistake of attempting to reduce norms to associations and feelings, the indeterminacy in the

The A-Deduction and the Nature of Intuitions

37. 38. 39. 40.

41. 42.

43. 44.

169

norms governing such representations on which Ginsborg relies undermines the role that Kant assigns to conceptual representations in making cognitions determinate. I.e., this “solution” undermines Kant’s response to the problem of complex representation that we outlined in the previous chapter. See fn. 28. See Wittgenstein, Philosophical Investigations, §219: “When I obey a rule, I do not choose. I obey blindly.” Sellars, “Some Reflections,” 207. Of course, the vast majority of language learning does not proceed via explicit training at all. In the previous chapter we encountered a similar example of Kant’s: that of the savage who encounters a house but who does not have the concept ‘house’ (Ak 9:33; Logic, 544–545). I argued there that the best way to understand this example is as illustrating the difference between two experiencing subjects who use different concepts to unite similar manifolds of sensation. Kant there articulates this difference as a difference in the form that each intuition takes and alludes to the similar sensations as a shared matter. Longuenesse, Kant and the Capacity to Judge, 203. Longuenesse does gesture toward the so-called synthesis speciosa in certain places as the faculty that will do this work. Longuenesse is certainly not the only scholar that makes this move, although I can find no one that describes the synthesis speciosa as anything other than a via negativa. This is an especially popular move for those who maintain that Kant is a non-conceptualist about representation: because something needs to do the work of forming a representation of a complex as complex, but it is puzzling what it is to do that work and how it gets done. Thus, the synthesis speciosa has gone from Kant’s name for the collective work of the imagination that he details in the A-Deduction, which I hope to have shown is clearly conceptual, to the deus ex machina of complex representation. One of the advantages that I hope to be able to claim here is having given reason to reject the notion that the synthesis speciosa is a non-conceptual faculty, and to have offered some detail of how it can nonetheless do the work credited it to it. Longuenesse, Kant and the Capacity to Judge, 118–119, emphasis added. The objection from the non-conceptualist that I will consider here is that there must be a worldly constraint on cognition, that there is a sense in which we must, e.g., see elephants before we can think about them. Of course, this is only one argument concerning the existence, nature, and function of nonconceptual content. It is outside the scope of the current study to address any of the several other arguments on this topic, although the conclusions reached here do provide the resources for doing just that. Hanna, “Beyond the Myth” provides an excellent catalogue of arguments for the existence of non-conceptual content. Below is his list of the titles that he has given each of these arguments (although not his very helpful summaries of them), followed by either the place here where each argument is, at least implicitly, addressed or a citation pointing to a rebuttal of the argument along lines sufficiently similar to the one presented here. 1. Phenomenal Richness—Schulte, “Fineness of Grain.” 2. Perceptual Discrimination—Rosenberg, Accessing Kant, Chapter 3. 3. From infant and nonhuman animal cognition—Sellars, “Mental Events.” 4. From the distinction between perception (or experience) and judgment (or thought)—in the current chapter the close of the previous section, along with the current section.

170

The A-Deduction and the Nature of Intuitions 5. From the knowing-how vs. knowing-that (or knowing-what) distinction— Rosenberg, Thinking Self, Chapter 6. 6. From the theory of concept-acquisition—this argument is a version of the worldly constraint argument that the current chapter is an attempt to address. 7. From the theory of demonstratives—Landy, “Incongruent Counterparts.”

45. 46. 47.

48. 49. 50. 51. 52.

Hanna also presents an additional argument in that article concerning the nature of spatial representation. This argument overlaps significantly, I think, with 7 here, and is accordingly addressed explicitly in Landy, “Incongruent Counterparts.” Allais, “Non-Conceptual Content” also provides an argument that concerns the nature of spatial representation. I will address Allais’s argument briefly in the following chapter. Sellars, Science and Metaphysics, 17. Sellars, Science and Metaphysics, 18. See also “Since the representation borrows its ground from the represented thing, it agrees with the latter in that it is composed out of its partial concepts in the same way that the whole represented thing is composed out of its parts.” (Ak 16:76; Notes and Fragments, 34) For Sellars, sense impressions are a theoretical posit; for Kant, sensations are a transcendental posit. Which are nonetheless applied blindly. In the next chapter we will see that Kant holds that objects are represented specifically as having their parts stand in necessary relations with each other. Sellars, Science and Metaphysics, 16. An interesting footnote to this dialectic is that Sellars himself seems to have realized the tenability of this position later on in his career. Here is Sellars, not in Science and Metaphysics, but in a later lecture subsequently published as “The Role of Imagination in Kant’s Theory of Experience,” talking about how images—the perspectival, or schematized, representations of which intuitions are the aperspectival counterparts—are formed. In the first place, the productive imagination is a unique blend of a capacity to form images in accordance with a recipe, and a capacity to conceive of objects in a way which supplies the relevant recipes. (Sellars, “The Role of Imagination,” §31) And again in, “Kant’s Transcendental Idealism”: Actually the most useful concept is that of a sequence of acts of representing which can reflectively be classified as conforming to a rule which is (at least in principle) graspable by thought. (Sellars, “Kant’s Transcendental Idealism,” 413)

The job of the productive imagination is to unite sensations into images in accordance with (not according to) a recipe (a set of conceptual-inferential rules). Sensations are both the causal antecedents to such images and their literal components. 53. The scare quotes here are because the issue of the ontological status of the self is a delicate one. Here the ontology of sensations is a kind of transcendental ontology, which is not the same as the transcendent one that the rational psychologist posits, nor the empirical one that the empirical psychologist investigates.

The A-Deduction and the Nature of Intuitions

171

REFERENCES Allais, Lucy. “Kant, Non-Conceptual Content and the Representation of Space.” Journal of the History of Philosophy 47 (2009): 383–413. Allison, Henry. Kant’s Transcendental Idealism. New Haven: Yale University Press, 1983. Becker, Wolfgang. Selbstbewusstein und Erfahrung: Zu Kants tranzendentaler Deduktion und ihrer argumentativen Rekonstruktion. Freiburg: Alber, 1984. Brook, Andrew. Kant and the Mind. Cambridge: Cambridge University Press, 1994. Descartes, Rene. The Philosophical Writings of Descartes: Volume 2. Translated by John Cottingham, Robert Stoothoff, and Dugald Murdoch. Cambridge: Cambridge University Press, 1985. Ginsborg, Hanna. “Kant and the Problem of Experience.” Philosophical Topics 34 (2006): 59–106. Ginsborg, Hanna. “Was Kant a Nonconceptualist?” Philosophical Studies 137 (2008): 65–77. Guyer, Paul. Kant and the Claims of Knowledge. Cambridge University Press, 1987. Guyer, Paul. “The Transcendental Deduction of the Categories.” In The Cambridge Companion to Kant, edited by Paul Guyer, 123–60. Cambridge: Cambridge University Press, 1992. Hanna, Robert. “Kant and Nonconceptual Content.” European Journal of Philosophy 13 (2005): 247–90. Hanna, Robert. “Kantian Nonconceptualism.” Philosophical Studies 137 (2008): 41–64. Hanna, Robert. “Beyond the Myth of the Myth: A Kantian Theory of Non-Conceptual Content.” International Journal of Philosophical Studies 19 (2011): 323–98. Henrich, Dieter. “Kant’s Notion of a Deduction and the Methodological Background of the First Critique.” In Kant’s Transcendental Deductions: The Three ‘Critiques’ and the ‘Opus Postumum.’ Stanford: Stanford University Press, 1989. Hume, David. A Treatise of Human Nature. Edited by L. A. Selby-Bigge. New York: Oxford University Press, 1974. Hume, David. A Treatise of Human Nature. Edited by David Fate Norton and Mary J. Norton. New York: Oxford University Press, 2000. Kant, Immanuel. Lectures on Logic. Translated and edited by J. Michael Young. Cambridge: Cambridge University Press, 1992. Kant, Immanuel. Critique of Pure Reason. Translated by Paul Guyer and Allen W. Wood. Cambridge: Cambridge University Press, 1998. Kant, Immanuel. Correspondence. Translated and edited by Arnulf Zweig. Cambridge: Cambridge University Press, 1999. Kant, Immanuel. Theoretical Philosophy After 1781. Edited by Henry Allison and Peter Heath. Translated by Gary Hatfield, Michael Friedman, Henry Allison, and Peter Heath. Cambridge: Cambridge University Press, 2002. Kant, Immanuel. Notes and Fragments. Edited by Paul Guyer. Translated by Curtis Bowman, Paul Guyer, and Frederick Rauscher. Cambridge: Cambridge University Press, 2005. Kitcher, Patricia. “Kant’s Real Self.” In Self and Nature in Kant’s Philosophy, edited by Allen Wood, 113–47. Ithaca: Cornell University Press, 1984. Kitcher, Patricia. Kant’s Transcendental Psychology. New York: Oxford University Press, 1990. Kitcher, Patricia. Kant’s Thinker. New York: Oxford University Press, 2011. Landy, David. “The Premise That Even Hume Must Accept.” In Self, Language, and World: Problems from Kant, Sellars, and Rosenberg, edited by Jim O’Shea and Eric Rubenstein, 28–46. Atascadero, CA: Ridgeview Publishing Co., 2010.

172

The A-Deduction and the Nature of Intuitions

Landy, David. “What Incongruent Counterparts Show.” European Journal of Philosophy 21 (2013): 507–24. Longuenesse, Beatrice. Kant and the Capacity to Judge. Translated by Charles T. Wolfe. Princeton: Princeton University Press, 1998. McCann, Edwin. “Skepticism and Kant’s B Deduction.” History of Philosophy Quarterly 2 (1985): 71–89. McDowell, John. Having the World in View. Cambridge, MA: Harvard University Press, 2009. Nelson, Alan. “Descartes’ Ontology of Thought.” Topoi 16 (1997): 163–78. O’Shea, James. Kant’s Critique of Pure Reason: An Introduction and Interpretation. London: Acumen Press, 2012. Pippin, Robert. Kant’s Theory of Form. New Haven: Yale University Press, 1982. Rosenberg, Jay. The Thinking Self. Philadelphia: Temple University Press, 1986. Rosenberg, Jay. Accessing Kant: A Relaxed Introduction to the Critique of Pure Reason. Oxford: Oxford University Press, 2005. Schulte, Thomas. “Fineness of Grain and the Myth of the Given.” Master’s thesis, San Francisco State University, 2012. Sellars, Wilfrid. “Some Reflections on Language Games.” Philosophy of Science 21 (1954): 204–28. Sellars, Wilfrid. “Induction as Vindication.” Philosophy of Science 31 (1964): 197–231. Sellars, Wilfrid. Science and Metaphysics. Atascadero, CA: Ridgeview Publishing Company, 1967. Sellars, Wilfrid. “The Role of Imagination in Kant’s Theory of Experience.” In Categories: A Colloquium, edited by Henry W. Johnstone, Jr., 231–45. University Park: Pennsylvania State University Press, 1978. Sellars, Wilfrid. “Mental Events.” Philosophical Studies 39 (1981): 325–45. Sellars, Wilfrid. “On Accepting First Principles.” In Philosophical Perspectives, Vol. 2: Epistemology, edited by James E. Tomberlin, 301–14. Atascadero, CA: Ridgeview Publishing Co., 1988. Wittgenstein, Ludwig. Philosophical Investigations. Translated by G. E. M. Anscombe. Oxford: Blackwell Publishers, 1958.

4

The Object of Representation

What we saw by the close of the previous chapter is that intuitions are complex representations of complex objects as such, and that they are, therefore, conceptually-inferentially structured. We also saw that the constituents of such intuitions are sensations, which are the non-conceptual representations of the parts of the object represented by the intuition. What all of this left open, however, is the question of precisely what this complex object represented by an intuition is represented as. We know that an object is represented as a complex of its parts, but there is more to Kant’s story here, and we have not said much more about this essential part of his system. It is the purpose of the current chapter to fill that lacuna. In particular, I will argue here that in the so-called objective side of the A-Deduction and in the B-Deduction, Kant makes use of a sophisticated inferentialist account of what an object is and how it is that we represent such objects. An object, according to that account, is first and foremost that which explains the consistency and coherence of manifold of sensations with which we find ourselves. Kant infers from this that it is also thereby a complex of parts each of which is necessarily connected to all the others. As such, objects necessarily exist independently of being perceived and are capable of continuing to exist when not perceived. Thus, if Kant can give a successful account of what it is that is represented by an intuition, this will constitute a direct rebuttal of Hume’s claims—(NE) and (EW)—that we cannot so much as represent either a necessary connection between two distinct items and that we cannot represent the external world qua that which exists independently of our perception of it and which continues to exist when we are not perceiving it. Kant’s account will be that intuitions represent objects, as a complex of parts necessarily connected and independently existing, by placing representations of these parts into a very particular kind of inferential relation with one another. As I will demonstrate, while Kant is not explicit on this point, everything he does say about this kind of representation points to a kind of inference is that valid not in virtue of its logical form but rather in virtue of the content of the concepts employed in the judgments that it concerns.

174

The Object of Representation

It should be noted at the outset that this is a fairly radical suggestion for at least two reasons. To see this we can once again consider the example of an intuition that we were using in the previous chapter: that of an elephant. Sadly, some elephant tails are not attached to any elephant body. Thus, there is no necessary connection between elephant tails and elephant bodies. Thus, the inferential license from ‘Here is an elephant’s tail’ to ‘Next to this is an elephant’s body’ is a bad one. It represents those two parts as being necessarily connected, though they are not. Thus, there is at least a sense in which this inferential license, despite being an entirely natural one to grant, is a mistake. If, however, such licenses are mistakes, one must wonder what licenses would be correct. The answer, of course, is that the correct ones are those that really do accurately portray the necessary connections among worldly objects, but given our elephant example, those would seem to be licenses of an entirely different order. One way of proceeding, which is not Kant’s, would be to hold that only inferences that accurately represent logical necessity will do.1 Kant, however, employs a broader view of necessity here according to which what we represent via the inferential structure of our mental representations is something very much like physical necessity, i.e., the lawful connections between worldly items. For Kant, this leads to two fairly radical conclusions. The first is that since what we represent in representing objects is the lawful connections of worldly items, all such representation is already a kind of theoretical activity. That is, in representing any object at all, we are already engaged in a project that requires hypothesizing about what the laws that govern our world are. All representation of objects is, in this sense, the representation of a kind of theoretical entity. As we will see farther along, Kant sees representation itself as a kind of theoretical-explanatory practice. This, in turn, implies that for Kant the distinction between the objects of everyday common sense and those of theoretical science will be more methodological than ontological. I.e., everyday objects are just as much theoretical-explanatory objects as are those of theoretical science: the difference between the two being merely that the former are where our explanations start and the latter are what replace such flawed attempts. Thus, what the items of intuition ought ultimately to be will be determined by what laws of nature actually govern our world. The second radical conclusion that we will encounter in delineating Kant’s theory concerns the way we represent such objects. What I will show is that Kant holds that the physically necessary connections between worldly objects are represented via the inferences we license, forbid, etc., concerning such objects. Representing such necessities includes representing the subjunctive conditionals that follow from them, and as has been argued in the contemporary literature, representing subjunctive conditionals requires allowing for inferences that are valid not merely in virtue of their logical form (e.g., modus ponens) but also in virtue of their content.2 That is, we ought not to see an argument such as

The Object of Representation

175

P1. There was a flash of lightning just now. C. There will be a clap of thunder soon. as an enthymeme, in need of an additional premise such as P2. If there was a flash of lightning just now, there will be a clap of thunder soon. Instead, we must think of the inference from P1 to C as valid in its own right. Kant’s theory of representation crucially involves the notion of what has been called material inference. As I said, these are fairly radical conclusions, and we must proceed slowly in working our way up to them. In fact, it is only in the next chapter, on the Analogies, that we will focus our attention on how the account of nonlogical, or material, inference that I will present here operates in Kant’s account of concept replacement. In this chapter we deal only with the matter of material inference itself. My procedure for doing that will be as follows. I will begin by situating this chapter’s discussion in its broader context by reviewing the place of the representations of objects in the argument of the Transcendental Deduction. I will then proceed to Kant’s claim that what makes a representation a representation of an object is that it represents the parts of its object as necessarily connected. Next, I will turn to the contemporary literature on inferentialism as a guide to understanding such representations, after which I will turn to the business of showing that a similar account of representation underlies Kant’s texts. Most importantly, I will show that Kant takes the primary object of our representations to be nature, the sum total of all lawfully connected objects of representation, and that he takes there to be a distinct form of inference that is used to represent the “real,” as opposed to merely logical, necessity that governs such a system. THE ROLE OF REPRESENTATIONS OF OBJECTS IN THE TRANSCENDENTAL DEDUCTION We can begin by briefly reviewing the outline of the argumentative structure of the Transcendental Deduction presented in the previous chapter. Recall that we there found Kant following Hume in rejecting Descartes’s inference from (D1) [I think x] and [I think y] and [I think z] to (D2) [The I that thinks x] = [the I that thinks y] = [the I that thinks z].

176

The Object of Representation

Descartes takes the fact that he can introspectively observe that he thinks x, and that he can introspectively observe that he thinks y, and that he can introspectively observe that he thinks z, to imply that it is one and the same thing, he, that has all of these thoughts. Hume, on the other hand, notices that when we introspect, we find exactly the matter that Descartes does— this or that perception—but that this is not sufficient to yield an experience of the self—something that endures through time and is the subject of these perceptions. Because Hume thinks that an experience of the self is the only ground that could warrant the further premise needed to make the argument valid, when he fails to find such an experience, he famously rejects Descartes’s conclusion. Kant, unlike Hume, does not but instead sets out to find its ground. What he discovers is (D2) does follow (and is only made possible by) (K) I think [x + y + z]. That is, while introspectively observing a manifold of various representations is not sufficient for conceiving of oneself as a single, unified subject persisting through time, forming a single representation, the content of which includes a manifold of representations, is sufficient. Otherwise put, he sees that we would be justified in inferring that one and the same thing experiences all of x, y, and z if we were justified in thinking that one and the same thing thinks something else whose elements included x, y, and z. Again, if x, y, and z were three parts of a single cognition had by a single individual, then it would follow trivially that the I that thinks x is the same as the I that thinks y and the same as the I that thinks z. As we noted in the previous chapter, the representation that plays this role for Kant is an intuition: a complex representation of a complex object, and we there examined the logical structure of such representations. The goal of the current chapter will be to shift our focus, as Kant does between the A-Deduction and B-Deduction, from that logical structure to what such a representation represents: an object. That is because this representation is just an intuition, a complex representation of a complex object as a complex of its necessarily connected parts. We can see this in the following passage, in which Kant outlines the path he will take from the necessity of conceiving oneself as a single subject of experience persisting through time to forming the kind of representation that employs the pure a priori concepts of the understanding. The supreme principle of all intuition in relation to the understanding is that all the manifold of intuition stand under conditions of the original synthetic unity of apperception. All the manifold representations of intuition stand under [this principle] insofar as they must be capable of being combined in one consciousness; for without that nothing could be thought or cognized through them, since the given representations

The Object of Representation

177

would not have in common the act of apperception, I think, and thereby would not be grasped together in a self consciousness. Understanding is, generally speaking, the faculty of cognitions. These consist in the determinate relation of given representations to an object. An object, however, is that in the concept of which the manifold of a given intuition is united. B136–B137 The first paragraph of this quotation is set up. An intuition, as we have seen, is a complex representation of a complex object. As such, it is a single representation had by a single subject of experience. So, in order for an intuition to be formed, it must meet whatever conditions there are on being a complex representation had by a single subject of experience. As we have seen, one such condition is that the intuition must be a kind of unity. It must bring together a manifold of representations (sensations) into a single representation of a complex. An intuition is what will play the role outlined by (K) above. It will be the [x + y + z] that is thought by a single self. What Kant adds to this in the second paragraph here is that it is in virtue of being a representation of an object that an intuition comes to meet these conditions. It is in virtue of being of an object that an intuition comes to be united, comes to be a single representation of a complex object as complex. The single cognition, [x + y + z], is the use of an object-concept whose elements include a manifold of sensations and that, because it is a single cognition, is had by a single thinker. Returning to our example from the previous chapter will help illustrate why it is that Kant thinks that this particular kind of cognition can do this work, while other kinds cannot. Suppose again that one is presented with the following diachronic manifold of intuitions. t1: This short gray tail. t2: This big gray body. t3: This big, flat gray ear. t4: This long gray trunk.3 According to Kant, it is by thinking of such a manifold of intuitions using an object-concept, such as ‘elephant,’ that we unite them in a single cognition. So, instead of merely having these representations each in turn, we think of all of them together by thinking something like, ‘This elephant is very large,’ where the representations are all themselves elements of the single complex representation ‘this elephant.’ Of course, as we saw in Chapter 2, it is not enough that these representations be, for instance, merely associated with one another. They must be united in a much stronger sense: they must be components of a single thought. It is Kant’s suggestion that this unity is only achieved in the thought of an object, and we must now ask after what it is about such a representation that makes it so particularly well suited to its task. This will be the topic of the following section.

178

The Object of Representation

THE REPRESENTATION OF OBJECTS We have now come to the main task of the current chapter: to understand Kant’s account of the representation of complex objects as complex. The question at the close of the previous section concerned the role that Kant sees such a representation playing in our cognition more generally: it serves to unite a manifold of representations into a single cognition, which is necessarily had by a single thinker. That provides one condition for what a successful theory of such a representation must be, at least for Kant. It must be able to serve this function. That, however, is not the only condition for such a theory. If Kant’s account is going to be considered a success, it must not only meet the conditions that are given by his own theoretical apparatus, but since it is specifically an account of what it is to represent an object, it must also answer to certain of our pre-theoretical notions of what such a representation is like. Once again, in this regard, Kant takes his cue from Hume. Here is Kant, in the A-Deduction, presenting a brief argument concerning the relation of the unity of the representation of objects to the unity of the self (about which more later). It is clear, however, that since we have to do only with the manifold of our representations, and that X which corresponds to them (the object), because it should be something distinct from all of our representations, is nothing for us, the unity that the object makes necessary can be nothing other than the formal unity of the consciousness in the synthesis of the manifold of representations. A105, emphasis added What is important from this quotation for current purposes is not the conclusion at which Kant arrives but the reasoning that leads to it. Kant clearly takes it as a condition of his account of representing objects that objects are “distinct from all of our representations.”4 That is noteworthy because it is precisely the second of the two conditions that Hume presents in the Treatise as what is required to provide an adequate account of the idea of body. We ought to examine apart those two questions, which are commonly confounded together, viz. why we attribute a CONTINU’D existence to objects, even when they are not present to the senses; and why we suppose them to have an existence DISTINCT from the mind and perception? T 1.4.2.2; SBN 187–8 Hume’s thought here is that to provide an adequate account of the idea of body, one needs to provide accounts of the two ideas that together constitute that idea: the idea of something that is distinct from our perceptions,

The Object of Representation

179

and the idea of something that continues to exist when we are not perceiving it. As Hume goes on to point out, the two questions are not unrelated. These two questions concerning the continu’d and distinct existence of body are intimately connected together. For if the objects of our senses continue to exist, even when they are not perceiv’d, their existence is of course independent of and distinct from the perception; and vice versa, if their existence be independent of the perception and distinct from it, they must continue to exist, even tho’ they be not perceiv’d. T 1.4.2.2; SBN 187–8 Of course, at this point in the Treatise Hume takes himself already to have shown that no idea can ever represent anything meeting either of these conditions.5 As we have already seen, the dialectic that unfolds between Hume and Kant is one in which Kant rejects Hume’s theory of mental representation and sets out to show exactly how we can have the kinds of representations that Hume was led to dismiss. So, it is appropriate here that Kant accepts Hume’s desiderata of what it would be to represent an object, even as he rejects Hume’s claim that we can represent no such thing. In fact, it is by accepting that desiderata that Kant’s account will meet Hume’s head on. If successful, Kant will show how it is that we can represent exactly what Hume argued we could not. So, Kant takes over Hume’s condition that what is represented in representing an object is something distinct from our representations of it. He also thereby takes on Hume’s second condition that what is represented in representing an object is something that can continue to exist independently of our representation of it. In presenting Kant’s account of the representation of objects, then, we will use these two additional conditions, along with the internal condition that such a representation must be able to account for the unity of the subject of experience, as the criteria of success of that account. For the sake of convenience, we can label each of these conditions as follows. DISTINCTNESS: The representation of an object must be a representation of something that is distinct from our representation of it. CONTINUITY: The representation of an object must be a representation of something that can continue to exist when we are not representing it. UNITY: The representation of an object must be such that the subject of that representation is necessarily a single subject of experience persisting through time. Kant must be able to account for how we form representations that meet each of these three conditions, and if he does this, his account will be a success.

180

The Object of Representation

To see how Kant’s account will meet each of these conditions, we can begin by determining what it is that Kant takes to be distinctive of the representation of an object. Kant’s answer is that by applying an object-concept to a manifold of representations, by thinking of these representations as being of, for instance, an elephant, what we crucially add to that manifold is an element of necessity. Here is Kant in a section of the Analogies where he makes his reasoning about this explicit. We have representations in us, of which we can also be conscious. But let this consciousness reach as far and be as exact and precise as one wants, still there always remain only representations, i.e., inner determinations of our mind in this or that temporal relation. A197/B242 The manifold of sensations alone is merely a complex of representations; it is not yet a representation of a complex, nor a representation of a complex object. Now how do we come to posit an object for these representations, or ascribe to their subjective reality, as modifications, some sort of objective reality? [. . .] If we investigate what new characteristic is given to our representations by the relation to an object, and what is the dignity that they thereby receive, we find that it does nothing beyond making the combination of representations necessary in a certain way, and subjecting them to a rule; and conversely that objective significance is conferred on our representations only insofar as a certain order in their temporal relation is necessary. A197/B242 To “confer objective significance” on our representations means to make our representations come collectively to signify objects, come to form a representation of an object, and it is Kant’s thesis here that we do this by representing those representations themselves as bearing certain necessary relations to one another. That is, what makes a mere manifold of representations into single representation of an object is our subjecting such representations to conceptual rules that determine their necessary (conceptual) relations to one another. Recall the following quotation: the concept of body makes necessary the representation of extension, and with it that of impenetrability, of shape, etc. A106, emphasis added We noted there that the notion of necessity operative here is not one concerning the associative tendencies of the human mind but is rather the necessity appropriate to a concept-qua-inferential-rule, i.e., a normative necessity.

The Object of Representation

181

For example, predicating the concept ‘body’ of some thing commits one to also predicating the concept ‘extension,’ of that thing and ‘attraction,’ etc. We can add to this now that what such inferential norms do is unite the manifold of representations that they govern into representations of objects. Here again is Kant: We find, however, that our thought of the relation of all cognition to its object carries something of necessity with it, since namely the latter is regarded as that which is opposed to our cognitions being determined at pleasure or arbitrarily rather than being determined a priori, since insofar as they are to relate to an object our cognitions must also necessarily agree with each other in relation to it, i.e., they must have that unity that constitutes the concept of an object. A104, emphasis added The normative necessity that governs our representations of objects prevents us from making just whatever judgments we want concerning those objects. The conceptual rules that govern such representations commit us to the truth of certain judgments and the falsity of others. It is such commitments that constitute our representation of the object because it is by undertaking such commitments that we come to represent our representations as being subject to not just our own judgmental whims but also by the constraints imposed on us by our contact with the world. In committing myself to the inferential norm that requires me to predicate ‘extension’ of anything of which I also predicate ‘body,’ I undertake a constraint on my judging activities; I oppose those activities as “being determined at pleasure or arbitrarily” with their being determined by something other than me, i.e., by the object that they thereby represent. Thus the original and necessary consciousness of the identity of oneself is at the same time a consciousness of an equally necessary unity of the synthesis of all appearances in accordance with concepts, i.e., in accordance with rules that not only make them necessarily reproducible, but also thereby determine an object for their intuition, i.e., the concept of something in which they are necessarily connected. A108, emphasis added Concepts, by uniting manifolds of sensations, determine an object for the intuition so formed. The concept of this object, that which is represented by such an intuition, is “the concept of something in which they are necessarily connected.” That is, the conceptual necessities connecting the manifold of representations not only transform that manifold into a representation of an object but also thereby come to posit that object as the ground of those necessities. The concept of an object, and the necessary connection of its parts, is the concept of that which explains the conceptual necessities among

182

The Object of Representation

the representations of those parts; it is the concept of that at which those conceptual necessities aim. The pure concept of this transcendental object (which in all of our cognitions is really always one and the same = X) is that which in all of our empirical concepts in general can provide relation to an object, i.e., objective reality. Now this concept cannot contain any determinate intuition at all, and therefore concerns nothing but that unity which must be encountered in a manifold of cognition insofar as it stands in relation to an object. A109 The concept of an object in general is the concept of that which relates to our representation of an object, and the unity that we represent such objects as having is just that unity that they need to have in order to explain the unity of the manifold of sensations qua representations of objects. To summarize, then, we represent objects by representing our representations of the parts of objects as being related to one another via conceptual necessities, rules of inference licensing, forbidding, and requiring certain judgments. In so representing such representations, we posit the object as that which explains these conceptual necessities, as that which these conceptual necessities are meant to picture. Before moving on to the question of how we represent the elements of the manifold of representations as being subject to these conceptual necessities, and thereby represent the parts of the object so represented as themselves being subject to causal laws, a few further observations are in order. To start, we need to recall that it is not just our actual manifold of sensations that is united in a representation of an object, but included in that representation will also be various possible sensations as well. That is, a representation of an elephant represents that elephant as having as parts more than just the parts of it that we perceive. We may not see the far side of the elephant, or its insides, or even the parts of it that we saw earlier, but that are obscured from us now. Still, when we represent an elephant, we represent it as having all of these parts. Notice that if Kant can account for how we manage to include these representations as parts of the more complex representation, he will have taken a large step in meeting Hume’s DISTINCTNESS and CONTINUITY conditions. If a representation of an object represents that object as having parts that we do not perceive, then at least these parts will be represented as having an existence that is distinct from our perception of them and as being capable of continuing to exist when unperceived. Recall also that the parts of the elephant that we do not perceive, and yet that are components of our representation of the elephant nonetheless, make their way into that representation by way of the concept ‘elephant.’ That is, it is the concept, ‘elephant’ that determines what representations must be included in a complex representation for that representation to

The Object of Representation

183

be of an elephant. The concept does this by serving as an inferential rule, a rule that licenses inferences from judgments concerning one part of an elephant to those concerning others. In particular, such inferential rules are, as Kant would put it, schematized concepts: they determine not only what an elephant is but also how it will appear via sensation to creatures like us.6 Now, for all that we have said, we still do not have Kant’s actual account before us, but this most recent observation gives us the necessary thread to follow in order to find that account. What we know is that according to the account, in order to represent the parts of objects as being necessarily connected to one another, we have to represent the representations of these parts as being subject to certain conceptual necessities. What we do not yet have is Kant’s account of how we represent two things, be they parts of objects or representations of parts of objects, as necessarily connected. The important clue that we have turned up is in remembering that concepts are inferential rules. If we also remember from the Metaphysical Deduction that concepts are a special kind of meta-representation for Kant,7 the pieces start to come together. This is because representing the parts of objects as necessarily connected by representing representations of those parts as being subject to certain conceptual necessities is to represent an object by employing a meta-representation: a concept-qua-inferential-rule. MATERIAL RULES OF INFERENCE What we must notice now is that the inferential rules at hand are of a very particular kind. Consider Hume’s notion of relations of ideas.8 There are certain propositions that are true, for Hume, in virtue of the relations of the ideas in those propositions.9 ‘If Joe went to the movies and Sam went to the movies, then Joe went to the movies’ is true because the idea ‘Joe went to the movies’ is literally contained in the idea ‘Joe went to the movies and Sam went to the movies.’ It is easy enough to extrapolate such an account to arrive at a theory of inference according to which the only valid inferences are also those made in virtue of certain relations of ideas. To put it in a more contemporary idiom, it would be easy to imagine Hume holding that the only valid inferences are those made in virtue of their logical form. In the contemporary literature on inferentialism, the point is often made that such a view of inference will be untenable for any theory of representation that includes among its explananda subjunctive conditionals. As we recently noticed, such conditionals will be essential to Kant’s own theory: it is a necessary part of representing an object that one represents the parts of those objects that one would see, were one encountering it from a perspective that is different from one’s actual one. So, it will be worth our while to explore this point. Consider, then, the following inference.10

184

The Object of Representation

1. x is to the north of y. 2. y is to the south of x. According to the account of inference that we have extracted from Hume, this inference is an enthymeme. It is not an example of modus ponens, modus tollens, conjunction elimination, or any other formal rule of inference. Since it is not valid in virtue of its logical form, the ring of validity that it has can only be due to a suppressed premise, the theory goes.11 The valid argument that this one stands in for is really the following. 1. x is to the north of y. 2. If x is to the north of y, then y is to the south of x. 3. y is to the south of x. The problem now is that while (3) does make for a formally valid inference when paired with (1) and (2), there is a closely related inference for which (3) is of no help. 4. Suppose x were to the north of y. 5. Then, y would be to the south of x. (3) is of no help here because it concerns only the actual relation of x to y. It concerns only what is the case if x actually is to the north of y, not what would be the case were x to be north of y. In order to validate the inference from (4) to (5), what is needed is a proposition that applies to these counterfactual situations as well. For this, our Humean might be tempted to offer, 6. If x is to the north of y, then necessarily, y is to the south of x. While this proposition would certainly be strong enough to complete the argument from (4) to (5), the inferentialist is happy to claim victory at this point. That is because if (6) can be adopted, so can the meta-level rule of inference, 7. ‘x is to the north of y’ implies ‘y is to the south of x.’12 That is, if (6) is true, then (4) can never be true where (5) is false. Thus, the inference from (4) to (5) is valid. Thus, the inferentialist takes himself to have shown that in any system of representation robust enough to encompass subjunctive conditionals, there will be inferences that are valid in virtue of something other than merely their logical form. Sellars called such inferences material inferences, and returning to Kant, we can see that such inferences seem to be just what he has in mind for the counterpart relations that structure the representations of objects as the necessary connections of their parts. Another example from the contemporary inferentialist literature will make this clearer. Consider, then, the following inference.

The Object of Representation

185

8. There was just a flash of lightning. 9. There will be a clap of thunder soon. On our Humean account of inference, this argument is an enthymeme, in need of a supporting premise such as 10. If there was just a flash of lightning, there will be a clap of thunder soon. Of course, what our inferentialist will point out is that while (10) might be sufficient to complete this argument, the closely related argument, 11. Suppose there had been a flash of lightning just now. 12. Then there would be a clap of thunder soon. requires instead, 13. If there is a flash of lightning, then, there will be a clap of thunder, which implies that meta-level rule of inference, 14. ‘There was just a flash of lightning’ implies ‘There will be a clap of thunder soon.’ The key point to notice here is that such rules of inference function precisely as meta-representations: they function by representing various object-level (or immediate, as Kant puts it) representations (of the lightning and thunder) as being related in a certain way. Namely, they represent the intuition ‘this lightning now’ and ‘the thunder soon’ as being necessarily connected.13 To see how all of this connects up with Kant’s account of the representation of objects as the necessary connection of their parts, consider one final argument. 15. Suppose there were a short gray tail in front of me now. 16. There would be a large gray body at the end of that tail. Qua enthymeme, this argument would depend on the further premise, 17. If there is a short gray tail in front of me, then necessarily there is a large gray body at the end of that tail. This premise, in turn, implies the rule of inference, 18. ‘There is short gray tail in front of me’ implies ‘There is a large gray body at the end of that tail.’

186

The Object of Representation

This rule of inference is part of the concept of an elephant, and as such, represents the parts of an elephant, those that are perceived as well as those that are not, as necessarily connected. It represents these parts as being necessarily connected by representing representations of these parts as being necessarily connected. It represents these representations as necessarily connected by licensing certain material inferences between judgments containing them. Thus, a concept is a meta-representation—a rule of inference—that represents objects as the necessary connection of their parts. Of course, some inferences are enthymematic and such arguments will need supplementation by a further premise to be made valid. Here is Kant’s description of such inferences from the Logic, which provides a helpful way of distinguishing these inferences from the material inferences just described. A formal inference of reason is one that not only contains everything required as to matter but also is expressed correctly and completely as to form. Opposed to formal inferences of reason are covert ones (cryptica), to which all those can be reckoned, in which either the premises are transposed, or one of the premises is left out, or, finally, the middle concept alone is combined with the conclusion. A covert inference of reason of the second kind, in which one premise is not expressed but only thought, is called a truncated one or an enthymeme. Ak 9:132; Logic, 625 An enthymeme is a covert inference in which a premise is not expressed but only thought. That is, it is an inference in which there is a suppressed premise that needs to be added to the argument to make it good. On the Humean view of inference with which we have been working, material inferences are just these: they need supplementation by some further premise, which we have recently uncovered will have to have the strength of a subjunctive conditional that will in turn itself be sufficient for validating a material rule of inference. So, we argued on Kant’s behalf that such subjunctive conditionals need not be added to these arguments as further premises because expressing them in this way is redundant. The material rule of inference suffices. Returning to the passage from the Logic, then, we see that such arguments are not enthymematic precisely because we do not need to regard them as requiring a premise that is left out or only thought but not expressed. Since the premise at issue is sufficient for establishing the validity of the corresponding material inference, requiring that it be included in the argument as a premise amounts to requiring that there should be a rule for how to apply rules of inference. Kant addresses this

The Object of Representation

187

inappropriate demand in the course of his explication of General Logic at the beginning of the Principles. If the understanding in general is explained as the faculty of rules, then the power of judgment is the faculty of subsuming under rules, i.e., of determining whether something stands under a given rule (casus datae legis) or not. General logic contains no precepts at all for the power of judgment, and moreover cannot contain them. [. . .] Now if it wanted to show generally how one ought to subsume under these rules, i.e., distinguish whether something stands under them or not, this could not happen except once again through a rule. But just because this is a rule, it would demand another instruction for the power of judgment, and so it becomes clear that although the understanding is certainly capable of being instructed and equipped through rules, the power of judgment is a special talent that cannot be taught but only practiced. A132–133/B171–B172 Once again we see the nature of rule following at the center of Kant’s concern, and here he anticipates some of Wittgenstein’s insight into that issue, although for our purposes it will be more instructive to compare him to a nineteenth-century ally: Lewis Carroll. As Carroll fantastically demonstrates in “What the Tortoise Said to Achilles” (Carroll, 1895), it cannot be a requirement of valid argument that the rules of inference governing that argument be featured as premises in it. What Carroll so vividly demonstrates is that if one were required to include in the premises of an argument the rules of inference that govern moving from the premises of that argument to the conclusion, one would thereby be required to add such premises ad infinitum. It is from exactly such considerations that Kant concludes that, “although the understanding is certainly capable of being instructed and equipped through rules, the power of judgment is a special talent that cannot be taught but only practiced.” One cannot require that all such rules be included: at some point one must simply draw an inference from the premises to the conclusion, i.e., one must practice, or exercise, the power of judgment. What we must now do is apply this lesson to the case of material rules of inference. The Humean view of inference that we have been considering violates precisely this cautionary note by requiring that the subjunctive conditionals that appear to be required to make good the inferences above be included as premises in those arguments. What such subjunctive conditionals state is that it is necessarily the case that if the premises of these arguments are true, then so will be their conclusions. As we have already noted, since what makes an argument valid is just that its conclusion cannot be false and its premises true, such subjunctive conditions are the equivalent of rules of inference that license a move from these premises

188

The Object of Representation

directly to their conclusions. The demand to have these expressed in the arguments themselves is the same absurd demand that the Tortoise makes of Achilles and that Kant frames as a demand “to show generally how one ought to subsume under these rules.” Such a demand leads to the infinite regress that both Kant and Carroll seek to avoid. Thus, such inferences should not be considered enthymematic because they do not leave out, or leave unexpressed, anything that itself must be cast as a premise in these arguments. At this point it is also worth recalling that we have already seen Kant endorse this kind of inference as we were following the trail from his thesis that concepts are rules, through his explication of rules as assertions under a condition, and these in turn as the major premises in syllogisms. We there noticed that Kant explicitly casts not only analytic judgments and their corresponding rules of inference as having this form but also synthetic ones, and thereby he endorses not only formal rules of inference but also material ones. An example of a synthetic proposition is, To everything x, to which the concept of body (a + b) belongs, belongs also attraction (c). Ak 9:111; Logic, 607 Thus in the appearance x, in which a is a concept, there must be, in addition to what is thought through a, conditions of its specification which make necessary a rule whose function is determined through b. Ak 17:665, 4680; Notes and Fragments, 172 In the first quotation, the condition ‘body’ serves as the major premise in an inference, not to a judgment that applies one of the concepts that are themselves part of the concept ‘body’ but to a judgment that applies the concept ‘attraction.’ Thus, applying the concept ‘body’ here represents the object to which that concept is applied as standing under a condition, which condition commits one to an inference to the conclusion that anything so represented also exhibits an attractive force. In the second quotation, we again have something, x, which is thought through one concept, a, which in turn requires that one also think x through a further distinct concept, b. It is not until the Second Analogy that we get Kant’s complete story about what such material inferences will involve for creatures like us, but one can already see here that Kant takes employing the concept of an object to require not only the thinking of that object through its own concept but also through the others to which it is connected via synthetic judgments and their counterparts, material rules of inference. Furthermore, the distinction between formal rules of inference and material ones plays an important systematic role for Kant. This distinction is one between two different ways that we might structure our representings, and it corresponds to a distinction in what is thereby represented. That is,

The Object of Representation

189

for Kant these two kinds of inference correspond to two different kinds of necessity represented by each. Necessity can be classified into real and into logical necessity. Ak 28:558; Metaphysics, 323 Formal rules of inference represent the objects that they encompass as bearing certain logically necessary relations to one another. Material rules of inference, on the other hand, represent a different brand of necessity, what Kant here calls real necessity, which, as we have seen, represents its objects as bearing something much more like physically necessary connections to one another. Our representing an elephant as consisting of its parts necessarily related to one another is not a matter of formal inference alone and thus does not represent these parts as logically related to one another. Rather, in relating the representations of elephant parts to one another via material rules of inference, we represent those parts as bearing lawful relations to one another. In fact, a passage from the Blomberg Logic explicitly connects these two facets of representation: the difference between logically formal inferences and what Kant calls “real subordination,” on the one hand, and what is represented by the latter inferences, lawful relations of cause and effect, on the other. The subordination of concepts, however, can occur both logice and realiter. Logical subordination consists in the fact that I take that which is common to many concepts and thereby form for myself a universal concept, under which I can subordinate the individual representations. In this way I make for myself various generaL and I subordinate the speciesL and individua to them. Real subordination, however, consists in the fact that I actually combine concepts with one another, so that not only is one contained under the other, but instead they also cohere as causes and effects. Ak 24:260; Logic, 208 Logical subordination is the mere subsumption of one concept under another concept that already contains it. Syllogisms represented by judgments of the form ‘All x that are (a + b) are a’ are examples of logical subordination. Real subordination, on the other hand, combines concepts with one another in a way that goes beyond mere logical subordination. Real subordination represents the relations of causes and effects to one another. It is syllogisms represented by judgments of the form ‘All x that are (a + b) are c’ that are the representings by which these lawful relations are represented. In the following passage, Kant is emphasizing the necessary unity of all such representations with one another but does in a way that ought to seem very familiar to us by now.

190

The Object of Representation There is only one experience in which all perceptions are represented as in thoroughgoing and lawlike connection, just as there is only one space and time, in which all forms of appearance and all relation of being or non-being take place. [. . .] The thoroughgoing and synthetic unity of perceptions is precisely what constitutes the form of experience, and it is nothing other than the synthetic unity of the appearances in accordance with concepts. A110

The synthetic (nonlogical) unity of perceptions is nothing other than the synthetic (lawful) unity of the appearances in accordance with concepts. It is by uniting intuitions with one another via material rules of inference that we come to represent the objects that such intuitions represent as bearing lawful relations to one another. What is represented by the unity affected among the manifold of representations is the lawful relations of worldly items. Once again, material inference allows us to explain Kant’s claim exactly. Material inferences are the counterpart relations that we use to represent the (lawfully) necessary connections of the parts of the objects so represented.14 Returning to the B-edition of the Deduction, we can see again that Kant’s account of what is represented by the representation of an object is real necessity, i.e., laws of nature. Now since all possible perception depends on the synthesis of apprehension, but the latter itself, this empirical synthesis, depends on the transcendental one, thus on the categories, all possible perceptions, hence everything that can ever reach empirical consciousness, i.e., all appearances of nature, as far as their combination is concerned, stand under the categories, on which nature (considered merely as nature in general) depends, as the original ground of its necessary lawfulness (as natura formaliter spectata). B164–165 Most of what Kant says here is familiar from what has already transpired in the Deduction: all possible perception, representation of a complex as complex, depends for its combination on the Categories. What is new here is Kant’s addition that what is represented by such perceptions is nature and that what the Categories do vis-à-vis nature is ground its necessary lawfulness. Kant continues: The pure faculty of understanding does not suffice, however, to prescribe to the appearances through mere categories a priori laws beyond those on which rests a nature in general, as lawfulness of appearances in space and time. Particular laws, because they concern empirically determined appearances, cannot be completely derived from the categories,

The Object of Representation

191

although they all stand under them. Experience must be added in order to come to know particular laws at all; but about experience in general, and about what can be cognized as an object of experience, only those a priori laws offer instruction. B165 In pointing out that particular natural laws require experience for their grounding, Kant also makes two telling remarks here. Firstly, we learn that nature is “lawfulness of appearances in space and time,” which lawfulness is, of course, grounded in the categories. Secondly, we learn that nature is an instantiation of “what can be cognized as an object of experience.” The picture that emerges from these reflections is that the “ultimate” representation that the Categories make possible is that of nature as the necessary (lawful) connection of items in space and time and that this is a representation that is itself a specification of the concept of an object in general. This specification is achieved by adding to the general concept of an object our forms of intuition—space and time—and the empirical content provided by sensation. Working backward, if these are what are added to the representation of an object in general to make it into a representation of a nature, what the concept of an object must itself be is simply the concept of a complex of parts necessarily connected to one another. A representation of nature is simply the representation of the sum total of spatio-temporal parts as necessarily (lawfully) related to one another. It is not difficult to find Kant expressing just this thesis: The whole of nature in general is really nothing but a connection of appearances according to rules Ak 9:11; Logic, 527 nature, as the sum total of objects of experience Bxix nature is nothing in itself but a sum of appearances, hence not a thing in itself but merely a multitude of representations of the mind A11415 without the understanding there would not be any nature at all, i.e., synthetic unity of the manifold of appearances in accordance with rules A126–127 So, combining two fundamental insights from the B-Deduction—that concepts are inferential rules and that nature, as the sum total of objects, is the necessary (lawful) connection of its parts—we arrive at precisely the theory that we were recently investigating: an object-concept represents the parts

192

The Object of Representation

of the object represented as necessarily connected by serving as a material rule of inference, the validity of which derives not from its logical form but rather from its content. Notice again the differences between this interpretation of Kant’s theory of concepts and the one that we considered in Chapter 2 from Longuenesse. Longuenesse and I agree that concepts are a kind of inferential rule. I have now given more detail to that thesis by suggesting that we represent objects as the necessary connection of their parts by licensing certain material rules of inference regarding these parts. E.g., I represent an elephant’s tail as necessarily connected to its body by licensing an inference from, e.g., ‘Here is an elephant tail’ to ‘There is an elephant body.’ This additional detail is important because it also brings out a more robust point of disagreement with Longuenesse. On Longuenesse’s interpretation what is represented by a concept is an immanent universal. I.e., while intuitions represent objects proper, concepts represent their own kind of “object”: the universals that are immanent to the objects represented by intuitions. I earlier argued that this leaves Longuenesse’s Kant subject to the problem of the unity of the proposition: concepts and intuitions are independently contentful and each serves a similar representative function in a judgment (to “name” an object and a universal, respectively). We can now also see that it saddles Kant with an unnecessary ontological burden. Kant’s distal goal for his theory of mental representation is to put it to use in explaining how it is that we represent nature, the sum total of objects represented as subject to causal laws. On the interpretation that I have given, Kant earns exactly this: we represent objects in nature as causally connected by licensing material inferences between representations of such objects. There is simply no need here for Kant to take on the additional baggage of supposing that in doing this we also represent the universals immanent to such objects. Kant can have the causal laws that he wants without also committing himself to the “real existence” of universals. As he repeatedly stresses, causal laws are the form of nature, whereas objects, or substance, are its object. Analogously, and for good reason, concepts are likewise the form of mental representation, whereas intuitions are its elements. It is also worth pausing here to notice the contrast between the account just offered and that suggested in Allais, “Non-Conceptual Content.” Allais there argues for a place in Kant’s theory of mental representation for non-conceptual representations, which she takes to represent not objects but instead what she calls ‘particulars.’ Allais is not entirely clear about what the distinction between objects and particulars is, but she does cite Strawson’s remark that “material objects, people, and their shadows are all particulars.”16 She also indicates that it is an essential feature of a particular that it have a location in space. What we can now see is that all of these will fall under the broad heading of ‘objects’ for Kant, or at the very least under the even broader heading ‘nature.’ That is, Allais’s particulars are all part of the law-governed world that Kant understands as

The Object of Representation

193

being what is represented by the picture formed by relating intuitions to one another via material inferential rules. We have already seen that to represent two items as spatially related, because this is a representation of a complex as complex, requires the use of concepts for Kant. (It is spatial concepts that represent what might be called the laws of geometry.) Additionally, though, once we take account of Kant’s claim that all experience represents only law-governed nature, it is clear that the representations of “materials objects, people, and their shadows” will be conceptually structured. That is because the representations of all of these particulars will each be part of the one representation of nature as the sum total of all objects of experience, and this representation is a picture the elements of which are intuitions and the structure of which is concepts-qua-materialinferential-rules. If I am right about the reading of Kant, then it immediately follows from this that Kant’s account meets both the CONTINUITY and DISTINCTNESS conditions that we prescribed at the outset of this section. Material rules of inference are broad enough in their scope to include not only representations of parts of an object that are actually perceived but also those that would be perceived in different circumstances. So, some of the judgments that are licensed by such rules of inference will be ones concerning parts of objects that continue to exist when we are not actually perceiving them, and parts that can are distinct from our representation of them. More importantly, however, such material rules of inference also license judgments about objects themselves that support such counterfactuals. For instance, because the concept ‘elephant’ consists in part of material rules of inference strong enough to support counterfactual conclusions, a judgment such as ‘This is an elephant before me’ licenses other judgments such as ‘Were my eyes closed, this elephant still would be here.’ Now, it might occur to the reader to object here on Hume’s behalf that, while we have earned for Kant material rules of inference with a theory meeting the CONTINUITY and DISTINCTNESS conditions, we have only done so by allowing him judgments employing subjunctive conditionals. Wouldn’t Hume object that such judgments are nonsensical, since they go beyond what can be copied from any impression? What we must keep in mind here is that the dialectic to this point has proceeded as follows. Hume proposed his theory of mental representation according to which all representation is copying. Kant rejects that theory on the grounds that it is inadequate for accounting for certain representational phenomena that are accepted by Hume as necessary explananda. Thus, Kant is no longer obligated to respond to charges that stem from the particulars of Hume’s theory. He is free to take on whatever objects of representation he likes and to attempt to explain how it is that we form mental representations of such objects. So, if he holds that we represent objects by making judgments that take the form of subjunctive conditionals, then so long as he can account for how we do this—using, for example, rules of material inference—it is not

194

The Object of Representation

part of what he needs to prove that Hume also takes subjunctive conditionals as an explanandum. What we have arrived at now is an account that answers to the various claims that we have thus far seen Kant make regarding his theory of the representation of objects (save for those concerning the UNITY condition). To review, here is a brief list of some of his most important relevant claims. (a) What makes a concept an object-concept is that it represents the parts of the object as being necessarily connected. (b) An object-concept represents the parts of the object as necessarily connected by representing the representations of these parts as themselves being necessarily connected. A concept is a meta-representation. (c) A representation of an object represents that object as having parts that are not actually, or currently, perceived by the experiencing subject. (d) The representation of an object must be a representation of something as distinct from our representation of it. (e) The representation of an object must be a representation of something as being capable of continuing to exist when we are not representing it. (f) Concepts are rules for judging.17 The thesis of this section has been that what ties all of these claims together is the notion of a concept as a material rule of inference (f). A rule of inference is a meta-representation (b) insofar as the intuitions contained in judgments it connects are thereby represented as being necessarily connected. It is by relating these intuitions to one another that such rules represent the parts of objects represented by such intuitions as themselves being necessarily connected (a). Since these material rules of inference concern not only actual representations but also possible representations, they include in their scope representations of the parts of objects that are not perceived (c). In doing so, they also represent objects as being distinct from our representations of them (d), and as being capable of continuing to exist when we are not representing them (e). Material rules of inference certainly seem to be what Kant has in mind as the centerpiece of his theory of mental representation. Of course, the condition that is conspicuously missing from this list is the one that, in a sense, is the driving force behind Kant’s theory: the UNITY condition, which requires that the representation of an object must be such that the subject of that representation is necessarily a single subject of experience persisting through time. While our discussion of such representations will not be complete until we have addressed the UNITY condition, that task must be postponed for one more chapter. This is because while we have now seen how material rules of inference can be used to explain Kant’s

The Object of Representation

195

theory of the representation of objects, and thereby how his refutation of Hume’s account of the external world and necessary connection, at their most general levels, as is well known, Kant’s specification of such representations, and of his refutation of Hume, do not occur until the Analogies, which we have yet to address. Thus, in the next chapter, I will reconstruct Kant’s arguments in the Analogies with an eye toward the role of conceptsqua-material-rules of inference in them. In the following chapter, we will turn, finally, to the UNITY condition, and to the representation that is the transcendental unity of apperception.

NOTES 1. This is Wittgenstein’s view in Wittgenstein, Tractatus. 2. Cf. Sellars, “Concepts as Involving Laws,” Rosenberg, “Wittgenstein’s Theory of Language,” and Brandom, Making It Explicit. 3. Notice that we could have run this example synchronically, if what we wanted was a unified self at a time. For instance, we could have made our manifold out of the synchronic experiences (1) I see a gray trunk, (2) I hear a loud trumpeting sound, (3) I feel leathery skin, (4) I smell dung, etc., or even synchronic experiences such as (1) I see a gray patch in such-and-such a portion of my visual field, (2) I see a darker gray patch in such-and-such other part of my visual field, etc. 4. That is, Kant holds that phenomenal objects are necessarily represented as distinct from our representations of them. The question of the ontological status of noumenal objects is a different one, a discussion of which I must postpone until the Postscript. I will there argue that for Kant there is no content to be given to the notion of an object apart from any representation of it, but that we can contrast our current way of representing objects with other ways of representing them, and that this contrast is one that falls within the scope of Kant’s transcendental idealism. 5. “Now since nothing is ever present to the mind but perceptions, and since all ideas are deriv’d from something antecedently present to the mind; it follows that ‘tis impossible for us so much as to conceive or form an idea of any thing specifically different from ideas and impressions.” (T 1.2.6.8; SBN 67–8) 6. The most relevant feature of us being that time is the form of all of our intuition. I.e., we represent both objects of outer sense as well as those of inner sense as occurring in time. So, the rule for uniting a manifold of sensations of elephant parts, for example, must determine how an elephant will appear via a temporally extended manifold of sensations. 7. “In every judgment, there is a concept that holds of many, and that among this many also comprehends a given representation, which is then related immediately to the object.” (A68/B93) 8. EHU 4.1; SBN 26. 9. As we saw in Chapter 2, Hume has a very specific view of propositions, or judgments. We can, for the most part, put the particulars of his account aside for current purposes. We can also put aside Hume’s account of inference as association. 10. The example seems to originate from Rosenberg, “Wittgenstein’s Theory of Language,” although it is more prominently found in Brandom, Making It Explicit.

196

The Object of Representation

11. Things are tricky for Hume here, since it is plausible that (1) and (2) here will turn out to be the same idea, the same picture. Of course, that would imply that (1) and (2) mean the same thing, and when we expand the scope of our investigation of material inference to include not only inferences concerning spatial relations but also inferences such as that from ‘lightning now’ to ‘thunder soon,’ that view will become much less tenable. 12. Since drawing an inference is a kind of action, a more perspicuous representation of this rule might be something like, From ‘x is to the north of y,’ it is always permissible to infer ‘y is to the south of x’ and never permissible to infer ‘y is not to the south of x.’

13.

14. 15. 16. 17.

The specifics of how this ought to work out are outside the scope of the current inquiry. Cf. Brandom, Making It Explicit. Of course, we don’t actually represent lightning and thunder as parts of the same object at all. That is because our representations, ‘lightning’ and ‘thunder,’ are not representations of parts at all. They are representations of the same object, and so the inference from ‘lightning now’ to ‘thunder soon’ is not a representation of a necessary connection between two parts of one object, but it is rather part of a representation of single, self identical object. Recall that we first saw Kant explicitly invoke counterpart relations in his notes on Meier’s Auszug aus der Vernunftlehre in Chapter 2. “Representations of the mind” here refers to that which is represented by the mind, of course. Strawson, Individuals, 15. Notice that each of these is a condition on the representation of an object. That accords with the thesis encountered earlier that Kant recasts the Categories from being object-level representations of special metaphysical kinds to meta-conceptual rules for what counts as an object-concept. It also leaves open the issue of how to understand Kant’s transcendental idealism, which has been taken to concern the notion of an object apart from all conditions of its representation. I will argue in the Postscript that this understanding of noumenal objects is fundamentally misguided and amounts to a form of transcendental realism.

REFERENCES Allais, Lucy. “Kant, Non-Conceptual Content and the Representation of Space.” Journal of the History of Philosophy 47 (2009): 383–413. Brandom, Robert. Making It Explicit. Cambridge, MA: Harvard University Press, 1994. Carroll, Lewis. “What the Tortoise Said to Achilles.” Mind 4 (1895): 278–80. Hume, David. A Treatise of Human Nature. Edited by L. A. Selby-Bigge. New York: Oxford University Press, 1974. Hume, David. Enquiries Concerning Human Understanding and Concerning the Principles of Morals. Edited by L. A. Selby-Bigge. Oxford: Oxford University Press, 1975. Hume, David. A Treatise of Human Nature. Edited by David Fate Norton and Mary J. Norton. New York: Oxford University Press, 2000. Kant, Immanuel. Lectures on Logic. Translated and edited by J. Michael Young. Cambridge: Cambridge University Press, 1992. Kant, Immanuel. Lectures on Metaphysics. Translated and edited by Karl Ameriks and Steve Naragon. Cambridge: Cambridge University Press, 1997.

The Object of Representation

197

Kant, Immanuel. Critique of Pure Reason. Translated by Paul Guyer and Allen W. Wood. Cambridge: Cambridge University Press, 1998. Kant, Immanuel. Notes and Fragments. Edited by Paul Guyer. Translated by Curtis Bowman, Paul Guyer, and Frederick Rauscher. Cambridge: Cambridge University Press, 2005. Longuenesse, Beatrice. Kant and the Capacity to Judge. Translated by Charles T. Wolfe. Princeton: Princeton University Press, 1998. Rosenberg, Jay. “Wittgenstein’s Theory of Language as Picture.” American Philosophical Quarterly 5 (1968): 18–30. Sellars, Wilfrid. “Concepts as Involving Laws and Inconceivable without Them.” Philosophy of Science 15 (1948): 287–315. Strawson, Peter. Individuals. London: Methuen, 1959. Wittgenstein, Ludwig. Tractatus Logico-Philosophicus. Translated by D. F. Pears and R. F. McGuiness. London: Routledge, 1922.

5

Self and World in the Analogies of Experience

We have now completed, in a very general way, our initial portrait of Kant’s theory of mental representation. Object-concepts are material inferential rules that picture the objects of possible experience as being related to one another according to natural laws. What we have not yet seen, however, is precisely how this theory applies to the mental representations of creatures like us, specifically sensorily passive concept users whose forms of intuition are Space and Time. This further specification of these concepts occurs throughout the Principles, but it is the Analogies, and in particular the First and Second Analogies, that will concern us here. This is because it is here that Kant’s conclusions are most directly relevant to completing his case against Hume. Recall the three important conclusions from Hume that we have been considering. (NC) We have no idea that is an idea of a necessary connection, (EW) We have no idea that is an idea of the external world, and (SSE) We have no idea that is an idea of a single subject of experience persisting through time. We saw the very general outline of the ground of Kant’s rejection of the first two of these theses in the previous chapter, wherein we delineated Kant’s account of the representation objects. We also there took out a promissory note for the explanation of Kant’s refutation of the third thesis but noted that given the structure of the Transcendental Deduction, representing oneself as a single subject of experience persisting through time must not only be possible for Kant, it must also be necessary. This assumption is again at work in the Analogies, where it grounds the specific account that Kant gives of how creatures like us represent both necessary connection and the external world and so thereby also grounds his final, specific refutation of the first two theses. In the First Analogy Kant attempts to show that undertaking the norms that constitute the representation of the self specifically through time requires that we represent the world as consisting of a single substance that is never created or destroyed, which implies the negation of (EW). In the Second Analogy, he attempts to show that undertaking them

Self and World in the Analogies of Experience

199

with respect to the deliverances of inner sense as occurring in a determinate temporal order requires representing our representations as being of a world governed entirely by laws specifying the necessary connections between distinct events, which famously constitutes one of Kant’s main arguments against (NC). One aim of the current chapter, then, will be to delineate these arguments and their relation to Kant’s inferential theory of mental representation that we have thus far been examining. Another of its aims, though, will be to draw another, underappreciated conclusion from these arguments, which will round our portrait of Kant’s inferentialism. As we saw in Chapter 4, Kant holds that the concept of an object is the concept of that which grounds the conceptual necessities that unite our manifold of representations. The concept of an object is the concept of something that we take to provide the worldly constraint on our representing activities. When I license the material inference from ‘This tail is the tail of an elephant’ to ‘if I turn my head, I will see the body of an elephant,’ I commit myself to the conceptual connection between representations of elephant tails and representations of elephant bodies. What explains this normative constraint on my thinking is that what these representational activities are aiming to picture is an elephant, an object that consists in, among other things, the necessary connection between its tail and its body. The implication of this account of objects, and the specific form that it takes in the Analogies, that I want to draw out here is that this constitutes a kind of theoreticalexplanatory scientific realism that is made more specific and explicit in the Analogies.1 What I will argue here is that Kant’s argument in the First Analogy yields a kind of scientific realism—one that gives ontological priority to theoreticalexplanatory entities over those of more “direct” experience—and that this realism is best understood in light of the kind of inferentialism that we have been outlining on Kant’s behalf. Representing oneself as a single subject of experience persisting through time requires representing the world as consisting of a single substance that is never created or destroyed. So, all apparent creation and destruction must be reconceptualized as merely apparent. This requires casting such apparent ontological changes (Wechsel) as mere alterations (Veränderung) of a more ontologically primitive substance. We have seen this sort of link between explanation and ontology before in the form of Kant’s account of the representation of an object as that which explains the necessary connections among our manifold of sensations. In Chapter 4 I argued that Kant is best understood as holding that we represent such objects by forming inferentially structured pictures of them using material rules of inference (rules of inference that are valid but not in virtue of their logical form). The argument of the First Analogy yields both an important condition regarding what such object-concepts will have to be like in order to count as providing an adequate explanation—they must represent the world as consisting of a single substance that is never created or

200

Self and World in the Analogies of Experience

destroyed—and a corresponding prescription for what to do when a potentially explanatory object-concept fails to meet this condition—it must be put aside in favor of a concept that is capable of explaining, now not only the necessary connections among the manifold of sensations but also how it is that the posits of the previously employed object-concepts can be recast as alterations of the substance posited by the new ones. That is, what Kant offers is what we would now call a very general criterion of rational theory succession, of how the conceptual structure of one theoretical-explanatory picture of the world must come to be replaced by that of another. The Second Analogy gives another such criterion. Whereas the First Analogy argues from the necessity of representing time as a unity to the necessity of representing substance as indestructible, the Second Analogy argues from the necessity of representing events in time as occurring in a determinate temporal order2 to the necessity of representing the alterations of substance as subject to causal laws.3 Thus Kant offers another criterion for the adequacy of the explanation provided by object-concepts—they must represent the alterations of substance as being subject to specific causal laws—and another corresponding prescription for what do when a potentially explanatory object-concept fails to meet this condition—it too must be put aside in favor of a concept that is capable of accounting for those alterations in this way, and also of accounting for the merely apparent success and ultimate failure of its predecessor concept. Again, this ontological priority of law-governed substance over the “objects” of common sense takes the form of the replacement of one set of concepts with another. It does so precisely because of the way that concepts function in Kant’s system: they represent the necessary connections among objects of experience by relating representations of these objects (intuitions) to one another via material rules of inference. These inferential pictures of the world can succeed or fail at representing either the ontological nature of the universe—intuitions might represent items that do not, in fact, constitute the most fundamental objects of the physical universe—or the laws of nature that connect these objects—the material rules of inference that structure such pictures might represent objects as being connected in ways that they are not. It is by replacing one picture with another that we commit ourselves to the existence of some set of fundamental ontological entities or the instantiation of some set of specific causal laws. What the Analogies provide is a set of meta-conceptual rules for carrying out this replacement.4 It will be the goal of this chapter to examine Kant’s arguments for these two conclusions and to delineate how each relates to Kant’s inferentialism. What I will argue is that in each case, Kant’s considerations do not yield any specific inferential norms but rather, appropriately, certain principles of inference: general meta-level imperatives concerning what kinds of inferences are allowable given the conditions of experience to which creatures like us are subject. The next two sections of this chapter will discuss the

Self and World in the Analogies of Experience

201

First Analogy and Second Analogy respectively. With my account of the bulk Kant’s refutation of (NC) and (EW) complete, in the final section I will turn finally to (SSE) as we make our way to the final piece of Kant’s grand argument. THE FIRST ANALOGY: SUBSTANCE We can begin our approach to the First Analogy with Kant’s distinction between change (Wechsel) and alteration (Veränderung). A change, as Kant understands it, is the event of something’s coming to be or ceasing to exist. ‘Change’ refers to an event that marks a difference in the ontological makeup of the world. In autumn when the green of a leaf ceases to exist and its orange begins to exist, the color of the leaf changes. Alteration, on the other hand, is “a way of existing that succeeds another way of existing of the very same object” (A187/B230). So, while the color of the leaves changes—the green ceases to be, and the orange comes to be—relative to these colors the leaves undergo alteration. They exist first as green, and then as orange. Kant’s thesis in the First Analogy is that all ostensible change, all seeming ontological difference, is actually mere alteration. Strictly speaking, there are no changes in the world—nothing comes to be or ceases to be; there are only alterations of the one, sempiternal substance. All appearances contain that which persists (substance) as the object itself, and that which can change as its mere determination, i.e., a way in which the object exists. A182 All apparent change is, in fact, mere alteration of the one, sempiternal substance. Kant’s use of the term ‘substance’ to refer to this one omnipresent matter in which all alteration occurs is in line with a certain historical tradition that runs through Kant’s early Modern predecessors: Locke, Descartes, Spinoza, Leibniz, Berkeley, and Hume. This is a distinctly ontological sense of ‘substance.’ There is another sense, however, of ‘substance’ stemming from a historical tradition tracing its roots back to at least Aristotle. This is the sense of a substance as the single subject of multiple predicables. While Aristotle and others draw ontological consequences from this use of ‘substance,’ in itself it is an essentially representational use of that term. Especially for Kant, these two senses of substance are not unconnected. In fact, he explicitly connects them. What cannot be thought [intuited] otherwise than as subject does not exist otherwise than as subject, and is therefore substance. B410

202

Self and World in the Analogies of Experience

What is an Aristotelian substance, the single subject of multiple predicables, is also a Lockean substance, the one, omnipresent matter that underlies all apparent change. As we have already seen, Kant shares with Hume a concern with the question of what constitutes the concept of an object. In the Deduction, Kant argues that the Categories are the meta-level rules that provide the answer to this question. Since an object is just that which is represented by an intuition, and the form of an intuition is essentially determined by its capacity for appearing in a judgment, all objects must be represented as having a form parallel to that of the logical forms of judgments. E.g., since judgments must be either universal, particular, or singular, the objects of those judgments must be either unities, pluralities, or totalities.5 So, regarding substance, Kant’s concern again is first and foremost with how we represent something as a substance (as opposed to an accident), and it is unsurprising that his answer is in terms of the formalcum-inferential properties of such a representation. Again, it is facts about the formal properties of certain representations—here that they can only be the subjects of judgments and never the predicates—that picture certain counterpart properties of that which is represented—here that a substance is the bearer of properties. This connection is significant here because it offers us a test to determine whether or not a purported Lockean substance, bit of matter, is in fact a genuine substance. If it can only be thought as subject, then it is; if it can be thought otherwise than as a subject, then it is not. What is really interesting about Kant’s argument in the First Analogy is that it implies that most of the objects that one might assume constitute the substance of the world—e.g., tables, chairs, elephants, planets—in fact fail this test and so, according to Kant, are not genuine substances. Such substances are mere determinations of the one, omnipresent substance. More on that in a moment. For now, what we need to do is see how it is that Kant arrives at this audacious conclusion. To do this, it will be helpful to first consider a quick reconstruction of Kant’s argument, along with a classic objection to this version of the argument. It will turn out that Kant himself has already considered this objection and has a telling response to it. Thus, in following this dialectic, we will come to the purpose of this section: to understand the argument of the First Analogy and how Kant’s inferentialist theory of mental representation can be used to account for its conclusion. So, we can begin with the following quick and dirty version of Kant’s argument in the First Analogy. 1. (In order to represent oneself as a single subject of experience persisting through time),6 one must represent time as a unity.7 2. Time itself is not perceived.8 3. Therefore, one can only represent time by marking time on the objects of experience.9

Self and World in the Analogies of Experience

203

4. Therefore, to represent time as a unity, one must mark time on a unitary substance.10 5. Therefore, we must represent the world as containing a single, sempiternal substance (which is neither created nor destroyed, and of which apparent changes are actually mere alterations). The idea here is that since the only way to represent time is by representing the changing states of the world, if one is to be able to represent time as unity, it cannot happen that there should be a time at which nothing exists. Were such a state to come to pass, the timeline that was marked on the objects before this state and the timeline that would be marked after it could not be related to one another by any intermediate time because there would, ex hypothesi, be no objects in existence during this intermediate span on which to mark the time between the end of the previous timeline and the beginning of the new one. So, there can be no time at which nothing exists. Melnick offers the following example to illustrate the point, to which we will see Van Cleve object in a moment. Suppose that the action (mechanism) of an ordinary faceclock is used to determine the magnitude of a time interval t1 to t2. We assume that at time t1 the hands on the clock read 4:00 A.M. and that at time t2 the hands on the clock read 4:05 A.M. We thus measure the time interval t1 to t2 as the time it takes for the action (the mechanism) to move the hands of the clock from a 4:00 reading to a 4:05 reading. Suppose that the clock that reads 4:00 at t1 does not have an uninterrupted existence up to time t2, i.e., suppose we have the following situation: At time t2 clock A reads 4:00. At time t’ between t1 and t2 clock A goes out of existence. At some time t” between t’ and t2 (where t” does not equal t’) clock B comes into existence and at t2 clock B reads 4:05. In order to determine the time interval between t1 and t2 we must be able to determine the interval between t’ and t”. It will not do in determining this interval to say, e.g., that since the last reading of clock A (at t’) was 4:02:25, and the first reading of clock B (at t”) was 4:02:27, that the interval t’ to t” was 2 seconds. [. . .] Thus, there can be no interval no matter how small (because we could not determine how small) between the times t1 and t2 at which there is a lacuna in the mechanism, if this mechanism is to be that in virtue of which we determine the magnitude of the interval t1 to t2.11 Since there is no substance in existence between t’ and t”, there is no way to mark the interval that passes between these two times. Thus Melnick earns for Kant the conclusion that in order to represent time as a unity, there must, at all times, exist some substance. As Van Cleve astutely points out, however, this principle is not the same as, and does not imply, the principle that in order to represent time as a unity, there exists some single substance

204

Self and World in the Analogies of Experience

that exists at all times. That is, it seems that if the kind of example that Melnick presents is all that there is to Kant’s argument, then Kant is guilty of a simple confusion regarding the scope of his quantifiers. Van Cleve proposes the following counterexample to make this clear.12 We could still measure the interval from t1 to t2 provided there were another clock that existed, say, from 4:02 until 4:03. This would enable us to verify that all three clocks were synchronized and to measure the interval from t’ to t” by means of the third.13 In this example, at every time there exists a clock by which to mark the time, but it is not the case that any single clock exists at all times, and it does not seem that there is any reason to believe that some such sempiternal clock must exist in order to bridge the gap in Melnick’s example. Despite its initial plausibility, Kant seems to have anticipated this kind of example and provided an argument against this sort of objection. Substances (in appearance) are the substrata of all time-determinations. The arising of some of them and the perishing of others would itself remove the sole condition of the empirical unity of time, and the appearances would then be related to two different times, in which existence flowed side by side, which is absurd. For there is only one time, in which all different times must not be placed simultaneously but only one after another. A188–9/B231–2, emphasis added Call the interposing clock from Van Cleve’s example Clock C. In that example we mark time first on the Clock A, then on both Clock A and Clock C for some time, then just on Clock C, then on Clock C and Clock B for some time, then on just Clock B. The situation would look something like this: The argument that Kant quickly articulates in the above passage concerns what we are to make of the intervals in which we are marking time on two different clocks: “the appearances would then be related to two

Figure 5.1

Self and World in the Analogies of Experience

205

different times, in which existence flowed side by side, which is absurd.” The problem that Kant is anticipating here is one of coordination. If time is being marked on two different clocks, or substances, then we have no criteria by which to judge that a time being marked on, e.g., Clock A is the same time that is being marked on Clock C. What we are tempted to say is that the hands on Clock A pointing to 4:02 occurs at the same time as the hands on Clock C pointing to 4:02, but this is to make precisely the mistake against which Kant warns: if the only way we have to mark time is by those clocks, then to the claim that the positions of the clocks match at a time is to say that two times occur at the same time, which, as Kant points out, is absurd. By way of illustration, consider the infamous meter stick in Paris that serves as the standard against which all lengths are measured. In the classic example, a meter is defined as whatsoever is the length of that particular stick.14 So, just as time cannot be marked except by the changing of substance, in this example there is no way to measure length independently of the meter stick. To see the absurdity to which Kant alludes in the above passage, consider a scenario in which there was not one such meter stick but two. This would mean that a meter would be defined as whatsoever is the same length as those two sticks. An example will help to bring out the absurdity of this situation. Consider a length of string that was first held up to one of these meter sticks and then against the other. Suppose further that the string measured exactly one meter according to the first stick. It is, then, one meter long. Now consider, however, what happens when we go to hold the string up against the second meter stick. Either it too will measure the string as being a meter long or it will not. Suppose, for example, that according to the second meter stick, the string is half of a meter long. It is, then, half of a meter long. Of course, what one is tempted to say here is that the two sticks are of different lengths. If, however, length itself is measured only by comparison with these two sticks, then that is not a thesis that is available: they are each, by definition, one meter long. So, in this case, the string would be two different lengths. Surprisingly, things are not much better in the alternate case, where the string measures a meter according to both sticks. That is because in that scenario, while the length of the string can be measured consistently, that it can is, in some sense, entirely contingent. That is because, as we have just seen, while there is a sense in which the two meter sticks are necessarily the same length, one meter, there is another sense in which this is entirely an accident. It could have been the case that the string was measured differently by each meter stick, and this contingency itself is absurd. Most importantly, however, each of these absurdities follows from the more basic one that is the analog of Kant’s claim about time: namely, that the length of a given magnitude would be determined by two different standards. The very idea of a standard of length against which all other lengths are measured itself precludes the idea of having two such standards.

206

Self and World in the Analogies of Experience

Back to the clocks: if time can only be represented via the alterations of substance, then in the situation that Van Cleve describes, in which he suggests we could represent time via overlapping substances, each such substance would have to serve as something like the meter stick. We would then, however, be left in the situation as above: we would have two standards against which to judge time—Clock A and Clock C—and no way to reconcile, even conceptually, the times that they each deliver. It would make no sense to say that Clock A reads 4:02 at the same time that Clock C does in just the same way that it would make no sense to say that both meter sticks are the same length, or that a centimeter on one stick is the same as on the other. Length is measured against those sticks; time is marked on those clocks; each stick is necessarily one meter long; each clock necessarily reads 4:02 at 4:02. There is no sense to be made of both readings occurring at the same time because there is no standard of time apart from those clocks. Thus, Kant’s argument does not proceed via a simple mistake in the scope of his quantifiers from a step that establishes that (a) every apparent change is a mere alteration of some substance to the conclusion that (b) there is some single substance of which every apparent change is a mere alteration. Rather, Kant concludes that the need to represent all of time as a unity requires that one represent the world as consisting of a single sempiternal substance that can never be created or destroyed because only a single such substance can provide the single standard needed on which to mark this time. If substance is the only possible standard of marking time, then there can be only one such standard. The only way to ensure this is to guarantee not only that there can be no time at which no substance exists but also that no two substances can ever “coexist.” Thus, to represent time as a unity, we must represent the world as consisting of a single, sempiternal substance (that is never created or destroyed). To return to the rough version of Kant’s argument above, the line of reasoning that we have been pursuing validates the inference in that argument from 3. Therefore, one can only represent time by marking time on the objects of experience to 4. Therefore, to represent time as a unity, one must mark time on a unitary substance,

Self and World in the Analogies of Experience

207

on the grounds that the unity of time that is required is not only the unity of earlier times with later ones but also the more general unity appropriate to representing a single continuous timeline (rather than one with multiple “concurrent” branches, which, as Kant points out, is absurd). Given that we can only represent time by marking it on the objects of experience, representing this more general kind of unity does, in fact, require a single substance that acts as the single standard on which we do just this. Of course, we can, in more ordinary circumstances, represent two clocks as displaying the same reading at the same time. Kant’s thesis is that we do so by representing the movements of both of those clocks against the background of the single, sempiternal substance. This, however, seems odd, as Kant says nothing about, and we seem to know nothing about, what this single, sempiternal substance is or is like. It is not as if we have one clock here, another there, and hold them both up against substance to make sure that they each display the same reading at the same time. One may wonder, then, what exactly Kant takes himself to have argued. What the First Analogy calls for is that we represent all apparent ontological change as the mere alteration of the one sempiternal substance, and while this can certainly look like a demand for representing substance as a kind of object distinct from, say, the colors of the leaves or the wood that is burnt or the clocks that may come and go, it is not. What I want to suggest is that we represent substance not by forming a representation that is distinct from the representation of ordinary objects but by representing these objects as alterations of substance. O’Shea puts it nicely: The reference to permanent substance is not a reference to some further content posited behind or beneath the changing contents of perception (the ‘accidental determinations’ of substance). Rather, the concept of substance is the rule that the changing contents encountered in sense experience must themselves be conceived as the successive constitutive characters of an identical substance that persists through such changes.15 The way that we represent this sempiternal substance, without knowing its nature, is by committing ourselves to the rule prescribed by the First Analogy: that all apparent change be represented as the mere alteration of such a substance. We have already seen, in our earlier examination of Kant’s conclusion, at least one feature of such a rule. The single, sempiternal substance of which all apparent change is merely an alteration cannot be represented other than as subject, i.e., it cannot be predicated of anything else. As we noted earlier, the converse of this principle is at least equally important. If a potential substance can be predicated of something else—the color green can be predicated of the leaf—then it is not a genuine substance. Similarly, if a potential substance can be created or destroyed—the piece of wood can be burnt—it is likewise not

208

Self and World in the Analogies of Experience

a genuine substance but a mere alteration of genuine substance. What Kant argues in the First Analogy is that we are committed to representing the world as consisting of a single, sempiternal substance and that we do so by committing ourselves to rules of inference corresponding to these conditionals, which constrain the picture of the world that we thereby form. Here is Kant noting that this is an important difference between the results of the Analogies and the other Principles: while the other Principles deliver rules that are constitutive of what it is to represent an object, the Analogies deliver only regulative imperatives about what such representations creatures like us ought to use. Things must be entirely different with those principles that are to bring the existence of appearances under rules a priori. For, since this existence cannot be constructed, these principles can concern only the relation of existence, and can yield nothing but merely regulative principles. A179/B221 Kant’s point here is that whereas the Axioms and Anticipations straightforwardly determine what it is for a representation to be a representation of an object at all—they must represent something of a determinate quantity and quality—because the Analogies deal with the relations between objects (or, strictly speaking, alterations of substance), they are not constitutive of such representations but rather regulative. The Analogies provide rules for the adequacy of our representations of the world. Thus, we do not represent substance itself, but rather we represent the manifold of appearances as alterations of substance, and we do this by subjecting the object-concepts that we use to the above rules. Unlike in Hume (or Berkeley whom Hume follows), the demand to form a fully determinate picture of what this substance is like is out of place here. Recall that for Hume, a complex idea represents the impressions of which its simple components are copies as being arranged in the same way that the complex idea is arranged. Since all such complex ideas must be fully determinate, whatever they represent must likewise be fully determinate. E.g., every idea of a triangle, because it is itself a triangle, must be either an idea of an isosceles triangle, an idea of an equilateral triangle, or an idea of a scalene triangle. ‘Tis a principle generally receiv’d in philosophy, that every thing in nature is individual, and that ‘tis utterly absurd to suppose a triangle really existent, which has no precise proportion of sides and angles. If this therefore be absurd in fact and reality, it must also be absurd in idea. T 1.1.7.6; SBN 19–20

Self and World in the Analogies of Experience

209

It is because the relations that structure mental representations for Hume are the same relations that structure the objects thereby pictured that Hume’s commitment to the principle that “every thing in nature is individual” commits him to the further principle that we can have no indeterminate ideas and cannot therefore represent anything indeterminate. As we have seen, Kant also holds a kind of picture theory, but one according to which the relations among the objects of representation are pictured via the inferences to which one is committed concerning judgments involving representations of those objects. This leaves room, where Hume’s theory does not, for indeterminate representation. One can commit oneself to an object’s being a triangle without also committing oneself to its being any particular kind of triangle (although one will still be committed to its being either isosceles, equilateral, or scalene). If one holds that some object is a triangle, one is thereby committed to its having three sides, its having three angles, the sum of its angle adding up to one hundred and eighty degrees, etc. That is what the content of the concept ‘triangle’ entails. One is not, however, thereby committed to its sides all being of equal length, etc. Thus the rules for representing an object as a triangle form a picture of the world that is indeterminate with respect to what kind of triangle there is. And there is nothing problematic about that in the way that there is on Hume’s account. Returning, then, to the representation of substance, what is needed to form such a representation, to form a picture of substance, is not a fully determinate representation that pictures precisely every feature of this substance but rather a set of rules of inference that relate the representations of elements of this substance to one another in a way that is adequate for representing the relevant features of this substance. In this case those relevant features are that all alterations are alterations of it, that it is never created or destroyed, and that time is marked against it. The first of these features manifests itself in a purely formal way: anything other than substance can be predicated of something else, while substance itself can only ever be represented by the subject of a judgment. The second is manifested by a commitment to a blanket forbearance of the judgment ‘there exists a time at which substance does not exist’ and its consequent conclusions regarding what is and is not a substance. The third is manifest in the commitment to the principle that all the alterations against which we mark time are the alterations of the single sempiternal substance. These inferential commitments together constitute the representation of substance for which Kant argues in the First Analogy. They together form a picture of the world as consisting of a single sempiternal substance that is never created or destroyed and of which all apparent change is actually an alteration. Consider again the question: how does the single sempiternal substance serve as the measure of time? Kant’s answer is that there is a sense in which it does so in what should be a familiar way. To represent time’s passage, we give a certain order to our representations by employing the concept ‘clock’:

210

Self and World in the Analogies of Experience

the clock read 4:00 a moment ago, now it reads 4:01, etc. We conceive of our representations as being representations of the alterations of the state of this clock. These local acts of ordering, though, are part of a more farreaching, and necessarily prospective, project of ordering all possible representations along a single, continuous timeline. Likewise, the concept ‘clock’ is part of a similar and connected project of representing nature as the sum total of all objects (as we saw in the previous chapter). It is by subjecting our representations of objects to the regulative concept ‘substance’ that we commit to carrying out both of these projects. We mark time by giving order to our representations using object-concepts; we represent time’s unity by committing to the prospective global unity of the objects so represented. The alterations that we now conceive as the alterations of these objects will ultimately be reconceived as the alterations of a single, sempiternal substance that can never be created or destroyed. What I want to do now is draw out a consequence of this reading of the First Analogy that has been largely overlooked. What we have seen thus far is that Kant’s conclusion in the First Analogy is that certain rules of inference are necessary for representing the world in the way that we must in order to conceive of oneself as a single subject of experience persisting through time. Furthermore, we have seen that some of these rules of inference serve as meta-rules: they demand that in certain circumstances, one change the first-order concepts that one uses to represent the world. So, for instance, if one had been employing the concepts ‘green’ or ‘piece of wood’ as object-concepts, once one sees that greens and pieces of wood can be destroyed, one is committed to representing these each as mere alterations of some other substance. Combining this picture of the succession of conceptual schemes with the thesis culled from the B-Deduction in the previous chapter that the concept of an object is the concept of something that explains the conceptual connections among our representations, what we arrive at is a kind of theoretical-explanatory realism. Consequently that which persists, in relation to which alone all temporal relations of appearances can be determined, is substance in the appearance, i.e., the real in the appearance, which as the substratum of all change always remains the same. A181/B225, emphasis added What is real in appearance is substance. We represent substance by replacing one set of object-concepts with another upon discovering that the predecessor set could not accurately meet the demands placed on it by the nature of our representative powers, and with the expectation that the successor set will more accurately picture the world. Compare this with a statement of the thesis of scientific realism by twentieth-century advocate of both scientific

Self and World in the Analogies of Experience

211

realism and a picture theory of representation, Wilfrid Sellars, regarding the language of physical science: [T]he Scientific Realist need only argue [. . .] that in principle this language could replace the common-sense framework in all its roles, with the result that the idea that scientific theory enables a more adequate picturing of the world could be taken at its face value.16 According to Sellars, the thesis of scientific realism—that the theoreticalexplanatory posits of the best scientific theory accurately represent the ontological makeup of the world, and that the material rules of inference used to structure these representations accurately represent the laws of nature that govern these worldly objects—can be cashed out entirely in terms of an adequate account of the process by which we replace one conceptualscheme, qua picture, with another. Of course, Sellars elsewhere recognizes that to make this form of realism tenable, one must supplement this kind of picture theory of representation with an account of the sense in which successive such pictures converge toward a limit that can plausibly lay claim to being the correct picture of the world. Thereby hangs the tale of the debates over the possibility and nature of rational theory succession that dominated philosophy of science for the middle portion of the previous century.17 It is not Kant’s purpose in the Critique to enter into that fray except in the most general way. What the thesis of the First Analogy does is provide one very general constraint on this process: since no theory is acceptable that takes as its basic entities objects that can be created or destroyed, all such theories ought to be abandoned in favor of a successor theory that can account for these substances as the alterations of another substance that itself cannot ever be created or destroyed. This criterion is applied most straightforwardly to the replacement of our ordinary conceptual scheme according to which the world consists of many middle-sized dry goods by an atomic theory according to which these objects are in fact mere configurations of a much greater number of atoms whose individual and combinatory properties account for the properties that we had attributed to the objects of common sense. When we later discover that these atoms, too, can be created and destroyed, these in turn are reconceptualized as mere alterations of a still more fundamental substance, and so forth. Clearly not all theories will fail of this criterion, but likewise not all theories will pass its muster—the requirement, for example, that the appearance of ontological change be reconceptualized as mere alteration is not a trivial condition of a successor theory. Farther along in the Critique, in the Appendix to the Dialectic ‘On the Regulative Use of the Ideas of Pure Reason,’ Kant describes in a more general way how he envisions this procedure as unfolding. The following passage paints a fairly detailed picture and is worth quoting at length.

212

Self and World in the Analogies of Experience Accordingly, this idea [that of the form of a whole of cognition] postulates complete unity of the understanding’s cognition, through which this cognition comes to be not merely a contingent aggregate but a system interconnected in accordance with necessary laws. One cannot properly say that this idea is the concept of an object, but only that of the thoroughgoing unity of these concepts, insofar as the idea serves the understanding as a rule. Such concepts of reason are not created by nature, rather we question nature according to these ideas, and we take our cognition to be defective as long as it is not adequate to them. Admittedly, it is hard to find pure earth, pure water, pure air, etc. Nevertheless, concepts of them are required (though as far as their complete purity is concerned, have their origin only in reason) in order appropriately to determine the share that each of these natural causes has in appearance; thus one reduces all materials to earths (mere weight, as it were), to salts and combustibles (as force), and finally to water and air as vehicles (machines, as it were, by means of which the aforementioned operate), in order to explain the chemical effects of materials in accordance with the idea of a mechanism. For even though it is not actually expressed this way, it is still very easy to discover the influence of reason on the classifications of students of nature. A645–646/B673–B674

The unity of nature—including its unity through time as in the First Analogy, and as we will see, its unity according to rules as in the Second Analogy— serves as a regulative ideal of all of our representative activity. It is that unity that we aim to picture by employing the object-concepts that we do. Thus, we “take our cognition to be defective” insofar as it does not allow for representing this unity, and we replace those cognitions, those objectconcepts, that are found to be lacking in this way. To use the example that Kant does here, on our first encounter with the world, we find it to consist of ordinary objects, but such objects do not admit of the kind of systematization that is required by the unity of nature. For one, they appear to be created and destroyed. For another, they seem to be independent substances existing “simultaneously.” So, we posit that these objects are, in fact, mere alterations of some more orderly and unified substance—in Kant’s example, pure earth, pure water, pure air, etc.—and we represent these as being related to one another in ways that can explain the properties and behaviors of the ordinary objects of our predecessor conceptual scheme. What justifies this replacement of one set of objectconcepts with another (and the corresponding set of inferential rules with another) is precisely the requirement that nature be conceived of as a unity. It is as part of this regulative ideal that ‘substance’ serves as the kind of meta-level conceptual norm that we have been outlining, and the result of deploying this norm is a reconceptualization of the ontological makeup of the world. First we think of it as being composed of ordinary objects,

Self and World in the Analogies of Experience

213

and then when that conceptual scheme fails, we think of it is as being composed of pure earth, pure air, pure water, etc. And each set of successor concepts, as Kant understands them, gets closer to representing the real substance of nature.18 Thus, what Kant offers in the First Analogy is a very general, transcendental form of scientific realism according to which one represents the real substance of the world by replacing one picture of this substance—constituted by a certain set of concepts-qua-material-rules of inference connecting intuitions to one another via judgments—with another in any case in which the predecessor conceptual scheme takes as its objects anything that can be created or destroyed and the successor scheme can reconceptualize such ontological changes as mere alterations of a single sempiternal substance. Thus, again it is Kant’s inferential theory of concepts that provides the key both for interpreting the ultimate conclusion of the First Analogy and for understanding the import of the kind of scientific realism that Kant thereby advocates. It is because Kant takes concepts-qua-inferential-rules to provide the structure of the pictures that we use to represent the world that the regulatory aspect of the conclusion of the First Analogy results in a very general theory of rational theory change that prescribes at least one condition for when such pictures must be replaced and takes this succession of conceptual schemes to aim at ever more accurate representations of the world. What we are about to see is that in the Second Analogy Kant argues for a second such general criterion and in a way that is very similar to the way that he argues for this first one, in which his inferentialism again plays a central role. THE SECOND ANALOGY: CAUSAL LAWS As in the First Analogy, in the Second Analogy Kant is once again concerned with our ability to represent ourselves as the single subject of experience persisting through time, though here Kant is no longer concerned with the unity of time but with its determinate order: that some events occur before others, others occur after those, etc. Still, the argument of the Second Analogy will share some very general features in common with the argument of the First. Both proceed from the premise that we do not perceive time itself and that nothing in our representations indicates their place in time (either as a unity or as determinately ordered). Both also draw from this the lesson that in order to represent time as we do, we must conceptualize our representations in a certain way. And finally both draw conclusions about the way that we must represent the world as being in order to make this representation of time possible. That said, I will begin my investigation of the Second Analogy not with Kant’s argument but with his conclusion, which has been the object of a longstanding debate in the secondary literature.

214

Self and World in the Analogies of Experience

Roughly, this debate has been between those who understand the main thesis of the Second Analogy as being relatively weak and those who believe it to be relatively strong. Proponents of the weak reading—such as Buchdal, Beck, Strawson, and Allison—argue that Kant concludes that what is required for representing a determinate temporal order is only the principle that every event has some cause from which it follows necessarily.19 Proponents of the strong reading—such as Guyer, Friedman, and Watkins—argue that Kant concludes that such representations also require a commitment to the existence of causal laws: it is not just individual events that are necessarily connected to one another but that the necessary connections between such events apply to those events as instances of certain kinds.20 The debate between proponents of the strong and the weak interpretations proceed in much the same way that we earlier saw debates about the argument of the Transcendental Deduction unfold, and for much the same reason. Proponents of the weak interpretation notice that establishing the strong conclusion requires a more robust argument in support of it, and they do not find Kant offering such an argument in the Second Analogy. Thus, they take the conclusion at which he is aiming to be only the weak one. Proponents of the strong interpretation seek to establish that there is such an argument to be found in the Second Analogy and that the strong conclusion is what Kant needs moving forward. Their main task is to show that Kant can move validly from the premises of the Second Analogy to the strong conclusion. As before, I will argue that adjudicating this debate actually requires properly understanding the argumentative work that Kant does well before he arrives at the Second Analogy. Specifically, with Hume’s theory of mental representation already refuted, and Kant’s own theory firmly in place, Kant begins the argument of the Second Analogy with more resources than either camp has previously credited to him. Specifically, he has already established, at least tentatively, that we represent the necessary connections among worldly objects by relating representations of those objects to one another via material rules of inference. As we will see in a moment, this thesis itself goes a long way toward establishing the strong conclusion, since such rules of inference will apply to these representations in virtue of the concepts that structure them, concepts are essentially general in their application, and so apply to these representations as kinds, rather than as individuals. So, once again, it is essential to understand Kant’s argument here that we first understand his theory of mental representation, since it is that theory that will do much of the work in that argument. Before we proceed to that argument, however, it will be worth pausing to consider how Kant himself presents his conclusion. Kant changes the way he phrases his conclusion from the A-edition to the B-edition—mostly, it would seem, to emphasize the findings of the First Analogy21—but both idioms provide fertile ground for the debate between proponents of the weak and the strong interpretations.

Self and World in the Analogies of Experience

215

Everything that happens (begins to be) presupposes something which it follows in accordance with a rule. A189 All alterations occur in accordance with the law of the connection of cause and effect. B232 The proponent of the weak interpretation can point to the fact that in the first passage Kant refers to individual events following from other individual events, and the necessity connecting them makes no explicit reference to causal laws that apply to kinds of events. Furthermore, the second version mentions the law of cause and effect, which the proponent of the weak interpretation can take to be the very general law that every event has some cause but does not make reference to specific causal laws, plural. Proponents of the strong interpretation, on the other hand, can point to the fact that in the A-edition version of the conclusion the events referred to follow in accordance with a rule and can emphasize that rules are general in their application and so are applicable to such events not just in particular instances, but also to all events of the same kind. Regarding the B-edition version, the proponent of the strong interpretation will point out that while Kant mentions only the general causal law, it would be natural to take his thesis to imply, especially given the details of the Second Analogy, that the way that alterations occur in accordance with that law is by occurring in accordance with specific casual laws and not just in virtue of standing in a necessary relation to some other one event. As I mentioned a moment ago, we have already encountered the notion of a causal law in the context of Kant’s account of what is represented by a representation of an object, and so we can make use of a fairly quick, straightforward, and novel argument for the conclusion that Kant holds that we must represent the world as governed by specific causal laws.22 1. We represent the world as consisting of objects bearing necessary connections to one another by picturing this world. 2. The elements of this picture are intuitions; its structure is conceptsqua-material-rules of inference that relate these intuitions to each other via the judgments in which they can appear. 3. Concepts-qua-material-rules of inference apply not only to particular intuitions but to kinds of intuitions via the concepts that provide the structure internal to an intuition (which essentially can be common to numerically distinct intuitions). 4. Causal laws are the necessary connections between kinds of objects.

216

Self and World in the Analogies of Experience

5. Material rules of inference are the counterpart in our picture of the world to the causal laws that structure that world and represent the world as being governed by these laws. (2, 3, 4) 6. Representing oneself as a single subject of experience requires representing the world. 7. Representing oneself as a single subject of experience requires representing the world as governed by specific causal laws. (1, 5, 6) Recall that we have already seen Kant argue against Hume that both representing complex states of affairs as complex and accounting for the unity of the proposition require the employment of material rules of inference. We have also seen him argue that representing complex states of affairs as complex is necessary for representing oneself as the single subject of manifold representations. Since causal laws are just the worldly counterpart of such material rules of inference—in relating our representations of objects via such rules, we represent these objects as being subject to the corresponding causal laws—by the time we reach the Second Analogy, Kant has already earned himself the conclusion that the strong interpretation attributes to him. However, both the weak and the strong interpretations, as applied to the Second Analogy alone, must be wrong. The weak interpretation is wrong because by the end of the Second Analogy Kant clearly does take himself to be entitled to at least the postulation of specific causal laws (and so it is not true that he postpones the justification of this practice until later works such as the Metaphysical Foundations of Natural Science). Notice, though, that according to this way of understanding the dialectic the strong interpretation is also wrong. The argument for the general conclusion that the world is governed by specific causal laws does not occur, except indirectly, in the Second Analogy. Kant establishes the necessity of causal laws in arguing for his theory of mental representation as a whole. This is because according to that theory, we represent causal laws by licensing the specific rules of material inference that constitute our representations of objects. So, if we are required to represent objects (as Kant holds that we are), and we represent objects by representing causal laws (as Kant holds that we do), then we are required to represent causal laws. That much is established back in the Metaphysical and Transcendental Deductions. What the Second Analogy adds to this is the specific form that such laws will take given the additional requirement that we represent a determinate temporal order. The general outline that we earlier gave of the Analogies should make this clear. The Analogies present how the Categories are applied by creatures likes us, specifically by creatures who represent the objects of both inner and outer sense as extended in time.23 So, what Kant argues for in the Second Analogy is the particular temporal form that causal laws must take for such creatures. Specifically, he argues that in order for us to represent events as occurring in a determinate temporal order, we must represent the

Self and World in the Analogies of Experience

217

world as being governed by causal laws that necessitate the objective order of events in time. Turning to that argument now, we can see that it has much the same form as the argument of the First Analogy. Both arguments begin with a premise concerning the conditions for representing oneself as the single subject of experience persisting through time; both then turn to eliminating various ways of meeting these conditions that are inconsistent with the particular form that experience takes for creatures like us; both then turn to the way that we represent the world for the ultimate surprising conclusion of the argument. 1. (In order to represent oneself as a single subject of experience persisting through time),24 one must represent events as occurring in a determinate temporal order.25 2. Time itself is not perceived.26 3. Our representations of events in time do not have intrinsic features that constitute them as representations occurring in any particular order.27 4. Our representations of events in time do not stand in any relation to one another that constitute them as representations of events occurring in any particular order.28 5. Therefore, there is nothing about the representations of events that constitute them as representations of events occurring in any particular order. (3, 4) 6. Since we do not perceive time and there is nothing about the representations of events that constitute them as representations of events occurring in any particular order, there must be something about the events represented that constitutes this representation.29 (2, 5) 7. To represent events themselves as determining their temporal order is to represent them as bearing certain necessary relations to one another.30 8. Therefore, to represent a determinate temporal order, one must represent events as bearing certain necessary relations to one another. (6, 7) 9. Therefore, we must represent events as bearing certain necessary relations to one another. (1, 8) Obviously, it is step 7 here that bears the greatest argumentative weight and consequently deserves the most careful scrutiny. The gist of the argument leading to it is simply that our representations themselves do not come time stamped, either intrinsically or via bearing any relations to one another. Examine our representations all you want, you won’t find anything about them that determines the order of occurrence of any event (inner or outer). So, if we are to represent events in time as occurring in a determinate temporal order (which we must), we must do so in virtue of something other than those representations. Kant’s suggestion is that we do so by representing the

218

Self and World in the Analogies of Experience

events represented by those representations as themselves determining their temporal order. That is, there is something about those events that makes it the case that they, unlike our representations, considered by themselves can only occur in one order and not any other. Again, rather than our representings having any feature that determines the temporal order of events, it must be something about that which is represented that does so. We must take our representation of the temporal order of events to be determined by the events represented themselves. Furthermore, it is Kant’s contention that the way that represented events determine their temporal order is by being represented as bearing necessary connections to one another. E.g., if B follows only from A, then A must come before B, etc. It will be instructive here to consider two places at which Kant considers the difference between representations that represent a determinate temporal order and those that do not. One of these contrasts comes in the form of his famous pair of examples: the sequence of representations that constitute a representation of a ship drifting downstream and the sequence of representations that constitute the representation of a house. The second contrast is Kant’s own explicit consideration of the difference between a representation of a determinate temporal order and a representation that is “only a subjective play of my imaginings.” We will consider each of these in turn, and, as has been our procedure up to this point, we will also note along the way how Kant’s inferential theory of mental representation is particularly well suited to accounting for these representations. In order to demonstrate what it is that he takes a representation of a determinate temporal order to consist in, Kant offers two contrasting examples. Consider the following manifolds of representations that respectively constitute the matter of the representation of a house and the representation of a ship drifting downstream. (a) a representation of a roof (a’) a representation of a ship upstream (b) a representation of a façade (b’) a representation of a ship midstream (c) a representation of a (c’) a representation of a ship downstream foundation In neither of these manifolds is there a representation of time itself; in neither is there a representation that in any way indicates its place in a determinate temporal order; and in neither is there a relation that these representations bear to one another that indicates the determinate temporal order of the whole. Nonetheless, Kant points out that the second manifold of representations (a’)–(c’) represents a determinate temporal order of events, while the first (a)–(c) does not. His suggestion is that what accounts for this difference, since it is not anything about the representations themselves, is something about the object represented. Specifically, it is the fact that the latter manifold is united by the concept of an object that makes necessary the order in which these must occur to the experiencing subject that

Self and World in the Analogies of Experience

219

makes it such a representation.31 That is, the object-concept that structures the representation of a ship moving downstream represents the order of this manifold of representations as being necessary in virtue of the nature of the ship’s movements. It is because we represent the ship’s being downstream as necessitated by its being midstream, which in turn we represent as necessitated by its being upstream, that our representation of the ship’s movement, but not the representation of the parts of the house, is a representation of a determinate temporal order. Consider the representation of the house. As Kant points out, while these representations might, in fact, occur in the order (a), (b), (c), there is nothing about the house itself that necessitates that they do so. The parts of the house could just as easily be seen in reverse order—(c), (b), (a)—or with a little creative eye movement in a different order entirely—say, (b), (c), (a). This is simply not so with the ship (barring, again, fanciful nonstandard perceptual scenarios). (a’), (b’), and (c’) are only perceptually available in that order because the ship is moving downstream, i.e., because its being upstream at one time necessitates its being midstream later and downstream later still. It is by representing these events as necessarily connected through time that we represent them as occurring in a determinate temporal order. We do not represent the roof, façade, and foundation of the house as so connected, and so we do not represent that roof, façade, and foundation as occurring in such an order.32 Notice that throughout the last few paragraphs, we have been using the idiom of the “events themselves” determining their temporal order and “objects represented as necessarily connected.” The time has now come to give some cash value to those idioms in terms of Kant’s inferential theory of mental representation, and by this point it should not be difficult to do so. What we already know is that for Kant we represent items in the world as bearing necessary connections to one another by licensing and forbidding certain inferences between judgments in which intuitions of such items appear. We represent the world by picturing it: the elements of the pictures are intuitions, and the structure of the picture is provided by the material rules of inference that relate these intuitions. So, for events themselves to determine their temporal order, or for objects represented to be represented as necessarily connected, will be for these events or objects to be pictured by intuitions related to one another by particular material inferential rules. Kant’s point in the Second Analogy is that these rules will have to take a temporal form. For example, in the case of the ship, the inferential rule in play might be something along the lines of the following. From ‘The ship is upstream now,’ one ought to infer, ‘The ship will be midstream soon’ and ‘The ship will be downstream later.’ Further, these judgments will forbid inferences to judgments such as ‘The ship will be upstream again later.’ It is the commitment to inferring the latter two judgments from the first that represents the necessary connection between the various temporal states of the ship. In undertaking such commitments, we thereby represent the ship as being subject to the corresponding causal law, e.g., ships move downstream.

220

Self and World in the Analogies of Experience

Contrast this again with the case of the house. Although both the manifold of sensations of the ship and of the house consist in a temporal series of representations, there is no set of inferences in the case of the house that correspond to the one just listed in the case of the ship. The concept ‘house’ does not contain any rule that determines the order in which one experiences the parts of a house.33 The concept ‘house’ does not require (or license) the inference from, for example, ‘The roof of the house is there,’ to ‘The roof of the house will be elsewhere later,’ or to ‘The roof of the house will not be there later.’ It is not just that the parts of houses do not typically change their location. Rather, the picture that is created by uniting a manifold of representations like (a)–(c) using the concept ‘house’ is not one that has a diachronic temporal dimension. The concept ‘house’ is used to unite a manifold of representations of the parts of the house not as successive events but as simultaneous parts. The material rules of inference that constitute that concept do not specify what perceptual judgments must follow what other perceptual judgments in time but rather specify only what the relation of various perceptual judgments at any time must be. E.g., the rules governing the representation of a house might, for example, be like the following. From ‘There is the roof of a house,’ one ought to infer that ‘below that there is the façade of the house.’ This last judgment does entail, ‘If I, in a moment, glance down, there would be the façade of the house,’ but it likewise entails, ‘If I had, a moment ago, glanced down, there would be the façade of the house’ and, ‘If I were, right now, glancing down, there would be the façade of the house.’ In the case of the ship, what is where at what time is not only specified but represented as necessary: not so in the case of the house. Thus, the set of inferences relating the elements of the manifold of representations of the ship constitutes a representation of a determinate temporal order, whereas those relating the elements of the manifold of representations of the house do not. By way of contrast, recall one of Hume’s most famous arguments for the conclusion that we have no idea of two distinct items being necessarily connected. Hume begins by noticing that we can imagine any two distinct ideas apart from one another.34 This is relevant to Hume because according to his theory of mental representation the relations that structure our complex mental representations are necessarily the same as the relations that structure what we thereby represent. E.g., to represent one item as being next to another, one places a representation of the one next to a representation of the other, as in a painting. Or, more relevantly to our current concern, to represent one occurrence as happening after another, one places a representation of the first after a representation of the second, as in a movie. Analogously, in order to represent two distinct items as necessarily connected, the representations of these items would likewise have themselves to be necessarily connected. In order to picture a necessary connection, we would have to have a representation that consists of two distinct ideas themselves necessarily connected. Since Hume begins this argument with the observation

Self and World in the Analogies of Experience

221

that we have no ideas related in this way—we can imagine any one idea without also imagining any other—he can conclude straightaway that we have no idea that is a representation of—a picture of—two objects as being necessarily connected. Having rejected this theory of mental representation, Kant can also reject Hume’s claim that our capacity to imagine two distinct items independently of one another offers any reason at all for thinking that we cannot represent them as being necessarily connected nonetheless. This is because, for Kant, to represent two items as being related, it is neither necessary nor sufficient to imagine them in any particular way. The relations that structure the pictures that Kant takes us to form of the world are inferential, not imaginative. Thus, when Kant argues in the Second Analogy that we must represent events as necessarily connected to one another in order to represent them as having a determinate temporal order, it will follow from this that we must form representings of these events that are structured by inferences with a temporal form. Whereas Hume’s theory predicts that we can represent a temporal complex merely by forming a temporal complex of representations, Kant sees that this cannot be right and sets out to give his own account of such representations. His focus in the Second Analogy is on what must be represented in order to represent events as occurring in a determinate temporal order, but we know from earlier in the Critique the answer to the equally important question of what kinds of representings must be used to do this work: concepts-quamaterial-inferential-rules. Another helpful contrast that Kant draws to illustrate this difference is between those representations that are structured by material inferences with temporal form and those that are not. Considering the latter, he writes, Contrariwise, if I were to posit that which precedes and the occurrence did not follow it necessarily, then I would have to hold it to be only a subjective play of my imaginings. [. . .] Thus the relation of appearances (as possible perceptions) in accordance with which the existence of that which succeeds (what happens) is determined in time necessarily and in accordance with a rule by something that precedes it, consequently the relation of cause to effect, is the condition of the objective validity of our empirical judgments with regard to the series of perceptions, thus of their empirical truth, and therefore of experience. A201/B246 Again, Kant’s aim here is to distinguish what makes a representation a representation of a determinate temporal order rather than a mere “subjective play of my imaginings,” and again he appeals to the representation of events as cause and effect to do so. In fact, though, Kant gives what appear to be two distinct conditions of such a representation: the effect must

222

Self and World in the Analogies of Experience

be “determined in time necessarily” by that which precedes it but also “in accordance with a rule.” It is worth asking here how these two conditions are meant to interact. It could be that Kant’s allusion to a rule here is merely a gesture toward causal laws, and so is a further specification of the form that the necessity of the first condition will take. That is entirely possible. More intriguing, though, is the suggestion that Kant’s choice of vocabulary here is more deliberate. In that case, these two conditions would correspond to what is represented by such by such a representation, on the one hand, and the means by which this is represented, on the other. Notice that the first of these conditions is couched in the idiom of necessary connection, a property of the objects of our representative activities. The second, however, is presented in a normative vocabulary—as being in accordance with a rule— which, as we have been understanding it, is the property of our representings that make them representations of items as being necessarily connected. So, Kant’s two conditions, while not identical, are necessarily coextensive: that which is represented as necessarily connected will be so represented in virtue of being in accordance with a rule that governs and constitutes an act of representing. The two conditions correspond to the two aspects of representation: that which is represented and the act of representing.35 Of course, one can only understand the relation of these two conditions in this way if one understands Kant’s theory of mental representation as we have: it is because necessary connections are represented by material rules of inference that necessary connection and rule-governed representation are two sides of the representative coin. So, Kant’s argument in the Second Analogy is that in order to represent events as occurring in a determinate temporal order, we must represent the world as subject to specific causal laws. The way that we do the latter, which is the crucial issue in Kant’s case against Hume, is by relating representations of events via material rules of inference with a specifically diachronic temporal form. Of course, as was the case in the First Analogy, we can be wrong about what is necessarily connected to what. The pictures that we construct of the world are not necessarily correct pictures. So, just as in the First Analogy, there is a distinctly prospective and regulatory aspect to Kant’s conclusion in the Second Analogy. While we must represent the world as subject to specific temporally structured causal laws, since there is no guarantee that such a representation will succeed, we are also obliged to replace any failed pictures with better ones. What Kant argues for is another meta-level inferential principle. Here is the full passage from which we drew in the previous section. Things must be entirely different with those principles that are to bring the existence of appearances under rules a priori. For, since this existence cannot be constructed, these principles can concern only the relation of existence, and can yield nothing but merely regulative principles. Here

Self and World in the Analogies of Experience

223

therefore neither axioms nor intuitions are to be thought of; rather, if a perception is given to us in a temporal relation to others (even though indeterminate), it cannot be said a priori which and how great this other perception is, but only how it is necessarily combined with the first, as regards its existence, in this modus of time. A179/B221 Kant’s point here is that the principles of the Analogies themselves to do not determine what concepts we must use to constitute experience but only that we must use some set of concepts that are adequate to the tasks at hand: representing the unity of time in the First Analogy and representing its determinate order in the Second. Again this yields a regulative principle, or a meta-level rule of inference, that prescribes a change in concepts whenever the goal set for by the conditions for representing oneself as the single subject of experience persisting through time cannot be met by the one’s current set of concepts. And again, this movement from one set of concepts to another is itself the form that Kant’s scientific realism takes. Now since all effect consists in that which happens, consequently in the changeable, which indicates succession in time, the ultimate subject of the changeable is therefore that which persists, as the substratum of everything that changes, i.e., the substance. A205/B250 We have already seen that substance is the real in appearance for Kant. Here Kant relates the causal laws that are the subject of the Second Analogy, as we have already seen him do, to alterations of that substance. The metalevel inferential principle of the Second Analogy thus prescribes a change in concepts in order to better represent that which is real. As we have already noted, this is in accordance with at least one form of scientific realism: that which takes the aim of the successive pictures that we form of the world to better represent how that world actually is. From all that we have seen from Kant to this point, this certainly seems to be in line with his theory of mental representation throughout the Critique. At the close of the previous chapter we saw that if we could complete Kant’s account of the representation of objects by considering the specification of it that he makes in the Analogies, this would leave only on item on our agenda: accounting for the UNITY condition of such representations, i.e., the condition that the representation of an object must be such that the subject of that representation is necessarily a single subject of experience persisting through time. As we have seen again here, meeting this condition provides Kant with an essential criteria of success in the Analogies. It is because we must represent ourselves as the single subject of

224

Self and World in the Analogies of Experience

experience persisting through time that we must represent time as a unity and as events occurring in a determinate temporal order. So, while I have allowed myself to use that criterion thus far with the promissory note of accounting for it later, the time has now come to cash out that note. The following section, therefore, will serve as a reminder of the importance of the UNITY condition to the work of the Transcendental Deduction and as a transition to the final chapter, which will address the representation of the single subject of experience persisting through time in the context of Kant’s inferentialism. THE UNITY OF THE EXPERIENCING SUBJECT The UNITY condition, recall, is that the representation of an object must be such that the subject of that representation is necessarily a single subject of experience persisting through time. What we will see in a moment is that the picture that we outlined earlier, while correct in its general form, is a bit misleading. Recall that the idea here was that Kant had followed Hume in noticing that the inference that they found in Descartes, from (D1) [I think x] and [I think y] and [I think z] to (D2) [The I that thinks x] = [the I that thinks y] = [the I that thinks z], is an invalid one. Taken on its own, the mere existence of a manifold of representations (D1) does not license the conclusion that such a manifold is had by a single subject of experience persisting through time (D2). Having rejected, however, the theory of mental representation that forces Hume to consequently abandon (D2), Kant looks to find a source of justification for it other than (D1). Thus, he is led to (K) I think [x + y + z]. If a manifold of representations could be united into a single cognition, that cognition would necessarily be had by a single subject, and thus one could infer (D2) from (K). Our work thus far in this chapter has been to make clear the nature of this kind of representation: an intuition, a complex representation of complex object as complex. Now that we have before us Kant’s account, we are in a position to see precisely how (K) does its work, and this turns out to be in a way slightly different from that first suggested. Consider an example, borrowed from Jay Rosenberg, meant to illuminate how (K) is supposed to function. Suppose I’m the sort of person who keeps a meticulous record of all of the quantities of food in my pantry. Suppose

Self and World in the Analogies of Experience

225

further that I come home one evening to find that a serving of peanut butter, two slices of bread, and a banana have gone missing from my store. With just this evidence, I do not have much reason to think that a single person ate all of these foods: it could just as easily have been three different people each eating one, or any number of other more obscure possibilities. Of course, there is an obvious reason for thinking that all three foods were eaten by a single person: the missing items, when properly assembled, constitute a peanut butter and banana sandwich. Thus, if I discovered, say, a half-eaten PB&B on the coffee table, I could conclude, ceteris paribus, from the fact that (PBB) Someone ate a peanut butter and banana sandwich that (I) [The person who ate the peanut butter] = [the person who ate the bread] = [the person who ate the banana]. That is, because I know that a single person ate a sandwich, the elements of which were the ingredients from my pantry, I can conclude, ceteris paribus, that the eater of each of those ingredients is identical. Here, the unity of the elements is the unity of a sandwich, and the unity of the “subject” is that of a consumer of such a sandwich. Such unities are familiar and clear enough. While there are several important disanalogies between the case of the missing ingredients and Kant’s consideration of the single subject of experience persisting through time,36 I want to focus here on just one: the kind of unity effected by the object and the unity of the subject that can be inferred from it. One might have supposed, upon a quick first pass through the argument, that Kant’s representation of an object would be much like the sandwich from the case of the missing ingredients: a manifold of representations physically united into a single, physically indissoluble one. In that case, it is easy to see why such a representation would have to be had by a single subject: it couldn’t be pried apart, so to speak. Of course, reflecting on the notion of a representation, one might have supposed that this strictly physical understanding of unity could not be what Kant has in mind. Clearly, Kant is not talking about a physical fusing of ideas, or even a fusing of nonphysical distinctly mental substances.37 Kant, in general, is no fan of mental substances at all. Still, the idea of some sort of cognitive fusion of representations remains. Isn’t Kant’s purpose here to show that a there is a single cognition that is necessarily had by a single subject of experience? Aren’t these both distinctly ontological notions? What we have recently seen is that Kant’s answer, at least concerning the representation of an object, is ‘no.’ The unity that binds a manifold of representations together into a single representation of a complex object as complex is an inferential unity. The “necessity” that binds together the elements

226

Self and World in the Analogies of Experience

of a manifold of representations is not a physical or mental necessity but a normative one. It is the necessity that is appropriate to being committed to one judgment in virtue of endorsing some other one. The consequence of this for Kant’s project in the Transcendental Deduction cannot be overemphasized. Because the unity of the representation of an object is not physical or mental but inferential, likewise, the unity of the subject of such a representation is not the unity of a physical or mental substance but rather the unity of the bearer of such inferential commitments. It is a normative unity. Investigating the details of this unity will be the subject of the next chapter, on the Paralogisms, but we can say a bit here about the broad-strokes picture in play. The first thing to notice is that whatever unity of the subject that the inferential unity of the representation of an object will effect can only be whatever kind of unity of the subject is necessary for this unity in the object. That is, just as the unity that is provided by the discovery of the PB&B sandwich is the unity of an eater of a sandwich, so the unity that will be provided by the inferential unity of the representation of an object can only be the unity of an inferrer. As Kant puts it in his notes, The I constitutes the substratum for a rule in general, and apprehension relates every appearance to it. Ak 17:656; Notes and Fragments, 166 The unity of the single subject of experience is the unity of the subject of a rule. Kant puts this same thesis more curtly elsewhere in his notes. The unity of apperception in relation to the faculty of imagination is the understanding. Rules. Ak 23:18; Notes and Fragments, 258 So, our question becomes: what kind of unity is the unity of a rule follower? Here is Kant distinguishing the kind of unity that we are looking to understand—I think [x + y + z]—from the empirical awareness of a manifold of representations—[I think x] + [I think y] + [I think z]. For the empirical consciousness that accompanies different representations is by itself dispersed and without relation to the identity of the subject. The latter relation therefore does not yet come about by my accompanying each representation with consciousness, but rather by my adding one representation to the other and being conscious of their synthesis. B133 The transcendental unity of apperception comes about by adding one representation to the other and being conscious of the synthesis of the manifold. In his description of the third phase of perceptual synthesis in the A-Deduction,

Self and World in the Analogies of Experience

227

Kant provides a helpful example for understanding exactly what he means here that involves both a literal addition of representations and the key phrase, “being conscious of their synthesis.” If, in counting, I forget that the units that now hover before my senses were successively added to each other by me, then I would not cognize the generation of the multitude through this successive addition of one to the other [. . .] for this concept consists solely in the consciousness of this unity of the synthesis. A103, emphases added The point that Kant is making here is about the necessity to retain previous units when adding for combination with subsequent ones, but the way he makes this point is illuminating. It is not enough that these previous units merely be remembered: they must not only “hover before my senses,” but it must also be remembered that they were added together by me. When I add together successive units, it is essential to that process, to following the rule for adding, that I see each successive unit as being combined by me. Counting is essentially an activity that is undertaken by a single individual. I have to be the one who adds together first this unit, then the next, then the next. If it is not a single counter that unites each subsequent unit, then the activity undertaken is not one of counting, and the units cannot be united. As Kant claims here, the concept of counting itself just is “the consciousness of this unity of the synthesis.” The rule for what it is for an activity to be a counting just is a rule for a single counter’s bringing together disparate units into a united whole. Thus, in adding together units, in proceeding according to this rule, I necessarily think of each unit as having been added together with all of the others by me, by the one following the rule for counting. This, then, is the nature of the unity of the subject that is effected by the representation of an object. Applying concepts-qua-inferential-rules to a manifold of representations is an activity that can only be accomplished by a single, unified rule follower. The I that makes the judgment ‘This body is metal’ must be the very same I that is thereby committed to the judgment ‘This metal is divisible’ just as the I that counts off 1, 2, and 3 must be the same I that later counts off 4. What makes these Is identical throughout these activities is just the undertaking of the activity itself. It is in virtue of being subject to the rules of counting that one is a unified counter. It is in virtue of being subject to the inferential rules required for representing a complex object as complex that one becomes a single subject of experience.38 As we will see in more detail in the next chapter, such activities do not carry with them any ontological commitments. We can count off units in turn and thereby together form a single counter. The key to doing so is simply that it is we, collectively, that are thereby subject to the rules of

228

Self and World in the Analogies of Experience

counting. Similarly, while William James’s famous explication of the unity of the self is certainly accurate, we must be careful not be misled by it. Take a sentence of a dozen words, and take twelve men and tell to each one word. Then stand the men in a row or jam them in a bunch, and let each man think of his word as intently as he will; nowhere will there be a consciousness of the whole sentence. James, 1890: 160 While it is certainly true in such a case that there is nowhere a consciousness of the whole sentence, if Kant is right, the reason why this is true is a surprising one. It is not because no single man has the entire sentence “running through his head.” Rather, it is because these men are not collectively subject to the norms that govern judgments containing the words that they have just been read aloud.39 As Kant puts it elsewhere, The consciousness of myself in the representation I is no intuition at all, but a merely intellectual representation of the self-activity of a thinking subject. B278 The unity of the subject of experience is what Kant calls a formal unity and also a form of representation. These are obscure phrases that are not obviously equivalent and warrant an investigation of their own. That is precisely the business of the chapter on the Paralogisms, and we will leave off our discussion of Kant’s account of the representation of the self until then.

NOTES 1. Thus, the interpretation offered here will contrast with offered in Kemp Smith, Commentary according to which the Analytic of Concepts is concerned only with “ordinary experience” and it is not until the Analytic of Judgments that we get Kant’s story about “scientific reorganization.” 2. I agree with Guyer, Kant and the Claims of Knowledge, 256 that the events in need of ordering are those of both outer and inner sense. 3. Thus, while the Deduction establishes that in order to represent an object at all, independently of our particular forms of intuition, requires representing its parts as being necessarily connected, and while Kant later expands this conclusion to encompass not just particular objects but the world as the sum total of all such objects, the Second Analogy, now incorporating time into this general conclusion, argues that the specific form that such necessary connections will take will be that of the lawful relations of alterations through time. The Third Analogy adds that it will also be necessary to represent the lawful relations of states of substance at a time. 4. So, while the so-called mathematical Categories (Quantity and Quality) serve as meta-conceptual rules that constitute what concepts count as object-concepts,

Self and World in the Analogies of Experience

5. 6.

7. 8. 9.

10.

11. 12. 13. 14. 15. 16. 17.

229

the Categories of Relation serve as regulative principles governing what such object-concepts are acceptable. Cf. Wolff, Vollständigkeit der kantischen Urteilstafel. “Now experience rests on the synthetic unity of appearances, i.e., on a synthesis according to concepts of the object of appearances in general, without which it would not even be cognition but rather a rhapsody of perceptions, which would not fit together in any context in accordance with rules of a thoroughly connected (possible) consciousness, thus not into the transcendental and necessary unity of apperception.” (A156/B195) “Different times are only parts of one and the same time.” (A32/B46) “Now time cannot be perceived by itself.” (A181/B225) “Consequently it is in the objects of perception, i.e., the appearances, that the substratum must be encountered that represents time in general” (A181/ B225). This does not strictly speaking follow from (2). One way to arrive at (3) would be to offer an argument that the two options listed here—that time is perceived and that time is marked on the objects of experience—exhaust the possibilities for representing time. Kant does not offer such an argument, but we might at least grant him that these are the two most obvious live options and that the more obscure ones can be dealt with in turn. “Consequently that which persists, in relation to which alone all temporal relations of appearances can be determined, is substance in the appearance, i.e., the real in the appearance, which as the substratum of all change always remains the same” (A181/B225). The objection that we are about to consider concerns the inference from (3) to (4). Melnick, Kant’s Analogies, 66. Similar arguments can be found in Strawson, Bounds of Sense and Bennett Kant’s Analytic, both of which can be dealt with in the same way as here. Van Cleve, “Substance, Matter, and Kant’s First Analogy,” 158. Cf. Wittgenstein, Philosophical Investigations, §50, although see also Kripke, Naming and Necessity, 55. O’Shea, “Kantian Matters,” 73. Sellars, Science and Metaphysics, 146. Popper, Logic of Scientific Discovery; Kuhn, Structure of Scientific Revolutions; Kuhn, “Objectivity, Value Judgment, and Theory Choice”; Lakatos, “Falsification and the Methodology of Scientific Research Programmes”; Feyerabend, Against Method; Sellars, “Scientific Realism or Irenic Instrumentalism”; Rosenberg. “Comparing the Incommensurable.” Unsurprisingly, Kant’s contribution to this dialectic gives a central role to the pure concepts of the understanding (as applied by creatures like us), which he takes to be meta-conceptual rules that determine what forms any conceptually structured intuition can take. This puts the Categories in a unique position in grappling with the incommensurability of distinct conceptual schemes. While it may be that the concepts employed in such schemes are not, e.g., intertranslatable, as we have just seen, each conceptual scheme’s being subject to the Categories does provide at least a very general standard by which one can be judged better than another: the standard of meeting the requirements, constituted by the Categories, for any adequate set of object-concepts. Here that is representing the world as consisting of a single sempiternal substance that is never created or destroyed. In the Second Analogy it is representing events in the world as governed by specific causal laws. Of course, none of this speaks to the issues raised by, e.g., Lakatos, about the conditions (or lack thereof) under which we ought to say that one conceptual scheme fails in these, or any regards. In the Postscript,

230

18.

19. 20. 21.

Self and World in the Analogies of Experience I defend Kant against Sellars’s charge that this leads to a kind of transcendental realism. In the Postscript I will defend Kant against Sellars’s charge that it is only because of a blurring of the distinction between perceptible material objects and the objects of theoretical science that Kant remains a Transcendental Idealist rather than a Transcendental Realist. That defense will consist in a demonstration that Kant has the resources to account for the progress of science that Sellars believes leads to Transcendental Realism entirely within the realm of appearances. I will there discuss certain passages from a variety of Kant’s texts that are relevant to the current discussion as well. Buchdal, Metaphysics and the Philosophy of Science; Beck, Essays; Strawson, Bounds of Sense; and Allison, Kant’s Transcendental Idealism. Guyer, Kant and the Claims of Knowledge; Friedman, “Causal Laws”; and Watkins, Kant and the Metaphysics of Causality. In his own copy of the first edition, Kant wrote the following in the margin of this page: All arising and perishing is only the alteration of that which endures (the substance), and this does not arise and perish (thus the world also does not). (Ak 23:30; Kant, 1998: 299)

So, Kant clearly thinks that the way he put this thesis in the first edition was a mistake, and one having to do precisely with the difference between change and alteration. That is, the mistake in the first edition was to use the idiom of change to describe the natural laws of the universe, when what these laws in fact concern are alterations. It is worth noting that it is events, the events of substance undergoing alteration that the causal laws of the Second Analogy must take these as their relata. Thus, the corrected second edition formulation: all alterations occur in accordance with the law of the connection of cause and effect. 22. It is worth noting here that the thesis of the Second Analogy clearly takes events as the proper relata of necessary connections, whereas the argument that we are about to consider concerns representations of objects and the necessary connections that they bear to one another. Strictly speaking, the necessary connections that we represent the parts of objects as bearing to one another fall under the Category of Community rather than Causality and Dependence, but Kant’s account of how we represent these different kinds of causal connections, in its general outline, is the same: necessary connections between both events and objects are represented by the material rules of inference that relate representations of these events or objects. The difference, as we are about to see, is the temporal form that these rules of inference take. 23. Having just concluded the section on the Schematism of the pure concepts of the understanding, which is time, Kant describes his task in the Principles as follows. In the previous chapter we have considered the transcendental power of judgment only in accordance with the general conditions under which alone it is authorized to use the pure concepts of the understanding for synthetic judgments. Now our task is to exhibit in systematic combination the judgments that the understanding actually brings about a priori subject to this critical warning. (A148/B187) The task of the Principles is to demonstrate what specific synthetic a priori judgments result from schematizing the Categories, i.e., from applying them to the experience of objects in time.

Self and World in the Analogies of Experience

231

24. “Now experience rests on the synthetic unity of appearances, i.e., on a synthesis according to concepts of the object of appearances in general, without which it would not even be cognition but rather a rhapsody of perceptions, which would not fit together in any context in accordance with rules of a thoroughly connected (possible) consciousness, thus not into the transcendental and necessary unity of apperception.” (A156/B195) 25. “I perceive that appearances succeed one another, i.e., that a state of things exists at one time the opposite of which existed in the previous state.” (A189/ B233) 26. “Now this determination of position cannot be borrowed from the relation of the appearances to absolute time (for that is not an object of perception).” (A200/B245) 27. “To all empirical cognition there belongs the synthesis of the manifold through the imagination, which is always successive; i.e., the representations always follow each other in it. But the order of the sequence (what must precede and what must follow) is not determined in the imagination at all, and the series of successive representations can be taken backwards just as well as forwards.” (A201/B246) 28. “Objective significance cannot consist in the relation to another representation (of that which one would call the object), for that would simply raise anew the question: How does this representation in turn go beyond itself and acquire objective significance in addition to the subjective significance that is proper to it as a determination of the state of mind?” (A197/B242) 29. “[T]he appearances themselves must determine their positions in time for each other, and make this determination in the temporal order necessary, i.e., that which follows or happens must succeed that which was contained in the previous state in accordance with a general rule, from which arises a series of appearances.” (A200/B245) 30. “If, therefore, my perception is to contain the cognition of an occurrence, namely that something actually happens, then it must be an empirical judgment in which one thinks that the sequence is determined, i.e., that it presupposes another appearance in time which it follows necessarily or in accordance with a rule.” (A201/B246) 31. This manifold of representations must occur to the experiencing subject in a determinate temporal order in standard conditions. Fanciful scenarios involving, say, a system of mirrors that delay the light from the upstream ship from reaching the perceiver until after the light from the downstream ship has reached him can only be made sense of using the object-concept ‘ship’ as deviant cases that are essentially parasitic on more standard ones. This is analogous to the way in which a white object’s seeming to be red in red light can only be made sense of by way of the concept of an object’s looking as it seems in standard conditions. 32. Importantly, we do represent the parts of the house as necessarily connected to each other—this is what it is to represent the house as an object—but we do not represent any as being the cause of the others. Or to put it another way, the concept ‘house’ is subsumed under the Category ‘Community’ rather than ‘Causality and Dependence.’ 33. The concept ‘house’ alone does not contain any such rule. Of course, when combined with certain facts about the location and behavior of the experiencing subject, such a rule will be produced. E.g., ‘I am standing in front of a house and running my eyes downward beginning from a certain angle’ commits one to ‘I will see first a roof, then a façade, then a foundation.’ 34. See Garrett, Cognition and Commitment, 58–75 on the senses in which Hume’s so-called Separability Principle is both empirically discovered and also

232

35.

36.

37. 38.

39.

Self and World in the Analogies of Experience a definition of distinctness, separation, and distinguishability. Compare also Garrett’s discussion with the discussion of ‘representation’ from Chapter 1. Notice that for Kant ‘appearance’ (der Erscheinungen), which is the term he employs in the above passage, is subject to the same ambiguity as ‘representation.’ ‘Appearance’ can indicate either that which appears or the act of being appeared to. I.e., it can be the object of an act of appearing or the act of appearing itself. One important such disanalogy being that in the case of the missing ingredients we have empirical criteria available to us against which we can check our conclusion. E.g., DNA samples taken from the sandwich, video surveillance footage of the pantry, etc. Hume’s point against Descartes is precisely that when it comes to the subject of experience, there can be no such criteria. As Hume takes empirical evidence to be the only possible warrant for such a conclusion, he must reject it. Nelson proposes something like this kind of fusion as lying at the heart of Descartes’s compositional theory of mental representation. See Nelson, “Descartes’ Ontology of Thought.” “Thus the original and necessary consciousness of the identity of oneself is at the same time a consciousness of an equally necessary unity of the synthesis of all appearances in accordance with concepts, i.e., in accordance with rules that not only make them necessarily reproducible, but also thereby determine an object for their intuition.” (A108) In preschool, my class sang Do-Re-Mi, each of us singing the part of one of the notes. I was Fa: a long, long way to run.

REFERENCES Allison, Henry. Kant’s Transcendental Idealism. New Haven: Yale University Press, 1983. Beck, Lewis White. Essays on Kant and Hume. New Haven: Yale University Press, 1978. Bennett, Jonathan. Kant’s Analytic. Cambridge: Cambridge University Press, 1966. Buchdahl, Gerd. Metaphysics and the Philosophy of Science. Oxford: Basil Blackwell, 1969. Feyerabend, Paul. Against Method. London: Verso, 1975. Friedman, Michael. “Causal Laws and the Foundations of Natural Science.” In The Cambridge Companion to Kant, edited by Paul Guyer, 161–99. New York: Cambridge University Press, 1992. Garrett, Don. Cognition and Commitment in Hume’s Philosophy. Oxford: Oxford University Press, 1997. Guyer, Paul. Kant and the Claims of Knowledge. Cambridge: Cambridge University Press, 1987. Hume, David. A Treatise of Human Nature. Edited by L. A. Selby-Bigge. New York: Oxford University Press, 1974. Hume, David. A Treatise of Human Nature. Edited by David Fate Norton and Mary J. Norton. New York: Oxford University Press, 2000. James, William. Principles of Psychology Volume 1. New York: Dover Publications, 1890. Kant, Immanuel. Critique of Pure Reason. Translated by Paul Guyer and Allen W. Wood. Cambridge: Cambridge University Press, 1998. Kant, Immanuel. Notes and Fragments. Edited by Paul Guyer. Translated by Curtis Bowman, Paul Guyer, and Frederick Rauscher. Cambridge: Cambridge University Press, 2005.

Self and World in the Analogies of Experience

233

Kemp Smith, Norman. A Commentary to Kant’s Critique of Pure Reason. Atlantic Highlands: Humanities Press, 1991. Kripke, Saul. Naming and Necessity. Cambridge, MA: Harvard University Press, 1980. Kuhn, Thomas. The Structure of Scientific Revolutions. Chicago: University of Chicago Press, 1962. Kuhn, Thomas. “Objectivity, Value Judgment, and Theory Choice.” In The Essential Tension, 320–39. Chicago: University of Chicago Press, 1977. Lakatos, Imre. “Falsification and the Methodology of Scientific Research Programmes.” In Criticism and the Growth of Knowledge, edited by Imre Lakatos and Alan Musgrave, 91–196. Cambridge: Cambridge University Press, 1970. Melnick, Arthur. Kant’s Analogies of Experience. Chicago: University of Chicago Press, 1973. Nelson, Alan. “Descartes’ Ontology of Thought.” Topoi 16 (1997): 163–78. O’Shea, James. “Kantian Matters: The Structure of Substance.” Acta Analytica 15 (1996): 67–88. Popper, Karl. The Logic of Scientific Discovery. London: Hutchinson, 1959. Rosenberg, Jay. “Comparing the Incommensurable: Another Look at Convergent Realism.” Philosophical Studies 54 (1988): 163–93. Sellars, Wilfrid. “Scientific Realism or Irenic Instrumentalism: A Critique of Nagel and Feyerabend on Theoretical Explanation.” In Boston Studies in the Philosophy of Science, Vol. II, edited by Robert Cohen and Max Wartofsky, 171–204. New York: Humanities Press, 1965. Sellars, Wilfrid. Science and Metaphysics. Atascadero, CA: Ridgeview Publishing Company, 1967. Strawson, Peter. The Bounds of Sense. London: Methuen Ltd., 1966. Van Cleve, James. “Substance, Matter, and Kant’s First Analogy.” Kant-Studien 70 (1979): 149–61. Van Cleve, James. Problems From Kant. New York: Oxford University Press, 1999. Watkins, Eric. Kant and the Metaphysics of Causality. Cambridge: Cambridge University Press, 2005. Wittgenstein, Ludwig. Philosophical Investigations. Translated by G. E. M. Anscombe. Oxford: Blackwell Publishers, 1958. Wolff, Michael. Die Vollständigkeit der kantischen Urteilstafel. Frankfurt a.M.: Klostermann, 1995.

6

The Inferential Self

In the previous chapter, we saw that the role that Kant’s account of how we represent objects is motivated in part by his solution to a problem that he found in Descartes and Hume concerning the representation of the self. Representations of complex objects play a unifying role in cognition: it is because we can form a single, unified representation of a complex object that we can think of ourselves as the single subject of experience persisting through time. Furthermore, we saw at the close of the previous chapter that the kind of unity that the representation of an object has is an inferential unity and that the kind of unity that the representation of the self has is, correspondingly, the unity appropriate to an inferrer: the unity appropriate to the bearer of conceptual norms. It is the purpose of the current chapter to flesh out that bare-bones suggestion. The subject of this chapter, then, will be Kant’s account of the representation of the self, and its focus will be on the Paralogisms, where Kant presents his account by way of contrast to what he calls Rational Psychology, “a putative science, which is built on the single proposition ‘I think’ ” (A342/B400). The Rational Psychologist draws conclusions about the nature of the self (as it is in itself) from facts about the representation of the self. Again, as we saw at the close of the previous chapter, this move—from semantic facts about the representation of the self to ontological conclusions about the nature of that which is represented—is a dubious one. In fact, Kant argues in the Paralogisms that the arguments of the Rational Psychologist all depend on a crucial equivocation. In critiquing the arguments of the Rational Psychologist, Kant outlines what he takes to be the entirely legitimate facts that together constitute our representation of the self, and unsurprisingly, these all turn out to be best understood in the context of Kant’s inferentialism. That Kant’s particular theory of mental representation comes into play in the Paralogisms is clear when we consider his famous remark that Through this I, or He, or It (the thing), which thinks, nothing further is represented than a transcendental subject of thoughts = x which is

The Inferential Self

235

recognized only through the thoughts that are its predicates, and about which, in abstraction, we can never have even the least concept; because of which we therefore turn in a constant circle, since we must always already avail ourselves of the representation of it at all times in order to judge anything about it; we cannot separate ourselves from this inconvenience, because the consciousness in itself is not even a representation distinguishing a particular object but rather a form of representation in general, insofar as it is to be called a cognition; for of it alone can I say that through it I think anything. A346/B404 The ‘I think’ is a representation, but it does not represent any object. It is not only a representation but is also form of representation, and a necessary one at that. We recognize this representation/form of representation only via the thoughts that are predicated of it. Given only what Kant says here, making sense of this representation would be difficult indeed. The purpose of the current chapter is to show that we can, however, easily understand all of these claims, and those articulated in the Paralogisms themselves, by using Kant’s inferentialism as our guide. That is, if we can understand the kind of representation the ‘I think’ is, then we can understand all of the strange things that Kant says about it and his arguments against the Rational Psychologist. To begin, we can notice something puzzling about the above quotation. Kant tells us that in order to judge anything about the ‘I think,’ we must already avail ourselves of the representation of it, through the thoughts that are its predicates, and that we therefore turn in a constant circle. On its own this seems like a strange claim. It looks like Kant is claiming that because the ‘I think’ must accompany all of our representations, we cannot say anything about it. Why should that be? There certainly doesn’t seem to be anything about the fact that we must always represent all of our thoughts as belonging to a single self that rules out our being able to think anything about this self. In fact, one might think that it is precisely this ubiquity of the self that demands that we be able to say something about it and that puts us in a uniquely good position to do so. In fact, Kant gives the key to understanding this puzzling line of reasoning in the opening clause of this passage. Through this I, or He, or It (the thing), which thinks, nothing further is represented than a transcendental subject of thoughts = x. The reason that we cannot say more about the self is not exactly because the self accompanies all of our representations but rather because this is all that it does. This is why what we end up doing in trying to describe the

236

The Inferential Self

self is turn in a constant circle. The only facts that we can articulate about the self are semantic facts, facts about the representation of the self in its role as the unifier of all of our other representations. Here are two other places later in the Paralogisms in which Kant makes strikingly similar sounding claims: this concept [of the self] merely revolves in a circle around itself and brings us no farther in regard to even one single question about synthetic cognition. A366 [The Rational Psychologist] makes his own representations into objects, and thus goes round and round in an eternal circle of ambiguities and contradictions. A395 Notice in the first quotation that it is the concept of the self that merely revolves in a circle around itself. I.e., all that the representation ‘I think’ represents is its own representative role. Similarly, in the second quotation, the mistake that the Rational Psychologist makes is taking what is represented by the ‘I think’ (a representing act) to be a full-fledged object. Kant’s thesis in the Paralogisms is that what we can say about the self are all and only tautologies, tautologies that concern solely the representation of the self. When we try to say what this representation is a representation of, we are simply thrown back onto certain semantic facts about the representation itself. All there is to the representation of the self are these semantic facts. It is imperative, therefore, that we come to understand these semantic facts, and my thesis is that the best way to do so is via Kant’s inferentialism. Here, then, is what the structure of the current chapter will be. I will begin by situating the kind of interpretation of the Paralogisms that I have just outlined within the secondary literature on the topic. That interpretation— according to which the ‘I think’ is a purely formal representation that represents only its own representative role rather than any sort of object—is what is called a formalist one, so this review of the literature will focus on nonformalist ones and the reasons for thinking that such approaches cannot properly account for Kant’s texts. Next, I will proceed to the three arguments that together form the misbegotten basis of Rational Psychology, and from these, we will extract the three facts that Kant claims constitute the representation of the self.1 Then, I will account for each of these facts about the representation of the self using the resources we that have heretofore gathered in the course of our discussions of Kant’s inferentialism. I will then conclude with a discussion of the relation of the representation of the self, the transcendental unity of apperception, to our empirical faculty for representing our thoughts, inner sense.

The Inferential Self

237

INTERPRETATIONS OF THE PARALOGISMS As just noted, a rough but serviceable way of dividing the recent literature on the Paralogisms is by grouping together those who understand Kant as holding that only a purely formal definition of the ‘I think’ is possible and those who understand Kant as holding that the ‘I think’ represents some sort of object, substance, or activity as well. Included in the former group are Kemp Smith, Sellars, Rosenberg, Bird, Theil, and Rosefeldt.2 The position that I will articulate on the Paralogisms will likewise be a version of this general approach with an emphasis on understanding it via Kant’s theory of concepts-qua-inferential-rules. What must be done, then, is to rebut the arguments of those on the other side of the divide: those scholars who understand the Paralogisms, and Kant’s account of the transcendental unity of apperception more generally, as implying specific material facts about the nature of that which is represented by the ‘I think.’ Of course, a diversity of interpretations fall under this general heading. Ameriks draws on the evolution of Kant’s views of the self from Kant’s earliest broadly naturalist position, through his rationalist and skeptical phases, to the critical phase, in which Ameriks argues that Kant retains significant sympathies with the rationalist metaphysician even if he critiques particular arguments that the latter employs. So, for Ameriks, the ‘I think’ has as its ultimate object an immaterial noumenal substance.3 Melnick likewise argues that “Kant believes there are ontological conclusions to be drawn from the cogito [. . .] only they are not those the rational psychologist draws.”4 He takes these conclusions to be that the ‘I think’ represents an activity, which is of necessity temporal and is therefore the activity of a phenomenal substance. Longuenesse also takes the error of the rational psychologist not to be in identifying the self as a substance but only in identifying it as the wrong kind of substance, a transcendent one, when it is, in fact, a mere empirical substance, the identity of which depends on “numerical identity through time of a spatio-temporal entity a living being endowed with mental states.”5 In addition to the divide among scholars over what is, in fact, implied by Kant’s account of the transcendental unity of apperception, the dialectic is further complicated by a debate orthogonal to that one over what Kant himself takes to be implied by that account. So, for example, Longuenesse readily admits that her interpretation of the ‘I think’ as referring to a living being is more Kantian than Kant’s; Kitcher holds that her interpretation is the best way to understand Kant even though it conflicts with a number of his explicit statements on the matter;6 Ameriks argues that in refuting the arguments of the Rational Psychologist Kant never intended to reject his conclusions, and that Kant’s texts are consistent with his taking the ‘I think’ to refer to a noumenal substance. Thus, the waters here are troubled, and while I cannot purport to calm them entirely, I can undertake to do the following. I will spend the remainder

238

The Inferential Self

of this section engaging each of these scholars in turn on the question of what Kant takes his theory to imply. I believe that there is ample evidence to support the view that Kant himself takes the ‘I think’ to be, as he puts it, a mere form of representation that represents no object (or noumenal substance, or activity of a phenomenal substance, etc.). Having tentatively established that Kant himself understands the transcendental unity of apperception as a purely formal representation, in the next section, I will turn to the details of the arguments of the Paralogisms from which I will argue one can extract an account of what it means to be such a purely formal representation, and what the specific formal content of the ‘I think’ is. I hope to thereby show that not only does Kant take the ‘I think’ to be a purely formal representation but also that this is itself a philosophically tenable position for him to occupy, especially in light of his theory of concepts-qua-inferential-rules. That is the distal plan; the proximate one is as follows. Longuenesse’s concession that her interpretation of the ‘I think’ is more Kantian than Kant’s will provide an instructive opportunity to review some of the relevant texts in favor of such an approach, so I will begin with that. Following that, I will turn to Ameriks’s account of the ‘I think’ as representing an immaterial noumenal substance and show that a careful examination of the texts that he uses to support his view actually point fairly clearly to a formalist position instead. Finally, I will turn to Melnick’s view, which is the one I will consider that is most like my own. Melnick rightly argues that interpretations that take the ‘I think’ refer to any kind of substance or object (empirical or noumenal) cannot be right. He further argues, though, that what is represented by the ‘I think’ is instead the activity of an empirical substance. I agree with Melnick that Kant takes the ‘I think’ to be a kind of activity, but whereas Melnick takes this to the activity of an empirical substance, I understand the ‘I think’ as representing a normative-inferential activity that is not per se temporal. Before turning to my explication of this difference with Melnick, however, I will first consider his arguments against such a purely formal approach to the ‘I think,’ which I will show are supported by a selection of Kant’s texts that is overly parochial. Reading those texts in their proper context again demonstrates that Kant’s position is that the ‘I think’ is a purely formal representation. That will conclude the business of this section, and in the next section I will turn from critiquing these nonformalist interpretations of the ‘I think’ to presenting the details of my own formalist approach. I will begin, then, with Longuenesse. According to Longuenesse, “Kant on the Identity of Persons,” while the representation ‘I think’ does play the formal role of uniting one’s representations as the representations of a single subject of experience persisting through time, such a representation only correctly represents a self insofar as it attributes those representations to a single empirical substance: the human body. According to Longuenesse,

The Inferential Self

239

What emerges from the preceding analysis is that it is a priori true of my thoughts (because it is a necessary condition for my engaging in any activity of thinking at all) that I attribute them to myself. This selfattribution expresses the unity and identity of an activity of ordering these thoughts through time, an ordering for which I hold myself, at any given time responsible and accountable. But this in no way guarantees that these thoughts I am at any given time referring to myself (synthesizing through time as my thoughts), have at different points in time indeed been thought by one and the same thinking being which I could identify as being me.7 Longuenesse goes on to suggest that what makes a representation of the self an accurate representation in this way is that the self that is so represented also turns out to be empirically identifiable as a single empirical object persisting through time: In the empirical world [. . .] the identity of persons depends on two factors [. . .] (1) unity of apperception, namely unity of consciousness of the contents of mental states, which makes it possible to accompany them with the thought ‘I think’; and (2) numerical identity through time of a spatio-temporal entity a living being endowed with mental states.8 So, whereas the Paralogisms’ three analytic propositions might constitute a representation of the self, as I will argue that they do, Longuenesse’s suggestion is that what is represented by such a representation is not, as I will argue, merely a fact about that representation itself but the self as a persisting empirical object. On this line, Kant’s critique of the Rationalist Psychologist is not that the latter takes the ‘I think’ to represent an object but only that he takes it to represent the wrong kind of object: a transcendent one. Longuenesse herself concedes that such an account might be more Kantian than it is Kant’s, but it is worth reviewing some of the most prominent reasons why Kant himself, at least, could not possibly hold such a view. To begin, Kant is explicit that representing oneself as a single subject of experience does not require representing oneself as any kind of object at all, metaphysically transcendent or empirical. Here is a representative quotation wherein Kant is explicit that what is represented by the ‘I think’ cannot be an empirical object. Now mere apperception (“I”) is substance in concept, simple in concept, etc. and thus all these psychological theorems are indisputably correct. Nevertheless, one by no means thereby cognizes anything about the soul that one really wants to know, for all these predicates are not valid of intuition at all, and therefore cannot have any consequences that could be applied to objects of experience. A400

240

The Inferential Self

The representation ‘I think’ does not tell us anything about the soul because it is not an intuition and therefore “cannot have any consequences that could be applied to objects of experience.” Not only does the ‘I think’ not represent an empirical (or noumenal) object, but it also cannot have any consequences that can be applied to these. So, Longuenesse’s thesis that the ‘I think’ is only a correct representation if what it, in fact, represents is an empirical body cannot be Kant’s, for this would surely be “a consequence that could be applied to objects of experience.” Furthermore, we have already encountered at least an inchoate version of Kant’s reasoning about why this must be the case: the ‘I think’ is not itself a representation of any object because it is a mere form of representation in general, i.e., what is represented by the ‘I think’ is only its own representational role. If anyone were to pose the question to me: What is the constitution of a thing that thinks? then I do not know the least thing to answer a priori, because the answer ought to be synthetic (for an analytic answer perhaps explains thinking, but gives no extended cognition of that on which thinking rests as to its possibility). But for every synthetic solution, intuition is necessary; but this is entirely left out of so universal a problem. [. . .] But now although I know no general answer to that question, yet it seems to me that I could give it in the individual case, in the proposition that expresses self-consciousness: “I think.” A398–399 There is no answer to the general question of what it is that answers to the ‘I think’ because the question is posed at such a high level of generality that nothing other than analytic propositions about the nature of thinking can be given. But notice that even in the more specific case in which we are free to appeal to intuition, the only answer that we can give is that our thinking is constituted by the proposition ‘I think’ itself. The ‘I think’ as a representation only represents its own representational activities.9 Recall the passages we examined in the introduction to this chapter declaring that the representation of the ‘I think’ revolves in circles around itself. The representation ‘I think’ only represents its own representing activities, and so attempting to determine the object of such a representation, as Longuenesse does, can only result in going “round and round in an eternal circle of ambiguities and contradictions.” Another important point to notice about Longuenesse’s account, in addition to the fact that no empirical object is represented by the ‘I think,’ is that a manifold of representations’ being attributable to such an empirical self cannot be among the correctness conditions for such a representation. Consider what Kant points out about the Rational Psychologist’s conclusion that the self is a simple noumenal object: the unity of a thought consisting of many representations is collective, and, as far as mere concepts are concerned, it can be related to the

The Inferential Self

241

collective unity of the substances cooperating in it (as the movement of a body is the composite movement of all its parts) just as easily as to the absolute unity of the subject. A353 While Kant is here addressing the supposition that the ‘I think’ refers to a simply noumenal substance, there is no reason why his point should not also stand with regard to empirical substances. As far as mere concepts are concerned, the representation ‘I think’ could just as easily refer to the collective unity of substances cooperating in an empirical body, e.g., the body of a living human being, as it could that body itself. As we have seen in our discussion of the Analogies, Kant’s theory of representation implies that, strictly speaking, there are no living human bodies but that such bodies are really only the alterations of the single sempiternal substance that itself is never created or destroyed. On Longuenesse’s account, it would follow from this that the representation ‘I think’ is only ever incorrectly used, but that is clearly not Kant’s own view. On Kant’s view that representation does not speak at all to what substance (noumenal or empirical), in fact, realizes it, and so the question of its application is independent of the question of the nature of empirical (or noumenal) substance. Of course, there must be a place for the empirical self in Kant’s account, but it does not make its appearance as what is represented by a representation of the self. As I have suggested, what is represented by such a representation is that which is subject to various conceptual/inferential rules, and while this must turn out to be some thing, or some several things, in the empirical world, the representation of such a self is not a representation of any such thing or things as such. Thus, the correctness conditions for such a representation cannot include that the self be any such object. If that is right, then Longuenesse is wrong to think that for Kant the representation of the self aims at representing some kind of object, and she is therefore also wrong to think that the object at which it aims is the empirical self. In fact, the representation of the self, as Kant says, does not represent any kind of object at all but rather a form of representation in general.10 Ameriks’s study of the Paralogisms is expansive and nuanced, and it deserves more thorough a response than can be given here. Nonetheless, by focusing on Ameriks’s treatment of Kant’s discussion of the immateriality of the soul, it will be possible to isolate a single point on which Ameriks’s interpretation depends and which the formalist rejects. This point concerns what Kant means in passages such as the following in which he claims that the ‘I think’ signifies a substance only in the idea but not in reality. Meanwhile, one can quite well allow the proposition The soul is substance to be valid, if only one admits that this concept of ours leads no further, that it cannot teach us any of the usual conclusions of the

242

The Inferential Self rationalistic doctrine of the soul [. . .] thus that it signifies a substance only in the idea but not in reality. A351

As Ameriks understands such claims, that the ‘I think’ signifies “a substance only in the idea but not in reality” does not contradict the further claim that that what is signified by such a representation is “truly substantial”: if a being is a substance ‘only in idea,’ that is, a being that meets the general necessary (‘transcendental’) conditions for being a substance, this hardly counts against its truly being substantial.11 For Ameriks, it is not what meets the empirical definition of substance— permanence—that is “truly” substantial but rather what meets the transcendental conditions—that the representation of which cannot be employed as a determination of another thing.12 So, since Ameriks understands ‘only in idea’ as signaling that the ‘I think’ meets this ontologically more basic criteria, he understands that representation as representing the subject of experience as being composed of the most basic ontological substance. It is noteworthy here that Ameriks leaves off the final clause of the passage from A351 quoted above. Ameriks’s claim is that to be a substance ‘only in idea’ is to meet the general necessary conditions for being a substance and that meeting these conditions is not only compatible with being ontologically basic but in fact implies being so. Kant’s complete claim about the ‘I think’ in the passage above, though, is that “it signifies a substance only in the idea but not in reality.” The phrase that Ameriks leaves out, ‘not in reality,’ on its most straightforward and natural reading, would certainly seem to imply that what the ‘I think’ represents is not something “truly substantial.” In fact, the inclusion of that phrase certainly makes it seem as if Kant means to contrast what the ‘I think’ represents with what is truly substantial. When Kant writes that the ‘I think’ represents a substance only in idea but not in reality, the formalist will hold, what he means is that the ‘I think’ has the formal characteristics of a representation of substance but not the material ones and so does not, in fact, represent anything as a substance at all. Ameriks does not give much consideration to formalist positions. Here, for example, is the entirety of his argument against Bennett’s version of such an approach. Something either meets the requirements for constituting substance or it does not; there is no merely empty or formal middle way such as Bennett suggests.13 And here is his argument against Chisholm’s claim that ‘I’ does not refer, which is a claim that is often part of a formalist reading (as it will be in the one I will present in the next section).

The Inferential Self

243

And of course Kant definitely does believe that the ‘I’ refers in that there is something substantial underlying it. [. . .] Kant is sure there is an underlying reality here and he even asserts that we are acquainted with its existence. (B429)14 Without a quotation, it is not obvious to which part of B429 Ameriks is referring, but the following passage seems to be the best contender as it is the only place in B429 in which Kant makes mention of the existence of the subject. But the proposition “I think,” insofar as it says only that I exist thinking, is not a merely logical function, but rather determines the subject (which is then at the same time an object) in regard to existence, and this cannot take place without inner sense, whose intuition always makes available the object not as thing in itself but merely as appearance. B429 The problem with Ameriks’s use of this passage, though, is that Kant is here contrasting the use of the ‘I think’ in inner sense with its use as the transcendental unity of apperception, and it is the latter that is of concern to both Chisholm and Ameriks. It is fairly uncontroversial that one can refer to oneself as an object of introspection, as what Kant calls the empirical unity of apperception. What is at issue is what the representation of the subject of experience, including the subject of experiences of inner sense, represents. In the passage that Ameriks cites, Kant is only concerned with the self as the object of inner sense. Notice that he is here considering the ‘I think’ in such a way that it “is not a merely logical function.” In the paragraph before this one, however, Kant explicitly casts his discussion there as concerning the ‘I think’ qua subject of experience and begins it by asserting, “Thinking, taken in itself, is merely the logical function” (B429). So, in the passage that Ameriks cites, Kant is clearly contrasting the discussion of the ‘I think’ qua subject of experience with the ‘I think’ qua the object of inner sense. Here is how the first passage continues. Thinking, taken in itself, is merely the logical function and hence the sheer spontaneity of combining the manifold of a merely possible intuition; and in no way does it present the subject of consciousness as appearance, merely because it takes no account at all of the kind of intuition, whether it is sensible or intellectual. [. . .] If here I represent myself as subject of a thought or even as ground of thinking then these ways of representing do not signify the categories of substance or cause, for these categories are those functions of thinking (of judging)

244

The Inferential Self applied to our sensible intuition, which would obviously be demanded if I wanted to cognize myself. B429

The point of this passage, in contrast to the one that Ameriks appears to cite, is precisely to make clear that in its role as the representation of the experiencing subject, the ‘I think’ “is merely the logical function” and that such a function does not require the deployment of the category ‘substance’ and so does not constitute a cognition of the self. Notice further that one does not deploy the category ‘substance’ precisely because “these categories are those functions of thinking [. . .] applied to our sensible intuition.” I.e., one does not deploy the category substance in thinking of oneself precisely because in thinking oneself one does not have available the matter, sensible intuition, necessary to do so. This is because the ‘I think’ is, as Kant states at the outset, a purely formal representation. Thus, the passage that Ameriks appears to cite as evidence that Kant not only denies that there is a purely formal representation of the self but also “even asserts that we are acquainted with its existence” is, in fact, a passage in which Kant is explicitly contrasting the self with which we are acquainted in inner sense with a purely formal representation of the self. The latter is what constitutes the transcendental unity of apperception and, in contrast to what Ameriks argues, does not deploy the category of substance at all. We can now turn to Melnick, who offers both a theory according to which the ‘I think’ represents not an empirical substance itself but rather the activity of such a substance and also arguments against the kind of understanding of the ‘I think’ (as a purely formal representation) that I have proposed to defend. At the outset, it is worth noting that the evidence that Melnick offers for his positive view is, as he takes it to be, strong evidence against the kinds of theories that we have already been considering: those that take the ‘I think’ to refer to a kind of substance or object (noumenal or empirical). So, Melnick and I agree on that much. We also agree that the ‘I think’ is best conceived as a kind of activity. Where we differ is firstly in regard to what the nature of this activity is—Melnick takes it to be the activity of a phenomenal substance; I take it to be a purely conceptual activity that is not, per se, the activity of any particular kind of thing—and secondly in regard to what kind of representation the ‘I think’ is—Melnick takes it to be the representation of the activity of a phenomenal substance; I take it be a purely formal representation. We can begin with Melnick’s arguments against this kind of formalist approach. Melnick’s first argument against the formalist approach begins with a passage from A353 that he takes as evidence that “[f]or Kant the simplicity of the ‘I’ is not derivable directly (independently of the cogito) by thinking of subjects of thought generally.”15

The Inferential Self

245

The proposition “A thought can only be the effect of the absolute unity of the thinking being” cannot be treated as analytic. A353 As Melnick understands this claim, Kant’s denial of the analyticity of this proposition shows that one cannot derive the simplicity of the representation ‘I think’ from there merely formal content of that representation.16 What is odd about Melnick’s choice of supporting texts, though, is that Kant is very clearly not talking about the representation ‘I think’ here but rather the attempt by the Rational Psychologist to demonstrate the ontological simplicity of the thinking being. That is, Kant is not making the point that Melnick attributes to him but is rather merely reiterating his thesis regarding the second Paralogism: that one cannot treat this claim as analytic if one also understands it as making a claim about the self qua noumenal substance. That this is what Kant is saying in this passage is first signaled by the phrases ‘absolute unity’ and ‘thinking being.’ The corresponding non-ontologically committing claim, ‘A thought can only be the effect of the logical unity of thinking’ is, as Kant repeatedly stresses, analytic. For example, just a few pages later Kant explicates what he takes to be the correct analysis of the analytic proposition ‘I am simple.’ But I am simple signifies no more than that this representation I encompasses not the least manifoldness within itself, and that it is an absolute (though merely logical) unity. A355 Note that here Kant modifies ‘absolute unity’ with ‘though merely logical’ and nullifies the ontological import of this analytic proposition by pointing out that it refers not to a thinking being but only to the representation ‘I think’ itself. That Melnick misunderstands Kant’s claim in the passage that he cites is also made clear by how that paragraph itself continues. Just a few sentences later, Kant concludes his argument for the claim above as follows. Thus there can be no insight into the necessity of presupposing a simple substance for a composite thought according to the rule of identity. A353 Kant’s point is that the purported proof of the ontological simplicity of the thinking subject that the Rational Psychologist offers is a mistake and that no such ontological conclusion can be reached using merely the rule of identity (from only analytic propositions). Kant is not denying that the claim ‘I am simple’ is analytic—he believes that it is—but rather he is denying the analyticity of the claim ‘I am a simple substance.’ Since it is claims such as the former, ‘I am simple,’ not the latter, that the formalist will use to

246

The Inferential Self

explicate the content of the representation ‘I think,’ there is nothing in the passage that Melnick presents that the formalist need contest. Putting aside Melnick’s his misreading of this passage, though, his more general point still needs to be addressed. That point is his claim that facts about the representation ‘I think’ cannot be derived from the concept of thinking in general but require nonformal supplementation. In support of this claim, Melnick cites a passage from A354 where Kant claims that the ‘I think’ cannot be regarded as the “concept of a thinking being in general” and that one does not represent oneself primarily as an instance of a kind but rather that one represents other thinking beings only derivatively by representing them as being selves like oneself. Melnick himself concludes straight away from this passage that: As that is my only representation of other beings, one could say it is a “logical” truth that I must represent all subjects as simple, but this is not a sense of logically true that contrasts with the simplicity or indivisibility being something real (as revealed in the cogito).17 Melnick’s thought here seems to be that if the ‘I think’ were a merely formal representation, then it would primarily be applied as a concept to all subjects rather than first to me and only derivatively to others. This conditional, however, depends entirely on what the formal role of the ‘I think’ actually is. As we will see in the following section, for example, it is entirely compatible with the ‘I think’ being a formal representation—with its representing nothing more than its own inferential role—that, as Kant claims, it be possible to accompany any and all representations that can properly be called mine with the ‘I think.’ That, in turn, is compatible with the thesis that it is primarily by so accompanying my representations that I come to use and understand the ‘I think’ and that this use of it is logically prior to using it to represent others as subjects. Nothing about this role, however, entails that the representation ‘I think’ has as its representational content anything nonformal at all. That is, there is no in-principle reason why the content of the ‘I think’ cannot be constituted entirely formally while at the same time having its primary use be first personal. Part of the formal role of the ‘I think’ is to accompany all of my representations, but this claim is merely analytic: there is no way to specify which representations are mine other than as those that can be accompanied by the ‘I think.’ That is the point of the second Paralogism: the representation ‘I think’ is not analyzable into any more simple representations, but we ought not to take this to imply any ontological facts about that which might play this role. Again, looking at the complete passage from which Melnick takes his excerpt makes Kant’s point more clear. In this case, when Kant warns against making the concept ‘I think’ into a concept of thinking being in general, he is really only warning against generalizing the invalid conclusion of

The Inferential Self

247

the Rational Psychologist, against moving from a merely formal fact about the representation ‘I think’ to an ontological fact about thinking beings in general. Here, therefore, as in the previous paralogism, the formal proposition of apperception, I think, remains the entire ground on which rational psychology ventures to extend its cognitions; this proposition is of course obviously not an experience, but rather the form of apperception, on which every experience depends and which precedes it, yet it must nevertheless always be regarded only in regard to a possible cognition in general, as its merely subjective condition, which we unjustly make into a condition of the possibility of the cognition of objects, namely into a concept of a thinking being in general, because we are unable to represent this being without positing ourselves along with the formula of our consciousness, in the place of every other intelligent being. A354 The Rational Psychologist infers from the “formal proposition of apperception, I think,” the mere “form of apperception” that he is a thinking being, a noumenal thinking substance. Kant notices that if one makes this mistake, since one also thinks of other subjects of experience by thinking of them as like oneself, one might be tempted to make the further mistake of generalizing this conclusion to the thesis that the concept of thinking in general is the concept of a thinking substance. Again, Kant is not making the claim that Melnick takes him to make—that the ‘I think’ is more than merely formal—but rather is again only making the much less controversial point that one should avoid the mistake of the Rational Psychologist of drawing ontological conclusions from what—as the complete quotation makes clear—are merely formal facts about the ‘I think.’ Melnick continues his argument against the formalist by next citing a passage in support of his claim that something must be added to the formal representation, ‘I think,’ in order to derive a positive account of the thinking subject. As Melnick puts it, “[F]or in the cogito Kant says, ‘I am the being itself’ ” (Melnick, 2009: 64). Here, once again, is the complete passage to which Melnick refers. But now I want to become conscious of myself only as thinking; I put to one side how my proper self is given in intuition, and then it could be a mere appearance that I think, but not insofar as I think; in the consciousness of myself in mere thinking I am the being itself, about which, however, nothing yet is thereby given to me for thinking. B429 Kant is not here expressing the claim that the ‘I think’ represents something nonformal but rather doing exactly the opposite. Kant’s point is that in

248

The Inferential Self

representing “the being itself” we do not represent any kind of object at all. Rather, we represent the thinking subject as a “being” only formally—the ‘I think’ can only play the role of a subject and never of a predicate—and this is precisely why “nothing yet is thereby given to me for thinking” by such a representation. This is clear once we again take into account the context of the passage. Here is what begins the paragraph from which this quotation is taken. Thinking, taken in itself, is merely the logical function and hence the sheer spontaneity of combining the manifold of a merely possible intuition; and in no way does it present the subject of consciousness as appearance, merely because it takes no account at all of the kind of intuition, whether it is sensible or intellectual. B428 Again, Kant’s point is that one must not be led by the formal role of the ‘I think’ to conclude that anything is represented by it other than that role itself. “In no way does it present the subject of consciousness as appearance.” The ‘I think’ does not represent an object, but is a “merely logical function.” Melnick concludes his case against the formalist by presenting his diagnosis of what has led the formalist astray and offering support for his own counterproposal. Melnick’s claim is that the analytic propositions noted by the formalist do not constitute Kant’s positive theory of the experiencing subject but are rather all that one can say if one “takes the ‘I think’ problematically.” Melnick cites as evidence Kant’s specific reference to the existence of the ‘I think’ the following passage. If, on the contrary, we follow the analytic procedure, grounded on the “I think” given as a proposition that already includes existence in itself . . . B418 Melnick’s parsing of this passage is that in it Kant “is setting out what can be concluded when the cogito is taken existentially rather than problematically. It is this and this alone I claim that constitutes Kant’s positive doctrine” (Melnick, 2009: 67). Once again, however, a consideration of the context in which this passage appears makes it clear that Melnick’s reading of it cannot be right. What Kant is actually doing here is showing that the analytic propositions that serve as the premises in the fallacious arguments of the Rational Psychologist can themselves be reached via two different ways of considering the experiencing subject: first via the very concept of such a subject itself, and secondly from considering what is left over of oneself as an experiencing subject “after everything empirical has been detached

The Inferential Self

249

from it” (B418). At the conclusion of both procedures, what is left over is the four-part proposition: (1) I think, (2) as subject, (3) as simple subject, (4) as identical subject, in every state of my thinking. B419 Here is what Kant has to say about this proposition—which is explicitly derived from “the ‘I think’ given as a proposition that already includes existence in itself,” which Melnick takes to present the positive content of Kant’s theory of the experiencing subject. Now because in the second proposition here it is not determined whether I could exist and be thought of only as subject and not as predicate of another thing, the concept of a subject is here taken merely logically, and it remains undetermined whether or not substance is to be understood by it. Yet in the third proposition the absolute unity of apperception, the simple I, in the representation to which every combination or separation constituting thought is related, also becomes important for its own sake, even if I have not settled anything about the subject’s constitution or subsistence. B419 Through the analysis of Melnick’s “ ‘I think’ given as a proposition that already includes existence in itself,” one reaches a conclusion in which ‘I am subject’ is “taken merely logically” and ‘I am simple subject’ leaves unsettled “anything about the subject’s constitution or subsistence.” So, even following Melnick’s recommendation to understand Kant’s positive theory of the experiencing subject as being presented only when we consider the ‘I think’ as already including existence, i.e., nonproblematically, the conclusion that we ought to reach is that all there is to such a representation is a series of analytic, or merely logical, propositions, i.e., a purely formal representation. Despite this fundamental point of disagreement with Melnick, as I indicated earlier, I believe that there is also a good deal that is right about the positive account of the ‘I think’ that Melnick eventually reaches. In particular, that positive account centers on the thesis that the ‘I think’ does not represent any substance but rather a kind of activity. Melnick resists the formalist interpretation of the ‘I think’ because he understands this activity to be one that is temporal and therefore the activity of a phenomenal substance. In support of understanding the ‘I think’ as a kind of activity, Melnick cites the following footnote. Only without any empirical representation, which provides the material for thinking, the act I think would not take place, and the empirical

250

The Inferential Self is only the condition of the application, or use, of the pure intellectual faculty. B423n

Melnick is right to notice that Kant does here take the ‘I think’ to be a kind of activity, but he overlooks and underappreciates its closing idiom. The act ‘I think’ is an application “of the pure intellectual faculty.” Space and time are our forms of sensibility and as such structure the “material for thinking.” Insofar as the ‘I think’ is applied temporally, this is only because outer and inner sense provide it with its material for thinking. The thinking of this material is not itself structured by space and time but by the intellectual faculty alone. Regarding that faculty, the understanding, we have long since been told at this point in the Critique that “the cognition of every, at least human, understanding is a cognition through concepts, not intuitive, but discursive” (A68/B93). Thus, the act ‘I think’ is not, per se, a temporal act at all but a conceptual one.18 Insofar as it is temporal, this is only because time is our form of inner and outer sense, and these provide the material for such acts of thinking. The acts of thinking themselves, though, are purely conceptual. As I have been arguing, conceptual for Kant means inferential. For an act to be inferential will be for that act to be understood in terms of the inferential norms that govern it, and the time has now come to turn to exactly what the inferential norms that govern the representation of the experiencing subject are, and how they are revealed in the analytic propositions that form the basis of the arguments of the Rational Psychologist. That is, I take it that it has now been at least tentatively established that the most prominent nonformalist approaches to the ‘I think,’ and to the Paralogisms specifically, are all untenable as interpretations of how Kant himself understands these. Longuenesse concedes as much about her approach. Ameriks and Melnick both defend their interpretations as delineating Kant’s own take on the ‘I think,’ but as we have now seen, these interpretations are supported almost entirely by unrepresentative selections of Kant’s texts, a more context-sensitive examination of which reveals that Kant does, as he repeatedly says he does, take the ‘I think’ to be a purely formal representation. Having established this much, the task of the next section will be to explain precisely what Kant means by this claim, and to do so within the framework of Kant’s inferentialism that we have been heretofore delineating. My thesis there will be that the three analytic propositions regarding the ‘I think,’ the misconstrual of which are the source of the fallacious arguments of the Rationalist Psychologist, together constitute the content of the representation ‘I think.’ THE ARGUMENTS OF THE RATIONAL PSYCHOLOGIST We can begin with the conclusions that Kant portrays the Rationalist Psychologist as drawing. Collectively, these are that the self is a substance that

The Inferential Self

251

is both indivisible and numerically identical to itself across time. That is, the self is (a) a substance, a basic ontological entity, not an aspect or mode of anything else; (b) simple, a basic ontological entity, not a composite of any other more basic ontological entities; and (c) single, a basic ontological entity, numerically identical across time. As Kant does in his rewriting of the Paralogisms for the second edition of the Critique, we can narrow our focus to just the first of these conclusions and its attendant argument and extrapolate from that to the second two arguments. So, here is the second-edition version of the Rational Psychologist’s argument for the conclusion that the self is a substance. 1. What cannot be thought otherwise than as subject does not exist otherwise than as subject, and is therefore substance. 2. Now a thinking being, considered merely as such, cannot be thought otherwise than as subject. 3. Therefore it also exists only as such a thing, i.e., as substance. (B411) Kant takes this conclusion to be “drawn per Sophisma figurae diction” (B411), i.e., via an equivocation, and explains in a footnote to that remark that the argument specifically equivocates on the meaning of the term ‘thought.’ “Thinking” is taken in an entirely different signification in the two premises: in the major premise, as it applies to an object in general (hence as it may be given in intuition); but in the minor premise only as it subsists in relation to self-consciousness, where, therefore, no object is thought, but only the relation to oneself as subject (as the form of thinking) is represented. B411n Rewriting the argument so that the equivocation is clear, we arrive at the following. 1. What cannot be intuited otherwise than as subject does not exist otherwise than as subject, and is therefore substance. 2. Now a thinking being, considered merely as such, cannot be thought otherwise than as subject. 3. Therefore it also exists only as such a thing, i.e., as substance. Here the argument is obviously fallacious because in Kant’s idiom, ‘intuit’ and ‘thought’ have entirely different meanings. ‘Thought’ is Kant’s most

252

The Inferential Self

general term for any mental representation; a thought that is an intuition, on the other hand, is subject to the Categories: the meta-conceptual rules that determine what counts as a representation of an object. This is a crucial difference here because while we can, according to Kant, think about the self, we cannot intuit it. This is the lesson that Kant takes from Hume: when I enter most intimately into what I call myself, I always stumble on some particular perception or other [. . .] I never catch myself at any time without a perception, and never can observe any thing but the perception. T1.4.6.3–4; SBN 252 Even in inner sense, we do not intuit the self itself. We encounter this or that perception but never the self. In fact, this is something to which the Rationalist Psychologist agrees. This is why he takes it to be necessary to draw the above inference about the nature of the self. It is because we do not intuit the self that our knowledge of the self must be inferred from the single proposition, ‘I think.’ The problem for the Rational Psychologist, then, is that, once he concedes that the self cannot be intuited, when we “consider” the thinking being merely as such in the second premise here, that considering can only be a thinking. A thinking being can never be thought otherwise than as a subject. Of course, the first premise of this argument does not say anything about thought. Its claim is that what cannot be intuited otherwise than as subject is therefore substance. The ‘intuited’ here is key because the only justification for accepting that premise that Kant regards as valid is that the concept ‘substance’ is one of the Categories and is therefore legitimately applicable to, and only to, objects of intuition (as demonstrated in the Transcendental Deduction). It is not true that what cannot be thought otherwise than as a subject is substance because there are thoughts that are not intuitions, and the Category ‘substance’ is not legitimately applicable to them at all. So, Kant accepts the first premise on the grounds that the Category ‘substance’ is applicable to all and only what cannot be intuited otherwise than as subject. He accepts the second premise only insofar as it concerns how a thinking being must be thought. The conclusion, however, that a thinking being is a substance does not follow from these two premises. It is worth noting here that Kant not only accepts the two premises of this argument but also takes each to be analytic. The reason that what cannot be intuited otherwise than as subject does not exist otherwise than as substance is that what intuitions represent is objects, and objects are necessarily represented by such intuitions as substances. So, if something must be intuited, must belong to the subject position of an intuition-concept judgment, then what is represented by that intuition is necessarily represented as a substance. Consequently, because of the demonstrated validity of the

The Inferential Self

253

Categories, what is represented by that intuition must also exist as a substance. So, when understood correctly, as applying to what is represented by an intuition, the first premise is analytic. More importantly for current purposes is to notice that so is the second. Consider the continuation of the footnote from B411. [T]he second premise, however, talks not about things, but about thinking (in that one abstracts from every object), in which the I always serves as subject of consciousness; hence in the conclusion it cannot follow that I cannot exist otherwise than as subject, but rather only that in thinking my existence I can use myself only as the subject of judgment, which is an identical proposition, that discloses absolutely nothing about the manner of my existence. B411n It is analytic (an identical proposition) that “in thinking my existence, I can use myself only as the subject of judgment.” That is, it is part of the content of the representation of the self that the ‘I think’ can never be predicated of anything else but must always be in the subject place of a subject-predicate judgment. Part of the content of this representation is this semantic fact about it. As I will argue later in this chapter, it is this semantic fact along with those expressed in the second and third Paralogisms that together constitute the entire content of the representation of the self. The next piece of business on our agenda, then, is to extrapolate Kant’s critique of the first Paralogism to the second and third. Like the first, each of these also depends on a crucial equivocation that overlooks the distinction between intuition and thought. Thus, the second Paralogism runs in the A-edition as follows. 1. That thing whose action can never be regarded as the concurrence of many acting things is simple. 2. Now the soul, or thinking I, is such a thing. 3. Thus, etc. (A351) Here again the first premise is true only when ‘regarded’ is taken to mean ‘intuited,’ and the second premise is true only insofar as we take the self to be a thing that cannot be thought as the concurrence of many acting things. Thus, the conclusion does not follow. Kant’s criticism of this Paralogism once again reveals important clues about his account of the representation of the self, clues that point toward the reading of that account that we have been advocating. Recall that at the close of the last chapter we noted that the unity of the self that is effected by the unity of the representation of an object is not an ontological but a normative unity.

254

The Inferential Self The proposition, “A thought can be only the effect of the absolute unity of a thinking being” cannot be treated as analytic. For the unity of a thought consisting of many representations is collective, and, as far as mere concepts are concerned, it can be related to the collective unity of the substances cooperating in it (as the movement of a body is the composite movement of all its parts) just as easily as to the absolute unity of the subject. A353

As we noted earlier, ontologically speaking, the self could very well be a composite, and its unified thoughts could very well be spread out over a composite of ontological parts. Again, what is interesting about William James’s famous example is not that a single thought cannot be had by a collection of human bodies; it could, but this requires that those bodies be collectively subject to the proper conceptual norms. As Kant puts it farther along, the unity of the self is a logical unity. But I am simple signifies no more than that this representation I encompasses not the least manifoldness within itself, and that it is an absolute (though merely logical) unity. A355 The Rational Psychologist holds that the self is a substance that cannot be analyzed into more simple substances. The truth corresponding to this unwarranted conclusion is that the representation of the self is a representation that cannot be analyzed into more simple representations. This is what Kant means by ‘logical unity’ here: the representation of the self does not have an internal logic to it that allows it to be analyzed into more simple semantic notions. It cannot, for instance, be analyzed (a la Hume) into a bundle or perceptions nor (a la Locke) as a succession of states each of which remembers all the previous ones. The representation of the self is representationally simple. We can provide no analysis of it.19 In the paragraph following the above Kant continues, Thus the so famous psychological proof is grounded merely on the indivisible unity of a representation, which governs the verb only in regard to a person. But it is obvious that the subject of inherence is designated only transcendentally through the I that is appended to thoughts, without noting the least property of it, or cognizing or knowing anything at all about it. It signifies only a Something in general (a transcendental subject), the representation of which must of course be simple just because one determines nothing at all about it; for certainly nothing can be represented as more simple than that which is represented through the concept of a mere Something. A355, emphasis added

The Inferential Self

255

Kant’s reasoning here is that it is because the content of the representation of the self is exhausted by its role as that which appended to thoughts that it has a logical unity. The representation of the self is a transcendental representation the content of which is described completely by the role we have seen it play as that which is subject to a conceptual-inferential rule. It cannot be analyzed into more semantically simple notions because this role is all there is to this representation. Thus, as in the first Paralogism, the key here is that a certain analytic proposition concerning the representative role of the ‘I think’ is mistaken for a proposition concerning the ontological properties of the substantial self. ‘I am simple’ is an analytic proposition that expresses the logical unity of the ‘I think’ in its role as that which appended to all thought. It expresses the logical unity of the follower of conceptual rules. Nothing, however, concerning the ontological makeup of the self follows from that proposition concerning the semantic properties of the ‘I think.’ The argument of the final Paralogism concerns the numerical identity of the self across time, an issue that we have already encountered in the previous chapter. In the A-edition it runs as follows. 1. What is conscious of the numerical identity of its Self in different times is to that extent a person. 2. Now the soul is etc. 3. Thus it is a person. (A361) Once again the key is to notice that the first premise ought to be understood to make a claim about what is intuited: if we could intuit ourselves as numerically identical through time, we would be numerically identical through time (again because of the objective validity of the Categories). The second premise, however, had Kant written it out, would concern only how we think about ourselves: we think of ourselves as numerically identical through time. Thus, again the conclusion does not follow from the premises. Here again, the second premise when properly understood turns out to be an analytic proposition about the representation of the self. It is part of the content of that representation, given as it is by the role of the ‘I think’ in uniting representations of objects, that this representation be univocal across time. In combining my manifold of sensations into a representation of an object, it is essential that I represent that combination as the activity of a single subject. In order to apply a rule to such a manifold, it must be one and the same rule follower that applies it first to one sensation, then the next. Recall the passage from the A-Deduction on counting that we have already had occasion to examine. If, in counting, I forget that the units that now hover before my senses were successively added to each other by me, then I would not cognize the generation of the multitude through this successive addition of one

256

The Inferential Self to the other [. . .] for this concept consists solely in the consciousness of this unity of the synthesis. A103, emphases added

The representation of the self that it presupposed by following conceptual norms (here the rule for counting) is necessarily a representation that is indivisible across time. In order to represent oneself as following such a rule, one must represent each successive act of rule following as carried out by the same self. Recall that this is the essence of Kant’s strategy in the Deduction. It is because (D2) [The I that thinks x] = [the I that thinks y] = [the I that thinks z] follows from (K) I think [x + y+ z] that we are able to represent ourselves as the single subject of experience persisting through time by forming a single cognition the elements of which are a manifold of sensation. Or, to run the argument in the direction that Kant does there, it is because one must think of oneself as the single subject of experience persisting through time that one is justified in applying the Categories to objects of experience (in order to form the cognition [x + y + z]). We are now in a position to see, at least superficially, why it is that Kant holds that doing the former—thinking of oneself as a single subject of experience persisting through time—is necessary. This is because, as in the previous two Paralogisms, the minor premise here—that we must think of, or represent, the self as numerically identical through time—is analytic. [T]he personality of the soul must be regarded not as inferred but rather as a completely identical proposition of self-consciousness in time, and that is also the cause of its being valid a priori. [. . .] The identity of the consciousness of Myself in different times is therefore only a formal condition of my thoughts and their connection, but it does not prove at all the numerical identity of my subject, in which—despite the logical identity of the I—a change can go on that does not allow it to keep its identity; and this even though all the while the identical-sounding “I” is assigned to it. A362–3, emphasis added Here again we have all of the elements that we have observed in Kant’s explication of the previous two Paralogisms. The minor premise is an identical proposition; it expresses a merely formal condition of the representation of the self; it does not carry with it any ontological consequences; and the

The Inferential Self

257

role of the ‘I think’ is exhausted by its being assigned to various thoughts. What remains to be done, then, is to explain all of these phenomena using Kant’s particular brand of inferentialism, and that is the task of the next section. THE NATURE OF THE REPRESENTATION ‘I THINK’ What we have seen thus far is that the Paralogisms are based on three analytic propositions concerning the nature of the representation of the self. Each expresses a semantic fact about that representation. What I will argue in this section is that taken together these three semantic facts alone constitute that representation. That is, these three propositions are each necessary and are together jointly sufficient conditions on what it is for a representation to be a representation of a self. In arguing for such a claim, it is always important to make explicit one’s criterion of success, and the previous chapter has supplied us with just such a criterion. What we learned from our examination of the Transcendental Deduction is that the unity of the representation of an object is an inferential unity and that the corresponding unity of representation of the self is the unity appropriate to an inferrer. As Kant puts it in his notes, “The I constitutes the substratum for a rule in general” (Ak 17:656; Notes and Fragments, 166). It is this representation of the self as the subject of conceptual-inferential norms for which we must now account. One way to do this is by answering the question: what makes a representation a representation of such a subject? Our answer will be that the three analytic propositions that motivate the Paralogisms are criteria for exactly this. What needs to be demonstrated, then, is that these three propositions are each necessary and are jointly sufficient for generating a representation of a subject of conceptual-inferential norms. To begin, here are those three analytic propositions, translated into a more contemporary idiom. 1. ‘I think’ does not have a predicative use. 2. ‘I think’ cannot be analyzed into any more simple semantic parts. 3. ‘I think’ is univocal with a given subject’s thought. We have already seen that each of these is regarded by Kant as an analytic proposition concerning the representation ‘I think.’ We have also seen that Kant holds a very particular inferentialist account of the nature of representation. What we need to do now is combine these two interpretive insights and examine the inferential account that Kant gives (or would give) of the representation constituted by these three propositions. Prima facie, there seems to be an immediate difficulty with this endeavor. As we have been portraying it thus far, Kant’s inferentialism is a kind of

258

The Inferential Self

picture theory. Objects are represented as complexes of necessarily connected parts by placing representations of these parts into inferential relations with each other. We earlier saw that Hume holds that to represent x as related to y, one places a representation of x into the same relation with a representation of y that x stands in to y. That is, for Hume, xRy is represented by ‘x’R‘y.’ We saw Kant object to Hume’s theory by noticing that using the same relation in the representation as that which is instantiated in what is represented led to crucial ambiguities in such representations. Consequently, Kant improves upon Hume’s theory not by abandoning the notion of thought as a picture but instead by introducing inferential counterpart relations as the structure of mental representation. So, Kant’s inferentialism instantiates the scheme xRy is represented by ‘x’R*‘y.’ What we saw in previous chapters is that, for Kant, the relations that items are represented as bearing to one another by this scheme are first and foremost lawful causal relations, and the counterpart inferential relations that represent these are rules of material inference. The problem, then, with the representation of the self is that the self is not represented as a complex of parts bearing lawful causal relations to each other. There are no xs and ys to be represented and thus no ‘x’s and ‘y’s to connect inferentially. The entire point of the Paralogisms is that the representation of the self is not the representation of any kind of object but is merely a “form of representation in general”: the consciousness in itself is not even a representation distinguishing a particular object but rather a form of representation in general, insofar as it is to be called a cognition; for of it alone can I say that through it I think anything. A346/B404 If, however, the representation of the self is not a representation of an object, and Kant’s inferentialism is a theory according to which concepts-quainferential-rules represent objects, then it is difficult to see how any inferential account of the representation of the self can possibly be given. It would seem that such an account is simply outside the scope of the theory. We can begin to see how this puzzle is solved by taking note of some of the specific differences between representations of objects and the representation of the self. Consider, then, the following quotation from the Metaphysical Deduction.

The Inferential Self

259

So in the judgment, e.g., “All bodies are divisible,” the concept of the divisible is related to various other concepts; among these, however, it is here particularly related to the concept of body, and this in turn is related to certain appearances that come before us. A68/B93 As we noted earlier, this quotation gives an important clue to understanding Kant’s inferentialism as it relates to intuition and material inference: the concept ‘divisible,’ he notes, is related not just to the concept ‘body’ but also to various other concepts. It is an inferential rule that connects the intuition of bodies to other intuitions in other judgments. What we did not note earlier is that the crucial final clause of the quotation in which Kant mentions that the concept ‘body’ is “in turn related to certain appearances that come before us.” This is noteworthy now because it gives a necessary condition for what it is for a concept to be an empirical concept: that concept must ultimately be a rule that is applicable to appearances, i.e., it must be a rule for uniting a manifold of sensations. Empirical concepts apply more or less directly to manifolds of sensation. The inferences that constitute the representation of the self, by contrast, have no such use. Empirical concepts are applied to manifolds of sensations in order to unite these into a single cognition of an empirical object. As we have seen, in order for this to be possible, such applications must be represented as having been carried out by a single rule-following subject. What the rules governing the use of the ‘I think’ do is ensure that this is done. For instance, (3)—the ‘I think’ is univocal with a given subject’s thought— licenses the crucial inference that leads from the fact that I am capable of accompanying each of my representations with an ‘I think’—that certain representations are possible objects of introspection for me—to the conclusion that all of my representations are the representations of a single subject. As we have seen, this is what Kant calls the analytic unity of apperception, and one of his main points in the Deduction is that it depends on the synthetic or transcendental unity of apperception. I.e., (D2) [The I that thinks x] = [the I that thinks y] = [the I that thinks z] depends on (K) I think [x + y + z]. That (D2) depends on (K), however, does not imply that (D2) itself it is not analytic, nor that the inferences constitute its content are not valid in virtue of their logical form. In fact, all three of the propositions that we have extracted from the Paralogisms are analytic and are constituted by such inferences. It is because one can, e.g., substitute ‘the I that thinks x’ for ‘the I that thinks y’ in any judgment (in which ‘x’ and ‘y’ are possible objects of

260

The Inferential Self

introspection for me) that the I that thinks x is identical to the I that thinks y, even if it is (K) that makes this intersubstitutivity possible. Licensing such inferences does not itself unite any manifold of sensations; it does not make such sensations into the representation of an object. Rather, it ensures that one of the conditions for carrying out such a synthesis is met. The key is that it does this by licensing this crucial inference from the possibility of the ‘I think’ accompanying these sensations, to their all being the sensations of a single subject. It is worth pausing for a moment here to be careful to avoid falling into the same trap that the Rational Psychologist does. The Rational Psychologist holds that this conclusion is a conclusion about a particular kind of numerically identical object: the self. Again, Kant’s conclusion is radically different. It is a conclusion about the univocality of a certain representation: the ‘I think.’ Clearly, we cannot cash out the univocality of that representation as “picking out the same object.” Kant’s suggestion about how to understand this univocality is a gesture toward the “identical-sounding I” (A363), but, as above, a more state-of-the-art inferentialism might render it as, for instance, the intersubstitutivity of the ‘I think’ in certain, but not other, judgments. Extrapolating to the other of Kant’s analytic propositions, we can see that (2) is likewise necessary to ensure the unity of the representation ‘I think.’ In particular, it ensures that the univocality of the ‘I think’ is a genuine one, insofar as that representation is not a mere semantic composite, decomposable into, for instance, a conjunction of more simple representations, each of which would accompany only a single element of the manifold. It accomplishes this by licensing the crucial, but again merely logical, inference from (K) I think [x + y + z] to (D2) [The I that thinks x] = [The I that thinks y] = [The I that thinks z]. Were the ‘I think’ semantically complex, this inference would not be valid because it could be the case that the ‘I think’ that accompanies [x + y + z] would decompose into parts, each of which only had the single thoughts, x, y, and z. Finally, (1)—that the ‘I think’ has no predicative use—ensures that in representing the self as the single subject of experience, it represents the right kind of unity. A representation that has a predicative use is one that has a use as a rule for uniting manifolds of sensation. Such representations have their own kind of unity: they represent, for instance, a number of objects as all being red, or elephantine, or parts of a sequence of a single act of counting. However, to represent oneself as the single subject of a manifold of intuitions is not to attribute this kind of unity to oneself (as Leibniz, for

The Inferential Self

261

example, seems to think that it does). One does not first and foremost represent oneself as a single kind with many possible instances.20 Rather, the unity of the self, the unity of the subject of conceptual-qua-inferential-norms is the unity of that which is subject to these norms, and this unity is affected in whatever that is in part by accompanying it with a representation that has no predicative use. While the representation of the self is not the representation of an object as the bearer of multiple properties, it is the representation of a single subject that is the bearer of multiple inferential commitments and is thus logico-grammatically a subject, rather than a predicate. What (1) does is ensure that the self is represented in this way by securing the ‘I think’ its proper in the judgments in which it appears: the ‘I think,’ like an intuition, “cannot be thought otherwise than as subject,” i.e., as a unity of the appropriate kind. For Kant, the most general way of thinking of something as a substance is by assigning this same logico-grammatical feature to it: representing as the subject of a judgment but never the appropriate predicate of one. Kant takes this to be the most general way of representing that a substance is a kind of unity: it is the one in which many properties inhere. The ‘I think’ shares this logico-grammatical feature of substance and is thereby represented as a similar kind of unity: it is the one subject in which many representations inhere. Again, one must be careful to understand this claim in the proper way: ‘inherence’ here must be taken in the most general possible sense. What we would normally recognize as inherence is a property only of what can be intuited as substance, and the experiencing subject is not intuited at all. What inherence amounts to in the case of the ‘I think’ is the fact that the norms that apply to the components of a thought only do so derivatively insofar as those components contribute to the content of what is represented by the experiencing subject. That is, just as properties must be conceived as the properties of some substance, so representations must be conceived as the representations of some experiencing subject. Any ontological multiplicity that may constitute whatever it is that is subject to these norms is incidental to the representation of this subject as a subject. In representing, e.g., a group of people as subject to the norms of counting, we necessarily represent the group as a whole as subject to those norms and only derivatively the members of that group as subject to corresponding individual norms. E.g., it is only because we are counting that I must say ‘three’ after he says ‘two.’ Similarly, it is only because I am counting that I-now must say ‘three’ after I-previously said ‘two.’ The norms that govern inferential commitments apply first and foremost to the single subject of those norms and only derivatively to its parts. Recall Kant’s discussion of this example in the A-Deduction. If, in counting, I forget that the units that now hover before my senses were successively added to each other by me, then I would not cognize the generation of the multitude through this successive addition of one

262

The Inferential Self to the other, and consequently I would not cognize the number; for this concept consists solely in the consciousness of this unity of the synthesis. The word “concept” itself could already lead us to this remark. A103–104

It is essential to understanding each of the acts that together constitute an act of counting as being parts of such an act that I represent them as being added together by me. It is because I am the bearer of the inferential commitments that constitute the concept ‘counting’ that the units that now hover before my senses constitute a representation of a number. The word ‘concept’ itself could already lead us to this remark precisely because a concept is a rule and as such presupposes exactly this kind of unity: the norms that govern these acts are only possible insofar as they are norms that apply first and foremost to the thoughts of a single subject of experience. That the representation ‘I think’ can only ever be the subject of a judgment and never a predicate of one is how this place in the logical structure of concept use is ensured. Thus, each of these three analytic propositions express a rule of inference that is essential to constituting the representation of the self as the single subject of experience persisting through time, or as we have understood this, as the single subject of temporally discursive conceptual-inferential norms. It is by being governed by such inferential norms that one comes to represent oneself as a single subject of experience persisting through time. To unite a manifold of sensation, one must represent oneself as the single subject of such empirical rules. One does this precisely by licensing the inferences expressed in the three analytic propositions that ground the Paralogisms. To return to Kant’s counting example, to represent certain tally marks, say, as being successively produced by me in accordance with the rules of counting, I must accompany each of these tally marks with a representation that has no predicative use, is semantically simple, and is univocal within my own thought. The next important difference to notice between the representation of the self and the representation of an empirical object is that the former is a kind of meta-representation. Now, as we noted in earlier chapters, concepts themselves are all a kind of meta-representation: they represent objects as being necessarily connected to one another by relating intuitions to one another inferentially. Furthermore, since we have also argued that intuitions themselves are conceptually structured—are composed of non-conceptual representations (sensations), which are themselves connected via relations modeled on the inferential ones between intuitions—even these are a kind of meta-representation. To understand the particular kind of metarepresentation the representation of the self is, we have to go a level higher than we have thus far. Concepts represent objects as bearing necessary connections to one another by relating intuitions to one another inferentially. The representation

The Inferential Self

263

of the self makes such representations possible by representing these conceptual activities as being the activities of a single conceptual agent. One counts as an inferential rule-following uniter of manifolds by being subject to the meta-level norms that govern the representation of the self. Given the iterations of meta-representations that we have already seen, one might expect the representation of the self to operate in much the same way that concepts operate at the levels of perception and intuition: as rules for uniting a manifold, now a manifold of representations of representations. One of the deeper points that Kant makes in the Paralogisms is that this is simply not so. A concept functions in a judgment by connecting intuitions, determinate singular representations, to one another inferentially. Intuitions themselves have a kind of proto-judgmental structure. ‘This-such’ bears an important structural relation to ‘This is such.’ So, concepts function in perception in a way roughly analogous to how they function in judgments: by connecting sensations, non-conceptual representations of the parts of empirical objects, to one another in a way modeled on the inferential connections between judgments. Each of these requires something close to subject-predicate, or intuition-concept, form to operate. There is a representation (the subject: intuition, or sensation) that is connected to other representations inferentially (via the concept-qua-inferential-rule). Part of what Kant wants to show in the Paralogisms is that, despite its appearance, ‘I think x’ does not have this form. Both intuitions and sensations are representations of something: determinate singular objects as the necessary connection of their parts and parts of objects (not represented as such, but represented nonetheless), respectively. Concepts operate over such representations by connecting them to one another inferentially. It is no accident, therefore, that the former are appropriate subjects and the latter appropriate predicates. Judgments of subject-predicate form operate by performing two simultaneous tasks: picking something out and saying something about it. The surface grammar of a judgment such as ‘I think x’ might lead one to suppose that it functions by picking out a self, ‘I,’ and saying something of this self, that it thinks x. In fact, though, its logical grammar is quite different. As we have already seen, we do not intuit the self. We do not have sensations that represent parts of the self. What Kant’s text suggests here is that ‘I’ is not a referring term.21 In fact, it is not, in any significant way, a term that picks out anything at all.22 What it does do is assign a certain status to the otherwise disparate representations of what thereby becomes a rule-following subject. It makes these activities collectively subject to the conceptual rules at hand. This is neither to pick out an object nor to attribute a property to that object. In licensing the three inferences above, one does not attribute a property to this thing, the self, but rather one makes the representations involved into the representations of a single subject of experience.

264

The Inferential Self

Insofar as I license the three inferences above regarding my representations, I take those representations as my own. Doing this is not to attribute the property of representing such-and-such to the object, the self. Rather, it is to make these representations subject to the appropriate governing norms. These norms, in turn, are defined not in terms of a single object that must behave in a certain way but rather entirely in terms of the locus of the inferential commitments undertaken.23 Hence of the thinking I (the soul), which [thus represents] itself as substance, simple, numerically identical in all time, and the correlate of all existence from which all other existence must be inferred, one can say not so much that it cognizes itself through the categories, but that it cognizes the categories, and through them all objects, in the absolute unity of apperception, and hence cognizes them through itself. A402 Endorsing the inferences that constitute representing oneself as the single subject of experience is, as J. L. Austin might put it, more like a making a performative utterance than it is like a uttering a constative.24 Still, once such inferential commitments are undertaken, a peculiar kind of meta-representation is formed. Specifically, one’s representations come to be represented as the rule-governed activities of a single subject of experience. We can see the significance of this by noticing again the difference between representations so represented and those that are not. Tally marks on a piece of paper are not a count of anything unless they are made subject to the rules of counting. Making them subject to such rules requires that they be represented as the products of a single act of counting undertaken by a single subject of such norms. Thus, in undertaking the commitments required by such norms, I do thereby represent these tally marks as being produced according to such rules. Kant’s insight in the Paralogisms is to see that this is a peculiar kind of representation but a kind of representation nonetheless. Consider again the three analytic propositions that we are discussing. 1. ‘I think’ does not have a predicative use. 2. ‘I think’ cannot be analyzed into any more simple semantic parts. 3. ‘I think’ is univocal with a given subject’s thought. What we have recently seen is that these three analytic propositions express inferential licenses that together imply that the subject of a single complex representation is also the subject of a manifold of representations. These propositions rule out that such a representation of such a subject is the representation of something that as merely a property of some other disparate object, a collection of some other disparate objects, and not strictly numerically identical to itself across time. What we can see now is that these all

The Inferential Self

265

express inferential commitments that concern one’s own representations. That is, the three analytic propositions above all concern inferences that crucially involve premises of the form ‘I think x,’ and what such a premise claims is simply that a certain representation, ‘x,’ is a representation that itself can be represented via this introspective faculty. Thus, the time has come to say something about Kant’s account of inner sense and its relation to the transcendental unity of apperception. INNER SENSE AND THE TRANSCENDENTAL UNITY OF APPERCEPTION In discussing the relation of inner sense to the transcendental unity of apperception, it will be important to distinguish what we can call two modes of discourse about inner sense. This is because inner sense is our faculty for representing our own representations, and those representations can be represented in either of two very different ways. We have for some time now been examining the nature of the content of our representations, and thus we have been representing our representations in terms of this content. Correspondingly, in one mode of discourse, inner sense consists of representations of our representations as having such-and-such content. It produces a functional classification of such representations as being of such-and-such a representative kind. E.g., ‘I think x’ can be used to contrast thoughts of xs with thoughts of ys. On the other hand, one might form meta-representations that treat such representations in terms of that which is so classified. That is, inner sense is a faculty for representing other representations, and these representations can be represented not just in terms of their content but also in terms of the mental items that have this content. This latter mode of discourse corresponds roughly to treating a chess piece as a hunk of wood with such-and-such a shape, treating it in terms of its naturalistic properties, whereas the former classifies the same piece functionally as being subject to such-and-such rules of chess, in terms of its normative properties. The reason that we must be particularly sensitive to these two modes of discourse is that Kant’s account of inner sense contains treatments of both kinds. Some of what will be relevant to the preceding discussion will concern inner sense qua representing representations as representations, and some of it will concern inner sense qua representing representations as modifications of the state of the subject. We can begin with the basics. Kant holds that we have a faculty for forming representations of “things outside of us” as a result of their affecting our sensory organs: outer sense. Our representative capacities, however, do not end there. We can also be affected by these representations in a way such that we are prompted to form representations of them in turn. This is what Kant calls inner sense, and it is an essential part of the account that he wants

266

The Inferential Self

to give of the self. This is not because he holds that inner sense gives us any special insight into the ontological nature of the self, but rather it is because the representations that are united according to the rules for representing oneself as single subject of experience persisting through time are precisely those that are produced by this faculty. The I think must be able to accompany all my representations; for otherwise something would be represented in me that could not be thought at all, which is as much as to say that the representation would either be impossible or else at least would be nothing for me. That representation that can be given prior to all thinking is called intuition. Thus all manifold of intuition has a necessary relation to the I think in the same subject in which this manifold is to be encountered. B131 After the work of the previous section, we are now in a position to make sense of this infamously difficult passage. The self is constituted by the three inferential norms discussed in the previous section. These norms each begin with a representation that is accompanied by the ‘I think.’ Thus, if there is a representation that could not be accompanied by the ‘I think,’ it would be a representation to which those norms did not apply. If, however, we are right that the self is constituted by these norms, then any representation to which these norms could not be applied “would be nothing for me.” They would be representations that could not be incorporated into the representation of the self that is constituted by such norms. The question before us, then, is this: what is the connection between the representations that can be accompanied by the ‘I think’ and inner sense? The key to answering this question is in recalling the suggestion from the previous section that ‘I think x’ is not, despite its surface grammar, a subjectpredicate judgment. ‘I’ does not pick out any object and ‘thinks x’ does not serve as a rule for uniting a manifold of representations. Rather, the function of ‘I think x’ is as a meta-representation. It represents ‘x’ as a representation. In particular, it represents ‘x’ as a representation of mine. That is, the products of inner sense—representations of my representations as my representations—are precisely ‘I think x’s. As such, it is the products of inner sense that are the premises for the inferences that together constitute the representation of the self. The next thing to notice about these ‘I think x’s is that just as ‘I’ is not the subject of this representation, neither is ‘x.’ That is, while the products of inner sense are a kind of representation, unsurprisingly, they are not the representation of any mental object, e.g., a thought. ‘The thought x’ is a grammatical nominalization of the expression ‘I think x,’ which, while it can be construed as having the surface grammar of a relational proposition, relating I to x, is in fact no such thing. ‘I think x’ functions instead as a kind of quotational device; it brackets the content of some other representation,

The Inferential Self

267

such as that of an object, and represents this content as the content of this representation. Notice that here we are using the mode of discourse that treats the products of inner sense as representations of representations having certain content, rather than as states of the experiencing subject. Given that we began our discussion of inner sense with the relation of the transcendental unity of apperception to the proposition ‘I think x,’ this is unsurprising. That is because, as we have understood it, the transcendental unity of apperception is a representation of the self as a rule-following subject, and the rules that it follows are precisely the rules of inference that determine the content of its representations. As we have seen, the transcendental unity of apperception is constituted by the three inferences that lead from (D1) [I think x] and [I think y] and [I think z] through (K) I think [x + y + z] to (D2) [The I that thinks x] = [The I that thinks y] = [The I that thinks z]. These inferences crucially involve the content of the representations that are accompanied by the ‘I think’ in each case. It is only insofar as one unites a manifold of representations into a single cognition, the [x + y + z] of (K), that one is licensed to infer the analytic unity of the apperception (D2). Thus, it is essential for these inferences that the representations that they take as their premises, such as ‘I think x,’ be representations of the content of the representations that they represent. They do precisely this in their role as quotational devices. Given this understanding of ‘I think x,’ our original question—what and how do the representations of inner sense represent?—is now transformed into the question of what and how a quotational device represents, and this is a question for which the means of providing an answer is clear. A quotational device is a way of representing the content of some other representation, and so if we know what the content to be represented is, we are well on our way to determining the nature of the quotational device. In the case at hand, the content to be represented is, first and foremost, although not only, the representations of outer sense, and we know what the content of these are. They are first and foremost representations of objects as the necessary connection of their parts. They have this content in virtue of being a complex of representations of the parts of such objects connected via material rules of inference. If those are the two sides of such representations, then what the meta-representations of inner sense represent should follow

268

The Inferential Self

directly: they are representations of the inferential connections internal to the representation ‘x.’ To judge ‘I think x’ is to classify ‘x’ functionally. It is to locate ‘x’ in the network of inferential commitments, forbearances, etc., that constitute that representation. That is what the representations of inner sense represent; the second half of our question is how they represent this. The idiom of the closing sentence of the above paragraph offers a clue. As we have seen, to literally locate an item in space and time, for Kant, involves forming a picture of it structured by certain material inferences that rely on the rules for constructing representations of spatial and temporal items. (For instance, rules governing the use of ‘north’ and ‘south’ in the case of space, and the rules for temporal determination that Kant outlines in the Analogies and Refutation of Idealism in the case of time.) To “locate” an item in a network of inferential connections, one would likewise form a kind of picture of that network using rules of inference to connect representations of the items in that network. What, at the object level, is the drawing of an inference, for instance, from ‘x is to the north of y’ to ‘y is to the south of x,’ is at the meta-level, the explicit licensing of such an inference: e.g., ‘x is to the north of y’ implies ‘y is to the south of x.’ Such licenses, of course, themselves stand in inferential relations to other licenses, forbearances, etc., and these all together form a picture of the representations they represent. It is worth noting that one key difference between representations of objects and representations of these representations as representations is that the latter, but not the former, makes use of an explicitly normative idiom. Kant, recall, holds that concepts are rules, and so there is a sense in which this seemingly radical conclusion ought to be unsurprising, at least as an exegesis of Kant’s own position. If concepts are rules, then representations of concepts will naturally be representations of a normative practice. Of course, we are concerned not just with quotational devices in general but more specifically with the representations of inner sense, and as we noted earlier these represent representations not just as representations but also as one’s own representations. Thus, in addition to representing the content of the representation at hand in terms of the inferential commitments, forbearances, etc., that constitute it, ‘I think x’ also attributes such commitments, etc., to oneself. ‘I think x,’ rather than represent a relation

The Inferential Self

269

between two items, I and x, delineates the content of a representation, ‘x,’ in terms of its place in a network of normative inferential relations and attributes the normative statuses that constitute this representation to oneself. These attributions in turn license the further inferences that constitute the representation of the self as the single subject of experience persisting through time as the locus of a manifold of such commitments, etc. Thus, the transition from the necessity of the ‘I think’ being capable of accompanying all of my representations, through inner sense, to the transcendental unity of apperception. This is how the representations that are the product of inner sense function as described in the first mode of discourse that we outlined. They represent one’s representations by bracketing the content of these representations and attributing that content to oneself by serving as premises in the inferences that constitute the representation of oneself. Before our account of inner sense is complete, however, we must also describe it in the second mode of discourse: as a faculty for producing representations of one’s representations, not in terms of their content, or normative properties, but rather in terms of the role that these representations play qua modifications of the state of the experiencing subject. The astute reader will recall that this idiom is not a new one; we encountered it previously as the description that Kant gives of sensations, the non-conceptual representations that are the causal antecedent to, and eventual constituents of, conceptually structured intuitions. As we will see in a moment, this is no coincidence. As we noted at the outset of this section, just as the representations of outer sense are prompted by one’s being affected by objects outside of us, the representations of inner sense are prompted by these representations. That is, the representations of inner sense are conceptual responses to worldly stimuli. Given the account that was presented in Chapter 3 of the role of sensations as the non-conceptual intermediaries between worldly objects and our conceptual responses to them, it would seem appropriate that a complete account of inner sense would likewise require the positing of such representations. Recall that we modeled Kant’s theory of sensation on Sellars’s sense impression inference, which was meant to account for the fact that normal perceivers have conceptual representations of a red and rectangular object both (a) when they are being affected in normal circumstances by a red and rectangular object and (b) when they are being affected in abnormal circumstances by objects that have other, but systematically related, characteristics.25 Similarly, since Kant, unlike Descartes, holds that our judgments about the contents of our own representations are fallible, the representations of inner sense are conceptual representations that are the product of being affected both by the representations that they correctly represent and also those that

270

The Inferential Self

they incorrectly represent “when they are being affected in abnormal circumstances by objects [representations] which have other, but systematically related characteristics.” I.e., the sense-impression inference is valid in the case of inner sense, just as it is in outer sense. Thus, we ought to conclude that just as we have sensations produced by outer objects, we likewise have sensations that are produced by our own mental states, not qua representations, of course, but qua “modifications of the state of the subject.” Thus we find Kant in the B-Deduction addressing the difference between the way in which the transcendental unity of apperception is a representation of the self and the way in which inner sense is. But how the I that I think is to differ from the I that intuits itself (for I can represent other kinds of intuition as at least possible) and yet be identical with the latter as the same subject, how therefore I can say that I as intelligence and thinking subject cognize my self as an object that is thought, insofar as I am also given to myself in intuition, only, like other phenomena, not as I am for the understanding but rather as I appear to myself, this is no more and no less difficult than how I can be an object for myself in general and indeed one of intuition and inner perceptions. B156, emphasis added As we have seen, by ‘object’ here, Kant can only mean object of a representation, since he does not hold that the representation of the self via inner sense is the representation of any ontological object. Analogously, by ‘intuition,’ Kant cannot mean that what inner sense represents is a determinate representation of an object as the necessary connection between its parts. Still, it does seem right that the representations formed by inner sense are similar to intuitions insofar as they are likewise a conceptual response to “inner perceptions,” the self-directed analog of sensations. If this analogy is to be carried through, we should expect these inner perceptions to be united by the understanding according to certain inferential rules into a single representation of a complex state of affairs. The question, then, is what the complex state of affairs is, if it is not an object. The answer is that it is a temporal complex of states of empirical consciousness, or phenomenal states. That is, what is represented by a complex of inner perceptions united according to the relevant inferential rules is simply that which we thereby experience as our temporally extended phenomenal consciousness. Recall Kant’s claim that while space is the form only of outer sense, time is the form of both outer and inner sense. I.e., when we represent objects, we represent them as existing in time and space, but when we represent our own inner states, we represent them only as existing in time. Given, however, that Kant is not willing to take on any commitments about the ontological nature of such phenomenal states, there is little more that he can say about them, or the perceptions that are their causal antecedents

The Inferential Self

271

and eventual constituents, other than that they are states of the subject, or person. In such a way the very same thing that is called a body in one relation would at the same time be a thinking being in another, whose thoughts, of course, we could not intuit, but only their signs in appearance. Thereby the expression that only souls (as a particular species of substances) think would be dropped; and instead it would be said, as usual, that human beings think, i.e., that the same being that as outer appearance is extended is inwardly (in itself) a subject, which is not composite, but is simple and thinks. But without allowing such hypotheses, one can remark generally that if by a “soul” I understand a thinking being in itself, then it is already in itself an unsuitable question to ask whether or not it is of the same species as matter (which is not a thing In itself at all, but only a species of representations in us); for it is already self-evident that a thing in itself is of another nature than the determinations that merely constitute its state. A359–60 What is represented by outer sense are empirical objects, substance, not things as they are in themselves. What is represented by inner sense, even in the mode of discourse in which we consider our representations as states of the subject, is not anything even as ontologically committing as this. Such representations do not represent thoughts as matter, or thinking substance, or any other kind of substance, but only as “the determinations that merely constitute its [the thinking being’s] state.” They are states of the perceiver, now not qua active uniter of manifolds of sensation and subject of conceptual norms but instead as whatever passive kind of thing (I, or He, or It) that this active this is. Inner perceptions and sensations are, in this sense, entirely transcendental: they are explanatory posits known to exist only because they are preconditions for the very possibility of our forming the kinds of representations that we do.

NOTES 1. It is worth noting at the outset that these three inferential norms constitute the most generic representation of the self that is possible in Kant’s system. That is, just as the Categories need to be schematized before they can be used to represent any possible object that can be experienced by creatures like us, similarly, what I will argue here is that the Paralogisms’ three inferential norms constitute a representation of the self in general, but that given the kinds of creatures that we are—specifically our sensory passivity and temporal discursivity—for such norms to actually be undertaken is a more complex matter. So, for example, as we have seen, in the First Analogy Kant attempts to

272

2. 3.

4. 5. 6.

7. 8. 9. 10. 11. 12.

13. 14. 15. 16.

The Inferential Self show that undertaking these norms through time requires that we represent our representations as being of the alterations of a world consisting of a single substance that is never created or destroyed. In the Second Analogy, he attempts to show that undertaking them with respect to the deliverances of inner sense as occurring in a determinate temporal order requires representing our representations as being of a world governed entirely by casual laws. Etc. Kemp Smith, Commentary; Sellars, “. . . this I or he or it (the thing) which thinks . . .”; Rosenberg, “ ‘I think’ ”; Bird, “Paralogisms”; Theil, “Critique of Rational Psychology”; and Rosefeldt, “Kant’s Self.” “In particular, Kant holds throughout to an immaterialist and idealist conception of the self in itself. Although his position is quite nuanced, the fact remains that Kant’s new extra emphasis on the critical limits of selfknowledge still contains a fundamentally rationalistic motive. Although the limits eventually require some retreat from the traditional claims that Kant was still trying to argue for within his earlier critical period, the limitations have the ultimate purpose of preserving a kind of traditional dualistic distinction between the knowable, physical, and merely external world on the one hand, and the transcendent, nonphysical, and ultimately significant self on the other hand.” (Ameriks, 2000: 227) Melnick, Kant’s Theory of the Self, viii. Longuenesse, “Kant on the Identity of Persons,” 159. Kitcher, “Kant’s Paralogisms.” Kitcher explicitly, albeit briefly, engages with the formalist approaches of Kemp Smith and Sellars, charging those authors with omitting an account of why the representation ‘I think’ has the special properties that attribute to it. I hope to remedy that omission here. Kitcher herself, however, goes well beyond sins of omission, to fairly serious ones of commission. Namely, Kitcher accuses Kant of a number of infelicities, including making his argument against a straw man (Kitcher, 1982: 531 and 538), including premises in those arguments claims that he is not “really serious about” (Kitcher, 1982: 526), and using terms in his argument either because he unthinkingly carries over the terminology of his day (Kitcher, 1982: 526) or for the purpose of obfuscation (Kitcher, 1982: 526). Longuenesse, “Kant on the Identity of Persons,” 154. Longuenesse, “Kant on the Identity of Persons,” 159. As Kant goes on to note, if the answer of ‘I think’ is misconstrued as it is by the Rational Psychologist, “my insight (so plausible at the start) into the nature of a thinking being [. . .] becomes suspicious.” (A399) Thus, the contention in Strawson, Bounds of Sense that the ‘I’ refers to an abstraction from the more fundamental notion of a person is also wrong. ‘I’ does not refer at all. Ameriks, Kant’s Theory of Mind, 67. It is helpful to note here that Ameriks holds that “In this case, the items falling under the schematized category are not—as generally is the case—a subclass of the items falling under the pure category.” (Ameriks, Kant’s Theory of Mind, 67) Ameriks, Kant’s Theory of Mind, 69. Ameriks, Kant’s Theory of Mind, 72. Melnick, Kant’s Theory of the Self, 63. This despite passages such as the following in which Kant explicitly denies precisely this claim: “That the I of apperception, consequently in every thought, is a single thing that cannot be resolved into a plurality of subjects, and hence a logically simple subject, lies already in the concept of thinking,

The Inferential Self

17. 18.

19.

20.

21.

273

and is consequently an analytic proposition; but that does not signify that the thinking I is a simple substance, which would be a synthetic proposition.” A407–408, emphasis added. Melnick, Kant’s Theory of the Self, 63. Other passages that Melnick cites in support of understanding the ‘I think’ as a kind of activity can be treated in precisely the same way. See “Thus I cannot determine my existence as that of a self-active being, rather I merely represent the spontaneity of my thought” (B158n, emphasis added). Melnick, Kant’s Theory of the Self, 5. One might worry here that this claim of Kant’s is subject to a simple counterexample. Consider, for example, that ‘I think’ does seem to be composed of ‘I,’ which contrasts with ‘he,’ ‘she,’ etc., and ‘think,’ which contrasts with an even greater variety of terms such as ‘hope,’ or ‘run.’ I will postpone discussing this worry until the next section, at which point we will see that the surface grammar of ‘I think’ is very different from its logical grammar. Specifically, I will there defend the thesis that while ‘I think’ appears to pick out an object, I, and say something about it, that it thinks, this is not the logic of the ‘I think’ at all and, in fact, that it is by being deceived by its surface grammar that is one of the ways in which the Rational Psychologist is led to the Paralogisms. So, e.g., ‘He thinks’ and ‘I run’ do not provide genuine contrasts for ‘I’ and ‘think’ because they are representations that have an entirely different kind of representational function. Recall that we saw Melnick make this point in the previous section. He took it to follow from this that the representation ‘I think’ cannot be purely formal, an inference that I resisted at the time precisely because of the possibility of Kant’s understanding the ‘I think’ as here. Brook, Kant and the Mind, 88 attributes this interpretive thesis to Allison, Rosenberg, Kitcher, and Powell and argues that “the view rests on nothing more than a simple confusion, the confusion of non-ascriptive reference for no reference.” Brook agrees that the ‘I think’ neither picks out an object nor says anything about it: Because ascribing properties distinguishes one thing from another and awareness of oneself as subject does not, using ‘I’ to refer to oneself on the basis of having a global representation could not be tied to ascribing any properties. (Brook, Kant and the Mind, 88) Brook denies, though, that this implies that the ‘I think’ does not refer. He cites Shoemaker: In making a judgment like ‘I feel pain’ one is aware of [no]thing less than the fact that one does, oneself feel pain. (Shoemaker, “Self-Reference and Self-Awareness,” 563) Finally, Brook understands this awareness as follows. Awareness of myself as subject is awareness of myself as nothing more than—myself. (Brook, Kant and the Mind, 88) Thus, what Brook appears to mean to say is that the ‘I think’ is the representation of the single subject of experience persisting through time but that it does not function as such either by picking out an object, the self, or attributing any properties to that object. Rather, the ‘I think’ functions purely formally in the role prescribed to it by the conditions for representing objects (or, more generally, nature) in a single cognition (a ‘global representation’ in

274

22.

23.

24. 25.

The Inferential Self Brook’s idiom). I find nothing objectionable in that thesis and so conclude that in this instance my difference from Brook is a mere quibble over the proper use of the term ‘refer.’ Thus when Kant writes that the transcendental subject is “recognized only through the thoughts that are its predicates” (A346/B404), he is intends to indicate not that thoughts are properties (or modes, etc.) of a substantial self, but rather he is referencing only the surface grammar of judgment of the form ‘I think x,’ in which of course ‘x’ is predicated (in a grammatical sense) of ‘I think.’ And what he is saying is that we do not form a representation of the ‘I think’ by any encounter with a substantial self (transcendent, empirical, or otherwise), but rather by forming a representation with precisely the grammatical-cum-inferential properties of the ‘I think.’ One way of putting this is to say that to represent oneself as the single subject of a manifold of representations is to locate oneself in the so-called logical space of reasons. That is, it is to think of oneself not as an object subject only to causal laws but also as a reasoner, subject to (in a different sense, of course) laws of thought. It is important here to keep in mind, though, that not representing oneself as caused is very different from representing oneself as not-caused. This, I think, is where Pippin, “Kant on the Spontaneity of Mind” goes wrong in interpreting the role of spontaneity in Kant’s account of the self. He takes Kant’s spontaneity thesis to be that the synthetic unity of apperception, and all conceptual activity, must be uncaused. Kant’s actual thesis is that the concept of such activity is not the concept of something caused (it is not, therefore, the concept of something uncaused, though). Austin, How to Do Things with Words. Sellars, Science and Metaphysics, 17.

REFERENCES Ameriks, Karl. Kant’s Theory of Mind: An Analysis of the Paralogisms of Pure Reason. Oxford: Clarendon Press, 2000. Austin, John. How to Do Things with Words. Oxford: Clarendon, 1962. Bird, Graham. “The Paralogisms and Kant’s Account of Psychology.” Kant-Studien 91 (1966): 129–45. Brook, Andrew. Kant and the Mind. Cambridge: Cambridge University Press, 1994. Hume, David. A Treatise of Human Nature. Edited by L. A. Selby-Bigge. New York: Oxford University Press, 1974. Hume, David. A Treatise of Human Nature. Edited by David Fate Norton and Mary J. Norton. New York: Oxford University Press, 2000. Kant, Immanuel. Critique of Pure Reason. Translated by Paul Guyer and Allen W. Wood. Cambridge: Cambridge University Press, 1998. Kemp Smith, Norman. A Commentary to Kant’s Critique of Pure Reason. Atlantic Highlands: Humanities Press, 1991. Kitcher, Patricia. “Kant’s Paralogisms.” Philosophical Review 91 (1982): 515–47. Longuenesse, Beatrice. “Kant on the Identity of Persons.” Proceedings of the Aristotelian Society 107 (2007): 149–67. Melnick, Arthur. Kant’s Theory of the Self. New York: Routledge, 2009. Pippin, Robert. “Kant on the Spontaneity of Mind.” Canadian Journal of Philosophy 17 (1987): 449–76. Rosefeldt, Tobias. “Kant’s Self: Real Entity and Logical Identity.” In Strawson and Kant, edited by Hans-Johann Glock, 141–54. Oxford: Oxford University Press, 2003.

The Inferential Self

275

Rosenberg, Jay. “ ‘I think’: Reflections on Kant’s ‘Paralogisms.’ ” In Midwest Studies in Philosophy, Vol. X: Philosophy of Mind, edited by Peter French et al., 496–523. Minneapolis: University of Minnesota Press, 1986. Sellars, Wilfrid. Science and Metaphysics. Atascadero, CA: Ridgeview Publishing Company, 1967. Sellars, Wilfrid. “. . . this I or he or it (the thing) which thinks . . .” Proceedings of the American Philosophical Association 44 (1972): 5–31. Shoemaker, Sydney. “Self-Reference and Self-Awareness.” Journal of Philosophy 65 (1968): 555–67. Strawson, Peter. The Bounds of Sense. London: Methuen Ltd., 1966. Theil, Udo. “The Critique of Rational Psychology.” In A Companion to Kant, edited by Graham Bird, 207–21. Malden, MA: Blackwell Publishing Ltd., 2006.

Postscript on Transcendental Idealism

In Chapter 3 I argued that sensations are transcendental-explanatory posits. In Chapters 4 and 5 I argued that Kant’s picture theory of mental representation and its accompanying account of the representations of the external world and necessary connection imply a robust kind of scientific realism about the entities of theoretical science. Each of these is a conclusion with a distinctly ontological component, and the time has now come to explain how I understand that component in each case. Such an explanation is necessary because Kant notoriously endorses both a certain kind of ontological realism, Empirical Realism, and what he takes to be a necessary complement to that realism, a certain kind of ontological anti-realism, Transcendental Idealism. As is usual with Kant, there has been a great deal of controversy over both the interpretive issue of what precisely Transcendental Idealism is and the philosophical issue of whether or not the thesis of Transcendental Idealism is at all plausible. Far from settling either of these debates, my modest goal here will be to delineate how I understand the two ontological theses above within the context of Kant’s Transcendental Idealism, touching on both the exegetical and the philosophical issues only by way of this more parochial discussion. Specifically, what I will argue here is that both of these conclusions are properly located within the scope of Kant’s Transcendental Idealism, i.e., that both concern ontological commitments regarding appearances, rather than noumena. In doing so, I will depend on a particular understanding of the notion of a noumenal object and a corresponding interpretation of Transcendental Idealism. Throughout my study to this point, I have been understanding ‘object’ as that which is represented by a cognition (a representation of a complex as complex). A natural, but I believe ultimately misguided, contrast to ‘object,’ indicated by Kant’s idiom, ‘the thing in itself’ (das Ding an sich), would be the notion of an object not as what is represented by some representation but rather as it is apart from all representation and its conditions. For Kant, this is an entirely empty notion of which little use can be made. His conception of a noumenal object, by contrast, is the notion of an object as that which is represented by a form of representation other than ours. While such a notion is not terribly more helpful than

Postscript on Transcendental Idealism

277

the one just considered, Kant insists that it is not an entirely empty one and does serve at least one transcendental philosophical purpose: it provides the contrast necessary to formulate and understand the thesis of Transcendental Idealism. Recently, much of the literature on Transcendental Idealism has concerned itself with the question of whether Transcendental Idealism is an epistemological or a metaphysical thesis: whether it is a claim about what we can or cannot know about noumenal objects, or a claim about what noumenal objects must or must not be. Unsurprisingly, I hold that both of these approaches miss the mark. Transcendental Idealism is primarily a thesis about what can and cannot be represented. Objects of representation are necessarily subject to the conditions through which we represent them, although a creature unlike us, e.g., an intuiti intellectuali, might have a different form of representation and therefore would represent “the same” object very differently. The scare quotes here signal that the question of in what sense the same object would be represented by two such different creatures is a delicate one that must be treated with some care, especially given the thesis that the notion of an object apart from all forms of representation is an empty one. All of this, however, will be addressed in its proper context farther along. Before turning to that business, however, it will be helpful to have a clearer sense of what the order of operations is to be. As I said, my primary goal here is to locate the two ontological theses established in earlier chapters—regarding substance and sensations— within Kant’s Transcendental Idealism, as theses regarding the nature of appearances. Here, once again, I believe that it will be instructive to draw on Sellars’s engagement with Kant, and once again I will find a great deal with which to agree in Sellars but also a crucial point of disagreement. In Science and Metaphysics, Sellars presents a series of arguments intended to show that Kant’s defense of Transcendental Idealism is a confused jumble of arguments and theses, only one version of which can be salvaged. Sellars argues that Kant ought to have been a Transcendental Idealist about perceptible material objects but a Transcendental Realist about the objects of theoretical science. Sellars will prove a particularly useful foil here because his arguments begin with the thesis just presented that the notion of a noumenon is that of an object as represented by a form of intuition or set of concepts other than ours. He then goes on to argue that the final representative framework of theoretical science will employ both concepts that are fundamentally different than ours and forms of intuition other than Space or Time. Each of the two parts of this claim—regarding concepts and intuitions, respectively—just so happens to engage each of the two ontological theses with which I am concerned—that regarding substance and causal laws in the Analogies, and that regarding the transcendental-explanatory positing of sensations. So, my procedure in locating those claims with respect to Kant’s Transcendental Idealism will be to defend the claim that each of these theses concern only phenomena, rather than noumena against

278

Postscript on Transcendental Idealism

Sellars’s arguments for Transcendental Realism. That defense will take the following form. In the first section, I will consider the thesis of Transcendental Idealism itself and review some of the recent literature on its interpretation. In the second section, I will turn to Sellars’s first set of arguments for Transcendental Realism: those concerning the ontology of theoretical science. In the third and final section, I will consider Sellars’s arguments for the transcendental reality of the posits, not of theoretical science but of theoretical philosophy. TRANSCENDENTAL IDEALISM By way of introduction to the topic of Transcendental Idealism, we can consider Karl Ameriks’s helpful categorization and discussion of the major interpretive positions of that thesis.1 He begins with two versions of what he calls epistemic interpretations. According to the first, which he attributes to Bird, Nagel, and Walker, among others, transcendental idealism is a thesis about that which can or cannot be subsumed under certain concepts. That which is not ideal (in the supposedly Kantian transcendental sense) would have to be something that transcends the concepts of all our possible theories.2 The first epistemic interpretation understands the notion of a noumenon as that which could not be subsumed under any possible concept.3 By contrast, the second epistemic interpretation, which Ameriks attributes primarily to Allison, takes Transcendental Idealism to be a thesis concerning what can be represented by our particular forms of intuition. But there is another type of “epistemic” interpretation, one which is unlike all these in so far as it understands the transcendentally real rather as that which would transcend our specific cognitive faculties. In particular, given Kant’s fundamental doctrine that for us objective cognition requires conception joined with intuition in warranted judgment, this view takes Kant’s transcendental idealism to arise not from the nature of concepts or theories as such but rather from specific features connected with the nature of our kind of intuition.4 According to this version of Transcendental Idealism, that which is transcendentally real is that which transcends not our ability to conceptualize but rather our ability to intuit. Specifically, this version of Transcendental Idealism holds that since Space and Time are our forms of intuition and not properties of noumena, representing objects as they are in themselves would require employing a form of intuition other than the ones that we necessarily do.

Postscript on Transcendental Idealism

279

Thus far, we have seen two versions of epistemic transcendental idealism: that we cannot know things in themselves because they transcend all of our possible concepts of them and that we cannot know things in themselves because they transcend all of our possible intuitions of them. Each of these interpretations is epistemic insofar as they understand Kant as advocating for limits on what can be known about transcendental reality while remaining agnostic with respect to what the properties of such reality are. Ameriks draws a contrast between these versions of Transcendental Idealism and Paul Guyer’s metaphysical interpretation by presenting an objection to the second epistemic interpretation. The epistemic interpretation, in understanding transcendental idealism as the claim that human knowledge is governed by certain sensible conditions, does not insist on Kant’s own stronger conclusion, which is that there are objects which in themselves have genuine ultimate properties that do not conform to those conditions.5 The metaphysical interpretation is one that does commit to this stronger conclusion: that noumena have properties that are other than those that they are represented as having when represented by our forms of intuition and that they do not have these latter properties (i.e., are not spatial or temporal).6 The trick for those defending this interpretation of Transcendental Idealism is to make this thesis consistent with Kant’s repeated claim that we can have no knowledge of things in themselves.7 As I mentioned in the introduction to this postscript, I will advocate a third interpretation of Transcendental Idealism, which will share important features of both the epistemic and metaphysical interpretations, but which begins with a different understanding of the key notion of a noumenal object. What I will call the representational interpretation takes the thesis of Transcendental Idealism to be that that which we represent using object-concepts is necessarily subject to the forms that such representations will necessarily take for creatures like us. That much should be relatively uncontroversial. What sets the representational interpretation apart is the additional thesis that the only contentful contrast to the forms of representation of creatures like us is the forms of representation (intuitional or conceptual) of creatures unlike us. More particularly, the notion of a noumenal object as considered apart from any form of representation whatsoever has no content and is therefore not the notion that we employ in wondering about the properties, relations, or statuses of noumenal objects. This thesis has important consequences for how the representational interpretation resembles and differs from both the epistemic and metaphysical interpretations. Here are a few: (a) The epistemic interpretation is that we cannot know what properties noumenal objects might have and so should remain agnostic regarding whether they will have, e.g., either spatio-temporal properties or

280

Postscript on Transcendental Idealism

categorical ones. According to the representational interpretation, since the only contentful contrast with our forms of representation is that of a form of representation other than ours, it trivially follows that noumenal objects will not be represented by such creatures as having either spatio-temporal or categorical form.8 (b) Thus, the representational interpretation shares with the metaphysical interpretation that noumenal objects will have properties other than the ones that we represent them as having, but this thesis becomes trivial and its defense takes a different form from that in the metaphysical interpretation. (c) Finally, the representational interpretation differs from the metaphysical interpretation insofar as it takes as entirely empty the notion that these are properties that things have “in themselves,” if this is understood as implying that these are properties had by objects independent of all representation. Even these three points, however, obscure how fundamentally different the representational interpretation is from either the epistemological or metaphysical ones. Notice that on the first two points above, the agreements and disagreements are generated from recasting the notion of a noumenal object from that of an object apart from all possible representation of it to that of an object as represented by some form of representation other than our own. While there are, therefore, ways of putting these theses such that agreement can be reached on certain very general points, the differences that lurk in the details are significant. Here is one way to bring those differences out. According to the representational interpretation that I am proposing, to ask whether the properties that we attribute to the objects of our representation are “really” properties of those things can be taken in only two ways.9 Taken in the first way, the question is an empirical one that is answered by applying the conceptual and intuitional norms that govern our use of our current conceptual scheme. Taken in the second way, the question can only concern the adequacy of that scheme itself and is not answered by appealing to the properties that objects have apart from any representation of them but rather by comparing the adequacy of our current conceptual scheme with that of a proposed successor scheme. All of this machinery is familiar enough to anyone acquainted with the debates in the philosophy of science that occupied much of the previous century, and I argued in Chapter 5 that Kant has a valuable contribution to make to that debate. The implications for understanding Kant’s Transcendental Idealism in this way, though, are surprising. On this line, both the epistemological and metaphysical interpretations turn out to be forms of transcendental realism. Both mistakenly suppose that there is an answer to the question of whether the properties that we attribute to the objects of our representation are “really” properties of those things that is independent of either of the questions presented above: the epistemological

Postscript on Transcendental Idealism

281

interpretation is that Kant holds that the answer to this question cannot be known, the metaphysical interpretation that the answer is no. As I understand him, Kant holds that the conception of a noumenal object on which this question rests is entirely empty, and so the question itself has no answer, even in principle. While we can trace the limits of our representational capacities from the inside, so to speak, the thesis of transcendental idealism is that the only contentful contrast to be made with those capacities is with capacities other than our own (or between our current capacities and our future ones). So, transcendental idealism is neither an epistemological thesis about what can be known about objects apart from any possible representation of them nor a metaphysical thesis about the properties of such objects. Both of these interpretations presuppose a notion of an object apart from all possible representation of it to which we simply cannot give content and so amount to a form of transcendental realism about such objects. Having said that, it is worth noting that there is at least one prominent advocate of a representational interpretation of Transcendental Idealism that takes it to imply a kind of a transcendental realism. Sellars argues not only that this interpretation of transcendental idealism leads to a kind of transcendental realism but also that that thesis is one to which Kant ought to have committed himself. That is, Sellars argues that it was only a certain impoverishment of resources that kept Kant a transcendental idealist and that a proper philosophical evaluation of the issue yields representational transcendental realism as the only tenable position. The remainder of this postscript will be a defense of representational transcendental idealism against Sellars’s arguments. THE REALITY OF THE OBJECTS OF THEORETICAL SCIENCE Sellars’s argument against Kant’s account of transcendental idealism and in favor of his own brand of transcendental realism unfolds over the course of the first five chapters of Science and Metaphysics and begins with two relatively straightforward objections. It is after presenting these two objections that Sellars first articulates his own position: that the perceptual world is transcendentally ideal but that the world as represented by theoretical science is transcendentally real. I will present Sellars’s two initial objections fairly quickly and without much commentary so as to clear the way to the hard core of the disagreement that I want to articulate between Sellars and Kant as I have presented him. The first of objection that Sellars raises rests on a distinction that he draws between “fine-grained” or “theoretical” Space consisting of a three-dimensional continuum of points and containing and infinity of lines, surfaces and regions, the chief characteristic of which, for our purposes, is that they are the willing subject-matter for

282

Postscript on Transcendental Idealism the mathematical techniques of Archimedes, Descartes and the inventors of the differential calculus10

and “course-grained” or “empirical” Space that is an individual the elements of which are possibilities—roughly, possible relations of perceptible material things.11 Sellars’s charge against Kant is that he mistakenly inferred the transcendental ideality of empirical Space— and, correspondingly, the transcendental ideality of perceptible physical things—from the transcendental ideality of their ideal counterparts.12 While Sellars is willing to grant to Kant that mathematical objects and the Space in which they are represented are transcendentally ideal—the infinitesimal continuum of points and the Space in which they exist that are the subject of mathematical theory are mere idealizations in the appropriate sense—his charge is that this Space is different from the Space of perceptible objects and therefore that one cannot infer the transcendental ideality of the latter from that of the former. While mathematical objects might be idealizations, perceptible objects are not. We can, for present purposes, grant as much. In his second objection to Kant’s arguments for transcendental idealism Sellars concedes the transcendental ideality of empirical Space but then objects to the inference from the ideality of empirical Space to the ideality of objects in such Space. He is willing to grant the former on the grounds that Kant’s view is that modality is not an object-level property of things but rather a meta-conceptual property of representables. As such, possibilities are transcendentally ideal: a possibility is also, although in a sense different from that of mathematical objects, a mere idealization. So if empirical Space consists of the possible relations of perceptible material things, then it too must be transcendentally ideal. Sellars notes, though, that while Space might be transcendentally ideal because it consists of possibilities, the objects or states of affairs that occupy such a Space do not themselves consist of such possibilities, but rather these objects realize those possibilities. Thus, such objects or states of affairs are not possibilities but actualities, and thus it does not follow that they too are transcendentally ideal. Again, what is transcendentally ideal is an idealization of a kind of a perceptible object. [I]t does not follow from the transcendental ideality of empirical Space as a system of possible relations that particular states of affairs involving spatial relations are transcendentally ideal. This is the Achilles’ heel of his argument.13

Postscript on Transcendental Idealism

283

Space consists of possibilities, so it is transcendentally ideal, but objects in Space do not consist of possibilities, and so their transcendental ideality cannot be inferred from that of Space. Again, we can grant as much for present purposes. It is at this point that Sellars reveals the outline of his own argument for the transcendental ideality of perceptible material objects and, by contrast, the transcendental reality of the objects of theoretical science. I shall tip my hand by saying that the true ground for the transcendental ideality of the perceptual world lies in the distinction between perceptible physical objects and the objects of theoretical science, a distinction which was blurred by Kant. Thus, his concept of physical appearance runs together not only the idealized counterparts of perceptible things (e.g. systems of point-masses whose velocities and accelerations are amenable to differential equations) but also the object of micro-physics which are as imperceptible as ideal objects, though for radically different reasons.14 As Sellars sees it, Kant runs together three different objects of representation— perceptible material objects, the ideal objects of mathematics, and the objects of theoretical science—and it is on the basis of this confusion that Kant concludes that all three objects of representation are transcendentally ideal. As Sellars sees it, Kant is two for three, but in only one case are his conclusions warranted by his premises. The ideal objects of mathematics are transcendentally ideal for the reasons that Kant gives. Perceptible material objects are transcendentally ideal, not for the reasons that Kant gives but rather, as we will see in a moment, because the conceptual structure by means of which they are represented is properly replaceable by the conceptual structure of theoretical science. Finally, Sellars will conclude, the objects of theoretical science are, in fact, transcendentally real. Before turning to that conclusion, it is worth pausing to point out that insofar as Kant does blur the distinction between perceptible physical objects and the objects of theoretical science, it is not for lack of attention to the issue. That is, Kant does not overlook this distinction but rather understands this distinction as being merely methodological. The postulate for cognizing the actuality of things requires perception, thus sensation of which one is conscious—not immediate perception of the object itself the existence of which is to be cognized, but still its connection with some actual perception in accordance with the analogies of experience, which exhibit all real connection in an experience in general. [. . .] Thus we cognize the existence of a magnetic matter penetrating all bodies from the perception of attracted iron filings, although an immediate perception of this matter is impossible for us given the

284

Postscript on Transcendental Idealism constitution of our organs. For in accordance with the laws of sensibility and the context of our perceptions we could also happen upon the immediate empirical intuition of it in an experience if our senses, the crudeness of which does not affect the form of possible experience in general, were finer. Thus wherever perception and whatever is appended to it in accordance with empirical laws reaches, there too reaches our cognition of the existence of things. A225/B272–A226/B273 Thus, according to the Critique, everything in an appearance is itself still appearance, however far the understanding may resolve it into its parts and demonstrate the actuality of parts which are no longer clearly perceptible to the senses; according to Mr. Eberhard, however, they then immediately cease to be appearances and are the thing itself. Ak 8:210; Theoretical Philosophy, 302

In the first passage, Kant clearly commits himself to the thesis that the objects of theoretical science have the same ontological status as perceptible material objects. It is just after this passage that Kant begins the Refutation of Idealism, which is in part meant to demonstrate exactly this: that the existence of objects in space (including those of theoretical science) is neither doubtful nor impossible. In the second passage, Kant not only reiterates this commitment but also demonstrates, in using that thesis as a bludgeon against Eberhard, that his commitment is no mere passing fancy or afterthought but one that he takes to be very important. He also there articulates at least one quick argument for this thesis and against understanding the objects of theoretical science as transcendentally real: roughly, that the difference between a perceptible material object and an object of theoretical science is a difference in degree that is represented by the transcendental realist as a difference in kind. The objects of theoretical science may be beyond our (current) perceptible capacities, but this is, according to Kant’s thinking here, merely because they are, e.g., smaller than what we can perceive. Differences in size, however, do not make some but not other objects transcendentally ideal, and thus if we gain knowledge of objects beyond our perceptual capacities as theoretical science allows us to do, this is not to gain knowledge of what is transcendentally real but only to extend the scope of our knowledge of appearances. As we will see farther along, Kant holds that all such objects will be subsumed under the Categories and necessarily represented in Space and Time. Kant’s attention to this issue so noted, we can pick up the thread that we were following in Science and Metaphysics almost one hundred pages later, after a great deal of preparatory stage setting, where Sellars returns to his conclusion: that the objects of theoretical science are transcendentally real. Here, then, is his final articulation of it.

Postscript on Transcendental Idealism

285

The claim that the common-sense framework is transcendentally ideal, i.e. that there really are no such things as the objects of which it speaks, can now be reassessed and reformulated. We must distinguish carefully between saying that these objects do not really exist and saying that these objects do not really exist as conceived in this framework. For they do really exist as conceived in what, omitting the qualifications which were introduced in the preceding section, we have called the Peirceian framework, the framework which is the regulative ideal which defines our concepts of ideal truth and reality. Sellars, 1967: 148 Perceptible material objects are transcendentally ideal because they do not exist as they are represented in the conceptual framework of common sense. They do, however, exist as they are represented by the conceptual framework used by the theoretical scientists at the Peirceian end of inquiry. To take a toy example, in the framework of common sense, I represent the table on which my computer is now sitting as solid and colored. In the Peirceian framework, the table is still represented, but it is represented as neither solid nor colored. As Sellars sees it, the table exists in both frameworks, but is only represented as what it really is in the Peirceian one. Earning the right to this conclusion takes up much of the difficult and elaborate formal work that occurs in the interim between Sellars’s first tip-of-the-hand and this later return to this conclusion. We can leave much of this aside, however, because while the question of whether what is represented by the framework of common sense can be said to be the same as what is represented by the Peirceian framework is an important one, as we will see, this is a point on which Sellars and Kant are in broad agreement. Where they differ is in how they understand what is represented by the Peirceian framework itself: Sellars takes the objects that it represents to be transcendentally real, whereas Kant takes those objects to be mere appearances and so transcendentally ideal. As Sellars sees it, his is not a conclusion that Kant himself could reach because the only resource that Kant has for articulating the notion of the in-itself is the idea of God, or of an intuiti intellectuali (A249/B305). Yet, according to the picture I have been sketching, the concepts in terms of which the objects of the manifest image are identified have ‘successor’ concepts in the scientific image, and, correspondingly, the individual concepts of the manifest image have counterparts in the scientific image which, however different in logical structure, can legitimately be regarded as their ‘successor.’ In this sense, which is not available to Kant, save with a theological twist, the objects of the manifest image do really exist.15 In addition to having Sellars’s articulation of transcendental realism in full dress here, we also have his assessment of why it is that Kant himself could

286

Postscript on Transcendental Idealism

not reach the same conclusion. Kant, according to Sellars, could only articulate the notion of a noumenon using a particular theological twist, whereas Sellars is able to take advantage of the additional apparatus of a succession of conceptual schemes, each of which more adequately pictures the world than does its successor. It will be worth our investigating the lead provided by Sellars—of Kant’s use of the theological twist—before returning to the claim that Sellars has more resources available to him than Kant does to properly articulate the notion of a noumenon and to his conclusion that he, but not Kant, can properly see that the objects of theoretical science are, in fact, things-in-themselves. We can begin this investigation by noticing that Kant approaches the notion of a noumenon by way of contrast to the concept of phenomenon. A phenomenon is that which is represented by an empirical intuition (and that is therefore subject to the conditions of the possibility of such representation: subsumption under the Categories and structured by our forms of intuition, Space and Time). The concept of a noumenon, by contrast, is the concept of that which is represented by some other form of representation. Now from this arises the concept of a noumenon, which, however, is not at all positive and does not signify a determinate cognition of any sort of thing, but rather only the thinking of something in general, in which I abstract from all form of sensible intuition. But in order for a noumenon to signify a true object, to be distinguished from all phenomena, it is not enough that I liberate my thoughts from all conditions of sensible intuition, but I must in addition have ground to assume another kind of intuition than this sensible one, under which such an object could be given; for otherwise my thought is empty, even though free from contradiction. A252 As Kant makes clear here, the concept of a noumenon is not the notion of an object independent of all representation and its conditions. Rather, it is the notion of an object as thought via a form of representation that is not our own, that is neither subsumed under the Categories nor represented as in Space or Time. While there is no contradiction in thinking of a noumenon as a via negativa, as that which is represented not by our form of intuition, the more complete notion of a noumenon requires specifying an alternative form of intuition by which such an object is thought. It is in service of that end that Kant goes on to introduce the idea of an intellectual intuition. The concept of a noumenon, i.e., of a thing that is not to be thought of as an object of the senses but rather as a thing in itself (solely through a pure understanding) A254/B310

Postscript on Transcendental Idealism

287

If the complaints that ‘we have no insight whatsoever into the intrinsic nature of things’ are supposed to mean that we cannot grasp by pure understanding what the things which appear to us may be in themselves, they are completely unreasonable and stupid. They want us to be able to be acquainted with things without senses, consequently they would have it that we have a faculty of cognition entirely distinct from the human. A277/B233 The full-dress concept of a noumenon is, as Sellars suggests, one that Kant takes essentially to rely on the idea of an object that is represented through pure understanding alone. Before continuing on to their point of disagreement, it is worth noting that Kant and Sellars share a commitment to the emptiness of the idea that we can think of an object in any sense other than as that which is the object of a representation. As we saw earlier, ‘object’ is not a first-order metaphysical concept that picks out a certain kind of entity, but instead it is a meta-level concept that serves to distinguish objectconcepts from other concepts. (It is the Categories that provide the criteria for what it is to be such an object-concept.) Furthermore, though, not only is the concept of a noumenon not a first-order concept of a particular kind of metaphysical entity, it is also not the concept of an object distinct from any form of representation whatsoever. As Kant and Sellars see it, neither of these notions has any content. The concept of a noumenon can only be the concept of that which is represented by some form of representation other than ours. What Sellars takes Kant to task for is thinking that the only determinate contrast to “our” form of intuition, and so the only way to make sense of the concept of a noumenon, is with God’s form of intuition: a nonsensible intuition of pure understanding. Sellars’s own proposal is that the proper contrast with “our” form of representation is the form of representation of an idealized version of us: the scientist at the end of inquiry. So, Sellars is at least right that Kant, in fact, explicates the notion of a noumenon with a theological twist. The question now is whether he is also right that this is because Kant does not have available to him the apparatus provided by the succession of increasingly adequate conceptual schemes and that that apparatus properly leads to the conclusion that the objects of theoretical science are transcendentally real. Here it is finally possible to tie Sellars’s notion of scientific realism back to one of the ontological claims that I have attributed to Kant. In Chapter 5 we saw Kant employ precisely the apparatus that Sellars leverages. That is, we saw that Kant takes the succession of increasingly adequate conceptual schemes, produced by taking a maximally adequate conceptual scheme as a regulative ideal, to correspond to an increasingly accurate representation of the real. As a reminder of these three points, we can reconsider the three most relevant passages from the Critique that we quoted there. The first is a

288

Postscript on Transcendental Idealism

portrayal of the role played by the successive conceptual schemes employed by theoretical science. Accordingly, this idea [that of the form of a whole of cognition] postulates complete unity of the understanding’s cognition, through which this cognition comes to be not merely a contingent aggregate but a system interconnected in accordance with necessary laws. One cannot properly say that this idea is the concept of an object, but only that of the thoroughgoing unity of these concepts, insofar as the idea serves the understanding as a rule. Such concepts of reason are not created by nature, rather we question nature according to these ideas, and we take our cognition to be defective as long as it is not adequate to them. Admittedly, it is hard to find pure earth, pure water, pure air, etc. Nevertheless, concepts of them are required (though as far as their complete purity is concerned, have their origin only in reason) in order appropriately to determine the share that each of these natural causes has in appearance; thus one reduces all materials to earths (mere weight, as it were), to salts and combustibles (as force), and finally to water and air as vehicles (machines, as it were, by means of which the aforementioned operate), in order to explain the chemical effects of materials in accordance with the idea of a mechanism. For even though it is not actually expressed this way, it is still very easy to discover the influence of reason on the classifications of students of nature. A645/B673–A646/B674 While we do not perceive pure earth, pure water, pure air, etc., “concepts of them are required [. . .] in order appropriately to determine the share that each of these natural causes has in appearance.” That is, the conceptual scheme of theoretical science necessarily replaces that of common sense because the former is more explanatorily adequate. The opening sentence of this passage also indicates that this succession proceeds in accord with the regulative ideal of the “complete unity of the understanding’s cognition.” In the passage below, in which Kant contrasts the mathematical principles that are the schematized versions of the Categories of Quantity and Quality with the principles articulated in the Analogies, we can see that it is these latter Categories that provide at least some of the criteria according to which such a regulative ideal is constructed. Things must be entirely different with those principles that are to bring the existence of appearances under rules a priori. For, since this existence cannot be constructed, these principles can concern only the relation of existence, and can yield nothing but merely regulative principles. A179/B221

Postscript on Transcendental Idealism

289

The concepts of Relation—Substance, Dependence, and Community— unlike those under of Quantity and Quality, are not rules by which a concept comes to be a concept of an object but rather are concepts that serve as regulative principles for the adequacy of such object-concepts. Notice that in both this passage and the one before what is pictured with increasing accuracy by the succession of increasingly adequate conceptual schemes is appearance. In the First Analogy, Kant indicates that this is the form that his empirical realism (which he takes to be the necessary complement of his transcendental idealism) will take. Consequently that which persists, in relation to which alone all temporal relations of appearances can be determined, is substance in the appearance, i.e., the real in the appearance, which as the substratum of all change always remains the same. A181/B225 The replacement of the conceptual framework of common sense by that of theoretical science in accord with the regulative ideal of the complete unity of the understanding’s cognition represents the real in appearance. It represents that which is subsumed under the Categories, in this passage the Category of Substance, but elsewhere those of Causation and Community as well. The conceptual framework of theoretical science, for Kant, represents that which is empirically, not transcendentally, real. So, it appears that, contra Sellars’s claim, it was not a lack of the relevant resources that forced Kant to articulate the notion of a noumenon with a theological twist. Not only does Kant have these resources—the apparatus of a succession of increasingly adequate conceptual schemes, produced by taking a maximally adequate conceptual scheme as a regulative ideal, which corresponds to an increasingly accurate representation of the real—but he uses them to articulate precisely the opposite notion that Sellars does: not that of the in-itself but rather that of the real in appearances. Whereas Sellars holds that what is represented by the ultimate conceptual scheme, that of the scientists at the ideal Peirceian end of inquiry, is noumenal, when Kant extrapolates his own account of theory succession to such a distance, he explicitly reaffirms his commitment to that which would be represented by such a conceptual scheme being phenomenal. Observation and analysis of the appearances penetrate into what is inner in nature, and one cannot know how far this will go in time. Those transcendental questions, however, that go beyond nature, we will never be able to answer, even if all of nature is revealed to us, since it is never given us to observe our own mind with any other intuition than that of our inner sense. A277/B233

290

Postscript on Transcendental Idealism

No matter how far we may extend our knowledge of nature—the real in appearance, substance—even to the very end of inquiry, what we thereby represent is only what is transcendentally ideal and not anything transcendentally real. What Sellars needs, then, is an argument that while Kant might be able to match his resources with respect to the progress of science, he cannot do so while still casting such progress as concerning only appearances. Sellars does seem to have such an argument at hand. As we have seen, Kant explicates the notion of the in-itself in terms of the objects of a divine form of intuition: a nonsensible intuition of pure understanding, or an intuiti intellectuali. Sellars also casts them in terms of a form of sensibility not our own: the objects of theoretical science are not represented, in the Peirceian framework, as existing in Space and Time but rather in a “duration that is not a time” and with a “presence” other than that in space.16 These are, in fact, Kant’s own words—Sellars finds the seeds of both of these notions in the Critique but believes that Kant did not make proper use of them—and it will be instructive to examine the context in which Kant uses them more closely. Here, then, is the first of the two passages that Sellars cites.17 Thus if one assumes an object of a non-sensible intuition as given, one can certainly represent it through all of the predicates that already lie in the presupposition that nothing belonging to sensible intuition pertains to it: thus it is not extended, or in space, that its duration is not a time, that no alteration (sequence of determinations in time) is to be encountered in it, etc. But it is not yet a genuine cognition if I indicate what the intuition of the object is not, without being able to say what is then contained in it; for then I have not represented the possibility of an object for my pure concept of the understanding at all, since I cannot give any intuition that would correspond to it, but could only say that ours is not valid for it. But what is most important here is that not even a single category could be applied to such a thing, e.g., the concept of a substance, i.e., that of something that could exist as a subject but never as a mere predicate; for I would not even know whether there could be anything that corresponded to this determination of thought if empirical intuition did not give me the case for its application. B149 Sellars’s thought about this passage is that it indicates that Kant sensed, but did not pursue thoroughly enough, that while the in-itself could not be temporal, it could nonetheless have a structure that would be similar to that of time, and that we could come to represent this structure by way of drawing an analogy between time and it. (Sellars follows Bergson in calling this structure duree; Rosenberg extrapolates this idiom to include the analog of Space, espace.) As this passage indicates, insofar as Kant could formulate the notions of duree and espace, he took them to be the same sort of mere

Postscript on Transcendental Idealism

291

via negativa that we encountered earlier: the notions of nothing more than that which is represented by a form of intuition that is not our own. If Sellars is right that such notions can be imbued with genuine content by way of analogy, then that would certainly curtail the objection that Kant raises above: it is not yet a genuine cognition if I indicate what the intuition of the object is not, without being able to say what is then contained in it; for then I have not represented the possibility of an object for my pure concept of the understanding at all, since I cannot give any intuition that would correspond to it, but could only say that ours is not valid for it. Kant’s point here is that without specifying anything about the form of intuition by which a noumenon is represented, such a representation cannot so much as purport to have a determinate object. If Sellars is right, however, that we can give some content to these forms of intuition, then Kant would have to give this objection up. Furthermore, it would likewise seem that Kant would have to concede the corollary that he draws from this objection. But what is most important here is that not even a single category could be applied to such a thing [. . .] for I would not even know whether there could be anything that corresponded to this determination of thought if empirical intuition did not give me the case for its application. It is because Kant holds that only empirical intuition can “give me the case for the application of the Categories,” that he also holds that the Categories cannot be applied to noumena. It is in representing a determinate object by way of intuition that the Categories are applied, so if the only form of intuition of which we can make sense is empirical (putting aside for present purposes pure intuition), then the only occasion for the application of the Categories is an empirical intuition. If Sellars is right, however, that we can employ a form of intuition that is non-empirical (in the sense that it is an analogical extension of empirical intuition, not that it does not involve receptivity at all), then there is no reason to think that such a form of intuition would not also provide an occasion for the application of the Categories, i.e., that the concepts used to represent such objects could not themselves be object-concepts. If Sellars is right, then, that we can give content to a form of intuition other than our own, and other than God’s, then he will have succeeded in undermining Kant’s arguments here that there is no possibility of cognizing a noumenon. There are two important points to note about the force of Sellars’s arguments. The first is that, as we saw earlier, what we will represent in cognizing such a noumenon is not an object apart from all representation but rather an object that is represented by a form of intuition other than our own. So, Sellars’s argument for transcendental realism

292

Postscript on Transcendental Idealism

aims at a more modest conclusion than one might have expected and in that sense is in line with Kant’s own surprisingly permissive conception of noumena. The second important point to notice about the force of Sellars’s argument is that, if it is good, what Sellars’s argument demonstrates is only that it is possible to cognize a noumenon. It is possible that the Peirceian framework employed by the scientists at the end of inquiry will make use of forms of intuition other than our own, espace and duree, understood by way of an analogy with Space and Time. What Sellars has not yet offered, though, is a reason for thinking that the Peirceian framework will actually make use of such forms of intuition. In particular, in light of the fact that Kant’s arguments in the Analogies seem to locate the objects of theoretical science among appearances, Sellars will need to show either that these arguments are unsound or that the objects for which Kant accounts in the Analogies will not be the sole objects of the Peirceian framework. Nowhere in Science and Metaphysics does Sellars undertake to demonstrate the former. He does present considerations of the second type, although not ones stemming from the progress of theoretical science. Rather, what demands the adoption of transcendental realism for Sellars are the posits of theoretical philosophy. Sellars hints that this will be at least one of the roads that he will take to transcendental realism early on. There are, of course, many who would say that it is the business of science to introduce hypothetical entities, and therefore not the business of philosophers to do so. The pragmatically useful division of intellectual labor, reflected in the proliferation of academic departments and disciplines, has been responsible for many necessary evils, but none more pernicious than this idea. Philosophy may perhaps be the chaste muse of clarity, but it is also the mother of hypothesis. Clarity is not to be confused with insight. It is the latter which is the true final cause of philosophy.18 Since Sellars holds that it is the explanatory force of a theoretical posit that warrants realism about its objects, the insight that is the true final cause of philosophy is likewise what warrants realism about that which it posits. Most importantly for current purposes is that it is on such philosophical grounds that Sellars himself makes what he calls the sense-impression inference, which we saw in Chapter 3 is a descendant of Kant’s own transcendental posit of sensations. So, it seems that we have traced the real grounds for Sellars’s transcendental realism not to the theoretical entities of natural science, which Kant has the resources to cast as entirely phenomenal, but rather to the theoretical entities of transcendental philosophy. It will be Sellars’s arguments for the transcendental reality of those entities that will be the subject of the next section.

Postscript on Transcendental Idealism

293

THE REALITY OF SENSATIONS In the context of adjudicating the debate between Sellars and Kant regarding the transcendental ontological status of sensations, we can stipulate from the start that sensations are posited on broadly explanatory grounds. (This, of course, is not something that could be merely stipulated in many other contexts!) We can also stipulate on Kant’s behalf Sellars’s thesis that in order to explain our conceptual responses to objects existing in space, we must suppose that sensations have a structure that is analogous to that of Space: a σ-structure. Likewise, we can stipulate that sensations must have a structure that is analogous to that of Time: a τ-structure. The latter two stipulations might seem by themselves to award the prize to Sellars because if we cede that sensations have σ- and τ-structures, then it seems that we must likewise have ceded the conclusion that the Peirceian framework will employ a form of intuition other than Space and Time, namely, it will employ σ and τ as its forms of intuition: it will represent its objects as being σ and τ related. Such a conclusion, however, is far too hasty. In order to earn that conclusion, Sellars will have to show not only that sensations will have σ- and τ-structures, but also either (a) that sensations’ having σ- and τ-structures is incompatible with their also having spatio-temporal structure or (b) that some essential role played by Space and Time in our current conceptual framework can only be played by σ- and τ-structures in the Peirceian framework. I will pursue each of these strategies in turn on Sellars’s behalf, beginning with (a). This first line of argumentation would aim to move from the fact, recently conceded, that the Peirceian framework will represent the world as containing sensations with σ- and τ-structures to the conclusion that such a framework cannot employ Space and Time as its forms of intuition. This move would be facilitated by a supposed incompatibility of sensations’ having σ- and τ-structures with their also having spatio-temporal structure. Sellars would have to rely on such an incompatibility because if objects that are represented as σ and τ related can also be represented in the very same conceptual framework as being related spatially and temporally, then σ and τ relations will not supplant spatial and temporal ones but complement them. Consider an example from the framework of common sense. Objects represented by that framework are represented not only as related spatially and temporally but also as standing in a multitude of other relations to one another, e.g., better and worse, hotter and colder, sweeter and more sour. That we represent commonsense objects as standing in such relations does not entail that the framework of common sense employs forms of intuition

294

Postscript on Transcendental Idealism

other than space and time. So, the question is whether sensations’ having σand τ-structures is incompatible with their also being spatially or temporally structured. If having σ- and τ-structure is compatible with being spatially or temporally structured, then having σ- and τ-structure does not by itself imply that sensations are transcendentally real. What we are in search of is an argument for the conclusion that the σand τ-structures that Sellars argues that sensations must have are incompatible with sensations’ also having spatial and temporal structures. We can begin that investigation with Sellars’s introduction of the sense-impression inference. Recall from Chapter 3 that this is an inference to the best explanation meant to explain why it is that we respond with similar conceptual representations to, e.g., both red objects and white objects in red light. The explanation is that sensations are the non-conceptual causal intermediaries between worldly stimuli and our conceptual responses to these and that they have a structure that is analogous, but not identical, to the structure of the colors to which we respond. The first point to notice from Sellars’s text is that for the sense-impression inference to do the explanatory work required of it, it must be the case that . . . all the possible ways in which conceptual representations of colour and shape can resemble and differ correspond to ways in which their immediate non-conceptual occasions, which must surely be construed as states of the perceiver, can resemble and differ.19 If sensations are to do the explanatory work that is required of them by the sense-impression inference, they will have to have a structure that is rich enough to explain all of our various actual and possible conceptual responses to worldly objects. So, for example, in the case of our conceptual representations of colors, sensations will have to have a structure that is analogous to the structure of colors. Thus, these non-conceptual states must have characteristics which, without being colors, are sufficiently analogous to colors to enable these states to play this guiding role.20 Sensations must have characteristics that are not colors but that are sufficiently analogous to colors to explain why it is that we respond conceptually to colored worldly items as we do. Notice that that alone does not rule out sensations’ also having characteristics that are colors. If sensations were colored, their colors would not be what does the important explanatory work that grounds the sense-impression inference, but that they have characteristics that allow them to serve that function does not rule out their also having color characteristics. (E.g., if sensations turned out to be, in some suitable sense, brain states, and if brains did not just so happen to turn out to be collections of colorless atoms.) What the sense-impression inference

Postscript on Transcendental Idealism

295

does is license the positive conclusion that sensations must have characteristics analogous to colors; it does not also license the negative conclusion that they do not have color characteristics. Consider an analogy. What explains how it is that a map represents the terrain that it does is that the map has characteristics that, without being identical to the characteristics of the terrain, are sufficiently analogous to these to play this representative role. So, for example, the distances on a map are sufficiently like (proportional to) the distances of the terrain without being identical to these distances. The map’s having these characteristics, though, does not rule out its sharing other characteristics with the terrain as well. E.g., the map might be made of dirt, or be green, or be spatial. The point is that an argument that shows that sensations must have a structure analogous, but not identical, to color structure does not by itself show that they cannot also be color structured. It is this latter conclusion, however, that we are in search of in support of Sellars’s transcendental realism. To continue, notice that the point that Sellars makes about the color-like structure of sensations above generalizes. As we have already seen, in the case of our conceptual representations of objects as existing in space and time, sensations will similarly have to have σ-and τ-structures. As the earlier quotation indicates, this more general principle is that “all the possible ways in which conceptual representations [. . .] can resemble and differ correspond to ways in which their immediate non-conceptual occasions [. . .] can resemble and differ.” Not only will sensations have structures analogous to the structure of colors, the structure of space, and the structure of time, they will also have structures that correspond to all of the possible ways in which conceptual representations can resemble and differ. So, as we concluded in Chapter 3, the structures of sensations will instantiate or be modeled on the structure of conceptual representations. The important takeaway from this in the present context is that sensations will have to stand in lots of relations to each other, be structured in lots of different ways, and these structures will presumably each be orthogonal to one another. That is, that sensations are structured in one way, e.g., in a way analogous to color, does not imply that they cannot also be structured in another, e.g., in a way analogous to space. So, Sellars is open to at least a very general version of the point just made: merely showing that sensations are structured in one way does not thereby rule out their also being structured in some other way. Sensations will necessarily be structured in many ways. What Sellars needs, then, is an argument to show that of all the structures that sensations must have, spatial and temporal structures are not among these. Unfortunately, though, what we find instead is Sellars just a short while after first presenting the sense-impression inference, and without further argument, summarizing his findings as follows. I shall begin with the contrary to fact assumption that Kant was clear about the radical difference between sense impressions proper and

296

Postscript on Transcendental Idealism the intuitions synthesized by the productive imagination. Such a Kant would then have distinguished between: (a) the non-conceptual representations of outer sense proper which, although conveniently described as impressions of spatial complexes, are strictly speaking non-spatial complexes of unextended and uncoloured impressions; (b) the intuitive (but conceptual) representations of extended structures located in space.21

Here Sellars is emphasizing the difference in what is represented by sensations and intuitions, but he does so in a way that makes clear where he finds Kant’s thinking about sensations to go wrong. Sensations are “nonspatial complexes of unextended and uncoloured impressions.” So, Sellars does seem to conclude from the sense-impression inference alone not just that sensations do have a structure that is analogous but not identical to spatial structure but also that they do not have spatial structure. As Sellars continues to explain the above distinction, though, we can once again see that what he has argued for does not support this stronger negative conclusion. On the one hand, there would be the intuitive (but conceptual) representations of Space (and Time) which serve as frameworks for the conceptual representation [. . .] of individual objects and events. On the other hand, there would be the attributes and relations between the impressions of pure receptivity. Though, as has been pointed out, we conceive of certain of these attributes and relations as counterparts of spatial attributes and relations proper, they would not literally be the spatial attributes and relations in terms of which we conceptually represent physical objects and events.22 Sellars’s point here that the attributes and relations that sensations must have to play their explanatory role—color-like structure, σ- and τ-structures, etc.—are not identical to the relations of worldly objects can be granted, but it does not follow from this, as Sellars seems to take it to, that the sensations that have these counterpart structures might not also themselves be colored spatio-temporal objects. That is, sensations might well turn out to be a certain subset of the objects and events that we conceptually represent as existing in space and time. If, however, sensations were to turn out to be such objects, then while they would, in the Peirceian framework, be represented as having σ- and τ-structures, they would also be represented as having spatial and temporal structures, and we would not need to understand that framework as utilizing forms of intuition other than our own. By way of analogy, consider again Kant’s own example of chemical theory that we encountered in the

Postscript on Transcendental Idealism

297

previous section. When the framework of commonsense physical objects is replaced by that of chemical theory, we thereby suppose there to exist a substance with a structure very different from the one of common sense. The relations of tables to chairs, for example, are replaced with the relations of elements held together by chemical bonds. Thus, part of the advance that the framework of chemical theory makes over the framework of common sense is to represent the world as having a kind of structure that is absent in the framework of common sense, a chemical structure. Merely representing the world as having this new kind of structure, however, does not by itself entail that in order to represent such a structure, we must use a form of intuition other than space or time. One can represent the entities posited by chemical theory as themselves existing in Space and Time even though such entities will bear other interesting (chemical) relations to one another. Not every conceptual change requires a change in form of intuition. Furthermore, there does not seem to be any additional reason to think that the conceptual framework that requires us to posit sensations as the causal intermediary between worldly objects and our conceptual responses to these will require an entirely new form of intuition of its own. So, the first line of argument that we supposed Sellars could take to Transcendental Idealism does not appear to bear any dialectical fruit. There simply is not anything about the idea of our sensations having σ- and τ-structures that is incompatible with their also having a spatio-temporal structure. Thus we can now turn to the he second line of argument: the suggestion that some essential role played by Space and Time in our current conceptual framework can only be played by σ- and τ-structures in the Peirceian framework. What Sellars will need to demonstrate is not just that the objects represented in the Peirceian framework are represented as having σ- and τ-structures but also that such structures necessarily come to replace the structures of Space and Time as the form of intuition employed in such a framework. On this line of argument, what Sellars will have to show is that there is some essential function played by Space and Time that can only be played by σ- and τ-structures in the Peirceian framework. Since my focus to this point has been on Kant’s theory of concepts and its role in his theory of mental representation, I have not previously had the occasion to give much attention to what it means that Space and Time are our forms of intuition, but that is precisely what is now called for, at least in a very general way. To do this, it will be helpful to have an example with which to work.23 Imagine, then, that the universe just so happens to exactly repeat itself every trillion years or so. There is a Big Bang; historical events unfold as they have up until the present; at some point in the future the universe collapses in on itself; the cycle starts anew without a single variation. Suppose further that this cycle has been going on for an infinite time and will perpetuate for an infinite time. Were this the case, no merely conceptual description of, say, the clock-radio on my desk would suffice to individuate that object

298

Postscript on Transcendental Idealism

uniquely. This is because whatever conceptual description one could give of that clock-radio would fit the clock-radio not only in this universe cycle but also all of those from past and future cycles as well. We can therefore make the difference between similar and equal but nonetheless incongruent things [. . .] intelligible through no concept alone, but only through the relation to right-hand and left-hand, which refers immediately to intuition. Ak 4:286, Theoretical Philosophy, 82 What Kant notices here is that despite the failure of purely a conceptual description to represent a particular object such as my clock-radio, we can represent that particular radio nonetheless, and we do so by employing our forms of intuition: Space and Time. We do so, that is, by making use of an essentially indexical representation of it, e.g., as the clock-radio that stands in such-and-such a spatio-temporal relation to me.24 What we have just noticed, then, is that we cannot identify this clockradio purely conceptually. In order to represent it as distinct from other, qualitatively identical clock-radios, there must be something non-conceptual to such a representation. What plays that non-conceptual role is our representation of the clock-radio as occupying a particular, indexically located place in space and time. I have, however, previously argued that the representations of space and time, and of objects located in space and time, are conceptually structured. So, what needs to be done now is to locate where the non-conceptuality of such a representation lies. To do that, we need to delve a little more deeply into Kant’s explanation of how we form such representations of objects as existing in spatial and temporal relations to us. For that, we will now have to consider my representation of my clock-radio in greater detail. We can begin by returning to our discussion of the threefold synthesis in Chapter 3. Recall that we noted there that while such representations are of objects, one does not see all of an object at once. One sees only the facing surface of it. Nonetheless, one does not represent it merely as the facing surface of a clock-radio but rather as a complete clock-radio (of which one is seeing only the facing surface). So, my representation of the clock-radio is of an object existing in space and time of which I see the facing surface but which also has other surfaces that I do not see (as well as insides that I do not see, electronic parts that I do not understand, LED crystals that change with time, etc.) So, my representation of my clock-radio is of that clockradio as having parts, some of which I actually see, and some of which I do not see but imagine.25 The important thing to notice about the formation of such representations for current purposes is that they are not only indexical but also perspectival. That is, the representation of the clock-radio, united according to the concept ‘clock-radio,’ is first made possible by taking the representation

Postscript on Transcendental Idealism

299

of the facing surface of the clock-radio as a representation of a clock-radio seen from a certain point of view by an experiencing subject. These objects are surely not representations of things as they in themselves, and as the pure understanding would cognize them, rather, they are sensory intuitions, i.e., appearances, whose possibility rests on the relation of certain things, unknown in themselves, to something else, namely our sensibility. Ak 4:286; Theoretical Philosophy, 82 In representing the clock-radio, I represent it as it appears to me via my forms of sensibility: as standing in such-and-such a relation to me in Space and Time. This is the sense in which space and time are our forms of intuition. It is by taking our sensations as representing parts of objects existing in space and time seen from the point of view of the experiencing subject that we come to represent not just any clock-radio but this clock-radio in front of me here and now.26 Returning to Sellars, then, what he will need to show is that in the Peirceian conceptual framework at the end of inquiry, it is not Space and Time that we use as the means by which we demonstratively and indexically locate the proper objects of our conceptual representations, but rather it will be σ and τ that serve this function. There is good reason, however, to think that this thesis is not one that we ought to accept. While it is obviously true that we can (and do) coordinate our various perspectives and so also represent such objects as existing in a Space and Time that is nonperspectival and distinct from the representational powers of any particular individual, it is Kant’s contention that this latter representational function is made possible by the original one. If σ and τ (or espace and duree) are to replace Space and Time as our forms of intuition, it would seem that they too must have as their primary use being the means by which we (or the users of the Peirceian framework) demonstratively locate objects from our (or their) individual perspectives. I.e., we must be able to unite our manifold of representations into a single complex representation of a complex object as complex existing in σ and persisting in τ. For example, I would have to be able to synthesize the sensations that are the constituents of the representation of the Peirceian analog of my clock-radio into a representation of a clock-radio (or the Peirceian analog thereof) standing in such-and-such σ and τ relations to me. What is odd about this is that Sellars conceives of σ and τ as analogical extensions of coordinatized Space and Time, not of perspectival Space and Time. So, at least methodologically, the first use to which σ and τ are put is as a system of nonperspectival coordinatized analogs of Space and Time, not as means for locating objects relative to a perceiver. Now, Sellars does argue that the methodological priority of commonsense concepts to those of theoretical science does not imply that the former have an epistemological priority over the latter. He also argues that,

300

Postscript on Transcendental Idealism

in fact, because theoretical science has greater explanatory force than does the framework of common sense, it is the Peirceian framework that most accurately represents the natural world. We have seen that Kant agrees with both of these theses. None of this, however, shows that σ and τ (or espace and duree) can or should replace Space and Time as our forms of intuition. What is true of concepts here is not necessarily also true of forms of intuition, and in fact what I want to suggest is that the methodological priority of Space and Time as our forms of intuition is precisely what prevents their ever being replaced in toto by any other form of intuition. Simply put, if σ and τ depend for their content on coordinatized Space and Time, and coordinatized Space and Time depend for their content on perspectival Space and Time, then Space and Time are ineliminable elements of all of our representations. It will be instructive here to draw a contrast with the sense in which the conceptual framework of common sense is methodologically but not epistemologically prior to that of theoretical science. Methodologically, the concepts of theoretical science are analogical extensions of the concepts of the commonsense framework, but once this analogical derivation is complete, the conceptual framework of theoretical science forms its own picture of the natural world. The elements of this picture are intuitions such as ‘this electron’ and the structure of the picture is the inferential structure of the conceptual framework itself. Sellars and Kant agree that there is no in-principle reason for thinking that this picture cannot serve its representative function without being parasitic on the commonsense picture of the world. For either of these pictures to serve this function, however, the intuitions that are their elements must represent determinate singular objects, and, as we have seen, it is perspectival Space and Time that makes possible such representations. So, while our manifold of representations might best be represented as having σ- and τ-structures, and while the objects of theoretical science might best be represented as existing in espace and persisting in duree, picturing such items as having such structures is itself made possible by employing perspectival Space and Time as our forms of intuition. Analogically formed concepts can replace their predecessor concepts because they come to form a picture the world that is independent of that formed by these predecessors. No successor form of intuition is possible, however, precisely because the concepts that structure any such form of intuition will necessarily presuppose the very relation to objects that Space and Time provide.27 If correct, this line of thinking demonstrates that creatures such as ourselves will never be in a position to employ any form of intuition other than Space and Time as the means by which we demonstratively locate objects relative to and from the perspective of the experiencing subject. The previous line of thinking that we pursued, if correct, showed that Sellars does not offer a compelling reason for thinking that representing our sensations as having σ- and τ-structures is incompatible with also representing them

Postscript on Transcendental Idealism

301

as having spatial and temporal structures. Thus, we have now eliminated the two remaining lines of argument open to Sellars for the conclusion that σ and τ manifolds are the transcendentally real structures of noumenal objects. Combined with the investigation of the previous section that showed that Kant uses the same resources that Sellars does to account for changes of conceptual frameworks but locates these changes within the realm of appearances, not the noumenal realm, we have now exhausted the prospects of finding a convincing argument for Transcendental Realism in Science and Metaphysics. That, in turn, casts the brand of scientific realism that I have advocated for on Kant’s behalf as a brand of his own declared Empirical Realism. As Kant conceives it, the explanatory force of theoretical science entails that the conceptual framework that it employs is the most adequate picture of the world that creatures like us can employ. This picture has as its elements intuitions of the objects posited by such science and as its structure the inferential rules that govern our thinking of such objects. Such pictures are the means by which we synthesize our manifold of representations into a single cognition necessarily had by a single experiencing subject, which in turn is the means by which we conceive the necessary formal identity of such a subject. Thus, Kant’s Transcendental Idealism and Scientific Empirical Realism are both essential parts of the radically new inferential theory of mental representation that he proposes and that earn him back the representations that we had seemed to have lost to Hume: of necessary connection, of the external world, and of the self.

NOTES 1. For those interested in the reasons counting in favor of each of the epistemic and metaphysical interpretations, Marshall, “Kant’s Appearances” also offers an excellent review of the now most current state of the debate. He, of course, ultimately takes the arguments against the epistemic interpretation to be decisive. I do not. I will leave it to the reader to render his or her own decision. 2. Ameriks, “Kant’s Idealism Today,” 333. 3. Watkins, “Kant’s Transcendental Idealism” investigates whether Kant makes an argument that the Categories cannot be applied to noumena and cannot locate any such argument in the Critique. 4. Ameriks, “Kant’s Idealism Today,” 334. 5. Ameriks, “Kant’s Idealism Today,” 334. 6. So, for example, both Langton, Kantian Humility and more recently Marshall, “Kant’s Appearances” are variants of the metaphysical interpretation. Langton holds that the properties of appearances are all relational properties and are therefore distinct from the intrinsic properties of things in themselves. (There is no overlap between the properties that things have in themselves and the properties that things have as appearances.) Marshall, by contrast, holds that the identity of an object is determined by which of its properties are essential to it and that things in themselves and appearances might have all of their properties in common, but not essentially so, and so are different objects.

302

Postscript on Transcendental Idealism

7. Langton attempts to do so by drawing a distinction between our knowledge that things in themselves have some intrinsic properties and our knowledge of what these properties are. Langton, Kantian Humility. 8. Unless we can make sense of a creature that uses additional forms of representation that are nonetheless compatible with ours. This is a topic that will arise in our discussion of Sellars’s arguments regarding alternative forms of intuition. 9. This is reminiscent of Carnap, “Empiricism, Semantics, and Ontology.” As I see it, Kant would agree with Carnap’s way of framing these two questions, but there are also a number of points in Carnap to which Kant would object, the most relevant here being how to answer questions about the comparative adequacy of distinct conceptual schemes (Chapter 5) but also regarding the structure of such schemes (Chapter 4) and their essentially representative function (Chapters 3 and 4). 10. Sellars, Science and Metaphysics, 53. 11. Sellars, Science and Metaphysics, 53. 12. Sellars, Science and Metaphysics, 54. 13. Sellars, Science and Metaphysics, 55. 14. Sellars, Science and Metaphysics, 56n. 15. Sellars, Science and Metaphysics, 150. 16. Sellars, Science and Metaphysics, 38. 17. The second is below. Since it makes the same points about space that the first passage makes about time, we need not consider it independently. Thus we are not allowed to think up any sort of new original forces, e.g., an understanding that is capable of intuiting its object without sense or an attractive force without any contact, or a new kind of substance, e.g., one which would be present in space without impenetrability; consequently we also cannot conceive of any community of substances that would be different from anything that experience provides; no presence except in space, no duration except merely in time. (A770/B798–A771/B799) 18. 19. 20. 21. 22. 23.

Sellars, Science and Metaphysics, 12. Sellars, Science and Metaphysics, 18. Sellars, Science and Metaphysics, 18. Sellars, Science and Metaphysics, 28. Sellars, Science and Metaphysics, 29. This example is essentially a temporal version of Strawson’s mirror chessboard universe, Strawson, 1959: 123, or Black’s bronze sphere universe, Black, “Identity of Indiscernables,” and it is, of course, also related to Kant’s own example of incongruent counterparts Ak 4:285; Theoretical Philosophy, 80. 24. This is one of the manifestations of Kant’s thesis of the essential co-dependence of representations of a single, unified self persisting through time and of the world as a network of causally related substances. 25. Recall that providing representations of the parts of spatio-temporal objects that we do not see but that nonetheless get synthesized into our representations of such objects is the role that Kant assigns to the imagination at B151. 26. This is what it means to schematize our concepts. The unschematized concept of a clock-radio is a rule of inference that represents a clock-radio, which of course is not itself perspectival. It will include inferences such as, e.g., from ‘All clock-radios use electricity’ to ‘All clock-radios are electrocution hazards.’ The schematized concept of a clock-radio will include inferences not included in the unschematized concept such as, ‘This screen is the

Postscript on Transcendental Idealism

303

facing surface of a clock-radio’ to ‘If I turn this around, I will see the back of a clock-radio.’ 27. Thus Kant is right that there is an analogy between the necessity of the Categories and that of our forms of intuition. The supreme principle of the possibility of all intuition in relation to sensibility was, according to the Transcendental Aesthetic, that all the manifold of sensibility stand under the formal conditions of space and time. The supreme principle of all intuition in relation to the understanding is that all the manifold of intuition stand under conditions of the original synthetic unity of apperception. (B136) Just as we have seen that the original synthetic unity of apperception provides one essential condition for representing objects, so Space and Time as our forms of intuition provide another. While it might be the case that there could be other forms of intuition, we can never employ those because we would have no way to gain access to them in the way that we gain access to new conceptual frameworks.

REFERENCES Ameriks, Karl. “Kantian Idealism Today.” History of Philosophy Quarterly 9 (1992): 329–42. Black, Max. ‘The Identity of Indiscernables.’ Mind 61 (1952): 153–61. Carnap, Rudolf. “Empiricism, Semantics, and Ontology.” Revue Internationale de Philosophie 4 (1950): 20–40. Kant, Immanuel. Critique of Pure Reason. Translated by Paul Guyer and Allen W. Wood. Cambridge: Cambridge University Press, 1998. Kant, Immanuel. Theoretical Philosophy After 1781. Edited by Henry Allison and Peter Heath. Translated by Gary Hatfield, Michael Friedman, Henry Allison, and Peter Heath. Cambridge: Cambridge University Press, 2002. Landy, David. “What Incongruent Counterparts Show.” European Journal of Philosophy 21 (2013): 507–24. Langton, Rae. Kantian Humility: Our Ignorance of Things in Themselves. Oxford: Clarendon Press, 1998. Marshall, Colin. “Kant’s Appearances and Things in Themselves as Qua Objects.” The Philosophical Quarterly 63 (2013): 520–45. Sellars, Wilfrid. Science and Metaphysics. Atascadero, CA: Ridgeview Publishing Company, 1967. Strawson, Peter. Individuals. London: Methuen, 1959. Watkins, Eric. “Kant’s Transcendental Idealism and the Categories.” History of Philosophy Quarterly 19 (2002): 191–215.

This page intentionally left blank

Index

Allais, Lucy 123, 128–9, 164, 192–3 Allison, Henry 15–16, 103, 278 alteration (Veränderung) 199–213, 215, 223, 230, 241, 272 Ameriks, Karl 237, 241–4, 250, 278–9 analytic propositions 68, 168, 188; Paralogisms 165, 239–40, 245, 248, 250, 255, 257, 260, 262, 264–5 analytic unity of apperception 113–15, 118, 259 apprehension 61, 123, 167, 226; in the imagination, synthesis of 134–6, 138, 190; rule of 96–100, 152 Beattie, James 12–13 Berkeley, George 32, 50, 71, 72, 80, 201, 208 blind 144–54, 160–1, 169, 170 body 68–71, 75–6, 78, 93, 95, 103–4, 120, 140–1, 168, 178–81, 188, 192, 227, 241, 259; see also body Brook, Andrew 165, 273–4 Carnap, Rudolf 302 Carroll, Lewis 187–8 Categories 74, 96, 116–20, 123–4, 190–1, 196, 202, 228–9, 244, 252–3, 286–7, 291, 303 causation see nature, laws of; necessary connection; necessity change: conceptual 104, 148–9, 210, 223, 297, 301; between versions of the Critique 117–18, 214, 230; (Wechsel) 199, 201–3, 206–13 circular concept 235–6, 240

clocks 203–10, 297–9, 302–3 cognition 70–1, 74–6, 94, 99–100, 103, 104, 114–19, 122–3, 127, 129–31, 154, 161–2, 165, 169, 176–8, 244, 289, 301 cognitive psychology 28, 133, 165 combination 59, 65, 67, 97, 111, 119, 122–30, 136–40, 144, 151, 159, 165, 166, 167, 168, 180, 190, 230, 249 complex: perceptions 42–8, 50, 51, 81–5, 91, 103, 104, 208, 220; representation as 55–64, 65–6, 70, 73, 76–80, 93, 102, 103, 104, 118–19, 121–132, 134–41, 153–4, 155–7, 161–3, 168–9, 173, 176–7, 178–83, 208, 221, 225–6, 228, 234, 270, 299 conceptualism 109, 129, 143, 154–5, 159, 163, 166, 168, 169 condition, of an inference 66–70, 76–8, 101, 168, 188 continuity 179, 182, 193–4 Copy Principle 21–3, 26–33; empirical generalization 21, 26–8, 32, 36, 39–41, 48 counterpart relations 64–5, 70, 75–7, 107, 184, 188, 190, 202, 216, 258 Descartes, Rene 25, 112–13, 115, 164–5, 175–6, 224, 232, 269 Dickerson, A. B. 102–3 distinctness 35, 48, 58–60, 68–9, 178, 179, 182, 188, 193–4, 195, 220–1, 232 divisible 68, 71, 75–6, 227, 259 duree 290–2, 299–300

306

Index

Eberhard, J. A. 284 elephants 8, 134, 137–8, 141, 147, 157–9, 169, 174, 177, 180, 182–3, 186, 189, 192–3, 262 enthymeme 175, 184–8 espace 290–2, 299–300 external world 11–13, 19, 20–1, 34–6, 38, 48, 163, 173, 195, 198; see also body Flew, Antony 26 Fodor, Jerry 30–1 form: of judgment 74, 91–2, 94–5, 116, 151, 202; of representation 235, 238, 240–1, 258, 276; space and time 276–80, 286–7 formal representation 164, 236, 238, 244, 246–50 Garrett, Don 22–36, 41, 48, 49, 50, 104, 231–2 general representation 38–9, 49, 63, 71, 80–92, 98, 104, 105 globe 83–4 Ginsborg, Hanna 164, 168–9 Guyer, Paul 14, 15, 102, 166, 228, 279 Hanna, Robert 123, 128–9, 164, 166, 169–70 Hatfield, Gary 13–14 house 42, 61–2, 64–5, 78, 97–9, 152, 158–9, 169, 218–20, 231 image 23–5, 28–32, 43, 49, 51, 98, 170, 285 indivisible 246, 250–1, 254, 256 inference, material rules of 175, 183–95, 196, 198, 199–200, 211, 213–16, 220–2, 230, 258–9, 267–8 inner sense 56, 145, 180, 195, 199, 216, 228, 243–4, 250, 265–71, 272; see also introspection introspection 112–15, 176, 243, 259–60, 265–71; see also inner sense intuiti intellectuali 277, 285, 290 intuitions 49, 55–60, 66, 68–73, 73–8, 89, 92–101, 103, 104, 105, 107–8, 116–17, 123–7, 132–54, 161–3, 165, 170, 173–4, 176–7, 181–2, 190, 215, 219, 229, 240, 243–4, 247–8, 251–3, 259, 262–3, 278, 286, 290–2, 299

judgment 25–6, 49, 67–70, 73–7, 80–101, 104, 105, 115–17, 126, 127, 133, 141, 147–54, 168, 169, 181–3, 187–8, 194, 195, 202, 219–21, 227, 252–3, 262, 263, 266 Kitcher, Patricia 133, 164, 165, 237, 272 Langton, Rae 301, 302 lists of names 90, 92–3, 101 Locke, John 32, 50, 59, 71–3, 75, 80–1, 97–100, 103, 104, 201–2, 254 Logic 66–8, 70–2, 99, 103, 152–3, 158, 166, 167, 168, 169, 186, 188, 189, 191 logical analysis 120, 133–4, 153, 163, 165 Longuenesse, Beatrice 3, 72, 96–101, 105, 117, 150–3, 165, 169, 192, 237–41, 250 McDowell, John 3, 163 manifold 55–6, 59–61, 67, 72, 113–15, 119–24, 127–8, 130–1, 134–41, 148–50, 154, 157–60, 177–8, 180–2, 195, 199–200, 218, 224–6, 255, 260, 263 Marshall, Colin 301–2 Meier, Friedrich 64–5, 73, 196 Melnick, Arthur 203–4, 237–8, 244–50, 273 mental proxies 156, 162–3 meta-representations 99–100, 105, 147, 183–6, 194, 196, 200, 202, 210, 212, 222–3, 228, 252, 262–4, 265–8, 28, 287 meter stick 205–6 Metaphysical Deduction 53, 73–77, 94, 117, 121–2, 144–5, 161–2, 258–9 modality 282 modelling 146–50, 153–4, 161, 262–3 nature 175, 190–3, 210, 212–13, 274, 288–90 nature, laws of 103–4, 174, 190, 200, 211; see also necessity, physical; necessity, real necessary connection 12, 19–21, 29–30, 35–6, 47–8, 53–4, 103–4, 173–4, 181–2, 184–6, 189–92, 196, 198–200, 214, 215, 218–22, 228, 230, 262, 263, 267, 270

Index

307

necessity: normative 180–1; physical 174, 189; real 189–190; see also nature, laws of New Jerusalem 42–4, 46–7 non-conceptual representation 109, 142–3, 145, 154–63, 166, 168–70 Notes and Fragments 64–5, 67–70, 74, 77, 103, 127, 131, 157, 162, 166, 167–8, 170, 188, 226, 257 noumenal object, nature of 276–81, 285–90 noumenal substance, self as 241–2, 247

regulative principle 208–12, 222–3, 228–9, 284–9 Representational Copy Principle 21–2, 29–48, 50, 51, 55, 81–4, 111 representation, different senses of 25–6, 133, 156 reproduction in the imagination, synthesis of 136–9, 145 Rosenberg, Jay 224–5, 290 rules, concepts as 2–3, 64–80, 89, 92–101, 103–4, 105, 138–41, 167, 168, 170, 180–3, 183–95, 258–9

O’Shea, James 207

savage 97–9, 152–3, 169 scientific realism 199, 210–13, 223, 276, 287, 292, 301 Sellars, Wilfrid 2–3, 64, 89–90, 108–9, 141–3, 146–7, 149, 154, 155–6, 159–61, 163, 170, 184, 210–11, 269–70, 277–8, 281–301 self, as rule-follower 226–8, 255, 264, 269 sensation 55–7, 98–9, 103, 105, 130–2, 142–5, 148–54, 154–63, 167–9, 170, 180–3, 195, 259–60, 269–70, 292, 293–301 sense-impression inference 155–6, 170, 269–70, 294–6 σ-/τ-manifolds 293–301 space 45, 51, 56–60, 108, 122–7, 268, 278, 281–3, 290–301 standing for vs. standing in for 156 Strawson, Peter 133, 192, 229, 272, 302 Stufenleiter 130 subjunctive conditionals 183–8, 193–4 substance 50, 201–13, 223, 228, 261, 289–90, 297; mental 225–6, 237–8, 240–4, 247–9, 250–3 succession, temporal 45, 61–3, 134–6, 139, 220, 227, 255–6, 261–2 syllogism 66–70, 76, 101, 188–9 synthesis 122–8, 144–5, 166–7, 168; consciousness of 226–7, 255–6; perceptual 132–41, 144–54, 158, 162–3; speciosa 117, 144, 169; see also apprehension; recognition; preproduction synthetic propositions 68–9, 78, 168, 188 synthetic unity of apperception 114–15, 117, 119, 190, 259, 274, 303; see also transcendental unity of apperception

Paris 42–3, 46–7, 205 peanut butter and banana sandwich 224–5 perception: Humean 22–4, 28, 35, 37, 39, 43, 47, 49, 50, 178–9; Kant’s theory of 108, 123, 129–31, 132–63, 164, 190, 207, 270–1, 283–4 pictorial content 22–32, 34, 36, 43–4, 48, 49, 81 picturing 5–9, 64–6, 73, 76, 79, 89, 96, 100, 104, 105, 107, 132, 139, 147, 156, 160, 192–3, 198–200, 202, 208–13, 215–16, 219–21, 222–3, 257–8, 268, 285–6, 289, 300–1 Pippin, Robert 72–3, 141–4, 146, 153, 156, 274 possibility 282–3 predication 26, 67, 68, 70, 76–7, 80–101, 105, 127, 181, 202, 207, 209, 235, 248–9, 253, 260–2, 263, 266, 274 Prolegomena 11, 13, 54, 104, 298, 299, 302 Quality 116, 208, 228–9, 288–9 Quantity 66, 74, 116, 208, 228–9, 288–9 quid juris 109 quotational device 266–8 real in appearance 9, 210–11, 223, 289–90 recognition in the concept, synthesis of 127, 139–41, 165 reconceptualization 199, 211–13; see also theory succession

308

Index

theoretical posits 145, 155–8, 170, 174, 181–2, 199–200, 210–12, 269–71, 292, 293, 297, 301 theory succession 200, 210–13, 280, 285–9, 300 this-such 108, 160, 263 time 45, 55–60, 108, 122–7, 134–9, 168, 195, 200, 203–10, 213, 217–20, 271–2, 278, 281–3, 290–301 Transcendental Aesthetic 55–60, 103, 123–5, 131, 303 transcendental idealism, representational 278–81

transcendental unity of apperception 114, 118, 150, 195, 226, 237–50, 259–60, 265–71 tree 78, 138, 147 triangle 76–7, 139–40, 168, 208–9 universals 71–3, 75, 96–101, 105, 192 Van Cleve, James 203–6 Walker, Ralph 71–2, 104 Watkins, Eric 14, 301 Wittgenstein, Ludwig 169, 187, 195, 229